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 a_ ( lowerCAmelCase_ : List[Any] ):
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = sum(lowerCAmelCase_ )
__lowerCAmelCase = [[False for x in range(s + 1 )] for y in range(n + 1 )]
for i in range(1, n + 1 ):
__lowerCAmelCase = True
for i in range(1, s + 1 ):
__lowerCAmelCase = False
for i in range(1, n + 1 ):
for j in range(1, s + 1 ):
__lowerCAmelCase = dp[i][j - 1]
if arr[i - 1] <= j:
__lowerCAmelCase = dp[i][j] or dp[i - 1][j - arr[i - 1]]
for j in range(int(s / 2 ), -1, -1 ):
if dp[n][j] is True:
__lowerCAmelCase = s - 2 * j
break
return diff
| 53 |
import json
from typing import List, Optional, Tuple
from tokenizers import normalizers
from ....tokenization_utils_fast import PreTrainedTokenizerFast
from ....utils import logging
from .tokenization_retribert import RetriBertTokenizer
_snake_case : Any = logging.get_logger(__name__)
_snake_case : int = {'vocab_file': 'vocab.txt', 'tokenizer_file': 'tokenizer.json'}
_snake_case : Optional[Any] = {
'vocab_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/vocab.txt'
),
},
'tokenizer_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/tokenizer.json'
),
},
}
_snake_case : str = {
'yjernite/retribert-base-uncased': 512,
}
_snake_case : Optional[int] = {
'yjernite/retribert-base-uncased': {'do_lower_case': True},
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = PRETRAINED_VOCAB_FILES_MAP
a_ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
a_ = PRETRAINED_INIT_CONFIGURATION
a_ = RetriBertTokenizer
a_ = ["""input_ids""", """attention_mask"""]
def __init__( self : Dict , lowerCAmelCase_ : Optional[Any]=None , lowerCAmelCase_ : Tuple=None , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : str="[UNK]" , lowerCAmelCase_ : Optional[Any]="[SEP]" , lowerCAmelCase_ : List[str]="[PAD]" , lowerCAmelCase_ : Optional[int]="[CLS]" , lowerCAmelCase_ : List[Any]="[MASK]" , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : List[str]=None , **lowerCAmelCase_ : List[Any] , ) -> Dict:
super().__init__(
lowerCAmelCase_ , tokenizer_file=lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , unk_token=lowerCAmelCase_ , sep_token=lowerCAmelCase_ , pad_token=lowerCAmelCase_ , cls_token=lowerCAmelCase_ , mask_token=lowerCAmelCase_ , tokenize_chinese_chars=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ , **lowerCAmelCase_ , )
__lowerCAmelCase = json.loads(self.backend_tokenizer.normalizer.__getstate__() )
if (
normalizer_state.get('lowercase' , lowerCAmelCase_ ) != do_lower_case
or normalizer_state.get('strip_accents' , lowerCAmelCase_ ) != strip_accents
or normalizer_state.get('handle_chinese_chars' , lowerCAmelCase_ ) != tokenize_chinese_chars
):
__lowerCAmelCase = getattr(lowerCAmelCase_ , normalizer_state.pop('type' ) )
__lowerCAmelCase = do_lower_case
__lowerCAmelCase = strip_accents
__lowerCAmelCase = tokenize_chinese_chars
__lowerCAmelCase = normalizer_class(**lowerCAmelCase_ )
__lowerCAmelCase = do_lower_case
def lowercase ( self : str , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Optional[int]=None ) -> Optional[int]:
__lowerCAmelCase = [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 lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[int] , lowerCAmelCase_ : Optional[List[int]] = None ) -> List[int]:
__lowerCAmelCase = [self.sep_token_id]
__lowerCAmelCase = [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 lowercase ( self : int , lowerCAmelCase_ : str , lowerCAmelCase_ : Optional[str] = None ) -> Tuple[str]:
__lowerCAmelCase = self._tokenizer.model.save(lowerCAmelCase_ , name=lowerCAmelCase_ )
return tuple(lowerCAmelCase_ )
| 53 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
_snake_case : int = logging.get_logger(__name__)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = """timm_backbone"""
def __init__( self : Union[str, Any] , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : Optional[Any]=3 , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : int=True , lowerCAmelCase_ : Any=None , **lowerCAmelCase_ : Union[str, Any] , ) -> Dict:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = backbone
__lowerCAmelCase = num_channels
__lowerCAmelCase = features_only
__lowerCAmelCase = use_pretrained_backbone
__lowerCAmelCase = True
__lowerCAmelCase = out_indices if out_indices is not None else (-1,)
| 53 |
import numpy as np
from cva import COLOR_BGR2GRAY, cvtColor, imread
from numpy import array, uinta
from PIL import Image
from digital_image_processing import change_contrast as cc
from digital_image_processing import convert_to_negative as cn
from digital_image_processing import sepia as sp
from digital_image_processing.dithering import burkes as bs
from digital_image_processing.edge_detection import canny
from digital_image_processing.filters import convolve as conv
from digital_image_processing.filters import gaussian_filter as gg
from digital_image_processing.filters import local_binary_pattern as lbp
from digital_image_processing.filters import median_filter as med
from digital_image_processing.filters import sobel_filter as sob
from digital_image_processing.resize import resize as rs
_snake_case : Union[str, Any] = imread(R'digital_image_processing/image_data/lena_small.jpg')
_snake_case : Optional[int] = cvtColor(img, COLOR_BGR2GRAY)
def a_ ( ):
__lowerCAmelCase = cn.convert_to_negative(lowerCAmelCase_ )
# assert negative_img array for at least one True
assert negative_img.any()
def a_ ( ):
with Image.open('digital_image_processing/image_data/lena_small.jpg' ) as img:
# Work around assertion for response
assert str(cc.change_contrast(lowerCAmelCase_, 110 ) ).startswith(
'<PIL.Image.Image image mode=RGB size=100x100 at' )
def a_ ( ):
__lowerCAmelCase = canny.gen_gaussian_kernel(9, sigma=1.4 )
# Assert ambiguous array
assert resp.all()
def a_ ( ):
__lowerCAmelCase = imread('digital_image_processing/image_data/lena_small.jpg', 0 )
# assert ambiguous array for all == True
assert canny_img.all()
__lowerCAmelCase = canny.canny(lowerCAmelCase_ )
# assert canny array for at least one True
assert canny_array.any()
def a_ ( ):
assert gg.gaussian_filter(lowerCAmelCase_, 5, sigma=0.9 ).all()
def a_ ( ):
# laplace diagonals
__lowerCAmelCase = array([[0.25, 0.5, 0.25], [0.5, -3, 0.5], [0.25, 0.5, 0.25]] )
__lowerCAmelCase = conv.img_convolve(lowerCAmelCase_, lowerCAmelCase_ ).astype(lowerCAmelCase_ )
assert res.any()
def a_ ( ):
assert med.median_filter(lowerCAmelCase_, 3 ).any()
def a_ ( ):
__lowerCAmelCase , __lowerCAmelCase = sob.sobel_filter(lowerCAmelCase_ )
assert grad.any() and theta.any()
def a_ ( ):
__lowerCAmelCase = sp.make_sepia(lowerCAmelCase_, 20 )
assert sepia.all()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg" ):
__lowerCAmelCase = bs.Burkes(imread(lowerCAmelCase_, 1 ), 120 )
burkes.process()
assert burkes.output_img.any()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg", ):
__lowerCAmelCase = rs.NearestNeighbour(imread(lowerCAmelCase_, 1 ), 400, 200 )
nn.process()
assert nn.output.any()
def a_ ( ):
__lowerCAmelCase = 'digital_image_processing/image_data/lena.jpg'
# Reading the image and converting it to grayscale.
__lowerCAmelCase = imread(lowerCAmelCase_, 0 )
# Test for get_neighbors_pixel function() return not None
__lowerCAmelCase = 0
__lowerCAmelCase = 0
__lowerCAmelCase = image[x_coordinate][y_coordinate]
__lowerCAmelCase = lbp.get_neighbors_pixel(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert neighbors_pixels is not None
# Test for local_binary_pattern function()
# Create a numpy array as the same height and width of read image
__lowerCAmelCase = np.zeros((image.shape[0], image.shape[1]) )
# Iterating through the image and calculating the local binary pattern value
# for each pixel.
for i in range(0, image.shape[0] ):
for j in range(0, image.shape[1] ):
__lowerCAmelCase = lbp.local_binary_value(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert lbp_image.any()
| 53 | 1 |
import argparse
import json
import numpy
import torch
from transformers.models.xlm.tokenization_xlm import VOCAB_FILES_NAMES
from transformers.utils import CONFIG_NAME, WEIGHTS_NAME, logging
logging.set_verbosity_info()
def a_ ( lowerCAmelCase_ : Dict, lowerCAmelCase_ : Tuple ):
# Load checkpoint
__lowerCAmelCase = torch.load(lowerCAmelCase_, map_location='cpu' )
__lowerCAmelCase = chkpt['model']
# We have the base model one level deeper than the original XLM repository
__lowerCAmelCase = {}
for k, v in state_dict.items():
if "pred_layer" in k:
__lowerCAmelCase = v
else:
__lowerCAmelCase = v
__lowerCAmelCase = chkpt['params']
__lowerCAmelCase = {n: v for n, v in config.items() if not isinstance(lowerCAmelCase_, (torch.FloatTensor, numpy.ndarray) )}
__lowerCAmelCase = chkpt['dico_word2id']
__lowerCAmelCase = {s + '</w>' if s.find('@@' ) == -1 and i > 13 else s.replace('@@', '' ): i for s, i in vocab.items()}
# Save pytorch-model
__lowerCAmelCase = pytorch_dump_folder_path + '/' + WEIGHTS_NAME
__lowerCAmelCase = pytorch_dump_folder_path + '/' + CONFIG_NAME
__lowerCAmelCase = pytorch_dump_folder_path + '/' + VOCAB_FILES_NAMES['vocab_file']
print(F"""Save PyTorch model to {pytorch_weights_dump_path}""" )
torch.save(lowerCAmelCase_, lowerCAmelCase_ )
print(F"""Save configuration file to {pytorch_config_dump_path}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, indent=2 ) + '\n' )
print(F"""Save vocab file to {pytorch_config_dump_path}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, indent=2 ) + '\n' )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--xlm_checkpoint_path', default=None, type=str, required=True, help='Path the official PyTorch dump.'
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : Any = parser.parse_args()
convert_xlm_checkpoint_to_pytorch(args.xlm_checkpoint_path, args.pytorch_dump_folder_path)
| 53 |
from typing import Dict, Iterable, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_DEFAULT_MEAN,
IMAGENET_DEFAULT_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, logging
_snake_case : List[Any] = logging.get_logger(__name__)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = ["""pixel_values"""]
def __init__( self : Optional[int] , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Union[int, float] = 1 / 2_5_5 , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_MEAN , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_STD , **lowerCAmelCase_ : Any , ) -> None:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = size if size is not None else {'shortest_edge': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else {'height': 2_2_4, 'width': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = do_resize
__lowerCAmelCase = size
__lowerCAmelCase = resample
__lowerCAmelCase = do_center_crop
__lowerCAmelCase = crop_size
__lowerCAmelCase = do_rescale
__lowerCAmelCase = rescale_factor
__lowerCAmelCase = do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else IMAGENET_DEFAULT_MEAN
__lowerCAmelCase = image_std if image_std is not None else IMAGENET_DEFAULT_STD
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : Optional[int] , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
# size_dict is a dict with either keys "height" and "width" or "shortest_edge"
if "shortest_edge" in size:
__lowerCAmelCase = int((2_5_6 / 2_2_4) * size['shortest_edge'] )
__lowerCAmelCase = get_resize_output_image_size(lowerCAmelCase_ , size=lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = {'height': output_size[0], 'width': output_size[1]}
if "height" not in size_dict or "width" not in size_dict:
raise ValueError(
f"""Size dict must have keys 'height' and 'width' or 'shortest_edge'. Got {size_dict.keys()}""" )
return resize(
lowerCAmelCase_ , size=(size_dict['height'], size_dict['width']) , resample=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : str , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : str , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ )
if "height" not in size or "width" not in size:
raise ValueError(f"""Size dict must have keys 'height' and 'width'. Got {size.keys()}""" )
return center_crop(lowerCAmelCase_ , size=(size['height'], size['width']) , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[int, float] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : int , ) -> np.ndarray:
return rescale(lowerCAmelCase_ , scale=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : int , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : List[str] , ) -> np.ndarray:
return normalize(lowerCAmelCase_ , mean=lowerCAmelCase_ , std=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : ImageInput , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : PILImageResampling = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[float] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[TensorType] = None , lowerCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **lowerCAmelCase_ : str , ) -> BatchFeature:
__lowerCAmelCase = do_resize if do_resize is not None else self.do_resize
__lowerCAmelCase = resample if resample is not None else self.resample
__lowerCAmelCase = do_center_crop if do_center_crop is not None else self.do_center_crop
__lowerCAmelCase = do_rescale if do_rescale is not None else self.do_rescale
__lowerCAmelCase = rescale_factor if rescale_factor is not None else self.rescale_factor
__lowerCAmelCase = do_normalize if do_normalize is not None else self.do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else self.image_mean
__lowerCAmelCase = image_std if image_std is not None else self.image_std
__lowerCAmelCase = size if size is not None else self.size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else self.crop_size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = make_list_of_images(lowerCAmelCase_ )
if not valid_images(lowerCAmelCase_ ):
raise ValueError(
'Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '
'torch.Tensor, tf.Tensor or jax.ndarray.' )
if do_resize and size is None:
raise ValueError('Size must be specified if do_resize is True.' )
if do_center_crop and crop_size is None:
raise ValueError('Crop size must be specified if do_center_crop is True.' )
if do_rescale and rescale_factor is None:
raise ValueError('Rescale factor must be specified if do_rescale is True.' )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError('Image mean and std must be specified if do_normalize is True.' )
# All transformations expect numpy arrays.
__lowerCAmelCase = [to_numpy_array(lowerCAmelCase_ ) for image in images]
if do_resize:
__lowerCAmelCase = [self.resize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_center_crop:
__lowerCAmelCase = [self.center_crop(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_rescale:
__lowerCAmelCase = [self.rescale(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_normalize:
__lowerCAmelCase = [self.normalize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = [to_channel_dimension_format(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = {'pixel_values': images}
return BatchFeature(data=lowerCAmelCase_ , tensor_type=lowerCAmelCase_ )
| 53 | 1 |
from arguments import InitializationArguments
from transformers import AutoConfig, AutoModelForCausalLM, AutoTokenizer, HfArgumentParser
# Configuration
_snake_case : int = HfArgumentParser(InitializationArguments)
_snake_case : str = parser.parse_args()
# Load codeparrot tokenizer trained for Python code tokenization
_snake_case : Optional[int] = AutoTokenizer.from_pretrained(args.tokenizer_name)
# Config: "scale_attn_by_layer_idx" and "reorder_and_upcast_attn" are Mistral stability tweaks
_snake_case : int = {
'vocab_size': len(tokenizer),
'scale_attn_by_inverse_layer_idx': True,
'reorder_and_upcast_attn': True,
}
# Load model config (GPT-2 large in this case)
_snake_case : Dict = AutoConfig.from_pretrained(args.config_name, **config_kwargs)
# Initialize new model with config
_snake_case : Union[str, Any] = AutoModelForCausalLM.from_config(config)
# Save model to the hub
model.save_pretrained(args.model_name, push_to_hub=args.push_to_hub)
| 53 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Dict=2 , lowerCAmelCase_ : Optional[int]=8 , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[Any]=9_9 , lowerCAmelCase_ : List[Any]=1_6 , lowerCAmelCase_ : int=5 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : str=3_6 , lowerCAmelCase_ : Optional[int]="gelu" , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : Optional[int]=0.0 , lowerCAmelCase_ : str=5_1_2 , lowerCAmelCase_ : List[str]=1_6 , lowerCAmelCase_ : str=2 , lowerCAmelCase_ : Tuple=0.02 , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : List[Any]=4 , lowerCAmelCase_ : List[str]=None , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_input_mask
__lowerCAmelCase = use_token_type_ids
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = type_vocab_size
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = num_labels
__lowerCAmelCase = num_choices
__lowerCAmelCase = scope
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = None
if self.use_input_mask:
__lowerCAmelCase = random_attention_mask([self.batch_size, self.seq_length] )
__lowerCAmelCase = None
if self.use_token_type_ids:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
__lowerCAmelCase = ids_tensor([self.batch_size] , self.num_choices )
__lowerCAmelCase = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowercase ( self : Any ) -> Union[str, Any]:
return MraConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , )
def lowercase ( self : Dict ) -> List[Any]:
__lowerCAmelCase = self.get_config()
__lowerCAmelCase = 3_0_0
return config
def lowercase ( self : Optional[int] ) -> Union[str, Any]:
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = self.prepare_config_and_inputs()
__lowerCAmelCase = True
__lowerCAmelCase = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
encoder_hidden_states,
encoder_attention_mask,
)
def lowercase ( self : Optional[int] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , ) -> Tuple:
__lowerCAmelCase = True
__lowerCAmelCase = MraModel(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , )
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , )
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Dict ) -> str:
__lowerCAmelCase = MraForQuestionAnswering(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , start_positions=lowerCAmelCase_ , end_positions=lowerCAmelCase_ , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowercase ( self : int , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : int , lowerCAmelCase_ : Dict ) -> Any:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForTokenClassification(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowercase ( self : int , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] ) -> List[Any]:
__lowerCAmelCase = self.num_choices
__lowerCAmelCase = MraForMultipleChoice(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowercase ( self : Tuple ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'input_ids': input_ids, 'token_type_ids': token_type_ids, 'attention_mask': input_mask}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
a_ = False
a_ = False
a_ = False
a_ = False
a_ = ()
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = MraModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : Tuple ) -> List[str]:
self.config_tester.run_common_tests()
def lowercase ( self : Optional[int] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
__lowerCAmelCase = type
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Any ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*lowerCAmelCase_ )
def lowercase ( self : List[str] ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*lowerCAmelCase_ )
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = MraModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
@unittest.skip(reason='MRA does not output attentions' )
def lowercase ( self : Optional[int] ) -> Tuple:
return
@require_torch
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@slow
def lowercase ( self : Optional[Any] ) -> List[str]:
__lowerCAmelCase = MraModel.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = torch.Size((1, 2_5_6, 7_6_8) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[-0.01_40, 0.08_30, -0.03_81], [0.15_46, 0.14_02, 0.02_20], [0.11_62, 0.08_51, 0.01_65]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : int ) -> Optional[int]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 2_5_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[9.25_95, -3.60_38, 11.88_19], [9.38_69, -3.26_93, 11.09_56], [11.85_24, -3.49_38, 13.12_10]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : Any ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-4096-8-d3' )
__lowerCAmelCase = torch.arange(4_0_9_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 4_0_9_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[5.47_89, -2.35_64, 7.50_64], [7.90_67, -1.33_69, 9.96_68], [9.07_12, -1.81_06, 7.03_80]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import json
import os
from typing import Optional, Tuple
from ...tokenization_utils import PreTrainedTokenizer
from ...utils import logging
_snake_case : Tuple = logging.get_logger(__name__)
_snake_case : int = {'vocab_file': 'vocab.json'}
_snake_case : List[str] = {
'vocab_file': {
'mgp-str': 'https://huggingface.co/alibaba-damo/mgp-str-base/blob/main/vocab.json',
}
}
_snake_case : Optional[int] = {'mgp-str': 27}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = PRETRAINED_VOCAB_FILES_MAP
a_ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
def __init__( self : int , lowerCAmelCase_ : str , lowerCAmelCase_ : Dict="[GO]" , lowerCAmelCase_ : List[Any]="[GO]" , lowerCAmelCase_ : List[str]="[s]" , lowerCAmelCase_ : Dict="[GO]" , **lowerCAmelCase_ : Optional[Any] ) -> List[Any]:
super().__init__(
unk_token=lowerCAmelCase_ , bos_token=lowerCAmelCase_ , eos_token=lowerCAmelCase_ , pad_token=lowerCAmelCase_ , **lowerCAmelCase_ , )
with open(lowerCAmelCase_ , encoding='utf-8' ) as vocab_handle:
__lowerCAmelCase = json.load(lowerCAmelCase_ )
__lowerCAmelCase = {v: k for k, v in self.vocab.items()}
@property
def lowercase ( self : str ) -> Any:
return len(self.vocab )
def lowercase ( self : Optional[Any] ) -> List[Any]:
return dict(self.vocab , **self.added_tokens_encoder )
def lowercase ( self : Dict , lowerCAmelCase_ : Optional[int] ) -> str:
__lowerCAmelCase = []
for s in text:
char_tokens.extend(lowerCAmelCase_ )
return char_tokens
def lowercase ( self : str , lowerCAmelCase_ : str ) -> Optional[int]:
return self.vocab.get(lowerCAmelCase_ , self.vocab.get(self.unk_token ) )
def lowercase ( self : Dict , lowerCAmelCase_ : Any ) -> Dict:
return self.decoder.get(lowerCAmelCase_ )
def lowercase ( self : str , lowerCAmelCase_ : str , lowerCAmelCase_ : Optional[str] = None ) -> Tuple[str]:
if not os.path.isdir(lowerCAmelCase_ ):
logger.error('Vocabulary path ({}) should be a directory'.format(lowerCAmelCase_ ) )
return
__lowerCAmelCase = os.path.join(
lowerCAmelCase_ , (filename_prefix + '-' if filename_prefix else '') + VOCAB_FILES_NAMES['vocab_file'] )
with open(lowerCAmelCase_ , 'w' , encoding='utf-8' ) as f:
f.write(json.dumps(self.vocab , indent=2 , sort_keys=lowerCAmelCase_ , ensure_ascii=lowerCAmelCase_ ) + '\n' )
return (vocab_file,)
| 53 |
import argparse
import json
import os
import re
import shutil
import torch
from transformers import BioGptConfig, BioGptForCausalLM
from transformers.models.biogpt.tokenization_biogpt import VOCAB_FILES_NAMES
from transformers.tokenization_utils_base import TOKENIZER_CONFIG_FILE
from transformers.utils import WEIGHTS_NAME, logging
logging.set_verbosity_warning()
_snake_case : Union[str, Any] = 2
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , *, # begin keyword-only arguments
lowerCAmelCase_ : str="<s>" , lowerCAmelCase_ : Dict="<pad>" , lowerCAmelCase_ : Any="</s>" , lowerCAmelCase_ : List[str]="<unk>" , lowerCAmelCase_ : Optional[Any]=None , ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = bos, unk, pad, eos
__lowerCAmelCase = []
__lowerCAmelCase = []
__lowerCAmelCase = {}
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
if extra_special_symbols:
for s in extra_special_symbols:
self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = len(self.symbols )
def __eq__( self : Dict , lowerCAmelCase_ : Dict ) -> str:
return self.indices == other.indices
def __getitem__( self : List[Any] , lowerCAmelCase_ : int ) -> Union[str, Any]:
if idx < len(self.symbols ):
return self.symbols[idx]
return self.unk_word
def __len__( self : Tuple ) -> List[Any]:
return len(self.symbols )
def __contains__( self : Optional[Any] , lowerCAmelCase_ : Dict ) -> Optional[int]:
return sym in self.indices
@classmethod
def lowercase ( cls : Dict , lowerCAmelCase_ : str ) -> str:
__lowerCAmelCase = cls()
d.add_from_file(lowerCAmelCase_ )
return d
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int=1 , lowerCAmelCase_ : Any=False ) -> Optional[Any]:
if word in self.indices and not overwrite:
__lowerCAmelCase = self.indices[word]
__lowerCAmelCase = self.count[idx] + n
return idx
else:
__lowerCAmelCase = len(self.symbols )
__lowerCAmelCase = idx
self.symbols.append(lowerCAmelCase_ )
self.count.append(lowerCAmelCase_ )
return idx
def lowercase ( self : str , lowerCAmelCase_ : Union[str, Any] ) -> Dict:
return 0
def lowercase ( self : Tuple , lowerCAmelCase_ : Union[str, Any] ) -> int:
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
try:
with open(lowerCAmelCase_ , 'r' , encoding='utf-8' ) as fd:
self.add_from_file(lowerCAmelCase_ )
except FileNotFoundError as fnfe:
raise fnfe
except UnicodeError:
raise Exception('Incorrect encoding detected in {}, please rebuild the dataset'.format(lowerCAmelCase_ ) )
return
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = self._load_meta(lowerCAmelCase_ )
for line in lines[indices_start_line:]:
try:
__lowerCAmelCase , __lowerCAmelCase = line.rstrip().rsplit(' ' , 1 )
if field == "#fairseq:overwrite":
__lowerCAmelCase = True
__lowerCAmelCase , __lowerCAmelCase = line.rsplit(' ' , 1 )
else:
__lowerCAmelCase = False
__lowerCAmelCase = int(lowerCAmelCase_ )
__lowerCAmelCase = line
if word in self and not overwrite:
raise RuntimeError(
'Duplicate word found when loading Dictionary: \'{}\'. '
'Duplicate words can overwrite earlier ones by adding the '
'#fairseq:overwrite flag at the end of the corresponding row '
'in the dictionary file. If using the Camembert model, please '
'download an updated copy of the model file.'.format(lowerCAmelCase_ ) )
self.add_symbol(lowerCAmelCase_ , n=lowerCAmelCase_ , overwrite=lowerCAmelCase_ )
except ValueError:
raise ValueError('Incorrect dictionary format, expected \'<token> <cnt> [flags]\'' )
def a_ ( lowerCAmelCase_ : List[str] ):
# (1) remove word breaking symbol, (2) add word ending symbol where the word is not broken up,
# e.g.: d = {'le@@': 5, 'tt@@': 6, 'er': 7} => {'le': 5, 'tt': 6, 'er</w>': 7}
__lowerCAmelCase = dict((re.sub(R'@@$', '', lowerCAmelCase_ ), v) if k.endswith('@@' ) else (re.sub(R'$', '</w>', lowerCAmelCase_ ), v) for k, v in d.items() )
__lowerCAmelCase = '<s> <pad> </s> <unk>'.split()
# restore the special tokens
for k in keep_keys:
del da[F"""{k}</w>"""]
__lowerCAmelCase = d[k] # restore
return da
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str] ):
# prep
if not os.path.exists(lowerCAmelCase_ ):
raise ValueError(F"""path {biogpt_checkpoint_path} does not exist!""" )
os.makedirs(lowerCAmelCase_, exist_ok=lowerCAmelCase_ )
print(F"""Writing results to {pytorch_dump_folder_path}""" )
# handle various types of models
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'checkpoint.pt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {checkpoint_file} does not exist!""" )
__lowerCAmelCase = torch.load(lowerCAmelCase_, map_location='cpu' )
__lowerCAmelCase = chkpt['cfg']['model']
# dicts
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'dict.txt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {dict_file} does not exist!""" )
__lowerCAmelCase = Dictionary.load(lowerCAmelCase_ )
__lowerCAmelCase = rewrite_dict_keys(src_dict.indices )
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['vocab_file'] )
print(F"""Generating {src_vocab_file} of {src_vocab_size} records""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# merges_file (bpecodes)
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'bpecodes' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {bpecodes_file} does not exist!""" )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['merges_file'] )
shutil.copyfile(lowerCAmelCase_, lowerCAmelCase_ )
# model config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'config.json' )
__lowerCAmelCase = {
'activation_dropout': args['activation_dropout'],
'architectures': ['BioGptForCausalLM'],
'attention_probs_dropout_prob': args['attention_dropout'],
'bos_token_id': 0,
'eos_token_id': 2,
'hidden_act': args['activation_fn'],
'hidden_dropout_prob': args['dropout'],
'hidden_size': args['decoder_embed_dim'],
'initializer_range': 0.02,
'intermediate_size': args['decoder_ffn_embed_dim'],
'layer_norm_eps': 1E-12,
'layerdrop': args['decoder_layerdrop'],
'max_position_embeddings': args['max_target_positions'],
'model_type': 'biogpt',
'num_attention_heads': args['decoder_attention_heads'],
'num_hidden_layers': args['decoder_layers'],
'pad_token_id': 1,
'scale_embedding': not args['no_scale_embedding'],
'tie_word_embeddings': args['share_decoder_input_output_embed'],
'vocab_size': src_vocab_size,
}
# good hparam defaults to start with
print(F"""Generating {biogpt_model_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# tokenizer config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = {
'bos_token': '<s>',
'eos_token': '</s>',
'model_max_length': 1024,
'pad_token': '<pad>',
'special_tokens_map_file': None,
'tokenizer_class': 'BioGptTokenizer',
'unk_token': '<unk>',
}
print(F"""Generating {biogpt_tokenizer_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# model
__lowerCAmelCase = chkpt['model']
# remove unneeded keys
__lowerCAmelCase = [
'decoder.version',
]
for k in ignore_keys:
model_state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = list(model_state_dict.keys() )
for layer_name in layer_names:
if layer_name.endswith('output_projection.weight' ):
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
else:
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
__lowerCAmelCase = BioGptConfig.from_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = BioGptForCausalLM(lowerCAmelCase_ )
# check that it loads ok
model_new.load_state_dict(lowerCAmelCase_ )
# save
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
print(F"""Generating {pytorch_weights_dump_path}""" )
torch.save(lowerCAmelCase_, lowerCAmelCase_ )
print('Conversion is done!' )
if __name__ == "__main__":
_snake_case : str = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--biogpt_checkpoint_path',
default=None,
type=str,
required=True,
help=(
'Path to the official PyTorch checkpoint file which is expected to reside in the dump dir with dicts,'
' bpecodes, etc.'
),
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_biogpt_checkpoint_to_pytorch(args.biogpt_checkpoint_path, args.pytorch_dump_folder_path)
| 53 | 1 |
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : str ):
def get_matched_characters(lowerCAmelCase_ : str, lowerCAmelCase_ : str ) -> str:
__lowerCAmelCase = []
__lowerCAmelCase = min(len(_stra ), len(_stra ) ) // 2
for i, l in enumerate(_stra ):
__lowerCAmelCase = int(max(0, i - limit ) )
__lowerCAmelCase = int(min(i + limit + 1, len(_stra ) ) )
if l in _stra[left:right]:
matched.append(lowerCAmelCase_ )
__lowerCAmelCase = F"""{_stra[0:_stra.index(lowerCAmelCase_ )]} {_stra[_stra.index(lowerCAmelCase_ ) + 1:]}"""
return "".join(lowerCAmelCase_ )
# matching characters
__lowerCAmelCase = get_matched_characters(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = get_matched_characters(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = len(lowerCAmelCase_ )
# transposition
__lowerCAmelCase = (
len([(ca, ca) for ca, ca in zip(lowerCAmelCase_, lowerCAmelCase_ ) if ca != ca] ) // 2
)
if not match_count:
__lowerCAmelCase = 0.0
else:
__lowerCAmelCase = (
1
/ 3
* (
match_count / len(lowerCAmelCase_ )
+ match_count / len(lowerCAmelCase_ )
+ (match_count - transpositions) / match_count
)
)
# common prefix up to 4 characters
__lowerCAmelCase = 0
for ca, ca in zip(stra[:4], stra[:4] ):
if ca == ca:
prefix_len += 1
else:
break
return jaro + 0.1 * prefix_len * (1 - jaro)
if __name__ == "__main__":
import doctest
doctest.testmod()
print(jaro_winkler('hello', 'world'))
| 53 |
from typing import Tuple, Union
from ...modeling_outputs import BackboneOutput
from ...modeling_utils import PreTrainedModel
from ...utils import is_timm_available, is_torch_available, requires_backends
from ...utils.backbone_utils import BackboneMixin
from .configuration_timm_backbone import TimmBackboneConfig
if is_timm_available():
import timm
if is_torch_available():
from torch import Tensor
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = """pixel_values"""
a_ = False
a_ = TimmBackboneConfig
def __init__( self : Tuple , lowerCAmelCase_ : Any , **lowerCAmelCase_ : Optional[int] ) -> Optional[Any]:
requires_backends(self , 'timm' )
super().__init__(lowerCAmelCase_ )
__lowerCAmelCase = config
if config.backbone is None:
raise ValueError('backbone is not set in the config. Please set it to a timm model name.' )
if config.backbone not in timm.list_models():
raise ValueError(f"""backbone {config.backbone} is not supported by timm.""" )
if hasattr(lowerCAmelCase_ , 'out_features' ) and config.out_features is not None:
raise ValueError('out_features is not supported by TimmBackbone. Please use out_indices instead.' )
__lowerCAmelCase = getattr(lowerCAmelCase_ , 'use_pretrained_backbone' , lowerCAmelCase_ )
if pretrained is None:
raise ValueError('use_pretrained_backbone is not set in the config. Please set it to True or False.' )
# We just take the final layer by default. This matches the default for the transformers models.
__lowerCAmelCase = config.out_indices if getattr(lowerCAmelCase_ , 'out_indices' , lowerCAmelCase_ ) is not None else (-1,)
__lowerCAmelCase = timm.create_model(
config.backbone , pretrained=lowerCAmelCase_ , features_only=config.features_only , in_chans=config.num_channels , out_indices=lowerCAmelCase_ , **lowerCAmelCase_ , )
# These are used to control the output of the model when called. If output_hidden_states is True, then
# return_layers is modified to include all layers.
__lowerCAmelCase = self._backbone.return_layers
__lowerCAmelCase = {layer['module']: str(lowerCAmelCase_ ) for i, layer in enumerate(self._backbone.feature_info.info )}
super()._init_backbone(lowerCAmelCase_ )
@classmethod
def lowercase ( cls : int , lowerCAmelCase_ : Dict , *lowerCAmelCase_ : Dict , **lowerCAmelCase_ : Union[str, Any] ) -> Optional[int]:
requires_backends(cls , ['vision', 'timm'] )
from ...models.timm_backbone import TimmBackboneConfig
__lowerCAmelCase = kwargs.pop('config' , TimmBackboneConfig() )
__lowerCAmelCase = kwargs.pop('use_timm_backbone' , lowerCAmelCase_ )
if not use_timm:
raise ValueError('use_timm_backbone must be True for timm backbones' )
__lowerCAmelCase = kwargs.pop('num_channels' , config.num_channels )
__lowerCAmelCase = kwargs.pop('features_only' , config.features_only )
__lowerCAmelCase = kwargs.pop('use_pretrained_backbone' , config.use_pretrained_backbone )
__lowerCAmelCase = kwargs.pop('out_indices' , config.out_indices )
__lowerCAmelCase = TimmBackboneConfig(
backbone=lowerCAmelCase_ , num_channels=lowerCAmelCase_ , features_only=lowerCAmelCase_ , use_pretrained_backbone=lowerCAmelCase_ , out_indices=lowerCAmelCase_ , )
return super()._from_config(lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Tuple , lowerCAmelCase_ : int ) -> Dict:
pass
def lowercase ( self : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : Any=None , **lowerCAmelCase_ : Dict ) -> Union[BackboneOutput, Tuple[Tensor, ...]]:
__lowerCAmelCase = return_dict if return_dict is not None else self.config.use_return_dict
__lowerCAmelCase = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__lowerCAmelCase = output_attentions if output_attentions is not None else self.config.output_attentions
if output_attentions:
raise ValueError('Cannot output attentions for timm backbones at the moment' )
if output_hidden_states:
# We modify the return layers to include all the stages of the backbone
__lowerCAmelCase = self._all_layers
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = self._return_layers
__lowerCAmelCase = tuple(hidden_states[i] for i in self.out_indices )
else:
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = tuple(lowerCAmelCase_ )
__lowerCAmelCase = tuple(lowerCAmelCase_ ) if hidden_states is not None else None
if not return_dict:
__lowerCAmelCase = (feature_maps,)
if output_hidden_states:
__lowerCAmelCase = output + (hidden_states,)
return output
return BackboneOutput(feature_maps=lowerCAmelCase_ , hidden_states=lowerCAmelCase_ , attentions=lowerCAmelCase_ )
| 53 | 1 |
import heapq
import sys
import numpy as np
_snake_case : Any = tuple[int, int]
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : str ) -> str:
__lowerCAmelCase = []
__lowerCAmelCase = set()
def lowercase ( self : Optional[Any] ) -> Tuple:
if not self.empty():
return self.elements[0][0]
else:
return float('inf' )
def lowercase ( self : Optional[int] ) -> List[str]:
return len(self.elements ) == 0
def lowercase ( self : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : List[str] ) -> Dict:
if item not in self.set:
heapq.heappush(self.elements , (priority, item) )
self.set.add(lowerCAmelCase_ )
else:
# update
# print("update", item)
__lowerCAmelCase = []
((__lowerCAmelCase) , (__lowerCAmelCase)) = heapq.heappop(self.elements )
while x != item:
temp.append((pri, x) )
((__lowerCAmelCase) , (__lowerCAmelCase)) = heapq.heappop(self.elements )
temp.append((priority, item) )
for pro, xxx in temp:
heapq.heappush(self.elements , (pro, xxx) )
def lowercase ( self : int , lowerCAmelCase_ : List[Any] ) -> Dict:
if item in self.set:
self.set.remove(lowerCAmelCase_ )
__lowerCAmelCase = []
((__lowerCAmelCase) , (__lowerCAmelCase)) = heapq.heappop(self.elements )
while x != item:
temp.append((pro, x) )
((__lowerCAmelCase) , (__lowerCAmelCase)) = heapq.heappop(self.elements )
for prito, yyy in temp:
heapq.heappush(self.elements , (prito, yyy) )
def lowercase ( self : Optional[Any] ) -> str:
return self.elements[0][1]
def lowercase ( self : str ) -> Dict:
((__lowerCAmelCase) , (__lowerCAmelCase)) = heapq.heappop(self.elements )
self.set.remove(lowerCAmelCase_ )
return (priority, item)
def a_ ( lowerCAmelCase_ : TPos, lowerCAmelCase_ : TPos ):
# euclidean distance
__lowerCAmelCase = np.array(lowerCAmelCase_ )
__lowerCAmelCase = np.array(lowerCAmelCase_ )
return np.linalg.norm(a - b )
def a_ ( lowerCAmelCase_ : TPos, lowerCAmelCase_ : TPos ):
# integer division by time variable
return consistent_heuristic(lowerCAmelCase_, lowerCAmelCase_ ) // t
def a_ ( lowerCAmelCase_ : TPos, lowerCAmelCase_ : TPos ):
# manhattan distance
return abs(p[0] - goal[0] ) + abs(p[1] - goal[1] )
def a_ ( lowerCAmelCase_ : TPos, lowerCAmelCase_ : int, lowerCAmelCase_ : TPos, lowerCAmelCase_ : dict[TPos, float] ):
__lowerCAmelCase = g_function[start] + Wa * heuristics[i](lowerCAmelCase_, lowerCAmelCase_ )
return ans
def a_ ( lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : Optional[int] ):
__lowerCAmelCase = np.chararray((n, n) )
for i in range(lowerCAmelCase_ ):
for j in range(lowerCAmelCase_ ):
__lowerCAmelCase = '*'
for i in range(lowerCAmelCase_ ):
for j in range(lowerCAmelCase_ ):
if (j, (n - 1) - i) in blocks:
__lowerCAmelCase = '#'
__lowerCAmelCase = '-'
__lowerCAmelCase = back_pointer[goal]
while x != start:
((__lowerCAmelCase) , (__lowerCAmelCase)) = x
# print(x)
__lowerCAmelCase = '-'
__lowerCAmelCase = back_pointer[x]
__lowerCAmelCase = '-'
for i in range(lowerCAmelCase_ ):
for j in range(lowerCAmelCase_ ):
if (i, j) == (0, n - 1):
print(grid[i][j], end=' ' )
print('<-- End position', end=' ' )
else:
print(grid[i][j], end=' ' )
print()
print('^' )
print('Start position' )
print()
print('# is an obstacle' )
print('- is the path taken by algorithm' )
print('PATH TAKEN BY THE ALGORITHM IS:-' )
__lowerCAmelCase = back_pointer[goal]
while x != start:
print(lowerCAmelCase_, end=' ' )
__lowerCAmelCase = back_pointer[x]
print(lowerCAmelCase_ )
sys.exit()
def a_ ( lowerCAmelCase_ : TPos ):
if p[0] < 0 or p[0] > n - 1:
return False
if p[1] < 0 or p[1] > n - 1:
return False
return True
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : List[Any], lowerCAmelCase_ : Any, lowerCAmelCase_ : str, lowerCAmelCase_ : int, lowerCAmelCase_ : Any, lowerCAmelCase_ : str, ):
for itera in range(lowerCAmelCase_ ):
open_list[itera].remove_element(lowerCAmelCase_ )
# print("s", s)
# print("j", j)
((__lowerCAmelCase) , (__lowerCAmelCase)) = s
__lowerCAmelCase = (x - 1, y)
__lowerCAmelCase = (x + 1, y)
__lowerCAmelCase = (x, y + 1)
__lowerCAmelCase = (x, y - 1)
for neighbours in [left, right, up, down]:
if neighbours not in blocks:
if valid(lowerCAmelCase_ ) and neighbours not in visited:
# print("neighbour", neighbours)
visited.add(lowerCAmelCase_ )
__lowerCAmelCase = -1
__lowerCAmelCase = float('inf' )
if valid(lowerCAmelCase_ ) and g_function[neighbours] > g_function[s] + 1:
__lowerCAmelCase = g_function[s] + 1
__lowerCAmelCase = s
if neighbours not in close_list_anchor:
open_list[0].put(lowerCAmelCase_, key(lowerCAmelCase_, 0, lowerCAmelCase_, lowerCAmelCase_ ) )
if neighbours not in close_list_inad:
for var in range(1, lowerCAmelCase_ ):
if key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ ) <= Wa * key(
lowerCAmelCase_, 0, lowerCAmelCase_, lowerCAmelCase_ ):
open_list[j].put(
lowerCAmelCase_, key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ ) )
def a_ ( ):
__lowerCAmelCase = []
for x in range(1, 5 ):
for y in range(1, 6 ):
some_list.append((x, y) )
for x in range(15, 20 ):
some_list.append((x, 17) )
for x in range(10, 19 ):
for y in range(1, 15 ):
some_list.append((x, y) )
# L block
for x in range(1, 4 ):
for y in range(12, 19 ):
some_list.append((x, y) )
for x in range(3, 13 ):
for y in range(16, 19 ):
some_list.append((x, y) )
return some_list
_snake_case : Dict = {0: consistent_heuristic, 1: heuristic_a, 2: heuristic_a}
_snake_case : int = [
(0, 1),
(1, 1),
(2, 1),
(3, 1),
(4, 1),
(5, 1),
(6, 1),
(7, 1),
(8, 1),
(9, 1),
(10, 1),
(11, 1),
(12, 1),
(13, 1),
(14, 1),
(15, 1),
(16, 1),
(17, 1),
(18, 1),
(19, 1),
]
_snake_case : Tuple = make_common_ground()
_snake_case : str = blocks_blk
# hyper parameters
_snake_case : Union[str, Any] = 1
_snake_case : Optional[Any] = 1
_snake_case : Dict = 20
_snake_case : List[Any] = 3 # one consistent and two other inconsistent
# start and end destination
_snake_case : int = (0, 0)
_snake_case : List[Any] = (n - 1, n - 1)
_snake_case : List[str] = 1
def a_ ( lowerCAmelCase_ : TPos, lowerCAmelCase_ : TPos, lowerCAmelCase_ : int ):
__lowerCAmelCase = {start: 0, goal: float('inf' )}
__lowerCAmelCase = {start: -1, goal: -1}
__lowerCAmelCase = []
__lowerCAmelCase = set()
for i in range(lowerCAmelCase_ ):
open_list.append(PriorityQueue() )
open_list[i].put(lowerCAmelCase_, key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ ) )
__lowerCAmelCase = []
__lowerCAmelCase = []
while open_list[0].minkey() < float('inf' ):
for i in range(1, lowerCAmelCase_ ):
# print(open_list[0].minkey(), open_list[i].minkey())
if open_list[i].minkey() <= Wa * open_list[0].minkey():
global t
t += 1
if g_function[goal] <= open_list[i].minkey():
if g_function[goal] < float('inf' ):
do_something(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
else:
__lowerCAmelCase , __lowerCAmelCase = open_list[i].top_show()
visited.add(lowerCAmelCase_ )
expand_state(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, )
close_list_inad.append(lowerCAmelCase_ )
else:
if g_function[goal] <= open_list[0].minkey():
if g_function[goal] < float('inf' ):
do_something(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
else:
__lowerCAmelCase = open_list[0].top_show()
visited.add(lowerCAmelCase_ )
expand_state(
lowerCAmelCase_, 0, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, )
close_list_anchor.append(lowerCAmelCase_ )
print('No path found to goal' )
print()
for i in range(n - 1, -1, -1 ):
for j in range(lowerCAmelCase_ ):
if (j, i) in blocks:
print('#', end=' ' )
elif (j, i) in back_pointer:
if (j, i) == (n - 1, n - 1):
print('*', end=' ' )
else:
print('-', end=' ' )
else:
print('*', end=' ' )
if (j, i) == (n - 1, n - 1):
print('<-- End position', end=' ' )
print()
print('^' )
print('Start position' )
print()
print('# is an obstacle' )
print('- is the path taken by algorithm' )
if __name__ == "__main__":
multi_a_star(start, goal, n_heuristic)
| 53 |
from __future__ import annotations
def a_ ( lowerCAmelCase_ : list[float] ):
if len(lowerCAmelCase_ ) < 2:
raise ValueError('Monogons and Digons are not polygons in the Euclidean space' )
if any(i <= 0 for i in nums ):
raise ValueError('All values must be greater than 0' )
__lowerCAmelCase = nums.copy()
copy_nums.sort()
return copy_nums[-1] < sum(copy_nums[:-1] )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 | 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 ViTConfig, ViTForImageClassification, ViTImageProcessor, ViTModel
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Union[str, Any] = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : int=False ):
__lowerCAmelCase = []
for i in range(config.num_hidden_layers ):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((F"""blocks.{i}.norm1.weight""", F"""vit.encoder.layer.{i}.layernorm_before.weight""") )
rename_keys.append((F"""blocks.{i}.norm1.bias""", F"""vit.encoder.layer.{i}.layernorm_before.bias""") )
rename_keys.append((F"""blocks.{i}.attn.proj.weight""", F"""vit.encoder.layer.{i}.attention.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.attn.proj.bias""", F"""vit.encoder.layer.{i}.attention.output.dense.bias""") )
rename_keys.append((F"""blocks.{i}.norm2.weight""", F"""vit.encoder.layer.{i}.layernorm_after.weight""") )
rename_keys.append((F"""blocks.{i}.norm2.bias""", F"""vit.encoder.layer.{i}.layernorm_after.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.weight""", F"""vit.encoder.layer.{i}.intermediate.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.bias""", F"""vit.encoder.layer.{i}.intermediate.dense.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.weight""", F"""vit.encoder.layer.{i}.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.bias""", F"""vit.encoder.layer.{i}.output.dense.bias""") )
# projection layer + position embeddings
rename_keys.extend(
[
('cls_token', 'vit.embeddings.cls_token'),
('patch_embed.proj.weight', 'vit.embeddings.patch_embeddings.projection.weight'),
('patch_embed.proj.bias', 'vit.embeddings.patch_embeddings.projection.bias'),
('pos_embed', 'vit.embeddings.position_embeddings'),
] )
if base_model:
# layernorm + pooler
rename_keys.extend(
[
('norm.weight', 'layernorm.weight'),
('norm.bias', 'layernorm.bias'),
] )
# if just the base model, we should remove "vit" from all keys that start with "vit"
__lowerCAmelCase = [(pair[0], pair[1][4:]) if pair[1].startswith('vit' ) else pair for pair in rename_keys]
else:
# layernorm + classification head
rename_keys.extend(
[
('norm.weight', 'vit.layernorm.weight'),
('norm.bias', 'vit.layernorm.bias'),
('head.weight', 'classifier.weight'),
('head.bias', 'classifier.bias'),
] )
return rename_keys
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Optional[int]=False ):
for i in range(config.num_hidden_layers ):
if base_model:
__lowerCAmelCase = ''
else:
__lowerCAmelCase = 'vit.'
# read in weights + bias of input projection layer (in timm, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[
: config.hidden_size, :
]
__lowerCAmelCase = in_proj_bias[: config.hidden_size]
__lowerCAmelCase = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
__lowerCAmelCase = in_proj_bias[
config.hidden_size : config.hidden_size * 2
]
__lowerCAmelCase = in_proj_weight[
-config.hidden_size :, :
]
__lowerCAmelCase = in_proj_bias[-config.hidden_size :]
def a_ ( lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = ['head.weight', 'head.bias']
for k in ignore_keys:
state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Optional[Any]=True ):
__lowerCAmelCase = ViTConfig()
# patch_size
if model_name[-1] == "8":
__lowerCAmelCase = 8
# set labels if required
if not base_model:
__lowerCAmelCase = 1000
__lowerCAmelCase = 'huggingface/label-files'
__lowerCAmelCase = 'imagenet-1k-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
__lowerCAmelCase = idalabel
__lowerCAmelCase = {v: k for k, v in idalabel.items()}
# size of the architecture
if model_name in ["dino_vits8", "dino_vits16"]:
__lowerCAmelCase = 384
__lowerCAmelCase = 1536
__lowerCAmelCase = 12
__lowerCAmelCase = 6
# load original model from torch hub
__lowerCAmelCase = torch.hub.load('facebookresearch/dino:main', lowerCAmelCase_ )
original_model.eval()
# load state_dict of original model, remove and rename some keys
__lowerCAmelCase = original_model.state_dict()
if base_model:
remove_classification_head_(lowerCAmelCase_ )
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_, base_model=lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_q_k_v(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# load HuggingFace model
if base_model:
__lowerCAmelCase = ViTModel(lowerCAmelCase_, add_pooling_layer=lowerCAmelCase_ ).eval()
else:
__lowerCAmelCase = ViTForImageClassification(lowerCAmelCase_ ).eval()
model.load_state_dict(lowerCAmelCase_ )
# Check outputs on an image, prepared by ViTImageProcessor
__lowerCAmelCase = ViTImageProcessor()
__lowerCAmelCase = image_processor(images=prepare_img(), return_tensors='pt' )
__lowerCAmelCase = encoding['pixel_values']
__lowerCAmelCase = model(lowerCAmelCase_ )
if base_model:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert torch.allclose(lowerCAmelCase_, outputs.last_hidden_state[:, 0, :], atol=1E-1 )
else:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert logits.shape == outputs.logits.shape
assert torch.allclose(lowerCAmelCase_, outputs.logits, atol=1E-3 )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
print(F"""Saving model {model_name} to {pytorch_dump_folder_path}""" )
model.save_pretrained(lowerCAmelCase_ )
print(F"""Saving image processor to {pytorch_dump_folder_path}""" )
image_processor.save_pretrained(lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : Dict = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='dino_vitb16',
type=str,
help='Name of the model trained with DINO you\'d like to convert.',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--base_model',
action='store_true',
help='Whether to only convert the base model (no projection head weights).',
)
parser.set_defaults(base_model=True)
_snake_case : List[Any] = parser.parse_args()
convert_vit_checkpoint(args.model_name, args.pytorch_dump_folder_path, args.base_model)
| 53 |
import inspect
import unittest
from transformers import RegNetConfig, is_flax_available
from transformers.testing_utils import require_flax, slow
from transformers.utils import cached_property, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor
if is_flax_available():
import jax
import jax.numpy as jnp
from transformers.models.regnet.modeling_flax_regnet import FlaxRegNetForImageClassification, FlaxRegNetModel
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int]=3 , lowerCAmelCase_ : Dict=3_2 , lowerCAmelCase_ : Tuple=3 , lowerCAmelCase_ : Union[str, Any]=1_0 , lowerCAmelCase_ : List[str]=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Optional[int]=[1, 1, 2, 1] , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : Tuple="relu" , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : Optional[int]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = embeddings_size
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = hidden_act
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = len(lowerCAmelCase_ )
def lowercase ( self : Optional[int] ) -> List[Any]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : Tuple ) -> List[Any]:
return RegNetConfig(
num_channels=self.num_channels , embeddings_size=self.embeddings_size , hidden_sizes=self.hidden_sizes , depths=self.depths , hidden_act=self.hidden_act , num_labels=self.num_labels , image_size=self.image_size , )
def lowercase ( self : List[str] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : List[str] ) -> str:
__lowerCAmelCase = FlaxRegNetModel(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
# Output shape (b, c, h, w)
self.parent.assertEqual(
result.last_hidden_state.shape , (self.batch_size, self.hidden_sizes[-1], self.image_size // 3_2, self.image_size // 3_2) , )
def lowercase ( self : str , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : int ) -> Tuple:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = FlaxRegNetForImageClassification(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_flax
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (FlaxRegNetModel, FlaxRegNetForImageClassification) if is_flax_available() else ()
a_ = False
a_ = False
a_ = False
def lowercase ( self : Dict ) -> None:
__lowerCAmelCase = FlaxRegNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : int ) -> Optional[int]:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : str ) -> Union[str, Any]:
return
def lowercase ( self : Dict ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCAmelCase_ )
@unittest.skip(reason='RegNet does not use inputs_embeds' )
def lowercase ( self : Union[str, Any] ) -> Any:
pass
@unittest.skip(reason='RegNet does not support input and output embeddings' )
def lowercase ( self : Tuple ) -> Tuple:
pass
def lowercase ( self : Optional[Any] ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.__call__ )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
def check_hidden_states_output(lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : str ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
with self.subTest(model_class.__name__ ):
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model_class(lowerCAmelCase_ )
@jax.jit
def model_jitted(lowerCAmelCase_ : Optional[int] , **lowerCAmelCase_ : Dict ):
return model(pixel_values=lowerCAmelCase_ , **lowerCAmelCase_ )
with self.subTest('JIT Enabled' ):
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
with self.subTest('JIT Disabled' ):
with jax.disable_jit():
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
self.assertEqual(len(lowerCAmelCase_ ) , len(lowerCAmelCase_ ) )
for jitted_output, output in zip(lowerCAmelCase_ , lowerCAmelCase_ ):
self.assertEqual(jitted_output.shape , output.shape )
def a_ ( ):
__lowerCAmelCase = Image.open('./tests/fixtures/tests_samples/COCO/000000039769.png' )
return image
@require_flax
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
return AutoImageProcessor.from_pretrained('facebook/regnet-y-040' ) if is_vision_available() else None
@slow
def lowercase ( self : Optional[Any] ) -> Union[str, Any]:
__lowerCAmelCase = FlaxRegNetForImageClassification.from_pretrained('facebook/regnet-y-040' )
__lowerCAmelCase = self.default_image_processor
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = image_processor(images=lowerCAmelCase_ , return_tensors='np' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
# verify the logits
__lowerCAmelCase = (1, 1_0_0_0)
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = jnp.array([-0.41_80, -1.50_51, -3.48_36] )
self.assertTrue(jnp.allclose(outputs.logits[0, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import unittest
from transformers import DebertaVaTokenizer, DebertaVaTokenizerFast
from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_tokenizers, slow
from ...test_tokenization_common import TokenizerTesterMixin
_snake_case : Tuple = get_tests_dir('fixtures/spiece.model')
@require_sentencepiece
@require_tokenizers
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = DebertaVaTokenizer
a_ = DebertaVaTokenizerFast
a_ = True
a_ = True
def lowercase ( self : List[Any] ) -> int:
super().setUp()
# We have a SentencePiece fixture for testing
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ , unk_token='<unk>' )
tokenizer.save_pretrained(self.tmpdirname )
def lowercase ( self : Any , lowerCAmelCase_ : Union[str, Any] ) -> Dict:
__lowerCAmelCase = 'this is a test'
__lowerCAmelCase = 'this is a test'
return input_text, output_text
def lowercase ( self : List[Any] ) -> Union[str, Any]:
__lowerCAmelCase = '<pad>'
__lowerCAmelCase = 0
self.assertEqual(self.get_tokenizer()._convert_token_to_id(lowerCAmelCase_ ) , lowerCAmelCase_ )
self.assertEqual(self.get_tokenizer()._convert_id_to_token(lowerCAmelCase_ ) , lowerCAmelCase_ )
def lowercase ( self : Optional[int] ) -> int:
__lowerCAmelCase = list(self.get_tokenizer().get_vocab().keys() )
self.assertEqual(vocab_keys[0] , '<pad>' )
self.assertEqual(vocab_keys[1] , '<unk>' )
self.assertEqual(vocab_keys[-1] , '[PAD]' )
self.assertEqual(len(lowerCAmelCase_ ) , 3_0_0_0_1 )
def lowercase ( self : Tuple ) -> Dict:
self.assertEqual(self.get_tokenizer().vocab_size , 3_0_0_0_0 )
def lowercase ( self : Dict ) -> Optional[Any]:
# fmt: off
__lowerCAmelCase = ' \tHeLLo!how \n Are yoU? '
__lowerCAmelCase = ['▁hello', '!', 'how', '▁are', '▁you', '?']
# fmt: on
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = DebertaVaTokenizerFast(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
@unittest.skip('There is an inconsistency between slow and fast tokenizer due to a bug in the fast one.' )
def lowercase ( self : str ) -> List[str]:
pass
@unittest.skip('There is an inconsistency between slow and fast tokenizer due to a bug in the fast one.' )
def lowercase ( self : Optional[Any] ) -> Tuple:
pass
def lowercase ( self : Tuple ) -> List[Any]:
# fmt: off
__lowerCAmelCase = 'I was born in 92000, and this is falsé.'
__lowerCAmelCase = ['▁', '<unk>', '▁was', '▁born', '▁in', '▁9', '2000', '▁', ',', '▁and', '▁this', '▁is', '▁fal', 's', '<unk>', '▁', '.', ]
# fmt: on
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = DebertaVaTokenizerFast(lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Optional[Any] ) -> int:
# fmt: off
__lowerCAmelCase = 'I was born in 92000, and this is falsé.'
__lowerCAmelCase = ['▁i', '▁was', '▁born', '▁in', '▁9', '2000', '▁', ',', '▁and', '▁this', '▁is', '▁fal', 's', '<unk>', '▁', '.', ]
# fmt: on
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = DebertaVaTokenizerFast(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Dict ) -> str:
# fmt: off
__lowerCAmelCase = 'I was born in 92000, and this is falsé.'
__lowerCAmelCase = ['▁i', '▁was', '▁born', '▁in', '▁9', '2000', ',', '▁and', '▁this', '▁is', '▁fal', 's', '<unk>', '.', ]
# fmt: on
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = DebertaVaTokenizerFast(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Tuple:
# fmt: off
__lowerCAmelCase = 'I was born in 92000, and this is falsé.'
__lowerCAmelCase = ['▁', '<unk>', '▁was', '▁born', '▁in', '▁9', '2000', '▁', ',', '▁and', '▁this', '▁is', '▁fal', 's', '<unk>', '▁', '.', ]
# fmt: on
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = DebertaVaTokenizerFast(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> int:
# fmt: off
__lowerCAmelCase = ' \tHeLLo!how \n Are yoU? '
__lowerCAmelCase = ['▁', '<unk>', 'e', '<unk>', 'o', '!', 'how', '▁', '<unk>', 're', '▁yo', '<unk>', '?']
# fmt: on
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = DebertaVaTokenizerFast(lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , split_by_punct=lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Dict:
__lowerCAmelCase = self.get_tokenizer()
__lowerCAmelCase = self.get_rust_tokenizer()
__lowerCAmelCase = 'I was born in 92000, and this is falsé.'
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ ) )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = self.get_rust_tokenizer()
__lowerCAmelCase = tokenizer.encode(lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.encode(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Optional[int] ) -> str:
__lowerCAmelCase = 'This is a test'
__lowerCAmelCase = [1_3, 1, 4_3_9_8, 2_5, 2_1, 1_2_8_9]
__lowerCAmelCase = ['▁', 'T', 'his', '▁is', '▁a', '▁test']
__lowerCAmelCase = ['▁', '<unk>', 'his', '▁is', '▁a', '▁test']
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ , keep_accents=lowerCAmelCase_ )
__lowerCAmelCase = DebertaVaTokenizerFast(lowerCAmelCase_ , keep_accents=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.tokenize(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.tokenize(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
# fmt: off
__lowerCAmelCase = 'I was born in 92000, and this is falsé.'
__lowerCAmelCase = [1_3, 1, 2_3, 3_8_6, 1_9, 5_6_1, 3_0_5_0, 1_5, 1_7, 4_8, 2_5, 8_2_5_6, 1_8, 1, 9]
__lowerCAmelCase = ['▁', 'I', '▁was', '▁born', '▁in', '▁9', '2000', ',', '▁and', '▁this', '▁is', '▁fal', 's', 'é', '.', ]
__lowerCAmelCase = ['▁', '<unk>', '▁was', '▁born', '▁in', '▁9', '2000', ',', '▁and', '▁this', '▁is', '▁fal', 's', '<unk>', '.', ]
# fmt: on
__lowerCAmelCase = tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.tokenize(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.convert_ids_to_tokens(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.encode(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.tokenize(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = rust_tokenizer.convert_ids_to_tokens(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Optional[Any] ) -> Union[str, Any]:
__lowerCAmelCase = DebertaVaTokenizer(lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.encode('sequence builders' )
__lowerCAmelCase = tokenizer.encode('multi-sequence build' )
__lowerCAmelCase = tokenizer.build_inputs_with_special_tokens(lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.build_inputs_with_special_tokens(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual([tokenizer.cls_token_id] + text + [tokenizer.sep_token_id] , lowerCAmelCase_ )
self.assertEqual(
[tokenizer.cls_token_id] + text + [tokenizer.sep_token_id] + text_a + [tokenizer.sep_token_id] , lowerCAmelCase_ , )
@slow
def lowercase ( self : Any ) -> Optional[int]:
# fmt: off
__lowerCAmelCase = {'input_ids': [[1, 3_9_8_6_7, 3_6, 1_9_3_9_0, 4_8_6, 2_7, 3_5_0_5_2, 8_1_4_3_6, 1_8, 6_0_6_8_5, 1_2_2_5, 7, 3_5_0_5_2, 8_1_4_3_6, 1_8, 9_3_6_7, 1_6_8_9_9, 1_8, 1_5_9_3_7, 5_3, 5_9_4, 7_7_3, 1_8, 1_6_2_8_7, 3_0_4_6_5, 3_6, 1_5_9_3_7, 6, 4_1_1_3_9, 3_8, 3_6_9_7_9, 6_0_7_6_3, 1_9_1, 6, 3_4_1_3_2, 9_9, 6, 5_0_5_3_8, 3_9_0, 4_3_2_3_0, 6, 3_4_1_3_2, 2_7_7_9, 2_0_8_5_0, 1_4, 6_9_9, 1_0_7_2, 1_1_9_4, 3_6, 3_8_2, 1_0_9_0_1, 5_3, 7, 6_9_9, 1_0_7_2, 2_0_8_4, 3_6, 2_0_4_2_2, 6_3_0, 5_3, 1_9, 1_0_5, 3_0_4_9, 1_8_9_6, 1_0_5_3, 1_6_8_9_9, 1_5_0_6, 1_1, 3_7_9_7_8, 4_2_4_3, 7, 1_2_3_7, 3_1_8_6_9, 2_0_0, 1_6_5_6_6, 6_5_4, 6, 3_5_0_5_2, 8_1_4_3_6, 7, 5_5_6_3_0, 1_3_5_9_3, 4, 2], [1, 2_6, 1_5_0_1_1, 1_3, 6_6_7, 8, 1_0_5_3, 1_8, 2_3_6_1_1, 1_2_3_7, 7_2_3_5_6, 1_2_8_2_0, 3_4, 1_0_4_1_3_4, 1_2_0_9, 3_5, 1_3_3_1_3, 6_6_2_7, 2_1, 2_0_2, 3_4_7, 7, 1_6_4, 2_3_9_9, 1_1, 4_6, 4_4_8_5, 4, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 5, 1_2_3_2, 2_8_6_4, 1_5_7_8_5, 1_4_9_5_1, 1_0_5, 5, 8_5_8_1, 1_2_5_0, 4, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]], 'token_type_ids': [[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]], 'attention_mask': [[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]} # noqa: E501
# fmt: on
self.tokenizer_integration_test_util(
expected_encoding=lowerCAmelCase_ , model_name='microsoft/deberta-v2-xlarge' , revision='ad6e42c1532ddf3a15c39246b63f5559d558b670' , )
| 53 |
import json
import logging
import math
import os
import sys
from dataclasses import dataclass, field
from typing import Optional
from datasets import Dataset, load_dataset
import transformers
from transformers import (
CONFIG_MAPPING,
MODEL_FOR_MASKED_LM_MAPPING,
AutoConfig,
AutoModelForMaskedLM,
AutoTokenizer,
DataCollatorForWholeWordMask,
HfArgumentParser,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
_snake_case : Optional[int] = logging.getLogger(__name__)
_snake_case : Dict = list(MODEL_FOR_MASKED_LM_MAPPING.keys())
_snake_case : List[Any] = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """If training from scratch, pass a model type from the list: """ + """, """.join(_UpperCamelCase )} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Override some existing default config settings when a model is trained from scratch. Example: """
"""n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained config name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained tokenizer name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."""} , )
a_ = field(
default="""main""" , metadata={"""help""": """The specific model version to use (can be a branch name, tag name or commit id)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Will use the token generated when running `huggingface-cli login` (necessary to use this script """
"""with private models)."""
)
} , )
def lowercase ( self : List[Any] ) -> List[Any]:
if self.config_overrides is not None and (self.config_name is not None or self.model_name_or_path is not None):
raise ValueError(
'--config_overrides can\'t be used in combination with --config_name or --model_name_or_path' )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The name of the dataset to use (via the datasets library)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(default=_UpperCamelCase , metadata={"""help""": """The input training data file (a text file)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input evaluation data file to evaluate the perplexity on (a text file)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input train ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input validation ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached training and evaluation sets"""} )
a_ = field(
default=5 , metadata={
"""help""": """The percentage of the train set used as validation set in case there's no validation split"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The maximum total input sequence length after tokenization. Sequences longer """
"""than this will be truncated. Default to the max input length of the model."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=0.15 , metadata={"""help""": """Ratio of tokens to mask for masked language modeling loss"""} )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Whether to pad all samples to `max_seq_length`. """
"""If False, will pad the samples dynamically when batching to the maximum length in the batch."""
)
} , )
def lowercase ( self : int ) -> int:
if self.train_file is not None:
__lowerCAmelCase = self.train_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`train_file` should be a csv, a json or a txt file."
if self.validation_file is not None:
__lowerCAmelCase = self.validation_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`validation_file` should be a csv, a json or a txt file."
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : Union[str, Any] ):
with open(lowerCAmelCase_, 'r', encoding='utf-8' ) as f:
__lowerCAmelCase = [json.loads(lowerCAmelCase_ ) for line in f.read().splitlines() if (len(lowerCAmelCase_ ) > 0 and not line.isspace())]
assert len(lowerCAmelCase_ ) == len(lowerCAmelCase_ )
__lowerCAmelCase = {c: dataset[c] for c in dataset.column_names}
__lowerCAmelCase = refs
return Dataset.from_dict(lowerCAmelCase_ )
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
__lowerCAmelCase = load_dataset(data_args.dataset_name, data_args.dataset_config_name )
if "validation" not in datasets.keys():
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[:{data_args.validation_split_percentage}%]""", )
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[{data_args.validation_split_percentage}%:]""", )
else:
__lowerCAmelCase = {}
if data_args.train_file is not None:
__lowerCAmelCase = data_args.train_file
if data_args.validation_file is not None:
__lowerCAmelCase = data_args.validation_file
__lowerCAmelCase = data_args.train_file.split('.' )[-1]
if extension == "txt":
__lowerCAmelCase = 'text'
__lowerCAmelCase = load_dataset(lowerCAmelCase_, data_files=lowerCAmelCase_ )
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.config_name:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.config_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
__lowerCAmelCase = CONFIG_MAPPING[model_args.model_type]()
logger.warning('You are instantiating a new config instance from scratch.' )
if model_args.config_overrides is not None:
logger.info(F"""Overriding config: {model_args.config_overrides}""" )
config.update_from_string(model_args.config_overrides )
logger.info(F"""New config: {config}""" )
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'use_fast': model_args.use_fast_tokenizer,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.tokenizer_name:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.tokenizer_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
raise ValueError(
'You are instantiating a new tokenizer from scratch. This is not supported by this script.'
'You can do it from another script, save it, and load it from here, using --tokenizer_name.' )
if model_args.model_name_or_path:
__lowerCAmelCase = AutoModelForMaskedLM.from_pretrained(
model_args.model_name_or_path, from_tf=bool('.ckpt' in model_args.model_name_or_path ), config=lowerCAmelCase_, cache_dir=model_args.cache_dir, revision=model_args.model_revision, use_auth_token=True if model_args.use_auth_token else None, )
else:
logger.info('Training new model from scratch' )
__lowerCAmelCase = AutoModelForMaskedLM.from_config(lowerCAmelCase_ )
model.resize_token_embeddings(len(lowerCAmelCase_ ) )
# Preprocessing the datasets.
# First we tokenize all the texts.
if training_args.do_train:
__lowerCAmelCase = datasets['train'].column_names
else:
__lowerCAmelCase = datasets['validation'].column_names
__lowerCAmelCase = 'text' if 'text' in column_names else column_names[0]
__lowerCAmelCase = 'max_length' if data_args.pad_to_max_length else False
def tokenize_function(lowerCAmelCase_ : str ):
# Remove empty lines
__lowerCAmelCase = [line for line in examples['text'] if len(lowerCAmelCase_ ) > 0 and not line.isspace()]
return tokenizer(examples['text'], padding=lowerCAmelCase_, truncation=lowerCAmelCase_, max_length=data_args.max_seq_length )
__lowerCAmelCase = datasets.map(
lowerCAmelCase_, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, remove_columns=[text_column_name], load_from_cache_file=not data_args.overwrite_cache, )
# Add the chinese references if provided
if data_args.train_ref_file is not None:
__lowerCAmelCase = add_chinese_references(tokenized_datasets['train'], data_args.train_ref_file )
if data_args.validation_ref_file is not None:
__lowerCAmelCase = add_chinese_references(
tokenized_datasets['validation'], data_args.validation_ref_file )
# If we have ref files, need to avoid it removed by trainer
__lowerCAmelCase = data_args.train_ref_file or data_args.validation_ref_file
if has_ref:
__lowerCAmelCase = False
# Data collator
# This one will take care of randomly masking the tokens.
__lowerCAmelCase = DataCollatorForWholeWordMask(tokenizer=lowerCAmelCase_, mlm_probability=data_args.mlm_probability )
# Initialize our Trainer
__lowerCAmelCase = Trainer(
model=lowerCAmelCase_, args=lowerCAmelCase_, train_dataset=tokenized_datasets['train'] if training_args.do_train else None, eval_dataset=tokenized_datasets['validation'] if training_args.do_eval else None, tokenizer=lowerCAmelCase_, data_collator=lowerCAmelCase_, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif model_args.model_name_or_path is not None and os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model() # Saves the tokenizer too for easy upload
__lowerCAmelCase = os.path.join(training_args.output_dir, 'train_results.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Train results *****' )
for key, value in sorted(train_result.metrics.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
# Need to save the state, since Trainer.save_model saves only the tokenizer with the model
trainer.state.save_to_json(os.path.join(training_args.output_dir, 'trainer_state.json' ) )
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = math.exp(eval_output['eval_loss'] )
__lowerCAmelCase = perplexity
__lowerCAmelCase = os.path.join(training_args.output_dir, 'eval_results_mlm_wwm.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Eval results *****' )
for key, value in sorted(results.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
return results
def a_ ( lowerCAmelCase_ : Tuple ):
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()
| 53 | 1 |
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Features, Sequence, Value
from .base import TaskTemplate
@dataclass(frozen=_UpperCamelCase )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = field(default="""question-answering-extractive""" , metadata={"""include_in_asdict_even_if_is_default""": True} )
a_ = Features({"""question""": Value("""string""" ), """context""": Value("""string""" )} )
a_ = Features(
{
"""answers""": Sequence(
{
"""text""": Value("""string""" ),
"""answer_start""": Value("""int32""" ),
} )
} )
a_ = "question"
a_ = "context"
a_ = "answers"
@property
def lowercase ( self : Any ) -> Dict[str, str]:
return {self.question_column: "question", self.context_column: "context", self.answers_column: "answers"}
| 53 |
def a_ ( lowerCAmelCase_ : int = 200_0000 ):
__lowerCAmelCase = [0 for i in range(n + 1 )]
__lowerCAmelCase = 1
__lowerCAmelCase = 1
for i in range(2, int(n**0.5 ) + 1 ):
if primality_list[i] == 0:
for j in range(i * i, n + 1, lowerCAmelCase_ ):
__lowerCAmelCase = 1
__lowerCAmelCase = 0
for i in range(lowerCAmelCase_ ):
if primality_list[i] == 0:
sum_of_primes += i
return sum_of_primes
if __name__ == "__main__":
print(F"""{solution() = }""")
| 53 | 1 |
import argparse
import os
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
########################################################################
# This is a fully working simple example to use Accelerate,
# specifically showcasing the experiment tracking capability,
# and builds off the `nlp_example.py` script.
#
# This example trains a Bert base model on GLUE MRPC
# in any of the following settings (with the same script):
# - single CPU or single GPU
# - multi GPUS (using PyTorch distributed mode)
# - (multi) TPUs
# - fp16 (mixed-precision) or fp32 (normal precision)
#
# To help focus on the differences in the code, building `DataLoaders`
# was refactored into its own function.
# New additions from the base script can be found quickly by
# looking for the # New Code # tags
#
# To run it in each of these various modes, follow the instructions
# in the readme for examples:
# https://github.com/huggingface/accelerate/tree/main/examples
#
########################################################################
_snake_case : Tuple = 16
_snake_case : int = 32
def a_ ( lowerCAmelCase_ : Accelerator, lowerCAmelCase_ : int = 16 ):
__lowerCAmelCase = AutoTokenizer.from_pretrained('bert-base-cased' )
__lowerCAmelCase = load_dataset('glue', 'mrpc' )
def tokenize_function(lowerCAmelCase_ : str ):
# max_length=None => use the model max length (it's actually the default)
__lowerCAmelCase = tokenizer(examples['sentence1'], examples['sentence2'], truncation=lowerCAmelCase_, max_length=lowerCAmelCase_ )
return outputs
# Apply the method we just defined to all the examples in all the splits of the dataset
# starting with the main process first:
with accelerator.main_process_first():
__lowerCAmelCase = datasets.map(
lowerCAmelCase_, batched=lowerCAmelCase_, remove_columns=['idx', 'sentence1', 'sentence2'], )
# We also rename the 'label' column to 'labels' which is the expected name for labels by the models of the
# transformers library
__lowerCAmelCase = tokenized_datasets.rename_column('label', 'labels' )
def collate_fn(lowerCAmelCase_ : Tuple ):
# On TPU it's best to pad everything to the same length or training will be very slow.
__lowerCAmelCase = 128 if accelerator.distributed_type == DistributedType.TPU else None
# When using mixed precision we want round multiples of 8/16
if accelerator.mixed_precision == "fp8":
__lowerCAmelCase = 16
elif accelerator.mixed_precision != "no":
__lowerCAmelCase = 8
else:
__lowerCAmelCase = None
return tokenizer.pad(
lowerCAmelCase_, padding='longest', max_length=lowerCAmelCase_, pad_to_multiple_of=lowerCAmelCase_, return_tensors='pt', )
# Instantiate dataloaders.
__lowerCAmelCase = DataLoader(
tokenized_datasets['train'], shuffle=lowerCAmelCase_, collate_fn=lowerCAmelCase_, batch_size=lowerCAmelCase_ )
__lowerCAmelCase = DataLoader(
tokenized_datasets['validation'], shuffle=lowerCAmelCase_, collate_fn=lowerCAmelCase_, batch_size=lowerCAmelCase_ )
return train_dataloader, eval_dataloader
# For testing only
if os.environ.get('TESTING_MOCKED_DATALOADERS', None) == "1":
from accelerate.test_utils.training import mocked_dataloaders
_snake_case : Optional[Any] = mocked_dataloaders # noqa: F811
def a_ ( lowerCAmelCase_ : Dict, lowerCAmelCase_ : List[Any] ):
# For testing only
if os.environ.get('TESTING_MOCKED_DATALOADERS', lowerCAmelCase_ ) == "1":
__lowerCAmelCase = 2
# Initialize Accelerator
# New Code #
# We pass in "all" to `log_with` to grab all available trackers in the environment
# Note: If using a custom `Tracker` class, should be passed in here such as:
# >>> log_with = ["all", MyCustomTrackerClassInstance()]
if args.with_tracking:
__lowerCAmelCase = Accelerator(
cpu=args.cpu, mixed_precision=args.mixed_precision, log_with='all', project_dir=args.project_dir )
else:
__lowerCAmelCase = Accelerator(cpu=args.cpu, mixed_precision=args.mixed_precision )
# Sample hyper-parameters for learning rate, batch size, seed and a few other HPs
__lowerCAmelCase = config['lr']
__lowerCAmelCase = int(config['num_epochs'] )
__lowerCAmelCase = int(config['seed'] )
__lowerCAmelCase = int(config['batch_size'] )
set_seed(lowerCAmelCase_ )
__lowerCAmelCase , __lowerCAmelCase = get_dataloaders(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = evaluate.load('glue', 'mrpc' )
# If the batch size is too big we use gradient accumulation
__lowerCAmelCase = 1
if batch_size > MAX_GPU_BATCH_SIZE and accelerator.distributed_type != DistributedType.TPU:
__lowerCAmelCase = batch_size // MAX_GPU_BATCH_SIZE
__lowerCAmelCase = MAX_GPU_BATCH_SIZE
# Instantiate the model (we build the model here so that the seed also control new weights initialization)
__lowerCAmelCase = AutoModelForSequenceClassification.from_pretrained('bert-base-cased', return_dict=lowerCAmelCase_ )
# We could avoid this line since the accelerator is set with `device_placement=True` (default value).
# Note that if you are placing tensors on devices manually, this line absolutely needs to be before the optimizer
# creation otherwise training will not work on TPU (`accelerate` will kindly throw an error to make us aware of that).
__lowerCAmelCase = model.to(accelerator.device )
# Instantiate optimizer
__lowerCAmelCase = AdamW(params=model.parameters(), lr=lowerCAmelCase_ )
# Instantiate scheduler
__lowerCAmelCase = get_linear_schedule_with_warmup(
optimizer=lowerCAmelCase_, num_warmup_steps=100, num_training_steps=(len(lowerCAmelCase_ ) * num_epochs) // gradient_accumulation_steps, )
# 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = accelerator.prepare(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# New Code #
# We need to initialize the trackers we use. Overall configurations can also be stored
if args.with_tracking:
__lowerCAmelCase = os.path.split(lowerCAmelCase_ )[-1].split('.' )[0]
accelerator.init_trackers(lowerCAmelCase_, lowerCAmelCase_ )
# Now we train the model
for epoch in range(lowerCAmelCase_ ):
model.train()
# New Code #
# For our tracking example, we will log the total loss of each epoch
if args.with_tracking:
__lowerCAmelCase = 0
for step, batch in enumerate(lowerCAmelCase_ ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs.loss
# New Code #
if args.with_tracking:
total_loss += loss.detach().float()
__lowerCAmelCase = loss / gradient_accumulation_steps
accelerator.backward(lowerCAmelCase_ )
if step % gradient_accumulation_steps == 0:
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
model.eval()
for step, batch in enumerate(lowerCAmelCase_ ):
# We could avoid this line since we set the accelerator with `device_placement=True` (the default).
batch.to(accelerator.device )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs.logits.argmax(dim=-1 )
__lowerCAmelCase , __lowerCAmelCase = accelerator.gather_for_metrics((predictions, batch['labels']) )
metric.add_batch(
predictions=lowerCAmelCase_, references=lowerCAmelCase_, )
__lowerCAmelCase = metric.compute()
# Use accelerator.print to print only on the main process.
accelerator.print(F"""epoch {epoch}:""", lowerCAmelCase_ )
# New Code #
# To actually log, we call `Accelerator.log`
# The values passed can be of `str`, `int`, `float` or `dict` of `str` to `float`/`int`
if args.with_tracking:
accelerator.log(
{
'accuracy': eval_metric['accuracy'],
'f1': eval_metric['f1'],
'train_loss': total_loss.item() / len(lowerCAmelCase_ ),
'epoch': epoch,
}, step=lowerCAmelCase_, )
# New Code #
# When a run is finished, you should call `accelerator.end_training()`
# to close all of the open trackers
if args.with_tracking:
accelerator.end_training()
def a_ ( ):
__lowerCAmelCase = argparse.ArgumentParser(description='Simple example of training script.' )
parser.add_argument(
'--mixed_precision', type=lowerCAmelCase_, default=lowerCAmelCase_, choices=['no', 'fp16', 'bf16', 'fp8'], help='Whether to use mixed precision. Choose'
'between fp16 and bf16 (bfloat16). Bf16 requires PyTorch >= 1.10.'
'and an Nvidia Ampere GPU.', )
parser.add_argument('--cpu', action='store_true', help='If passed, will train on the CPU.' )
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=lowerCAmelCase_, default='logs', help='Location on where to store experiment tracking logs` and relevent project information', )
__lowerCAmelCase = parser.parse_args()
__lowerCAmelCase = {'lr': 2E-5, 'num_epochs': 3, 'seed': 42, 'batch_size': 16}
training_function(lowerCAmelCase_, lowerCAmelCase_ )
if __name__ == "__main__":
main()
| 53 |
import json
import logging
import os
import sys
from pathlib import Path
import finetune_rag
from transformers.file_utils import is_apex_available
from transformers.testing_utils import (
TestCasePlus,
execute_subprocess_async,
require_ray,
require_torch_gpu,
require_torch_multi_gpu,
)
logging.basicConfig(level=logging.DEBUG)
_snake_case : Tuple = logging.getLogger()
_snake_case : Any = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Any , lowerCAmelCase_ : Dict ) -> Optional[int]:
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = {'source': 'What is love ?', 'target': 'life'}
__lowerCAmelCase = {'train': 1_2, 'val': 2, 'test': 2}
for split in ["train", "test", "val"]:
for field in ["source", "target"]:
__lowerCAmelCase = '\n'.join([contents[field]] * n_lines[split] )
with open(os.path.join(lowerCAmelCase_ , f"""{split}.{field}""" ) , 'w' ) as f:
f.write(lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : int , lowerCAmelCase_ : str = "pytorch" ) -> List[str]:
__lowerCAmelCase = self.get_auto_remove_tmp_dir()
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'output' )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'data' )
self._create_dummy_data(data_dir=lowerCAmelCase_ )
__lowerCAmelCase = f"""
--data_dir {data_dir} \
--output_dir {output_dir} \
--model_name_or_path facebook/rag-sequence-base \
--model_type rag_sequence \
--do_train \
--do_predict \
--n_val -1 \
--val_check_interval 1.0 \
--train_batch_size 2 \
--eval_batch_size 1 \
--max_source_length 25 \
--max_target_length 25 \
--val_max_target_length 25 \
--test_max_target_length 25 \
--label_smoothing 0.1 \
--dropout 0.1 \
--attention_dropout 0.1 \
--weight_decay 0.001 \
--adam_epsilon 1e-08 \
--max_grad_norm 0.1 \
--lr_scheduler polynomial \
--learning_rate 3e-04 \
--num_train_epochs 1 \
--warmup_steps 4 \
--gradient_accumulation_steps 1 \
--distributed-port 8787 \
--use_dummy_dataset 1 \
--distributed_retriever {distributed_retriever} \
""".split()
if gpus > 0:
testargs.append(f"""--gpus={gpus}""" )
if is_apex_available():
testargs.append('--fp16' )
else:
testargs.append('--gpus=0' )
testargs.append('--distributed_backend=ddp_cpu' )
testargs.append('--num_processes=2' )
__lowerCAmelCase = [sys.executable, str(Path(finetune_rag.__file__ ).resolve() )] + testargs
execute_subprocess_async(lowerCAmelCase_ , env=self.get_env() )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'metrics.json' )
with open(lowerCAmelCase_ ) as f:
__lowerCAmelCase = json.load(lowerCAmelCase_ )
return result
@require_torch_gpu
def lowercase ( self : str ) -> int:
__lowerCAmelCase = self._run_finetune(gpus=1 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
def lowercase ( self : List[str] ) -> Dict:
__lowerCAmelCase = self._run_finetune(gpus=2 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_gpu
@require_ray
def lowercase ( self : int ) -> Tuple:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
@require_ray
def lowercase ( self : List[Any] ) -> str:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
| 53 | 1 |
import argparse
import json
from collections import OrderedDict
from functools import partial
from pathlib import Path
import timm
import torch
from huggingface_hub import hf_hub_download
from transformers import LevitConfig, LevitForImageClassificationWithTeacher, LevitImageProcessor
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Dict = logging.get_logger()
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : LevitConfig, lowerCAmelCase_ : Path, lowerCAmelCase_ : bool = True ):
print(F"""Converting {name}...""" )
with torch.no_grad():
if hidden_sizes == 128:
if name[-1] == "S":
__lowerCAmelCase = timm.create_model('levit_128s', pretrained=lowerCAmelCase_ )
else:
__lowerCAmelCase = timm.create_model('levit_128', pretrained=lowerCAmelCase_ )
if hidden_sizes == 192:
__lowerCAmelCase = timm.create_model('levit_192', pretrained=lowerCAmelCase_ )
if hidden_sizes == 256:
__lowerCAmelCase = timm.create_model('levit_256', pretrained=lowerCAmelCase_ )
if hidden_sizes == 384:
__lowerCAmelCase = timm.create_model('levit_384', pretrained=lowerCAmelCase_ )
from_model.eval()
__lowerCAmelCase = LevitForImageClassificationWithTeacher(lowerCAmelCase_ ).eval()
__lowerCAmelCase = OrderedDict()
__lowerCAmelCase = from_model.state_dict()
__lowerCAmelCase = list(from_model.state_dict().keys() )
__lowerCAmelCase = list(our_model.state_dict().keys() )
print(len(lowerCAmelCase_ ), len(lowerCAmelCase_ ) )
for i in range(len(lowerCAmelCase_ ) ):
__lowerCAmelCase = weights[og_keys[i]]
our_model.load_state_dict(lowerCAmelCase_ )
__lowerCAmelCase = torch.randn((2, 3, 224, 224) )
__lowerCAmelCase = from_model(lowerCAmelCase_ )
__lowerCAmelCase = our_model(lowerCAmelCase_ ).logits
assert torch.allclose(lowerCAmelCase_, lowerCAmelCase_ ), "The model logits don't match the original one."
__lowerCAmelCase = name
print(lowerCAmelCase_ )
if push_to_hub:
our_model.save_pretrained(save_directory / checkpoint_name )
__lowerCAmelCase = LevitImageProcessor()
image_processor.save_pretrained(save_directory / checkpoint_name )
print(F"""Pushed {checkpoint_name}""" )
def a_ ( lowerCAmelCase_ : Path, lowerCAmelCase_ : str = None, lowerCAmelCase_ : bool = True ):
__lowerCAmelCase = 'imagenet-1k-id2label.json'
__lowerCAmelCase = 1000
__lowerCAmelCase = (1, num_labels)
__lowerCAmelCase = 'huggingface/label-files'
__lowerCAmelCase = num_labels
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
__lowerCAmelCase = idalabel
__lowerCAmelCase = {v: k for k, v in idalabel.items()}
__lowerCAmelCase = partial(lowerCAmelCase_, num_labels=lowerCAmelCase_, idalabel=lowerCAmelCase_, labelaid=lowerCAmelCase_ )
__lowerCAmelCase = {
'levit-128S': 128,
'levit-128': 128,
'levit-192': 192,
'levit-256': 256,
'levit-384': 384,
}
__lowerCAmelCase = {
'levit-128S': ImageNetPreTrainedConfig(
hidden_sizes=[128, 256, 384], num_attention_heads=[4, 6, 8], depths=[2, 3, 4], key_dim=[16, 16, 16], drop_path_rate=0, ),
'levit-128': ImageNetPreTrainedConfig(
hidden_sizes=[128, 256, 384], num_attention_heads=[4, 8, 12], depths=[4, 4, 4], key_dim=[16, 16, 16], drop_path_rate=0, ),
'levit-192': ImageNetPreTrainedConfig(
hidden_sizes=[192, 288, 384], num_attention_heads=[3, 5, 6], depths=[4, 4, 4], key_dim=[32, 32, 32], drop_path_rate=0, ),
'levit-256': ImageNetPreTrainedConfig(
hidden_sizes=[256, 384, 512], num_attention_heads=[4, 6, 8], depths=[4, 4, 4], key_dim=[32, 32, 32], drop_path_rate=0, ),
'levit-384': ImageNetPreTrainedConfig(
hidden_sizes=[384, 512, 768], num_attention_heads=[6, 9, 12], depths=[4, 4, 4], key_dim=[32, 32, 32], drop_path_rate=0.1, ),
}
if model_name:
convert_weight_and_push(
names_to_hidden_sizes[model_name], lowerCAmelCase_, names_to_config[model_name], lowerCAmelCase_, lowerCAmelCase_ )
else:
for model_name, config in names_to_config.items():
convert_weight_and_push(names_to_hidden_sizes[model_name], lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
return config, expected_shape
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default=None,
type=str,
help='The name of the model you wish to convert, it must be one of the supported Levit* architecture,',
)
parser.add_argument(
'--pytorch_dump_folder_path',
default='levit-dump-folder/',
type=Path,
required=False,
help='Path to the output PyTorch model directory.',
)
parser.add_argument('--push_to_hub', action='store_true', help='Push model and image processor to the hub')
parser.add_argument(
'--no-push_to_hub',
dest='push_to_hub',
action='store_false',
help='Do not push model and image processor to the hub',
)
_snake_case : List[Any] = parser.parse_args()
_snake_case : Path = args.pytorch_dump_folder_path
pytorch_dump_folder_path.mkdir(exist_ok=True, parents=True)
convert_weights_and_push(pytorch_dump_folder_path, args.model_name, args.push_to_hub)
| 53 |
import copy
import inspect
import unittest
from transformers import AutoBackbone
from transformers.configuration_utils import PretrainedConfig
from transformers.testing_utils import require_timm, require_torch, torch_device
from transformers.utils.import_utils import is_torch_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor
if is_torch_available():
import torch
from transformers import TimmBackbone, TimmBackboneConfig
from ...test_pipeline_mixin import PipelineTesterMixin
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : Optional[int]=None , lowerCAmelCase_ : Union[str, Any]="resnet50" , lowerCAmelCase_ : str=3 , lowerCAmelCase_ : List[str]=3_2 , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : Dict=True , lowerCAmelCase_ : Optional[Any]=True , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = out_indices if out_indices is not None else [4]
__lowerCAmelCase = stage_names
__lowerCAmelCase = out_features
__lowerCAmelCase = backbone
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = use_pretrained_backbone
__lowerCAmelCase = is_training
def lowercase ( self : List[str] ) -> Any:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : List[Any] ) -> Union[str, Any]:
return TimmBackboneConfig(
image_size=self.image_size , num_channels=self.num_channels , out_features=self.out_features , out_indices=self.out_indices , stage_names=self.stage_names , use_pretrained_backbone=self.use_pretrained_backbone , backbone=self.backbone , )
def lowercase ( self : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ) -> int:
__lowerCAmelCase = TimmBackbone(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.feature_map[-1].shape , (self.batch_size, model.channels[-1], 1_4, 1_4) , )
def lowercase ( self : List[str] ) -> str:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
@require_timm
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (TimmBackbone,) if is_torch_available() else ()
a_ = {"""feature-extraction""": TimmBackbone} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Tuple ) -> int:
__lowerCAmelCase = TimmBackboneModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : Dict ) -> List[str]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Union[str, Any] ) -> Optional[int]:
__lowerCAmelCase = 'resnet18'
__lowerCAmelCase = 'microsoft/resnet-18'
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(len(timm_model.stage_names ) , len(transformers_model.stage_names ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
# Out indices are set to the last layer by default. For timm models, we don't know
# the number of layers in advance, so we set it to (-1,), whereas for transformers
# models, we set it to [len(stage_names) - 1] (kept for backward compatibility).
self.assertEqual(timm_model.out_indices , (-1,) )
self.assertEqual(transformers_model.out_indices , [len(timm_model.stage_names ) - 1] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ , out_indices=[1, 2, 3] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , out_indices=[1, 2, 3] )
self.assertEqual(timm_model.out_indices , transformers_model.out_indices )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
@unittest.skip('TimmBackbone doesn\'t support feed forward chunking' )
def lowercase ( self : List[str] ) -> Tuple:
pass
@unittest.skip('TimmBackbone doesn\'t have num_hidden_layers attribute' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('TimmBackbone initialization is managed on the timm side' )
def lowercase ( self : str ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Any ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Optional[Any]:
pass
@unittest.skip('TimmBackbone model cannot be created without specifying a backbone checkpoint' )
def lowercase ( self : Dict ) -> Any:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Any ) -> Optional[int]:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : Union[str, Any] ) -> Tuple:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : List[str] ) -> Optional[int]:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Tuple ) -> List[str]:
pass
@unittest.skip('TimmBackbone doesn\'t have hidden size info in its configuration.' )
def lowercase ( self : int ) -> Optional[int]:
pass
@unittest.skip('TimmBackbone doesn\'t support output_attentions.' )
def lowercase ( self : Union[str, Any] ) -> str:
pass
@unittest.skip('Safetensors is not supported by timm.' )
def lowercase ( self : Dict ) -> str:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : List[str] ) -> Optional[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = True
__lowerCAmelCase = self.has_attentions
# no need to test all models as different heads yield the same functionality
__lowerCAmelCase = self.all_model_classes[0]
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs[0][-1]
# Encoder-/Decoder-only models
__lowerCAmelCase = outputs.hidden_states[0]
hidden_states.retain_grad()
if self.has_attentions:
__lowerCAmelCase = outputs.attentions[0]
attentions.retain_grad()
output.flatten()[0].backward(retain_graph=lowerCAmelCase_ )
self.assertIsNotNone(hidden_states.grad )
if self.has_attentions:
self.assertIsNotNone(attentions.grad )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , len(config.out_indices ) )
self.assertEqual(len(model.channels ) , len(config.out_indices ) )
# Check output of last stage is taken if out_features=None, out_indices=None
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , 1 )
self.assertEqual(len(model.channels ) , 1 )
# Check backbone can be initialized with fresh weights
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = False
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
| 53 | 1 |
import torch
from diffusers import UnCLIPScheduler
from .test_schedulers import SchedulerCommonTest
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = (UnCLIPScheduler,)
def lowercase ( self : str , **lowerCAmelCase_ : str ) -> str:
__lowerCAmelCase = {
'num_train_timesteps': 1_0_0_0,
'variance_type': 'fixed_small_log',
'clip_sample': True,
'clip_sample_range': 1.0,
'prediction_type': 'epsilon',
}
config.update(**lowerCAmelCase_ )
return config
def lowercase ( self : Optional[Any] ) -> Optional[int]:
for timesteps in [1, 5, 1_0_0, 1_0_0_0]:
self.check_over_configs(num_train_timesteps=lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> List[Any]:
for variance in ["fixed_small_log", "learned_range"]:
self.check_over_configs(variance_type=lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
for clip_sample in [True, False]:
self.check_over_configs(clip_sample=lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> List[Any]:
for clip_sample_range in [1, 5, 1_0, 2_0]:
self.check_over_configs(clip_sample_range=lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Any:
for prediction_type in ["epsilon", "sample"]:
self.check_over_configs(prediction_type=lowerCAmelCase_ )
def lowercase ( self : int ) -> Tuple:
for time_step in [0, 5_0_0, 9_9_9]:
for prev_timestep in [None, 5, 1_0_0, 2_5_0, 5_0_0, 7_5_0]:
if prev_timestep is not None and prev_timestep >= time_step:
continue
self.check_over_forward(time_step=lowerCAmelCase_ , prev_timestep=lowerCAmelCase_ )
def lowercase ( self : Any ) -> Dict:
__lowerCAmelCase = self.scheduler_classes[0]
__lowerCAmelCase = self.get_scheduler_config(variance_type='fixed_small_log' )
__lowerCAmelCase = scheduler_class(**lowerCAmelCase_ )
assert torch.sum(torch.abs(scheduler._get_variance(0 ) - 1.00_00e-10 ) ) < 1e-5
assert torch.sum(torch.abs(scheduler._get_variance(4_8_7 ) - 0.0_54_96_25 ) ) < 1e-5
assert torch.sum(torch.abs(scheduler._get_variance(9_9_9 ) - 0.9_99_49_87 ) ) < 1e-5
def lowercase ( self : List[Any] ) -> int:
__lowerCAmelCase = self.scheduler_classes[0]
__lowerCAmelCase = self.get_scheduler_config(variance_type='learned_range' )
__lowerCAmelCase = scheduler_class(**lowerCAmelCase_ )
__lowerCAmelCase = 0.5
assert scheduler._get_variance(1 , predicted_variance=lowerCAmelCase_ ) - -10.1_71_27_90 < 1e-5
assert scheduler._get_variance(4_8_7 , predicted_variance=lowerCAmelCase_ ) - -5.7_99_80_52 < 1e-5
assert scheduler._get_variance(9_9_9 , predicted_variance=lowerCAmelCase_ ) - -0.0_01_00_11 < 1e-5
def lowercase ( self : Dict ) -> Optional[int]:
__lowerCAmelCase = self.scheduler_classes[0]
__lowerCAmelCase = self.get_scheduler_config()
__lowerCAmelCase = scheduler_class(**lowerCAmelCase_ )
__lowerCAmelCase = scheduler.timesteps
__lowerCAmelCase = self.dummy_model()
__lowerCAmelCase = self.dummy_sample_deter
__lowerCAmelCase = torch.manual_seed(0 )
for i, t in enumerate(lowerCAmelCase_ ):
# 1. predict noise residual
__lowerCAmelCase = model(lowerCAmelCase_ , lowerCAmelCase_ )
# 2. predict previous mean of sample x_t-1
__lowerCAmelCase = scheduler.step(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , generator=lowerCAmelCase_ ).prev_sample
__lowerCAmelCase = pred_prev_sample
__lowerCAmelCase = torch.sum(torch.abs(lowerCAmelCase_ ) )
__lowerCAmelCase = torch.mean(torch.abs(lowerCAmelCase_ ) )
assert abs(result_sum.item() - 2_52.2_68_24_95 ) < 1e-2
assert abs(result_mean.item() - 0.3_28_47_43 ) < 1e-3
def lowercase ( self : List[str] ) -> Optional[int]:
__lowerCAmelCase = self.scheduler_classes[0]
__lowerCAmelCase = self.get_scheduler_config()
__lowerCAmelCase = scheduler_class(**lowerCAmelCase_ )
scheduler.set_timesteps(2_5 )
__lowerCAmelCase = scheduler.timesteps
__lowerCAmelCase = self.dummy_model()
__lowerCAmelCase = self.dummy_sample_deter
__lowerCAmelCase = torch.manual_seed(0 )
for i, t in enumerate(lowerCAmelCase_ ):
# 1. predict noise residual
__lowerCAmelCase = model(lowerCAmelCase_ , lowerCAmelCase_ )
if i + 1 == timesteps.shape[0]:
__lowerCAmelCase = None
else:
__lowerCAmelCase = timesteps[i + 1]
# 2. predict previous mean of sample x_t-1
__lowerCAmelCase = scheduler.step(
lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , prev_timestep=lowerCAmelCase_ , generator=lowerCAmelCase_ ).prev_sample
__lowerCAmelCase = pred_prev_sample
__lowerCAmelCase = torch.sum(torch.abs(lowerCAmelCase_ ) )
__lowerCAmelCase = torch.mean(torch.abs(lowerCAmelCase_ ) )
assert abs(result_sum.item() - 2_58.2_04_49_83 ) < 1e-2
assert abs(result_mean.item() - 0.3_36_20_38 ) < 1e-3
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
pass
def lowercase ( self : Tuple ) -> Tuple:
pass
| 53 |
# Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import os
import platform
import numpy as np
import psutil
import torch
from accelerate import __version__ as version
from accelerate.commands.config import default_config_file, load_config_from_file
from ..utils import is_npu_available, is_xpu_available
def a_ ( lowerCAmelCase_ : str=None ):
if subparsers is not None:
__lowerCAmelCase = subparsers.add_parser('env' )
else:
__lowerCAmelCase = argparse.ArgumentParser('Accelerate env command' )
parser.add_argument(
'--config_file', default=lowerCAmelCase_, help='The config file to use for the default values in the launching script.' )
if subparsers is not None:
parser.set_defaults(func=lowerCAmelCase_ )
return parser
def a_ ( lowerCAmelCase_ : Optional[int] ):
__lowerCAmelCase = torch.__version__
__lowerCAmelCase = torch.cuda.is_available()
__lowerCAmelCase = is_xpu_available()
__lowerCAmelCase = is_npu_available()
__lowerCAmelCase = 'Not found'
# Get the default from the config file.
if args.config_file is not None or os.path.isfile(lowerCAmelCase_ ):
__lowerCAmelCase = load_config_from_file(args.config_file ).to_dict()
__lowerCAmelCase = {
'`Accelerate` version': version,
'Platform': platform.platform(),
'Python version': platform.python_version(),
'Numpy version': np.__version__,
'PyTorch version (GPU?)': F"""{pt_version} ({pt_cuda_available})""",
'PyTorch XPU available': str(lowerCAmelCase_ ),
'PyTorch NPU available': str(lowerCAmelCase_ ),
'System RAM': F"""{psutil.virtual_memory().total / 1024 ** 3:.2f} GB""",
}
if pt_cuda_available:
__lowerCAmelCase = torch.cuda.get_device_name()
print('\nCopy-and-paste the text below in your GitHub issue\n' )
print('\n'.join([F"""- {prop}: {val}""" for prop, val in info.items()] ) )
print('- `Accelerate` default config:' if args.config_file is None else '- `Accelerate` config passed:' )
__lowerCAmelCase = (
'\n'.join([F"""\t- {prop}: {val}""" for prop, val in accelerate_config.items()] )
if isinstance(lowerCAmelCase_, lowerCAmelCase_ )
else F"""\t{accelerate_config}"""
)
print(lowerCAmelCase_ )
__lowerCAmelCase = accelerate_config
return info
def a_ ( ):
__lowerCAmelCase = env_command_parser()
__lowerCAmelCase = parser.parse_args()
env_command(lowerCAmelCase_ )
return 0
if __name__ == "__main__":
raise SystemExit(main())
| 53 | 1 |
import gc
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModelWithProjection, CLIPTokenizer
from diffusers import HeunDiscreteScheduler, PriorTransformer, ShapEPipeline
from diffusers.pipelines.shap_e import ShapERenderer
from diffusers.utils import load_numpy, slow
from diffusers.utils.testing_utils import require_torch_gpu, torch_device
from ..test_pipelines_common import PipelineTesterMixin, assert_mean_pixel_difference
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = ShapEPipeline
a_ = ["""prompt"""]
a_ = ["""prompt"""]
a_ = [
"""num_images_per_prompt""",
"""num_inference_steps""",
"""generator""",
"""latents""",
"""guidance_scale""",
"""frame_size""",
"""output_type""",
"""return_dict""",
]
a_ = False
@property
def lowercase ( self : int ) -> Any:
return 3_2
@property
def lowercase ( self : int ) -> Optional[int]:
return 3_2
@property
def lowercase ( self : Union[str, Any] ) -> Optional[int]:
return self.time_input_dim * 4
@property
def lowercase ( self : Union[str, Any] ) -> List[str]:
return 8
@property
def lowercase ( self : Optional[int] ) -> List[str]:
__lowerCAmelCase = CLIPTokenizer.from_pretrained('hf-internal-testing/tiny-random-clip' )
return tokenizer
@property
def lowercase ( self : Optional[Any] ) -> Optional[int]:
torch.manual_seed(0 )
__lowerCAmelCase = CLIPTextConfig(
bos_token_id=0 , eos_token_id=2 , hidden_size=self.text_embedder_hidden_size , projection_dim=self.text_embedder_hidden_size , intermediate_size=3_7 , layer_norm_eps=1e-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1_0_0_0 , )
return CLIPTextModelWithProjection(lowerCAmelCase_ )
@property
def lowercase ( self : str ) -> Optional[Any]:
torch.manual_seed(0 )
__lowerCAmelCase = {
'num_attention_heads': 2,
'attention_head_dim': 1_6,
'embedding_dim': self.time_input_dim,
'num_embeddings': 3_2,
'embedding_proj_dim': self.text_embedder_hidden_size,
'time_embed_dim': self.time_embed_dim,
'num_layers': 1,
'clip_embed_dim': self.time_input_dim * 2,
'additional_embeddings': 0,
'time_embed_act_fn': 'gelu',
'norm_in_type': 'layer',
'encoder_hid_proj_type': None,
'added_emb_type': None,
}
__lowerCAmelCase = PriorTransformer(**lowerCAmelCase_ )
return model
@property
def lowercase ( self : Union[str, Any] ) -> List[Any]:
torch.manual_seed(0 )
__lowerCAmelCase = {
'param_shapes': (
(self.renderer_dim, 9_3),
(self.renderer_dim, 8),
(self.renderer_dim, 8),
(self.renderer_dim, 8),
),
'd_latent': self.time_input_dim,
'd_hidden': self.renderer_dim,
'n_output': 1_2,
'background': (
0.1,
0.1,
0.1,
),
}
__lowerCAmelCase = ShapERenderer(**lowerCAmelCase_ )
return model
def lowercase ( self : Tuple ) -> Dict:
__lowerCAmelCase = self.dummy_prior
__lowerCAmelCase = self.dummy_text_encoder
__lowerCAmelCase = self.dummy_tokenizer
__lowerCAmelCase = self.dummy_renderer
__lowerCAmelCase = HeunDiscreteScheduler(
beta_schedule='exp' , num_train_timesteps=1_0_2_4 , prediction_type='sample' , use_karras_sigmas=lowerCAmelCase_ , clip_sample=lowerCAmelCase_ , clip_sample_range=1.0 , )
__lowerCAmelCase = {
'prior': prior,
'text_encoder': text_encoder,
'tokenizer': tokenizer,
'renderer': renderer,
'scheduler': scheduler,
}
return components
def lowercase ( self : List[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[int]=0 ) -> int:
if str(lowerCAmelCase_ ).startswith('mps' ):
__lowerCAmelCase = torch.manual_seed(lowerCAmelCase_ )
else:
__lowerCAmelCase = torch.Generator(device=lowerCAmelCase_ ).manual_seed(lowerCAmelCase_ )
__lowerCAmelCase = {
'prompt': 'horse',
'generator': generator,
'num_inference_steps': 1,
'frame_size': 3_2,
'output_type': 'np',
}
return inputs
def lowercase ( self : List[str] ) -> str:
__lowerCAmelCase = 'cpu'
__lowerCAmelCase = self.get_dummy_components()
__lowerCAmelCase = self.pipeline_class(**lowerCAmelCase_ )
__lowerCAmelCase = pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = pipe(**self.get_dummy_inputs(lowerCAmelCase_ ) )
__lowerCAmelCase = output.images[0]
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (2_0, 3_2, 3_2, 3)
__lowerCAmelCase = np.array(
[
0.00_03_92_16,
0.00_03_92_16,
0.00_03_92_16,
0.00_03_92_16,
0.00_03_92_16,
0.00_03_92_16,
0.00_03_92_16,
0.00_03_92_16,
0.00_03_92_16,
] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def lowercase ( self : int ) -> List[str]:
# NOTE: Larger batch sizes cause this test to timeout, only test on smaller batches
self._test_inference_batch_consistent(batch_sizes=[1, 2] )
def lowercase ( self : Any ) -> str:
__lowerCAmelCase = torch_device == 'cpu'
__lowerCAmelCase = True
self._test_inference_batch_single_identical(
batch_size=2 , test_max_difference=lowerCAmelCase_ , relax_max_difference=lowerCAmelCase_ , )
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = self.get_dummy_components()
__lowerCAmelCase = self.pipeline_class(**lowerCAmelCase_ )
__lowerCAmelCase = pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = 1
__lowerCAmelCase = 2
__lowerCAmelCase = self.get_dummy_inputs(lowerCAmelCase_ )
for key in inputs.keys():
if key in self.batch_params:
__lowerCAmelCase = batch_size * [inputs[key]]
__lowerCAmelCase = pipe(**lowerCAmelCase_ , num_images_per_prompt=lowerCAmelCase_ )[0]
assert images.shape[0] == batch_size * num_images_per_prompt
@slow
@require_torch_gpu
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Dict ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def lowercase ( self : int ) -> Dict:
__lowerCAmelCase = load_numpy(
'https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main'
'/shap_e/test_shap_e_np_out.npy' )
__lowerCAmelCase = ShapEPipeline.from_pretrained('openai/shap-e' )
__lowerCAmelCase = pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = torch.Generator(device=lowerCAmelCase_ ).manual_seed(0 )
__lowerCAmelCase = pipe(
'a shark' , generator=lowerCAmelCase_ , guidance_scale=15.0 , num_inference_steps=6_4 , frame_size=6_4 , output_type='np' , ).images[0]
assert images.shape == (2_0, 6_4, 6_4, 3)
assert_mean_pixel_difference(lowerCAmelCase_ , lowerCAmelCase_ )
| 53 |
import importlib
import sys
from argparse import REMAINDER, ArgumentParser
from pathlib import Path
import torch_xla.distributed.xla_multiprocessing as xmp
def a_ ( ):
__lowerCAmelCase = ArgumentParser(
description=(
'PyTorch TPU distributed training launch '
'helper utility that will spawn up '
'multiple distributed processes'
) )
# Optional arguments for the launch helper
parser.add_argument('--num_cores', type=lowerCAmelCase_, default=1, help='Number of TPU cores to use (1 or 8).' )
# positional
parser.add_argument(
'training_script', type=lowerCAmelCase_, help=(
'The full path to the single TPU training '
'program/script to be launched in parallel, '
'followed by all the arguments for the '
'training script'
), )
# rest from the training program
parser.add_argument('training_script_args', nargs=lowerCAmelCase_ )
return parser.parse_args()
def a_ ( ):
__lowerCAmelCase = parse_args()
# Import training_script as a module.
__lowerCAmelCase = Path(args.training_script )
sys.path.append(str(script_fpath.parent.resolve() ) )
__lowerCAmelCase = script_fpath.stem
__lowerCAmelCase = importlib.import_module(lowerCAmelCase_ )
# Patch sys.argv
__lowerCAmelCase = [args.training_script] + args.training_script_args + ['--tpu_num_cores', str(args.num_cores )]
xmp.spawn(mod._mp_fn, args=(), nprocs=args.num_cores )
if __name__ == "__main__":
main()
| 53 | 1 |
from math import sqrt
def a_ ( lowerCAmelCase_ : int ):
if 1 < number < 4:
# 2 and 3 are primes
return True
elif number < 2 or number % 2 == 0 or number % 3 == 0:
# Negatives, 0, 1, all even numbers, all multiples of 3 are not primes
return False
# All primes number are in format of 6k +/- 1
for i in range(5, int(sqrt(lowerCAmelCase_ ) + 1 ), 6 ):
if number % i == 0 or number % (i + 2) == 0:
return False
return True
def a_ ( lowerCAmelCase_ : int = 1_0001 ):
__lowerCAmelCase = 0
__lowerCAmelCase = 1
while count != nth and number < 3:
number += 1
if is_prime(lowerCAmelCase_ ):
count += 1
while count != nth:
number += 2
if is_prime(lowerCAmelCase_ ):
count += 1
return number
if __name__ == "__main__":
print(F"""{solution() = }""")
| 53 |
import inspect
import unittest
from huggingface_hub import hf_hub_download
from transformers import ConvNextConfig, UperNetConfig
from transformers.testing_utils import require_torch, require_torch_multi_gpu, require_vision, slow, torch_device
from transformers.utils import is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import UperNetForSemanticSegmentation
from transformers.models.upernet.modeling_upernet import UPERNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Dict=1_3 , lowerCAmelCase_ : str=3_2 , lowerCAmelCase_ : Optional[Any]=3 , lowerCAmelCase_ : Any=4 , lowerCAmelCase_ : str=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Tuple=[2, 2, 3, 2] , lowerCAmelCase_ : str=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[int]=3_7 , lowerCAmelCase_ : Dict="gelu" , lowerCAmelCase_ : List[Any]=1_0 , lowerCAmelCase_ : str=0.02 , lowerCAmelCase_ : Dict=["stage2", "stage3", "stage4"] , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : List[Any]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = num_stages
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = out_features
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = num_stages
def lowercase ( self : Dict ) -> List[str]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = self.get_config()
return config, pixel_values, labels
def lowercase ( self : List[str] ) -> Union[str, Any]:
return ConvNextConfig(
num_channels=self.num_channels , num_stages=self.num_stages , hidden_sizes=self.hidden_sizes , depths=self.depths , is_training=self.is_training , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , out_features=self.out_features , )
def lowercase ( self : Dict ) -> List[str]:
return UperNetConfig(
backbone_config=self.get_backbone_config() , hidden_size=5_1_2 , pool_scales=[1, 2, 3, 6] , use_auxiliary_head=lowerCAmelCase_ , auxiliary_loss_weight=0.4 , auxiliary_in_channels=4_0 , auxiliary_channels=2_5_6 , auxiliary_num_convs=1 , auxiliary_concat_input=lowerCAmelCase_ , loss_ignore_index=2_5_5 , num_labels=self.num_labels , )
def lowercase ( self : List[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : int ) -> Optional[Any]:
__lowerCAmelCase = UperNetForSemanticSegmentation(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.logits.shape , (self.batch_size, self.num_labels, self.image_size, self.image_size) )
def lowercase ( self : Union[str, Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (UperNetForSemanticSegmentation,) if is_torch_available() else ()
a_ = {"""image-segmentation""": UperNetForSemanticSegmentation} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = UperNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Tuple ) -> Union[str, Any]:
return
def lowercase ( self : Optional[int] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_semantic_segmentation(*lowerCAmelCase_ )
@unittest.skip(reason='UperNet does not use inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Dict:
pass
@unittest.skip(reason='UperNet does not support input and output embeddings' )
def lowercase ( self : Optional[Any] ) -> Dict:
pass
@unittest.skip(reason='UperNet does not have a base model' )
def lowercase ( self : Optional[int] ) -> List[Any]:
pass
@unittest.skip(reason='UperNet does not have a base model' )
def lowercase ( self : str ) -> Dict:
pass
@require_torch_multi_gpu
@unittest.skip(reason='UperNet has some layers using `add_module` which doesn\'t work well with `nn.DataParallel`' )
def lowercase ( self : Optional[Any] ) -> Optional[int]:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : Tuple ) -> List[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Tuple:
def check_hidden_states_output(lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
# ConvNext's feature maps are of shape (batch_size, num_channels, height, width)
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [self.model_tester.image_size // 4, self.model_tester.image_size // 4] , )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Any ) -> Tuple:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = _config_zero_init(lowerCAmelCase_ )
__lowerCAmelCase = _config_zero_init(configs_no_init.backbone_config )
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(config=lowerCAmelCase_ )
for name, param in model.named_parameters():
if param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() , [0.0, 1.0] , msg=f"""Parameter {name} of model {model_class} seems not properly initialized""" , )
@unittest.skip(reason='UperNet does not have tied weights' )
def lowercase ( self : Any ) -> int:
pass
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in UPERNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
def a_ ( ):
__lowerCAmelCase = hf_hub_download(
repo_id='hf-internal-testing/fixtures_ade20k', repo_type='dataset', filename='ADE_val_00000001.jpg' )
__lowerCAmelCase = Image.open(lowerCAmelCase_ ).convert('RGB' )
return image
@require_torch
@require_vision
@slow
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Dict ) -> Union[str, Any]:
__lowerCAmelCase = AutoImageProcessor.from_pretrained('openmmlab/upernet-swin-tiny' )
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained('openmmlab/upernet-swin-tiny' ).to(lowerCAmelCase_ )
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = torch.Size((1, model.config.num_labels, 5_1_2, 5_1_2) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-7.59_58, -7.59_58, -7.43_02], [-7.59_58, -7.59_58, -7.43_02], [-7.47_97, -7.47_97, -7.30_68]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, 0, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = AutoImageProcessor.from_pretrained('openmmlab/upernet-convnext-tiny' )
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained('openmmlab/upernet-convnext-tiny' ).to(lowerCAmelCase_ )
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = torch.Size((1, model.config.num_labels, 5_1_2, 5_1_2) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-8.81_10, -8.81_10, -8.65_21], [-8.81_10, -8.81_10, -8.65_21], [-8.77_46, -8.77_46, -8.61_30]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, 0, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import collections
import inspect
import unittest
from typing import Dict, List, Tuple
from transformers import MaskFormerSwinConfig
from transformers.testing_utils import require_torch, require_torch_multi_gpu, torch_device
from transformers.utils import is_torch_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import MaskFormerSwinBackbone
from transformers.models.maskformer import MaskFormerSwinModel
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : str=1_3 , lowerCAmelCase_ : Union[str, Any]=3_2 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : str=3 , lowerCAmelCase_ : Union[str, Any]=1_6 , lowerCAmelCase_ : Any=[1, 2, 1] , lowerCAmelCase_ : Optional[Any]=[2, 2, 4] , lowerCAmelCase_ : str=2 , lowerCAmelCase_ : str=2.0 , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : List[str]=0.0 , lowerCAmelCase_ : Optional[int]=0.0 , lowerCAmelCase_ : int=0.1 , lowerCAmelCase_ : Any="gelu" , lowerCAmelCase_ : Optional[int]=False , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Optional[int]=0.02 , lowerCAmelCase_ : int=1e-5 , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : Optional[Any]=None , lowerCAmelCase_ : int=True , lowerCAmelCase_ : List[Any]=1_0 , lowerCAmelCase_ : Tuple=8 , lowerCAmelCase_ : Optional[Any]=["stage1", "stage2", "stage3"] , lowerCAmelCase_ : Dict=[1, 2, 3] , ) -> Optional[int]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = patch_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = embed_dim
__lowerCAmelCase = depths
__lowerCAmelCase = num_heads
__lowerCAmelCase = window_size
__lowerCAmelCase = mlp_ratio
__lowerCAmelCase = qkv_bias
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = drop_path_rate
__lowerCAmelCase = hidden_act
__lowerCAmelCase = use_absolute_embeddings
__lowerCAmelCase = patch_norm
__lowerCAmelCase = layer_norm_eps
__lowerCAmelCase = initializer_range
__lowerCAmelCase = is_training
__lowerCAmelCase = scope
__lowerCAmelCase = use_labels
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = encoder_stride
__lowerCAmelCase = out_features
__lowerCAmelCase = out_indices
def lowercase ( self : Dict ) -> List[Any]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = self.get_config()
return config, pixel_values, labels
def lowercase ( self : Dict ) -> Optional[int]:
return MaskFormerSwinConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , embed_dim=self.embed_dim , depths=self.depths , num_heads=self.num_heads , window_size=self.window_size , mlp_ratio=self.mlp_ratio , qkv_bias=self.qkv_bias , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , drop_path_rate=self.drop_path_rate , hidden_act=self.hidden_act , use_absolute_embeddings=self.use_absolute_embeddings , path_norm=self.patch_norm , layer_norm_eps=self.layer_norm_eps , initializer_range=self.initializer_range , encoder_stride=self.encoder_stride , out_features=self.out_features , out_indices=self.out_indices , )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Optional[int] ) -> Union[str, Any]:
__lowerCAmelCase = MaskFormerSwinModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
__lowerCAmelCase = ((config.image_size // config.patch_size) ** 2) // (4 ** (len(config.depths ) - 1))
__lowerCAmelCase = int(config.embed_dim * 2 ** (len(config.depths ) - 1) )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, expected_seq_len, expected_dim) )
def lowercase ( self : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[Any] ) -> Optional[int]:
__lowerCAmelCase = MaskFormerSwinBackbone(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) , len(config.out_features ) )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) , [1_3, 1_6, 1_6, 1_6] )
# verify channels
self.parent.assertEqual(len(model.channels ) , len(config.out_features ) )
self.parent.assertListEqual(model.channels , [1_6, 3_2, 6_4] )
# verify ValueError
with self.parent.assertRaises(lowerCAmelCase_ ):
__lowerCAmelCase = ['stem']
__lowerCAmelCase = MaskFormerSwinBackbone(config=lowerCAmelCase_ )
def lowercase ( self : str ) -> List[str]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (
(
MaskFormerSwinModel,
MaskFormerSwinBackbone,
)
if is_torch_available()
else ()
)
a_ = {"""feature-extraction""": MaskFormerSwinModel} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : List[Any] ) -> List[Any]:
__lowerCAmelCase = MaskFormerSwinModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , embed_dim=3_7 )
@require_torch_multi_gpu
@unittest.skip(
reason=(
'`MaskFormerSwinModel` outputs `hidden_states_spatial_dimensions` which doesn\'t work well with'
' `nn.DataParallel`'
) )
def lowercase ( self : Optional[Any] ) -> List[Any]:
pass
def lowercase ( self : Any ) -> Optional[Any]:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : List[Any] ) -> List[str]:
return
def lowercase ( self : List[Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> List[str]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_backbone(*lowerCAmelCase_ )
@unittest.skip('Swin does not use inputs_embeds' )
def lowercase ( self : Optional[Any] ) -> str:
pass
@unittest.skip('Swin does not support feedforward chunking' )
def lowercase ( self : Optional[int] ) -> List[Any]:
pass
def lowercase ( self : Any ) -> List[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
__lowerCAmelCase = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCAmelCase_ , nn.Linear ) )
def lowercase ( self : str ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
@unittest.skip(reason='MaskFormerSwin is only used as backbone and doesn\'t support output_attentions' )
def lowercase ( self : List[Any] ) -> Optional[int]:
pass
@unittest.skip(reason='MaskFormerSwin is only used as an internal backbone' )
def lowercase ( self : Dict ) -> Any:
pass
def lowercase ( self : Dict , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any ) -> Union[str, Any]:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.hidden_states
__lowerCAmelCase = getattr(
self.model_tester , 'expected_num_hidden_layers' , len(self.model_tester.depths ) + 1 )
self.assertEqual(len(lowerCAmelCase_ ) , lowerCAmelCase_ )
# Swin has a different seq_length
__lowerCAmelCase = (
config.patch_size
if isinstance(config.patch_size , collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
__lowerCAmelCase = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [num_patches, self.model_tester.embed_dim] , )
def lowercase ( self : int ) -> Any:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size , collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
for model_class in self.all_model_classes:
__lowerCAmelCase = True
self.check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
self.check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = 3
__lowerCAmelCase = (
self.model_tester.image_size
if isinstance(self.model_tester.image_size , collections.abc.Iterable )
else (self.model_tester.image_size, self.model_tester.image_size)
)
__lowerCAmelCase = (
config.patch_size
if isinstance(config.patch_size , collections.abc.Iterable )
else (config.patch_size, config.patch_size)
)
__lowerCAmelCase = image_size[0] + patch_size[0] - (image_size[0] % patch_size[0])
__lowerCAmelCase = image_size[1] + patch_size[1] - (image_size[1] % patch_size[1])
for model_class in self.all_model_classes:
__lowerCAmelCase = True
self.check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , (padded_height, padded_width) )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
self.check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , (padded_height, padded_width) )
@unittest.skip(reason='MaskFormerSwin doesn\'t have pretrained checkpoints' )
def lowercase ( self : Optional[int] ) -> Optional[Any]:
pass
@unittest.skip(reason='This will be fixed once MaskFormerSwin is replaced by native Swin' )
def lowercase ( self : Optional[Any] ) -> Dict:
pass
@unittest.skip(reason='This will be fixed once MaskFormerSwin is replaced by native Swin' )
def lowercase ( self : Tuple ) -> List[str]:
pass
def lowercase ( self : Union[str, Any] ) -> Dict:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
def set_nan_tensor_to_zero(lowerCAmelCase_ : Dict ):
__lowerCAmelCase = 0
return t
def check_equivalence(lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Any={} ):
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ , return_dict=lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = model(**lowerCAmelCase_ , return_dict=lowerCAmelCase_ , **lowerCAmelCase_ ).to_tuple()
def recursive_check(lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[str] ):
if isinstance(lowerCAmelCase_ , (List, Tuple) ):
for tuple_iterable_value, dict_iterable_value in zip(lowerCAmelCase_ , lowerCAmelCase_ ):
recursive_check(lowerCAmelCase_ , lowerCAmelCase_ )
elif isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
for tuple_iterable_value, dict_iterable_value in zip(
tuple_object.values() , dict_object.values() ):
recursive_check(lowerCAmelCase_ , lowerCAmelCase_ )
elif tuple_object is None:
return
else:
self.assertTrue(
torch.allclose(
set_nan_tensor_to_zero(lowerCAmelCase_ ) , set_nan_tensor_to_zero(lowerCAmelCase_ ) , atol=1e-5 ) , msg=(
'Tuple and dict output are not equal. Difference:'
f""" {torch.max(torch.abs(tuple_object - dict_object ) )}. Tuple has `nan`:"""
f""" {torch.isnan(lowerCAmelCase_ ).any()} and `inf`: {torch.isinf(lowerCAmelCase_ )}. Dict has"""
f""" `nan`: {torch.isnan(lowerCAmelCase_ ).any()} and `inf`: {torch.isinf(lowerCAmelCase_ )}."""
) , )
recursive_check(lowerCAmelCase_ , lowerCAmelCase_ )
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
check_equivalence(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ , return_labels=lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ , return_labels=lowerCAmelCase_ )
check_equivalence(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
check_equivalence(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , {'output_hidden_states': True} )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ , return_labels=lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ , return_labels=lowerCAmelCase_ )
check_equivalence(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , {'output_hidden_states': True} )
@require_torch
class _UpperCAmelCase ( unittest.TestCase , _UpperCamelCase ):
"""simple docstring"""
a_ = (MaskFormerSwinBackbone,) if is_torch_available() else ()
a_ = MaskFormerSwinConfig
def lowercase ( self : List[str] ) -> Dict:
__lowerCAmelCase = MaskFormerSwinModelTester(self )
def lowercase ( self : Tuple ) -> List[str]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = inputs_dict['pixel_values'].shape[0]
for backbone_class in self.all_model_classes:
__lowerCAmelCase = backbone_class(lowerCAmelCase_ )
backbone.to(lowerCAmelCase_ )
backbone.eval()
__lowerCAmelCase = backbone(**lowerCAmelCase_ )
# Test default outputs and verify feature maps
self.assertIsInstance(outputs.feature_maps , lowerCAmelCase_ )
self.assertTrue(len(outputs.feature_maps ) == len(backbone.channels ) )
for feature_map, n_channels in zip(outputs.feature_maps , backbone.channels ):
self.assertTrue(feature_map.shape[:2] , (batch_size, n_channels) )
self.assertIsNone(outputs.hidden_states )
self.assertIsNone(outputs.attentions )
# Test output_hidden_states=True
__lowerCAmelCase = backbone(**lowerCAmelCase_ , output_hidden_states=lowerCAmelCase_ )
self.assertIsNotNone(outputs.hidden_states )
self.assertTrue(len(outputs.hidden_states ) , len(backbone.stage_names ) )
# We skip the stem layer
for hidden_states, n_channels in zip(outputs.hidden_states[1:] , backbone.channels ):
for hidden_state in hidden_states:
# Hidden states are in the format (batch_size, (height * width), n_channels)
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = hidden_state.shape
self.assertTrue((h_batch_size, h_n_channels) , (batch_size, n_channels) )
# Test output_attentions=True
if self.has_attentions:
__lowerCAmelCase = backbone(**lowerCAmelCase_ , output_attentions=lowerCAmelCase_ )
self.assertIsNotNone(outputs.attentions )
| 53 |
import pytest
from datasets import Dataset, DatasetDict, Features, NamedSplit, Value
from datasets.io.text import TextDatasetReader
from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : Optional[Any] ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ )
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('keep_in_memory', [False, True] )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : int ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_, keep_in_memory=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize(
'features', [
None,
{'text': 'string'},
{'text': 'int32'},
{'text': 'float32'},
], )
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Dict, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = features.copy() if features else default_expected_features
__lowerCAmelCase = (
Features({feature: Value(lowerCAmelCase_ ) for feature, dtype in features.items()} ) if features is not None else None
)
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, features=lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize('split', [None, NamedSplit('train' ), 'train', 'test'] )
def a_ ( lowerCAmelCase_ : List[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Any ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_, split=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
assert dataset.split == split if split else "train"
@pytest.mark.parametrize('path_type', [str, list] )
def a_ ( lowerCAmelCase_ : Dict, lowerCAmelCase_ : Any, lowerCAmelCase_ : Dict ):
if issubclass(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = text_path
elif issubclass(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = [text_path]
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : int, lowerCAmelCase_ : Tuple=("train",) ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ )
for split in splits:
__lowerCAmelCase = dataset_dict[split]
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('keep_in_memory', [False, True] )
def a_ ( lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : Dict ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__lowerCAmelCase = TextDatasetReader({'train': text_path}, cache_dir=lowerCAmelCase_, keep_in_memory=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize(
'features', [
None,
{'text': 'string'},
{'text': 'int32'},
{'text': 'float32'},
], )
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[Any] ):
__lowerCAmelCase = tmp_path / 'cache'
# CSV file loses col_1 string dtype information: default now is "int64" instead of "string"
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = features.copy() if features else default_expected_features
__lowerCAmelCase = (
Features({feature: Value(lowerCAmelCase_ ) for feature, dtype in features.items()} ) if features is not None else None
)
__lowerCAmelCase = TextDatasetReader({'train': text_path}, features=lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize('split', [None, NamedSplit('train' ), 'train', 'test'] )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : Optional[int] ):
if split:
__lowerCAmelCase = {split: text_path}
else:
__lowerCAmelCase = 'train'
__lowerCAmelCase = {'train': text_path, 'test': text_path}
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_, splits=list(path.keys() ) )
assert all(dataset[split].split == split for split in path.keys() )
| 53 | 1 |
from functools import lru_cache
def a_ ( lowerCAmelCase_ : int ):
__lowerCAmelCase = 2
__lowerCAmelCase = set()
while i * i <= n:
if n % i:
i += 1
else:
n //= i
factors.add(lowerCAmelCase_ )
if n > 1:
factors.add(lowerCAmelCase_ )
return factors
@lru_cache
def a_ ( lowerCAmelCase_ : int ):
return len(unique_prime_factors(lowerCAmelCase_ ) )
def a_ ( lowerCAmelCase_ : list ):
return len(set(lowerCAmelCase_ ) ) in (0, 1)
def a_ ( lowerCAmelCase_ : int ):
__lowerCAmelCase = 2
while True:
# Increment each value of a generated range
__lowerCAmelCase = [base + i for i in range(lowerCAmelCase_ )]
# Run elements through out unique_prime_factors function
# Append our target number to the end.
__lowerCAmelCase = [upf_len(lowerCAmelCase_ ) for x in group]
checker.append(lowerCAmelCase_ )
# If all numbers in the list are equal, return the group variable.
if equality(lowerCAmelCase_ ):
return group
# Increment our base variable by 1
base += 1
def a_ ( lowerCAmelCase_ : int = 4 ):
__lowerCAmelCase = run(lowerCAmelCase_ )
return results[0] if len(lowerCAmelCase_ ) else None
if __name__ == "__main__":
print(solution())
| 53 |
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 ViTConfig, ViTForImageClassification, ViTImageProcessor, ViTModel
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Union[str, Any] = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : int=False ):
__lowerCAmelCase = []
for i in range(config.num_hidden_layers ):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((F"""blocks.{i}.norm1.weight""", F"""vit.encoder.layer.{i}.layernorm_before.weight""") )
rename_keys.append((F"""blocks.{i}.norm1.bias""", F"""vit.encoder.layer.{i}.layernorm_before.bias""") )
rename_keys.append((F"""blocks.{i}.attn.proj.weight""", F"""vit.encoder.layer.{i}.attention.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.attn.proj.bias""", F"""vit.encoder.layer.{i}.attention.output.dense.bias""") )
rename_keys.append((F"""blocks.{i}.norm2.weight""", F"""vit.encoder.layer.{i}.layernorm_after.weight""") )
rename_keys.append((F"""blocks.{i}.norm2.bias""", F"""vit.encoder.layer.{i}.layernorm_after.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.weight""", F"""vit.encoder.layer.{i}.intermediate.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.bias""", F"""vit.encoder.layer.{i}.intermediate.dense.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.weight""", F"""vit.encoder.layer.{i}.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.bias""", F"""vit.encoder.layer.{i}.output.dense.bias""") )
# projection layer + position embeddings
rename_keys.extend(
[
('cls_token', 'vit.embeddings.cls_token'),
('patch_embed.proj.weight', 'vit.embeddings.patch_embeddings.projection.weight'),
('patch_embed.proj.bias', 'vit.embeddings.patch_embeddings.projection.bias'),
('pos_embed', 'vit.embeddings.position_embeddings'),
] )
if base_model:
# layernorm + pooler
rename_keys.extend(
[
('norm.weight', 'layernorm.weight'),
('norm.bias', 'layernorm.bias'),
] )
# if just the base model, we should remove "vit" from all keys that start with "vit"
__lowerCAmelCase = [(pair[0], pair[1][4:]) if pair[1].startswith('vit' ) else pair for pair in rename_keys]
else:
# layernorm + classification head
rename_keys.extend(
[
('norm.weight', 'vit.layernorm.weight'),
('norm.bias', 'vit.layernorm.bias'),
('head.weight', 'classifier.weight'),
('head.bias', 'classifier.bias'),
] )
return rename_keys
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Optional[int]=False ):
for i in range(config.num_hidden_layers ):
if base_model:
__lowerCAmelCase = ''
else:
__lowerCAmelCase = 'vit.'
# read in weights + bias of input projection layer (in timm, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[
: config.hidden_size, :
]
__lowerCAmelCase = in_proj_bias[: config.hidden_size]
__lowerCAmelCase = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
__lowerCAmelCase = in_proj_bias[
config.hidden_size : config.hidden_size * 2
]
__lowerCAmelCase = in_proj_weight[
-config.hidden_size :, :
]
__lowerCAmelCase = in_proj_bias[-config.hidden_size :]
def a_ ( lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = ['head.weight', 'head.bias']
for k in ignore_keys:
state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Optional[Any]=True ):
__lowerCAmelCase = ViTConfig()
# patch_size
if model_name[-1] == "8":
__lowerCAmelCase = 8
# set labels if required
if not base_model:
__lowerCAmelCase = 1000
__lowerCAmelCase = 'huggingface/label-files'
__lowerCAmelCase = 'imagenet-1k-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
__lowerCAmelCase = idalabel
__lowerCAmelCase = {v: k for k, v in idalabel.items()}
# size of the architecture
if model_name in ["dino_vits8", "dino_vits16"]:
__lowerCAmelCase = 384
__lowerCAmelCase = 1536
__lowerCAmelCase = 12
__lowerCAmelCase = 6
# load original model from torch hub
__lowerCAmelCase = torch.hub.load('facebookresearch/dino:main', lowerCAmelCase_ )
original_model.eval()
# load state_dict of original model, remove and rename some keys
__lowerCAmelCase = original_model.state_dict()
if base_model:
remove_classification_head_(lowerCAmelCase_ )
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_, base_model=lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_q_k_v(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# load HuggingFace model
if base_model:
__lowerCAmelCase = ViTModel(lowerCAmelCase_, add_pooling_layer=lowerCAmelCase_ ).eval()
else:
__lowerCAmelCase = ViTForImageClassification(lowerCAmelCase_ ).eval()
model.load_state_dict(lowerCAmelCase_ )
# Check outputs on an image, prepared by ViTImageProcessor
__lowerCAmelCase = ViTImageProcessor()
__lowerCAmelCase = image_processor(images=prepare_img(), return_tensors='pt' )
__lowerCAmelCase = encoding['pixel_values']
__lowerCAmelCase = model(lowerCAmelCase_ )
if base_model:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert torch.allclose(lowerCAmelCase_, outputs.last_hidden_state[:, 0, :], atol=1E-1 )
else:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert logits.shape == outputs.logits.shape
assert torch.allclose(lowerCAmelCase_, outputs.logits, atol=1E-3 )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
print(F"""Saving model {model_name} to {pytorch_dump_folder_path}""" )
model.save_pretrained(lowerCAmelCase_ )
print(F"""Saving image processor to {pytorch_dump_folder_path}""" )
image_processor.save_pretrained(lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : Dict = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='dino_vitb16',
type=str,
help='Name of the model trained with DINO you\'d like to convert.',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--base_model',
action='store_true',
help='Whether to only convert the base model (no projection head weights).',
)
parser.set_defaults(base_model=True)
_snake_case : List[Any] = parser.parse_args()
convert_vit_checkpoint(args.model_name, args.pytorch_dump_folder_path, args.base_model)
| 53 | 1 |
import bza
import gzip
import lzma
import os
import shutil
import struct
import tarfile
import warnings
import zipfile
from abc import ABC, abstractmethod
from pathlib import Path
from typing import Dict, List, Optional, Type, Union
from .. import config
from .filelock import FileLock
from .logging import get_logger
_snake_case : Union[str, Any] = get_logger(__name__)
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : int , lowerCAmelCase_ : Optional[str] = None ) -> Optional[Any]:
__lowerCAmelCase = (
os.path.join(lowerCAmelCase_ , config.EXTRACTED_DATASETS_DIR ) if cache_dir else config.EXTRACTED_DATASETS_PATH
)
__lowerCAmelCase = Extractor
def lowercase ( self : List[Any] , lowerCAmelCase_ : str ) -> str:
from .file_utils import hash_url_to_filename
# Path where we extract compressed archives
# We extract in the cache dir, and get the extracted path name by hashing the original path"
__lowerCAmelCase = os.path.abspath(lowerCAmelCase_ )
return os.path.join(self.extract_dir , hash_url_to_filename(lowerCAmelCase_ ) )
def lowercase ( self : Dict , lowerCAmelCase_ : str , lowerCAmelCase_ : bool ) -> bool:
return force_extract or (
not os.path.isfile(lowerCAmelCase_ ) and not (os.path.isdir(lowerCAmelCase_ ) and os.listdir(lowerCAmelCase_ ))
)
def lowercase ( self : List[str] , lowerCAmelCase_ : str , lowerCAmelCase_ : bool = False ) -> str:
__lowerCAmelCase = self.extractor.infer_extractor_format(lowerCAmelCase_ )
if not extractor_format:
return input_path
__lowerCAmelCase = self._get_output_path(lowerCAmelCase_ )
if self._do_extract(lowerCAmelCase_ , lowerCAmelCase_ ):
self.extractor.extract(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
return output_path
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
@classmethod
@abstractmethod
def lowercase ( cls : Optional[Any] , lowerCAmelCase_ : Union[Path, str] , **lowerCAmelCase_ : List[str] ) -> bool:
...
@staticmethod
@abstractmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
...
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = []
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : int ) -> List[Any]:
with open(lowerCAmelCase_ , 'rb' ) as f:
return f.read(lowerCAmelCase_ )
@classmethod
def lowercase ( cls : str , lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : bytes = b"" ) -> bool:
if not magic_number:
__lowerCAmelCase = max(len(lowerCAmelCase_ ) for cls_magic_number in cls.magic_numbers )
try:
__lowerCAmelCase = cls.read_magic_number(lowerCAmelCase_ , lowerCAmelCase_ )
except OSError:
return False
return any(magic_number.startswith(lowerCAmelCase_ ) for cls_magic_number in cls.magic_numbers )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
@classmethod
def lowercase ( cls : Tuple , lowerCAmelCase_ : Union[Path, str] , **lowerCAmelCase_ : str ) -> bool:
return tarfile.is_tarfile(lowerCAmelCase_ )
@staticmethod
def lowercase ( lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[Any] ) -> str:
def resolved(lowerCAmelCase_ : str ) -> str:
return os.path.realpath(os.path.abspath(lowerCAmelCase_ ) )
def badpath(lowerCAmelCase_ : str , lowerCAmelCase_ : str ) -> bool:
# joinpath will ignore base if path is absolute
return not resolved(os.path.join(lowerCAmelCase_ , lowerCAmelCase_ ) ).startswith(lowerCAmelCase_ )
def badlink(lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : str ) -> bool:
# Links are interpreted relative to the directory containing the link
__lowerCAmelCase = resolved(os.path.join(lowerCAmelCase_ , os.path.dirname(info.name ) ) )
return badpath(info.linkname , base=lowerCAmelCase_ )
__lowerCAmelCase = resolved(lowerCAmelCase_ )
for finfo in members:
if badpath(finfo.name , lowerCAmelCase_ ):
logger.error(f"""Extraction of {finfo.name} is blocked (illegal path)""" )
elif finfo.issym() and badlink(lowerCAmelCase_ , lowerCAmelCase_ ):
logger.error(f"""Extraction of {finfo.name} is blocked: Symlink to {finfo.linkname}""" )
elif finfo.islnk() and badlink(lowerCAmelCase_ , lowerCAmelCase_ ):
logger.error(f"""Extraction of {finfo.name} is blocked: Hard link to {finfo.linkname}""" )
else:
yield finfo
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = tarfile.open(lowerCAmelCase_ )
tar_file.extractall(lowerCAmelCase_ , members=TarExtractor.safemembers(lowerCAmelCase_ , lowerCAmelCase_ ) )
tar_file.close()
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = [b"""\x1F\x8B"""]
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
with gzip.open(lowerCAmelCase_ , 'rb' ) as gzip_file:
with open(lowerCAmelCase_ , 'wb' ) as extracted_file:
shutil.copyfileobj(lowerCAmelCase_ , lowerCAmelCase_ )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = [
b"""PK\x03\x04""",
b"""PK\x05\x06""", # empty archive
b"""PK\x07\x08""", # spanned archive
]
@classmethod
def lowercase ( cls : List[str] , lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : bytes = b"" ) -> bool:
if super().is_extractable(lowerCAmelCase_ , magic_number=lowerCAmelCase_ ):
return True
try:
# Alternative version of zipfile.is_zipfile that has less false positives, but misses executable zip archives.
# From: https://github.com/python/cpython/pull/5053
from zipfile import (
_CD_SIGNATURE,
_ECD_DISK_NUMBER,
_ECD_DISK_START,
_ECD_ENTRIES_TOTAL,
_ECD_OFFSET,
_ECD_SIZE,
_EndRecData,
sizeCentralDir,
stringCentralDir,
structCentralDir,
)
with open(lowerCAmelCase_ , 'rb' ) as fp:
__lowerCAmelCase = _EndRecData(lowerCAmelCase_ )
if endrec:
if endrec[_ECD_ENTRIES_TOTAL] == 0 and endrec[_ECD_SIZE] == 0 and endrec[_ECD_OFFSET] == 0:
return True # Empty zipfiles are still zipfiles
elif endrec[_ECD_DISK_NUMBER] == endrec[_ECD_DISK_START]:
fp.seek(endrec[_ECD_OFFSET] ) # Central directory is on the same disk
if fp.tell() == endrec[_ECD_OFFSET] and endrec[_ECD_SIZE] >= sizeCentralDir:
__lowerCAmelCase = fp.read(lowerCAmelCase_ ) # CD is where we expect it to be
if len(lowerCAmelCase_ ) == sizeCentralDir:
__lowerCAmelCase = struct.unpack(lowerCAmelCase_ , lowerCAmelCase_ ) # CD is the right size
if centdir[_CD_SIGNATURE] == stringCentralDir:
return True # First central directory entry has correct magic number
return False
except Exception: # catch all errors in case future python versions change the zipfile internals
return False
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
with zipfile.ZipFile(lowerCAmelCase_ , 'r' ) as zip_file:
zip_file.extractall(lowerCAmelCase_ )
zip_file.close()
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = [b"""\xFD\x37\x7A\x58\x5A\x00"""]
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
with lzma.open(lowerCAmelCase_ ) as compressed_file:
with open(lowerCAmelCase_ , 'wb' ) as extracted_file:
shutil.copyfileobj(lowerCAmelCase_ , lowerCAmelCase_ )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = [b"""Rar!\x1a\x07\x00""", b"""Rar!\x1a\x07\x01\x00"""] # RAR_ID # RAR5_ID
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
if not config.RARFILE_AVAILABLE:
raise ImportError('Please pip install rarfile' )
import rarfile
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = rarfile.RarFile(lowerCAmelCase_ )
rf.extractall(lowerCAmelCase_ )
rf.close()
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = [b"""\x28\xb5\x2F\xFD"""]
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
if not config.ZSTANDARD_AVAILABLE:
raise ImportError('Please pip install zstandard' )
import zstandard as zstd
__lowerCAmelCase = zstd.ZstdDecompressor()
with open(lowerCAmelCase_ , 'rb' ) as ifh, open(lowerCAmelCase_ , 'wb' ) as ofh:
dctx.copy_stream(lowerCAmelCase_ , lowerCAmelCase_ )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = [b"""\x42\x5A\x68"""]
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
with bza.open(lowerCAmelCase_ , 'rb' ) as compressed_file:
with open(lowerCAmelCase_ , 'wb' ) as extracted_file:
shutil.copyfileobj(lowerCAmelCase_ , lowerCAmelCase_ )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = [b"""\x37\x7A\xBC\xAF\x27\x1C"""]
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
if not config.PY7ZR_AVAILABLE:
raise ImportError('Please pip install py7zr' )
import pyazr
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
with pyazr.SevenZipFile(lowerCAmelCase_ , 'r' ) as archive:
archive.extractall(lowerCAmelCase_ )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = [b"""\x04\x22\x4D\x18"""]
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] ) -> None:
if not config.LZ4_AVAILABLE:
raise ImportError('Please pip install lz4' )
import lza.frame
with lza.frame.open(lowerCAmelCase_ , 'rb' ) as compressed_file:
with open(lowerCAmelCase_ , 'wb' ) as extracted_file:
shutil.copyfileobj(lowerCAmelCase_ , lowerCAmelCase_ )
class _UpperCAmelCase :
"""simple docstring"""
a_ = {
"tar": TarExtractor,
"gzip": GzipExtractor,
"zip": ZipExtractor,
"xz": XzExtractor,
"rar": RarExtractor,
"zstd": ZstdExtractor,
"bz2": BzipaExtractor,
"7z": SevenZipExtractor, # <Added version="2.4.0"/>
"lz4": LzaExtractor, # <Added version="2.4.0"/>
}
@classmethod
def lowercase ( cls : int ) -> Any:
return max(
len(lowerCAmelCase_ )
for extractor in cls.extractors.values()
if issubclass(lowerCAmelCase_ , lowerCAmelCase_ )
for extractor_magic_number in extractor.magic_numbers )
@staticmethod
def lowercase ( lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : int ) -> Optional[Any]:
try:
return MagicNumberBaseExtractor.read_magic_number(lowerCAmelCase_ , magic_number_length=lowerCAmelCase_ )
except OSError:
return b""
@classmethod
def lowercase ( cls : Dict , lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : bool = False ) -> bool:
warnings.warn(
'Method \'is_extractable\' was deprecated in version 2.4.0 and will be removed in 3.0.0. '
'Use \'infer_extractor_format\' instead.' , category=lowerCAmelCase_ , )
__lowerCAmelCase = cls.infer_extractor_format(lowerCAmelCase_ )
if extractor_format:
return True if not return_extractor else (True, cls.extractors[extractor_format])
return False if not return_extractor else (False, None)
@classmethod
def lowercase ( cls : str , lowerCAmelCase_ : Union[Path, str] ) -> str: # <Added version="2.4.0"/>
__lowerCAmelCase = cls._get_magic_number_max_length()
__lowerCAmelCase = cls._read_magic_number(lowerCAmelCase_ , lowerCAmelCase_ )
for extractor_format, extractor in cls.extractors.items():
if extractor.is_extractable(lowerCAmelCase_ , magic_number=lowerCAmelCase_ ):
return extractor_format
@classmethod
def lowercase ( cls : Union[str, Any] , lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Union[Path, str] , lowerCAmelCase_ : Optional[str] = None , lowerCAmelCase_ : Optional[BaseExtractor] = "deprecated" , ) -> None:
os.makedirs(os.path.dirname(lowerCAmelCase_ ) , exist_ok=lowerCAmelCase_ )
# Prevent parallel extractions
__lowerCAmelCase = str(Path(lowerCAmelCase_ ).with_suffix('.lock' ) )
with FileLock(lowerCAmelCase_ ):
shutil.rmtree(lowerCAmelCase_ , ignore_errors=lowerCAmelCase_ )
if extractor_format or extractor != "deprecated":
if extractor != "deprecated" or not isinstance(lowerCAmelCase_ , lowerCAmelCase_ ): # passed as positional arg
warnings.warn(
'Parameter \'extractor\' was deprecated in version 2.4.0 and will be removed in 3.0.0. '
'Use \'extractor_format\' instead.' , category=lowerCAmelCase_ , )
__lowerCAmelCase = extractor if extractor != 'deprecated' else extractor_format
else:
__lowerCAmelCase = cls.extractors[extractor_format]
return extractor.extract(lowerCAmelCase_ , lowerCAmelCase_ )
else:
warnings.warn(
'Parameter \'extractor_format\' was made required in version 2.4.0 and not passing it will raise an '
'exception in 3.0.0.' , category=lowerCAmelCase_ , )
for extractor in cls.extractors.values():
if extractor.is_extractable(lowerCAmelCase_ ):
return extractor.extract(lowerCAmelCase_ , lowerCAmelCase_ )
| 53 |
import unittest
from diffusers.pipelines.pipeline_utils import is_safetensors_compatible
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Union[str, Any] ) -> List[str]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Any:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
# Removed: 'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> int:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
# Removed: 'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> str:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[str]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Union[str, Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
# Removed: 'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : List[str] ) -> List[Any]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : Optional[Any] ) -> Optional[Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
# 'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
| 53 | 1 |
import json
import logging
import os
import sys
from pathlib import Path
import finetune_rag
from transformers.file_utils import is_apex_available
from transformers.testing_utils import (
TestCasePlus,
execute_subprocess_async,
require_ray,
require_torch_gpu,
require_torch_multi_gpu,
)
logging.basicConfig(level=logging.DEBUG)
_snake_case : Tuple = logging.getLogger()
_snake_case : Any = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Any , lowerCAmelCase_ : Dict ) -> Optional[int]:
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = {'source': 'What is love ?', 'target': 'life'}
__lowerCAmelCase = {'train': 1_2, 'val': 2, 'test': 2}
for split in ["train", "test", "val"]:
for field in ["source", "target"]:
__lowerCAmelCase = '\n'.join([contents[field]] * n_lines[split] )
with open(os.path.join(lowerCAmelCase_ , f"""{split}.{field}""" ) , 'w' ) as f:
f.write(lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : int , lowerCAmelCase_ : str = "pytorch" ) -> List[str]:
__lowerCAmelCase = self.get_auto_remove_tmp_dir()
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'output' )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'data' )
self._create_dummy_data(data_dir=lowerCAmelCase_ )
__lowerCAmelCase = f"""
--data_dir {data_dir} \
--output_dir {output_dir} \
--model_name_or_path facebook/rag-sequence-base \
--model_type rag_sequence \
--do_train \
--do_predict \
--n_val -1 \
--val_check_interval 1.0 \
--train_batch_size 2 \
--eval_batch_size 1 \
--max_source_length 25 \
--max_target_length 25 \
--val_max_target_length 25 \
--test_max_target_length 25 \
--label_smoothing 0.1 \
--dropout 0.1 \
--attention_dropout 0.1 \
--weight_decay 0.001 \
--adam_epsilon 1e-08 \
--max_grad_norm 0.1 \
--lr_scheduler polynomial \
--learning_rate 3e-04 \
--num_train_epochs 1 \
--warmup_steps 4 \
--gradient_accumulation_steps 1 \
--distributed-port 8787 \
--use_dummy_dataset 1 \
--distributed_retriever {distributed_retriever} \
""".split()
if gpus > 0:
testargs.append(f"""--gpus={gpus}""" )
if is_apex_available():
testargs.append('--fp16' )
else:
testargs.append('--gpus=0' )
testargs.append('--distributed_backend=ddp_cpu' )
testargs.append('--num_processes=2' )
__lowerCAmelCase = [sys.executable, str(Path(finetune_rag.__file__ ).resolve() )] + testargs
execute_subprocess_async(lowerCAmelCase_ , env=self.get_env() )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'metrics.json' )
with open(lowerCAmelCase_ ) as f:
__lowerCAmelCase = json.load(lowerCAmelCase_ )
return result
@require_torch_gpu
def lowercase ( self : str ) -> int:
__lowerCAmelCase = self._run_finetune(gpus=1 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
def lowercase ( self : List[str] ) -> Dict:
__lowerCAmelCase = self._run_finetune(gpus=2 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_gpu
@require_ray
def lowercase ( self : int ) -> Tuple:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
@require_ray
def lowercase ( self : List[Any] ) -> str:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
| 53 |
import math
def a_ ( lowerCAmelCase_ : list, lowerCAmelCase_ : int ):
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = int(math.floor(math.sqrt(lowerCAmelCase_ ) ) )
__lowerCAmelCase = 0
while arr[min(lowerCAmelCase_, lowerCAmelCase_ ) - 1] < x:
__lowerCAmelCase = step
step += int(math.floor(math.sqrt(lowerCAmelCase_ ) ) )
if prev >= n:
return -1
while arr[prev] < x:
__lowerCAmelCase = prev + 1
if prev == min(lowerCAmelCase_, lowerCAmelCase_ ):
return -1
if arr[prev] == x:
return prev
return -1
if __name__ == "__main__":
_snake_case : List[str] = input('Enter numbers separated by a comma:\n').strip()
_snake_case : Optional[Any] = [int(item) for item in user_input.split(',')]
_snake_case : List[str] = int(input('Enter the number to be searched:\n'))
_snake_case : Optional[int] = jump_search(arr, x)
if res == -1:
print('Number not found!')
else:
print(F"""Number {x} is at index {res}""")
| 53 | 1 |
from collections.abc import Callable
from math import pi, sqrt
from random import uniform
from statistics import mean
def a_ ( lowerCAmelCase_ : int ):
# A local function to see if a dot lands in the circle.
def is_in_circle(lowerCAmelCase_ : float, lowerCAmelCase_ : float ) -> bool:
__lowerCAmelCase = sqrt((x**2) + (y**2) )
# Our circle has a radius of 1, so a distance
# greater than 1 would land outside the circle.
return distance_from_centre <= 1
# The proportion of guesses that landed in the circle
__lowerCAmelCase = mean(
int(is_in_circle(uniform(-1.0, 1.0 ), uniform(-1.0, 1.0 ) ) )
for _ in range(lowerCAmelCase_ ) )
# The ratio of the area for circle to square is pi/4.
__lowerCAmelCase = proportion * 4
print(F"""The estimated value of pi is {pi_estimate}""" )
print(F"""The numpy value of pi is {pi}""" )
print(F"""The total error is {abs(pi - pi_estimate )}""" )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : Callable[[float], float], lowerCAmelCase_ : float = 0.0, lowerCAmelCase_ : float = 1.0, ):
return mean(
function_to_integrate(uniform(lowerCAmelCase_, lowerCAmelCase_ ) ) for _ in range(lowerCAmelCase_ ) ) * (max_value - min_value)
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : float = 0.0, lowerCAmelCase_ : float = 1.0 ):
def identity_function(lowerCAmelCase_ : float ) -> float:
return x
__lowerCAmelCase = area_under_curve_estimator(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = (max_value * max_value - min_value * min_value) / 2
print('******************' )
print(F"""Estimating area under y=x where x varies from {min_value} to {max_value}""" )
print(F"""Estimated value is {estimated_value}""" )
print(F"""Expected value is {expected_value}""" )
print(F"""Total error is {abs(estimated_value - expected_value )}""" )
print('******************' )
def a_ ( lowerCAmelCase_ : int ):
def function_to_integrate(lowerCAmelCase_ : float ) -> float:
return sqrt(4.0 - x * x )
__lowerCAmelCase = area_under_curve_estimator(
lowerCAmelCase_, lowerCAmelCase_, 0.0, 2.0 )
print('******************' )
print('Estimating pi using area_under_curve_estimator' )
print(F"""Estimated value is {estimated_value}""" )
print(F"""Expected value is {pi}""" )
print(F"""Total error is {abs(estimated_value - pi )}""" )
print('******************' )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 |
import argparse
import torch
from transformers import RemBertConfig, RemBertModel, load_tf_weights_in_rembert
from transformers.utils import logging
logging.set_verbosity_info()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[Any], lowerCAmelCase_ : str ):
# Initialise PyTorch model
__lowerCAmelCase = RemBertConfig.from_json_file(lowerCAmelCase_ )
print('Building PyTorch model from configuration: {}'.format(str(lowerCAmelCase_ ) ) )
__lowerCAmelCase = RemBertModel(lowerCAmelCase_ )
# Load weights from tf checkpoint
load_tf_weights_in_rembert(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# Save pytorch-model
print('Save PyTorch model to {}'.format(lowerCAmelCase_ ) )
torch.save(model.state_dict(), lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--tf_checkpoint_path', default=None, type=str, required=True, help='Path to the TensorFlow checkpoint path.'
)
parser.add_argument(
'--rembert_config_file',
default=None,
type=str,
required=True,
help=(
'The config json file corresponding to the pre-trained RemBERT model. \n'
'This specifies the model architecture.'
),
)
parser.add_argument(
'--pytorch_dump_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_rembert_tf_checkpoint_to_pytorch(args.tf_checkpoint_path, args.rembert_config_file, args.pytorch_dump_path)
| 53 | 1 |
import inspect
import unittest
from transformers import RegNetConfig, is_flax_available
from transformers.testing_utils import require_flax, slow
from transformers.utils import cached_property, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor
if is_flax_available():
import jax
import jax.numpy as jnp
from transformers.models.regnet.modeling_flax_regnet import FlaxRegNetForImageClassification, FlaxRegNetModel
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int]=3 , lowerCAmelCase_ : Dict=3_2 , lowerCAmelCase_ : Tuple=3 , lowerCAmelCase_ : Union[str, Any]=1_0 , lowerCAmelCase_ : List[str]=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Optional[int]=[1, 1, 2, 1] , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : Tuple="relu" , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : Optional[int]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = embeddings_size
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = hidden_act
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = len(lowerCAmelCase_ )
def lowercase ( self : Optional[int] ) -> List[Any]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : Tuple ) -> List[Any]:
return RegNetConfig(
num_channels=self.num_channels , embeddings_size=self.embeddings_size , hidden_sizes=self.hidden_sizes , depths=self.depths , hidden_act=self.hidden_act , num_labels=self.num_labels , image_size=self.image_size , )
def lowercase ( self : List[str] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : List[str] ) -> str:
__lowerCAmelCase = FlaxRegNetModel(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
# Output shape (b, c, h, w)
self.parent.assertEqual(
result.last_hidden_state.shape , (self.batch_size, self.hidden_sizes[-1], self.image_size // 3_2, self.image_size // 3_2) , )
def lowercase ( self : str , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : int ) -> Tuple:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = FlaxRegNetForImageClassification(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_flax
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (FlaxRegNetModel, FlaxRegNetForImageClassification) if is_flax_available() else ()
a_ = False
a_ = False
a_ = False
def lowercase ( self : Dict ) -> None:
__lowerCAmelCase = FlaxRegNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : int ) -> Optional[int]:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : str ) -> Union[str, Any]:
return
def lowercase ( self : Dict ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCAmelCase_ )
@unittest.skip(reason='RegNet does not use inputs_embeds' )
def lowercase ( self : Union[str, Any] ) -> Any:
pass
@unittest.skip(reason='RegNet does not support input and output embeddings' )
def lowercase ( self : Tuple ) -> Tuple:
pass
def lowercase ( self : Optional[Any] ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.__call__ )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
def check_hidden_states_output(lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : str ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
with self.subTest(model_class.__name__ ):
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model_class(lowerCAmelCase_ )
@jax.jit
def model_jitted(lowerCAmelCase_ : Optional[int] , **lowerCAmelCase_ : Dict ):
return model(pixel_values=lowerCAmelCase_ , **lowerCAmelCase_ )
with self.subTest('JIT Enabled' ):
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
with self.subTest('JIT Disabled' ):
with jax.disable_jit():
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
self.assertEqual(len(lowerCAmelCase_ ) , len(lowerCAmelCase_ ) )
for jitted_output, output in zip(lowerCAmelCase_ , lowerCAmelCase_ ):
self.assertEqual(jitted_output.shape , output.shape )
def a_ ( ):
__lowerCAmelCase = Image.open('./tests/fixtures/tests_samples/COCO/000000039769.png' )
return image
@require_flax
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
return AutoImageProcessor.from_pretrained('facebook/regnet-y-040' ) if is_vision_available() else None
@slow
def lowercase ( self : Optional[Any] ) -> Union[str, Any]:
__lowerCAmelCase = FlaxRegNetForImageClassification.from_pretrained('facebook/regnet-y-040' )
__lowerCAmelCase = self.default_image_processor
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = image_processor(images=lowerCAmelCase_ , return_tensors='np' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
# verify the logits
__lowerCAmelCase = (1, 1_0_0_0)
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = jnp.array([-0.41_80, -1.50_51, -3.48_36] )
self.assertTrue(jnp.allclose(outputs.logits[0, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 |
import argparse
import json
import pickle
from pathlib import Path
import requests
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import MaskFormerConfig, MaskFormerForInstanceSegmentation, MaskFormerImageProcessor, SwinConfig
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Any = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = SwinConfig.from_pretrained(
'microsoft/swin-tiny-patch4-window7-224', out_features=['stage1', 'stage2', 'stage3', 'stage4'] )
__lowerCAmelCase = MaskFormerConfig(backbone_config=lowerCAmelCase_ )
__lowerCAmelCase = 'huggingface/label-files'
if "ade20k-full" in model_name:
# this should be ok
__lowerCAmelCase = 847
__lowerCAmelCase = 'maskformer-ade20k-full-id2label.json'
elif "ade" in model_name:
# this should be ok
__lowerCAmelCase = 150
__lowerCAmelCase = 'ade20k-id2label.json'
elif "coco-stuff" in model_name:
# this should be ok
__lowerCAmelCase = 171
__lowerCAmelCase = 'maskformer-coco-stuff-id2label.json'
elif "coco" in model_name:
# TODO
__lowerCAmelCase = 133
__lowerCAmelCase = 'coco-panoptic-id2label.json'
elif "cityscapes" in model_name:
# this should be ok
__lowerCAmelCase = 19
__lowerCAmelCase = 'cityscapes-id2label.json'
elif "vistas" in model_name:
# this should be ok
__lowerCAmelCase = 65
__lowerCAmelCase = 'mapillary-vistas-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
return config
def a_ ( lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = []
# stem
# fmt: off
rename_keys.append(('backbone.patch_embed.proj.weight', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.weight') )
rename_keys.append(('backbone.patch_embed.proj.bias', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.bias') )
rename_keys.append(('backbone.patch_embed.norm.weight', 'model.pixel_level_module.encoder.model.embeddings.norm.weight') )
rename_keys.append(('backbone.patch_embed.norm.bias', 'model.pixel_level_module.encoder.model.embeddings.norm.bias') )
# stages
for i in range(len(config.backbone_config.depths ) ):
for j in range(config.backbone_config.depths[i] ):
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_bias_table""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_bias_table""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_index""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_index""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.bias""") )
if i < 3:
rename_keys.append((F"""backbone.layers.{i}.downsample.reduction.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.reduction.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.bias""") )
rename_keys.append((F"""backbone.norm{i}.weight""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.weight""") )
rename_keys.append((F"""backbone.norm{i}.bias""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.bias""") )
# FPN
rename_keys.append(('sem_seg_head.layer_4.weight', 'model.pixel_level_module.decoder.fpn.stem.0.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.weight', 'model.pixel_level_module.decoder.fpn.stem.1.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.bias', 'model.pixel_level_module.decoder.fpn.stem.1.bias') )
for source_index, target_index in zip(range(3, 0, -1 ), range(0, 3 ) ):
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.0.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.bias""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.0.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.bias""") )
rename_keys.append(('sem_seg_head.mask_features.weight', 'model.pixel_level_module.decoder.mask_projection.weight') )
rename_keys.append(('sem_seg_head.mask_features.bias', 'model.pixel_level_module.decoder.mask_projection.bias') )
# Transformer decoder
for idx in range(config.decoder_config.decoder_layers ):
# self-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.bias""") )
# cross-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.bias""") )
# MLP 1
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc1.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc1.bias""") )
# MLP 2
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc2.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc2.bias""") )
# layernorm 1 (self-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.bias""") )
# layernorm 2 (cross-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.bias""") )
# layernorm 3 (final layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.weight""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.bias""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.bias""") )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.weight', 'model.transformer_module.decoder.layernorm.weight') )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.bias', 'model.transformer_module.decoder.layernorm.bias') )
# heads on top
rename_keys.append(('sem_seg_head.predictor.query_embed.weight', 'model.transformer_module.queries_embedder.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.weight', 'model.transformer_module.input_projection.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.bias', 'model.transformer_module.input_projection.bias') )
rename_keys.append(('sem_seg_head.predictor.class_embed.weight', 'class_predictor.weight') )
rename_keys.append(('sem_seg_head.predictor.class_embed.bias', 'class_predictor.bias') )
for i in range(3 ):
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.weight""", F"""mask_embedder.{i}.0.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.bias""", F"""mask_embedder.{i}.0.bias""") )
# fmt: on
return rename_keys
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int ):
__lowerCAmelCase = [int(backbone_config.embed_dim * 2**i ) for i in range(len(backbone_config.depths ) )]
for i in range(len(backbone_config.depths ) ):
__lowerCAmelCase = num_features[i]
for j in range(backbone_config.depths[i] ):
# fmt: off
# read in weights + bias of input projection layer (in original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[:dim, :]
__lowerCAmelCase = in_proj_bias[: dim]
__lowerCAmelCase = in_proj_weight[
dim : dim * 2, :
]
__lowerCAmelCase = in_proj_bias[
dim : dim * 2
]
__lowerCAmelCase = in_proj_weight[
-dim :, :
]
__lowerCAmelCase = in_proj_bias[-dim :]
# fmt: on
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : Dict ):
# fmt: off
__lowerCAmelCase = config.decoder_config.hidden_size
for idx in range(config.decoder_config.decoder_layers ):
# read in weights + bias of self-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# read in weights + bias of cross-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# fmt: on
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : bool = False ):
__lowerCAmelCase = get_maskformer_config(lowerCAmelCase_ )
# load original state_dict
with open(lowerCAmelCase_, 'rb' ) as f:
__lowerCAmelCase = pickle.load(lowerCAmelCase_ )
__lowerCAmelCase = data['model']
# for name, param in state_dict.items():
# print(name, param.shape)
# rename keys
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_swin_q_k_v(lowerCAmelCase_, config.backbone_config )
read_in_decoder_q_k_v(lowerCAmelCase_, lowerCAmelCase_ )
# update to torch tensors
for key, value in state_dict.items():
__lowerCAmelCase = torch.from_numpy(lowerCAmelCase_ )
# load 🤗 model
__lowerCAmelCase = MaskFormerForInstanceSegmentation(lowerCAmelCase_ )
model.eval()
for name, param in model.named_parameters():
print(lowerCAmelCase_, param.shape )
__lowerCAmelCase , __lowerCAmelCase = model.load_state_dict(lowerCAmelCase_, strict=lowerCAmelCase_ )
assert missing_keys == [
"model.pixel_level_module.encoder.model.layernorm.weight",
"model.pixel_level_module.encoder.model.layernorm.bias",
]
assert len(lowerCAmelCase_ ) == 0, F"""Unexpected keys: {unexpected_keys}"""
# verify results
__lowerCAmelCase = prepare_img()
if "vistas" in model_name:
__lowerCAmelCase = 65
elif "cityscapes" in model_name:
__lowerCAmelCase = 6_5535
else:
__lowerCAmelCase = 255
__lowerCAmelCase = True if 'ade' in model_name else False
__lowerCAmelCase = MaskFormerImageProcessor(ignore_index=lowerCAmelCase_, reduce_labels=lowerCAmelCase_ )
__lowerCAmelCase = image_processor(lowerCAmelCase_, return_tensors='pt' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
print('Logits:', outputs.class_queries_logits[0, :3, :3] )
if model_name == "maskformer-swin-tiny-ade":
__lowerCAmelCase = torch.tensor(
[[3.6353, -4.4770, -2.6065], [0.5081, -4.2394, -3.5343], [2.1909, -5.0353, -1.9323]] )
assert torch.allclose(outputs.class_queries_logits[0, :3, :3], lowerCAmelCase_, atol=1E-4 )
print('Looks ok!' )
if pytorch_dump_folder_path is not None:
print(F"""Saving model and image processor to {pytorch_dump_folder_path}""" )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
model.save_pretrained(lowerCAmelCase_ )
image_processor.save_pretrained(lowerCAmelCase_ )
if push_to_hub:
print('Pushing model and image processor to the hub...' )
model.push_to_hub(F"""nielsr/{model_name}""" )
image_processor.push_to_hub(F"""nielsr/{model_name}""" )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='maskformer-swin-tiny-ade',
type=str,
help=('Name of the MaskFormer model you\'d like to convert',),
)
parser.add_argument(
'--checkpoint_path',
default='/Users/nielsrogge/Documents/MaskFormer_checkpoints/MaskFormer-Swin-tiny-ADE20k/model.pkl',
type=str,
help='Path to the original state dict (.pth file).',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--push_to_hub', action='store_true', help='Whether or not to push the converted model to the 🤗 hub.'
)
_snake_case : List[str] = parser.parse_args()
convert_maskformer_checkpoint(
args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub
)
| 53 | 1 |
# Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import os
import platform
import numpy as np
import psutil
import torch
from accelerate import __version__ as version
from accelerate.commands.config import default_config_file, load_config_from_file
from ..utils import is_npu_available, is_xpu_available
def a_ ( lowerCAmelCase_ : str=None ):
if subparsers is not None:
__lowerCAmelCase = subparsers.add_parser('env' )
else:
__lowerCAmelCase = argparse.ArgumentParser('Accelerate env command' )
parser.add_argument(
'--config_file', default=lowerCAmelCase_, help='The config file to use for the default values in the launching script.' )
if subparsers is not None:
parser.set_defaults(func=lowerCAmelCase_ )
return parser
def a_ ( lowerCAmelCase_ : Optional[int] ):
__lowerCAmelCase = torch.__version__
__lowerCAmelCase = torch.cuda.is_available()
__lowerCAmelCase = is_xpu_available()
__lowerCAmelCase = is_npu_available()
__lowerCAmelCase = 'Not found'
# Get the default from the config file.
if args.config_file is not None or os.path.isfile(lowerCAmelCase_ ):
__lowerCAmelCase = load_config_from_file(args.config_file ).to_dict()
__lowerCAmelCase = {
'`Accelerate` version': version,
'Platform': platform.platform(),
'Python version': platform.python_version(),
'Numpy version': np.__version__,
'PyTorch version (GPU?)': F"""{pt_version} ({pt_cuda_available})""",
'PyTorch XPU available': str(lowerCAmelCase_ ),
'PyTorch NPU available': str(lowerCAmelCase_ ),
'System RAM': F"""{psutil.virtual_memory().total / 1024 ** 3:.2f} GB""",
}
if pt_cuda_available:
__lowerCAmelCase = torch.cuda.get_device_name()
print('\nCopy-and-paste the text below in your GitHub issue\n' )
print('\n'.join([F"""- {prop}: {val}""" for prop, val in info.items()] ) )
print('- `Accelerate` default config:' if args.config_file is None else '- `Accelerate` config passed:' )
__lowerCAmelCase = (
'\n'.join([F"""\t- {prop}: {val}""" for prop, val in accelerate_config.items()] )
if isinstance(lowerCAmelCase_, lowerCAmelCase_ )
else F"""\t{accelerate_config}"""
)
print(lowerCAmelCase_ )
__lowerCAmelCase = accelerate_config
return info
def a_ ( ):
__lowerCAmelCase = env_command_parser()
__lowerCAmelCase = parser.parse_args()
env_command(lowerCAmelCase_ )
return 0
if __name__ == "__main__":
raise SystemExit(main())
| 53 |
import json
import logging
import os
import re
import sys
from dataclasses import dataclass, field
from typing import Any, Dict, List, Optional, Union
import datasets
import numpy as np
import torch
import torchaudio
from packaging import version
from torch import nn
import transformers
from transformers import (
HfArgumentParser,
Trainer,
TrainingArguments,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaForCTC,
WavaVecaProcessor,
is_apex_available,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
if is_apex_available():
from apex import amp
if version.parse(version.parse(torch.__version__).base_version) >= version.parse('1.6'):
_snake_case : List[Any] = True
from torch.cuda.amp import autocast
_snake_case : Dict = logging.getLogger(__name__)
def a_ ( lowerCAmelCase_ : str=None, lowerCAmelCase_ : str=None ):
return field(default_factory=lambda: default, metadata=lowerCAmelCase_ )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
metadata={"""help""": """Path to pretrained model or model identifier from huggingface.co/models"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to freeze the feature extractor layers of the model."""} )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout ratio for the attention probabilities."""} )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout ratio for activations inside the fully connected layer."""} )
a_ = field(
default=0.1 , metadata={
"""help""": """The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler."""
} , )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout probabilitiy for all 1D convolutional layers in feature extractor."""} , )
a_ = field(
default=0.05 , metadata={
"""help""": (
"""Propability of each feature vector along the time axis to be chosen as the start of the vector"""
"""span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"""
"""vectors will be masked along the time axis. This is only relevant if ``apply_spec_augment is True``."""
)
} , )
a_ = field(default=0.0 , metadata={"""help""": """The LayerDrop probability."""} )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(
default="""train+validation""" , metadata={
"""help""": """The name of the training data set split to use (via the datasets library). Defaults to 'train'"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached preprocessed datasets or not."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""For debugging purposes or quicker training, truncate the number of training examples to this """
"""value if set."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""For debugging purposes or quicker training, truncate the number of validation examples to this """
"""value if set."""
)
} , )
a_ = list_field(
default=[""",""", """?""", """.""", """!""", """-""", """;""", """:""", """\"\"""", """%""", """'""", """\"""", """�"""] , metadata={"""help""": """A list of characters to remove from the transcripts."""} , )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = 42
a_ = True
a_ = None
a_ = None
a_ = None
a_ = None
def __call__( self : int , lowerCAmelCase_ : List[Dict[str, Union[List[int], torch.Tensor]]] ) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# different padding methods
__lowerCAmelCase = [{'input_values': feature['input_values']} for feature in features]
__lowerCAmelCase = [{'input_ids': feature['labels']} for feature in features]
__lowerCAmelCase = self.processor.pad(
lowerCAmelCase_ , padding=self.padding , max_length=self.max_length , pad_to_multiple_of=self.pad_to_multiple_of , return_tensors='pt' , )
__lowerCAmelCase = self.processor.pad(
labels=lowerCAmelCase_ , padding=self.padding , max_length=self.max_length_labels , pad_to_multiple_of=self.pad_to_multiple_of_labels , return_tensors='pt' , )
# replace padding with -100 to ignore loss correctly
__lowerCAmelCase = labels_batch['input_ids'].masked_fill(labels_batch.attention_mask.ne(1 ) , -1_0_0 )
__lowerCAmelCase = labels
return batch
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Tuple , lowerCAmelCase_ : nn.Module , lowerCAmelCase_ : Dict[str, Union[torch.Tensor, Any]] ) -> torch.Tensor:
model.train()
__lowerCAmelCase = self._prepare_inputs(lowerCAmelCase_ )
if self.use_amp:
with autocast():
__lowerCAmelCase = self.compute_loss(lowerCAmelCase_ , lowerCAmelCase_ )
else:
__lowerCAmelCase = self.compute_loss(lowerCAmelCase_ , lowerCAmelCase_ )
if self.args.n_gpu > 1:
if model.module.config.ctc_loss_reduction == "mean":
__lowerCAmelCase = loss.mean()
elif model.module.config.ctc_loss_reduction == "sum":
__lowerCAmelCase = loss.sum() / (inputs['labels'] >= 0).sum()
else:
raise ValueError(f"""{model.config.ctc_loss_reduction} is not valid. Choose one of ['mean', 'sum']""" )
if self.args.gradient_accumulation_steps > 1:
__lowerCAmelCase = loss / self.args.gradient_accumulation_steps
if self.use_amp:
self.scaler.scale(lowerCAmelCase_ ).backward()
elif self.use_apex:
with amp.scale_loss(lowerCAmelCase_ , self.optimizer ) as scaled_loss:
scaled_loss.backward()
elif self.deepspeed:
self.deepspeed.backward(lowerCAmelCase_ )
else:
loss.backward()
return loss.detach()
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets:
__lowerCAmelCase = datasets.load_dataset(
'common_voice', data_args.dataset_config_name, split=data_args.train_split_name )
__lowerCAmelCase = datasets.load_dataset('common_voice', data_args.dataset_config_name, split='test' )
# Create and save tokenizer
__lowerCAmelCase = F"""[{"".join(data_args.chars_to_ignore )}]"""
def remove_special_characters(lowerCAmelCase_ : Any ):
__lowerCAmelCase = re.sub(lowerCAmelCase_, '', batch['sentence'] ).lower() + ' '
return batch
__lowerCAmelCase = train_dataset.map(lowerCAmelCase_, remove_columns=['sentence'] )
__lowerCAmelCase = eval_dataset.map(lowerCAmelCase_, remove_columns=['sentence'] )
def extract_all_chars(lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = ' '.join(batch['text'] )
__lowerCAmelCase = list(set(lowerCAmelCase_ ) )
return {"vocab": [vocab], "all_text": [all_text]}
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, batched=lowerCAmelCase_, batch_size=-1, keep_in_memory=lowerCAmelCase_, remove_columns=train_dataset.column_names, )
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, batched=lowerCAmelCase_, batch_size=-1, keep_in_memory=lowerCAmelCase_, remove_columns=eval_dataset.column_names, )
__lowerCAmelCase = list(set(vocab_train['vocab'][0] ) | set(vocab_test['vocab'][0] ) )
__lowerCAmelCase = {v: k for k, v in enumerate(lowerCAmelCase_ )}
__lowerCAmelCase = vocab_dict[' ']
del vocab_dict[" "]
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = len(lowerCAmelCase_ )
with open('vocab.json', 'w' ) as vocab_file:
json.dump(lowerCAmelCase_, lowerCAmelCase_ )
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = WavaVecaCTCTokenizer(
'vocab.json', unk_token='[UNK]', pad_token='[PAD]', word_delimiter_token='|', )
__lowerCAmelCase = WavaVecaFeatureExtractor(
feature_size=1, sampling_rate=1_6000, padding_value=0.0, do_normalize=lowerCAmelCase_, return_attention_mask=lowerCAmelCase_ )
__lowerCAmelCase = WavaVecaProcessor(feature_extractor=lowerCAmelCase_, tokenizer=lowerCAmelCase_ )
__lowerCAmelCase = WavaVecaForCTC.from_pretrained(
model_args.model_name_or_path, cache_dir=model_args.cache_dir, activation_dropout=model_args.activation_dropout, attention_dropout=model_args.attention_dropout, hidden_dropout=model_args.hidden_dropout, feat_proj_dropout=model_args.feat_proj_dropout, mask_time_prob=model_args.mask_time_prob, gradient_checkpointing=training_args.gradient_checkpointing, layerdrop=model_args.layerdrop, ctc_loss_reduction='mean', pad_token_id=processor.tokenizer.pad_token_id, vocab_size=len(processor.tokenizer ), )
if data_args.max_train_samples is not None:
__lowerCAmelCase = min(len(lowerCAmelCase_ ), data_args.max_train_samples )
__lowerCAmelCase = train_dataset.select(range(lowerCAmelCase_ ) )
if data_args.max_val_samples is not None:
__lowerCAmelCase = eval_dataset.select(range(data_args.max_val_samples ) )
__lowerCAmelCase = torchaudio.transforms.Resample(4_8000, 1_6000 )
# Preprocessing the datasets.
# We need to read the aduio files as arrays and tokenize the targets.
def speech_file_to_array_fn(lowerCAmelCase_ : int ):
__lowerCAmelCase , __lowerCAmelCase = torchaudio.load(batch['path'] )
__lowerCAmelCase = resampler(lowerCAmelCase_ ).squeeze().numpy()
__lowerCAmelCase = 1_6000
__lowerCAmelCase = batch['text']
return batch
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, remove_columns=train_dataset.column_names, num_proc=data_args.preprocessing_num_workers, )
__lowerCAmelCase = eval_dataset.map(
lowerCAmelCase_, remove_columns=eval_dataset.column_names, num_proc=data_args.preprocessing_num_workers, )
def prepare_dataset(lowerCAmelCase_ : Union[str, Any] ):
# check that all files have the correct sampling rate
assert (
len(set(batch['sampling_rate'] ) ) == 1
), F"""Make sure all inputs have the same sampling rate of {processor.feature_extractor.sampling_rate}."""
__lowerCAmelCase = processor(
audio=batch['speech'], text=batch['target_text'], sampling_rate=batch['sampling_rate'][0] )
batch.update(lowerCAmelCase_ )
return batch
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, remove_columns=train_dataset.column_names, batch_size=training_args.per_device_train_batch_size, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, )
__lowerCAmelCase = eval_dataset.map(
lowerCAmelCase_, remove_columns=eval_dataset.column_names, batch_size=training_args.per_device_train_batch_size, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, )
# Metric
__lowerCAmelCase = datasets.load_metric('wer' )
def compute_metrics(lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = pred.predictions
__lowerCAmelCase = np.argmax(lowerCAmelCase_, axis=-1 )
__lowerCAmelCase = processor.tokenizer.pad_token_id
__lowerCAmelCase = processor.batch_decode(lowerCAmelCase_ )
# we do not want to group tokens when computing the metrics
__lowerCAmelCase = processor.batch_decode(pred.label_ids, group_tokens=lowerCAmelCase_ )
__lowerCAmelCase = wer_metric.compute(predictions=lowerCAmelCase_, references=lowerCAmelCase_ )
return {"wer": wer}
if model_args.freeze_feature_extractor:
model.freeze_feature_extractor()
# Data collator
__lowerCAmelCase = DataCollatorCTCWithPadding(processor=lowerCAmelCase_, padding=lowerCAmelCase_ )
# Initialize our Trainer
__lowerCAmelCase = CTCTrainer(
model=lowerCAmelCase_, data_collator=lowerCAmelCase_, args=lowerCAmelCase_, compute_metrics=lowerCAmelCase_, train_dataset=train_dataset if training_args.do_train else None, eval_dataset=eval_dataset if training_args.do_eval else None, tokenizer=processor.feature_extractor, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
# Save the feature_extractor and the tokenizer
if is_main_process(training_args.local_rank ):
processor.save_pretrained(training_args.output_dir )
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model()
__lowerCAmelCase = train_result.metrics
__lowerCAmelCase = (
data_args.max_train_samples if data_args.max_train_samples is not None else len(lowerCAmelCase_ )
)
__lowerCAmelCase = min(lowerCAmelCase_, len(lowerCAmelCase_ ) )
trainer.log_metrics('train', lowerCAmelCase_ )
trainer.save_metrics('train', lowerCAmelCase_ )
trainer.save_state()
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = data_args.max_val_samples if data_args.max_val_samples is not None else len(lowerCAmelCase_ )
__lowerCAmelCase = min(lowerCAmelCase_, len(lowerCAmelCase_ ) )
trainer.log_metrics('eval', lowerCAmelCase_ )
trainer.save_metrics('eval', lowerCAmelCase_ )
return results
if __name__ == "__main__":
main()
| 53 | 1 |
import os
def a_ ( ):
__lowerCAmelCase = os.path.dirname(os.path.realpath(lowerCAmelCase_ ) )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'triangle.txt' )
with open(lowerCAmelCase_ ) as f:
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = []
for line in triangle:
__lowerCAmelCase = []
for number in line.strip().split(' ' ):
numbers_from_line.append(int(lowerCAmelCase_ ) )
a.append(lowerCAmelCase_ )
for i in range(1, len(lowerCAmelCase_ ) ):
for j in range(len(a[i] ) ):
__lowerCAmelCase = a[i - 1][j] if j != len(a[i - 1] ) else 0
__lowerCAmelCase = a[i - 1][j - 1] if j > 0 else 0
a[i][j] += max(lowerCAmelCase_, lowerCAmelCase_ )
return max(a[-1] )
if __name__ == "__main__":
print(solution())
| 53 |
import json
from typing import List, Optional, Tuple
from tokenizers import normalizers
from ....tokenization_utils_fast import PreTrainedTokenizerFast
from ....utils import logging
from .tokenization_retribert import RetriBertTokenizer
_snake_case : Any = logging.get_logger(__name__)
_snake_case : int = {'vocab_file': 'vocab.txt', 'tokenizer_file': 'tokenizer.json'}
_snake_case : Optional[Any] = {
'vocab_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/vocab.txt'
),
},
'tokenizer_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/tokenizer.json'
),
},
}
_snake_case : str = {
'yjernite/retribert-base-uncased': 512,
}
_snake_case : Optional[int] = {
'yjernite/retribert-base-uncased': {'do_lower_case': True},
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = PRETRAINED_VOCAB_FILES_MAP
a_ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
a_ = PRETRAINED_INIT_CONFIGURATION
a_ = RetriBertTokenizer
a_ = ["""input_ids""", """attention_mask"""]
def __init__( self : Dict , lowerCAmelCase_ : Optional[Any]=None , lowerCAmelCase_ : Tuple=None , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : str="[UNK]" , lowerCAmelCase_ : Optional[Any]="[SEP]" , lowerCAmelCase_ : List[str]="[PAD]" , lowerCAmelCase_ : Optional[int]="[CLS]" , lowerCAmelCase_ : List[Any]="[MASK]" , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : List[str]=None , **lowerCAmelCase_ : List[Any] , ) -> Dict:
super().__init__(
lowerCAmelCase_ , tokenizer_file=lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , unk_token=lowerCAmelCase_ , sep_token=lowerCAmelCase_ , pad_token=lowerCAmelCase_ , cls_token=lowerCAmelCase_ , mask_token=lowerCAmelCase_ , tokenize_chinese_chars=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ , **lowerCAmelCase_ , )
__lowerCAmelCase = json.loads(self.backend_tokenizer.normalizer.__getstate__() )
if (
normalizer_state.get('lowercase' , lowerCAmelCase_ ) != do_lower_case
or normalizer_state.get('strip_accents' , lowerCAmelCase_ ) != strip_accents
or normalizer_state.get('handle_chinese_chars' , lowerCAmelCase_ ) != tokenize_chinese_chars
):
__lowerCAmelCase = getattr(lowerCAmelCase_ , normalizer_state.pop('type' ) )
__lowerCAmelCase = do_lower_case
__lowerCAmelCase = strip_accents
__lowerCAmelCase = tokenize_chinese_chars
__lowerCAmelCase = normalizer_class(**lowerCAmelCase_ )
__lowerCAmelCase = do_lower_case
def lowercase ( self : str , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Optional[int]=None ) -> Optional[int]:
__lowerCAmelCase = [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 lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[int] , lowerCAmelCase_ : Optional[List[int]] = None ) -> List[int]:
__lowerCAmelCase = [self.sep_token_id]
__lowerCAmelCase = [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 lowercase ( self : int , lowerCAmelCase_ : str , lowerCAmelCase_ : Optional[str] = None ) -> Tuple[str]:
__lowerCAmelCase = self._tokenizer.model.save(lowerCAmelCase_ , name=lowerCAmelCase_ )
return tuple(lowerCAmelCase_ )
| 53 | 1 |
# this script reports modified .py files under the desired list of top-level sub-dirs passed as a list of arguments, e.g.:
# python ./utils/get_modified_files.py utils src tests examples
#
# it uses git to find the forking point and which files were modified - i.e. files not under git won't be considered
# since the output of this script is fed into Makefile commands it doesn't print a newline after the results
import re
import subprocess
import sys
_snake_case : Tuple = subprocess.check_output('git merge-base main HEAD'.split()).decode('utf-8')
_snake_case : List[Any] = (
subprocess.check_output(F"""git diff --diff-filter=d --name-only {fork_point_sha}""".split()).decode('utf-8').split()
)
_snake_case : Optional[int] = '|'.join(sys.argv[1:])
_snake_case : str = re.compile(RF"""^({joined_dirs}).*?\.py$""")
_snake_case : List[Any] = [x for x in modified_files if regex.match(x)]
print(' '.join(relevant_modified_files), end='')
| 53 |
import numpy as np
from cva import COLOR_BGR2GRAY, cvtColor, imread
from numpy import array, uinta
from PIL import Image
from digital_image_processing import change_contrast as cc
from digital_image_processing import convert_to_negative as cn
from digital_image_processing import sepia as sp
from digital_image_processing.dithering import burkes as bs
from digital_image_processing.edge_detection import canny
from digital_image_processing.filters import convolve as conv
from digital_image_processing.filters import gaussian_filter as gg
from digital_image_processing.filters import local_binary_pattern as lbp
from digital_image_processing.filters import median_filter as med
from digital_image_processing.filters import sobel_filter as sob
from digital_image_processing.resize import resize as rs
_snake_case : Union[str, Any] = imread(R'digital_image_processing/image_data/lena_small.jpg')
_snake_case : Optional[int] = cvtColor(img, COLOR_BGR2GRAY)
def a_ ( ):
__lowerCAmelCase = cn.convert_to_negative(lowerCAmelCase_ )
# assert negative_img array for at least one True
assert negative_img.any()
def a_ ( ):
with Image.open('digital_image_processing/image_data/lena_small.jpg' ) as img:
# Work around assertion for response
assert str(cc.change_contrast(lowerCAmelCase_, 110 ) ).startswith(
'<PIL.Image.Image image mode=RGB size=100x100 at' )
def a_ ( ):
__lowerCAmelCase = canny.gen_gaussian_kernel(9, sigma=1.4 )
# Assert ambiguous array
assert resp.all()
def a_ ( ):
__lowerCAmelCase = imread('digital_image_processing/image_data/lena_small.jpg', 0 )
# assert ambiguous array for all == True
assert canny_img.all()
__lowerCAmelCase = canny.canny(lowerCAmelCase_ )
# assert canny array for at least one True
assert canny_array.any()
def a_ ( ):
assert gg.gaussian_filter(lowerCAmelCase_, 5, sigma=0.9 ).all()
def a_ ( ):
# laplace diagonals
__lowerCAmelCase = array([[0.25, 0.5, 0.25], [0.5, -3, 0.5], [0.25, 0.5, 0.25]] )
__lowerCAmelCase = conv.img_convolve(lowerCAmelCase_, lowerCAmelCase_ ).astype(lowerCAmelCase_ )
assert res.any()
def a_ ( ):
assert med.median_filter(lowerCAmelCase_, 3 ).any()
def a_ ( ):
__lowerCAmelCase , __lowerCAmelCase = sob.sobel_filter(lowerCAmelCase_ )
assert grad.any() and theta.any()
def a_ ( ):
__lowerCAmelCase = sp.make_sepia(lowerCAmelCase_, 20 )
assert sepia.all()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg" ):
__lowerCAmelCase = bs.Burkes(imread(lowerCAmelCase_, 1 ), 120 )
burkes.process()
assert burkes.output_img.any()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg", ):
__lowerCAmelCase = rs.NearestNeighbour(imread(lowerCAmelCase_, 1 ), 400, 200 )
nn.process()
assert nn.output.any()
def a_ ( ):
__lowerCAmelCase = 'digital_image_processing/image_data/lena.jpg'
# Reading the image and converting it to grayscale.
__lowerCAmelCase = imread(lowerCAmelCase_, 0 )
# Test for get_neighbors_pixel function() return not None
__lowerCAmelCase = 0
__lowerCAmelCase = 0
__lowerCAmelCase = image[x_coordinate][y_coordinate]
__lowerCAmelCase = lbp.get_neighbors_pixel(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert neighbors_pixels is not None
# Test for local_binary_pattern function()
# Create a numpy array as the same height and width of read image
__lowerCAmelCase = np.zeros((image.shape[0], image.shape[1]) )
# Iterating through the image and calculating the local binary pattern value
# for each pixel.
for i in range(0, image.shape[0] ):
for j in range(0, image.shape[1] ):
__lowerCAmelCase = lbp.local_binary_value(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert lbp_image.any()
| 53 | 1 |
from typing import Dict
import numpy as np
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, GenericTensor, Pipeline, PipelineException
if is_tf_available():
import tensorflow as tf
from ..tf_utils import stable_softmax
if is_torch_available():
import torch
_snake_case : List[str] = logging.get_logger(__name__)
@add_end_docstrings(
_UpperCamelCase , R"""
top_k (`int`, defaults to 5):
The number of predictions to return.
targets (`str` or `List[str]`, *optional*):
When passed, the model will limit the scores to the passed targets instead of looking up in the whole
vocab. If the provided targets are not in the model vocab, they will be tokenized and the first resulting
token will be used (with a warning, and that might be slower).
""" , )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : List[str] , lowerCAmelCase_ : GenericTensor ) -> np.ndarray:
if self.framework == "tf":
__lowerCAmelCase = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()
elif self.framework == "pt":
__lowerCAmelCase = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=lowerCAmelCase_ )
else:
raise ValueError('Unsupported framework' )
return masked_index
def lowercase ( self : Tuple , lowerCAmelCase_ : GenericTensor ) -> np.ndarray:
__lowerCAmelCase = self.get_masked_index(lowerCAmelCase_ )
__lowerCAmelCase = np.prod(masked_index.shape )
if numel < 1:
raise PipelineException(
'fill-mask' , self.model.base_model_prefix , f"""No mask_token ({self.tokenizer.mask_token}) found on the input""" , )
def lowercase ( self : Optional[int] , lowerCAmelCase_ : GenericTensor ) -> List[Any]:
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
for model_input in model_inputs:
self._ensure_exactly_one_mask_token(model_input['input_ids'][0] )
else:
for input_ids in model_inputs["input_ids"]:
self._ensure_exactly_one_mask_token(lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Union[str, Any]=None , **lowerCAmelCase_ : List[Any] ) -> Dict[str, GenericTensor]:
if return_tensors is None:
__lowerCAmelCase = self.framework
__lowerCAmelCase = self.tokenizer(lowerCAmelCase_ , return_tensors=lowerCAmelCase_ )
self.ensure_exactly_one_mask_token(lowerCAmelCase_ )
return model_inputs
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : str ) -> List[Any]:
__lowerCAmelCase = self.model(**lowerCAmelCase_ )
__lowerCAmelCase = model_inputs['input_ids']
return model_outputs
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Union[str, Any]=5 , lowerCAmelCase_ : int=None ) -> List[Any]:
# Cap top_k if there are targets
if target_ids is not None and target_ids.shape[0] < top_k:
__lowerCAmelCase = target_ids.shape[0]
__lowerCAmelCase = model_outputs['input_ids'][0]
__lowerCAmelCase = model_outputs['logits']
if self.framework == "tf":
__lowerCAmelCase = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()[:, 0]
__lowerCAmelCase = outputs.numpy()
__lowerCAmelCase = outputs[0, masked_index, :]
__lowerCAmelCase = stable_softmax(lowerCAmelCase_ , axis=-1 )
if target_ids is not None:
__lowerCAmelCase = tf.gather_nd(tf.squeeze(lowerCAmelCase_ , 0 ) , target_ids.reshape(-1 , 1 ) )
__lowerCAmelCase = tf.expand_dims(lowerCAmelCase_ , 0 )
__lowerCAmelCase = tf.math.top_k(lowerCAmelCase_ , k=lowerCAmelCase_ )
__lowerCAmelCase , __lowerCAmelCase = topk.values.numpy(), topk.indices.numpy()
else:
__lowerCAmelCase = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=lowerCAmelCase_ ).squeeze(-1 )
# Fill mask pipeline supports only one ${mask_token} per sample
__lowerCAmelCase = outputs[0, masked_index, :]
__lowerCAmelCase = logits.softmax(dim=-1 )
if target_ids is not None:
__lowerCAmelCase = probs[..., target_ids]
__lowerCAmelCase , __lowerCAmelCase = probs.topk(lowerCAmelCase_ )
__lowerCAmelCase = []
__lowerCAmelCase = values.shape[0] == 1
for i, (_values, _predictions) in enumerate(zip(values.tolist() , predictions.tolist() ) ):
__lowerCAmelCase = []
for v, p in zip(_values , _predictions ):
# Copy is important since we're going to modify this array in place
__lowerCAmelCase = input_ids.numpy().copy()
if target_ids is not None:
__lowerCAmelCase = target_ids[p].tolist()
__lowerCAmelCase = p
# Filter padding out:
__lowerCAmelCase = tokens[np.where(tokens != self.tokenizer.pad_token_id )]
# Originally we skip special tokens to give readable output.
# For multi masks though, the other [MASK] would be removed otherwise
# making the output look odd, so we add them back
__lowerCAmelCase = self.tokenizer.decode(lowerCAmelCase_ , skip_special_tokens=lowerCAmelCase_ )
__lowerCAmelCase = {'score': v, 'token': p, 'token_str': self.tokenizer.decode([p] ), 'sequence': sequence}
row.append(lowerCAmelCase_ )
result.append(lowerCAmelCase_ )
if single_mask:
return result[0]
return result
def lowercase ( self : List[str] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[Any]=None ) -> List[str]:
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
__lowerCAmelCase = [targets]
try:
__lowerCAmelCase = self.tokenizer.get_vocab()
except Exception:
__lowerCAmelCase = {}
__lowerCAmelCase = []
for target in targets:
__lowerCAmelCase = vocab.get(lowerCAmelCase_ , lowerCAmelCase_ )
if id_ is None:
__lowerCAmelCase = self.tokenizer(
lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ , return_attention_mask=lowerCAmelCase_ , return_token_type_ids=lowerCAmelCase_ , max_length=1 , truncation=lowerCAmelCase_ , )['input_ids']
if len(lowerCAmelCase_ ) == 0:
logger.warning(
f"""The specified target token `{target}` does not exist in the model vocabulary. """
'We cannot replace it with anything meaningful, ignoring it' )
continue
__lowerCAmelCase = input_ids[0]
# XXX: If users encounter this pass
# it becomes pretty slow, so let's make sure
# The warning enables them to fix the input to
# get faster performance.
logger.warning(
f"""The specified target token `{target}` does not exist in the model vocabulary. """
f"""Replacing with `{self.tokenizer.convert_ids_to_tokens(id_ )}`.""" )
target_ids.append(id_ )
__lowerCAmelCase = list(set(lowerCAmelCase_ ) )
if len(lowerCAmelCase_ ) == 0:
raise ValueError('At least one target must be provided when passed.' )
__lowerCAmelCase = np.array(lowerCAmelCase_ )
return target_ids
def lowercase ( self : int , lowerCAmelCase_ : Any=None , lowerCAmelCase_ : Any=None ) -> Tuple:
__lowerCAmelCase = {}
if targets is not None:
__lowerCAmelCase = self.get_target_ids(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = target_ids
if top_k is not None:
__lowerCAmelCase = top_k
if self.tokenizer.mask_token_id is None:
raise PipelineException(
'fill-mask' , self.model.base_model_prefix , 'The tokenizer does not define a `mask_token`.' )
return {}, {}, postprocess_params
def __call__( self : List[Any] , lowerCAmelCase_ : Optional[Any] , *lowerCAmelCase_ : Dict , **lowerCAmelCase_ : Union[str, Any] ) -> Optional[int]:
__lowerCAmelCase = super().__call__(lowerCAmelCase_ , **lowerCAmelCase_ )
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ) and len(lowerCAmelCase_ ) == 1:
return outputs[0]
return outputs
| 53 |
from typing import Dict, Iterable, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_DEFAULT_MEAN,
IMAGENET_DEFAULT_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, logging
_snake_case : List[Any] = logging.get_logger(__name__)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = ["""pixel_values"""]
def __init__( self : Optional[int] , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Union[int, float] = 1 / 2_5_5 , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_MEAN , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_STD , **lowerCAmelCase_ : Any , ) -> None:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = size if size is not None else {'shortest_edge': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else {'height': 2_2_4, 'width': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = do_resize
__lowerCAmelCase = size
__lowerCAmelCase = resample
__lowerCAmelCase = do_center_crop
__lowerCAmelCase = crop_size
__lowerCAmelCase = do_rescale
__lowerCAmelCase = rescale_factor
__lowerCAmelCase = do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else IMAGENET_DEFAULT_MEAN
__lowerCAmelCase = image_std if image_std is not None else IMAGENET_DEFAULT_STD
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : Optional[int] , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
# size_dict is a dict with either keys "height" and "width" or "shortest_edge"
if "shortest_edge" in size:
__lowerCAmelCase = int((2_5_6 / 2_2_4) * size['shortest_edge'] )
__lowerCAmelCase = get_resize_output_image_size(lowerCAmelCase_ , size=lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = {'height': output_size[0], 'width': output_size[1]}
if "height" not in size_dict or "width" not in size_dict:
raise ValueError(
f"""Size dict must have keys 'height' and 'width' or 'shortest_edge'. Got {size_dict.keys()}""" )
return resize(
lowerCAmelCase_ , size=(size_dict['height'], size_dict['width']) , resample=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : str , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : str , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ )
if "height" not in size or "width" not in size:
raise ValueError(f"""Size dict must have keys 'height' and 'width'. Got {size.keys()}""" )
return center_crop(lowerCAmelCase_ , size=(size['height'], size['width']) , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[int, float] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : int , ) -> np.ndarray:
return rescale(lowerCAmelCase_ , scale=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : int , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : List[str] , ) -> np.ndarray:
return normalize(lowerCAmelCase_ , mean=lowerCAmelCase_ , std=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : ImageInput , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : PILImageResampling = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[float] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[TensorType] = None , lowerCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **lowerCAmelCase_ : str , ) -> BatchFeature:
__lowerCAmelCase = do_resize if do_resize is not None else self.do_resize
__lowerCAmelCase = resample if resample is not None else self.resample
__lowerCAmelCase = do_center_crop if do_center_crop is not None else self.do_center_crop
__lowerCAmelCase = do_rescale if do_rescale is not None else self.do_rescale
__lowerCAmelCase = rescale_factor if rescale_factor is not None else self.rescale_factor
__lowerCAmelCase = do_normalize if do_normalize is not None else self.do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else self.image_mean
__lowerCAmelCase = image_std if image_std is not None else self.image_std
__lowerCAmelCase = size if size is not None else self.size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else self.crop_size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = make_list_of_images(lowerCAmelCase_ )
if not valid_images(lowerCAmelCase_ ):
raise ValueError(
'Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '
'torch.Tensor, tf.Tensor or jax.ndarray.' )
if do_resize and size is None:
raise ValueError('Size must be specified if do_resize is True.' )
if do_center_crop and crop_size is None:
raise ValueError('Crop size must be specified if do_center_crop is True.' )
if do_rescale and rescale_factor is None:
raise ValueError('Rescale factor must be specified if do_rescale is True.' )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError('Image mean and std must be specified if do_normalize is True.' )
# All transformations expect numpy arrays.
__lowerCAmelCase = [to_numpy_array(lowerCAmelCase_ ) for image in images]
if do_resize:
__lowerCAmelCase = [self.resize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_center_crop:
__lowerCAmelCase = [self.center_crop(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_rescale:
__lowerCAmelCase = [self.rescale(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_normalize:
__lowerCAmelCase = [self.normalize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = [to_channel_dimension_format(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = {'pixel_values': images}
return BatchFeature(data=lowerCAmelCase_ , tensor_type=lowerCAmelCase_ )
| 53 | 1 |
from __future__ import annotations
import requests
_snake_case : Any = set(
'approved_at_utc approved_by author_flair_background_color\nauthor_flair_css_class author_flair_richtext author_flair_template_id author_fullname\nauthor_premium can_mod_post category clicked content_categories created_utc downs\nedited gilded gildings hidden hide_score is_created_from_ads_ui is_meta\nis_original_content is_reddit_media_domain is_video link_flair_css_class\nlink_flair_richtext link_flair_text link_flair_text_color media_embed mod_reason_title\nname permalink pwls quarantine saved score secure_media secure_media_embed selftext\nsubreddit subreddit_name_prefixed subreddit_type thumbnail title top_awarded_type\ntotal_awards_received ups upvote_ratio url user_reports'.split()
)
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : int = 1, lowerCAmelCase_ : str = "new", lowerCAmelCase_ : list | None = None ):
__lowerCAmelCase = wanted_data or []
if invalid_search_terms := ", ".join(sorted(set(lowerCAmelCase_ ) - valid_terms ) ):
__lowerCAmelCase = F"""Invalid search term: {invalid_search_terms}"""
raise ValueError(lowerCAmelCase_ )
__lowerCAmelCase = requests.get(
F"""https://reddit.com/r/{subreddit}/{age}.json?limit={limit}""", headers={'User-agent': 'A random string'}, )
if response.status_code == 429:
raise requests.HTTPError
__lowerCAmelCase = response.json()
if not wanted_data:
return {id_: data["data"]["children"][id_] for id_ in range(lowerCAmelCase_ )}
__lowerCAmelCase = {}
for id_ in range(lowerCAmelCase_ ):
__lowerCAmelCase = {
item: data['data']['children'][id_]['data'][item] for item in wanted_data
}
return data_dict
if __name__ == "__main__":
# If you get Error 429, that means you are rate limited.Try after some time
print(get_subreddit_data('learnpython', wanted_data=['title', 'url', 'selftext']))
| 53 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Dict=2 , lowerCAmelCase_ : Optional[int]=8 , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[Any]=9_9 , lowerCAmelCase_ : List[Any]=1_6 , lowerCAmelCase_ : int=5 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : str=3_6 , lowerCAmelCase_ : Optional[int]="gelu" , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : Optional[int]=0.0 , lowerCAmelCase_ : str=5_1_2 , lowerCAmelCase_ : List[str]=1_6 , lowerCAmelCase_ : str=2 , lowerCAmelCase_ : Tuple=0.02 , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : List[Any]=4 , lowerCAmelCase_ : List[str]=None , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_input_mask
__lowerCAmelCase = use_token_type_ids
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = type_vocab_size
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = num_labels
__lowerCAmelCase = num_choices
__lowerCAmelCase = scope
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = None
if self.use_input_mask:
__lowerCAmelCase = random_attention_mask([self.batch_size, self.seq_length] )
__lowerCAmelCase = None
if self.use_token_type_ids:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
__lowerCAmelCase = ids_tensor([self.batch_size] , self.num_choices )
__lowerCAmelCase = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowercase ( self : Any ) -> Union[str, Any]:
return MraConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , )
def lowercase ( self : Dict ) -> List[Any]:
__lowerCAmelCase = self.get_config()
__lowerCAmelCase = 3_0_0
return config
def lowercase ( self : Optional[int] ) -> Union[str, Any]:
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = self.prepare_config_and_inputs()
__lowerCAmelCase = True
__lowerCAmelCase = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
encoder_hidden_states,
encoder_attention_mask,
)
def lowercase ( self : Optional[int] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , ) -> Tuple:
__lowerCAmelCase = True
__lowerCAmelCase = MraModel(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , )
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , )
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Dict ) -> str:
__lowerCAmelCase = MraForQuestionAnswering(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , start_positions=lowerCAmelCase_ , end_positions=lowerCAmelCase_ , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowercase ( self : int , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : int , lowerCAmelCase_ : Dict ) -> Any:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForTokenClassification(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowercase ( self : int , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] ) -> List[Any]:
__lowerCAmelCase = self.num_choices
__lowerCAmelCase = MraForMultipleChoice(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowercase ( self : Tuple ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'input_ids': input_ids, 'token_type_ids': token_type_ids, 'attention_mask': input_mask}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
a_ = False
a_ = False
a_ = False
a_ = False
a_ = ()
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = MraModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : Tuple ) -> List[str]:
self.config_tester.run_common_tests()
def lowercase ( self : Optional[int] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
__lowerCAmelCase = type
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Any ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*lowerCAmelCase_ )
def lowercase ( self : List[str] ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*lowerCAmelCase_ )
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = MraModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
@unittest.skip(reason='MRA does not output attentions' )
def lowercase ( self : Optional[int] ) -> Tuple:
return
@require_torch
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@slow
def lowercase ( self : Optional[Any] ) -> List[str]:
__lowerCAmelCase = MraModel.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = torch.Size((1, 2_5_6, 7_6_8) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[-0.01_40, 0.08_30, -0.03_81], [0.15_46, 0.14_02, 0.02_20], [0.11_62, 0.08_51, 0.01_65]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : int ) -> Optional[int]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 2_5_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[9.25_95, -3.60_38, 11.88_19], [9.38_69, -3.26_93, 11.09_56], [11.85_24, -3.49_38, 13.12_10]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : Any ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-4096-8-d3' )
__lowerCAmelCase = torch.arange(4_0_9_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 4_0_9_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[5.47_89, -2.35_64, 7.50_64], [7.90_67, -1.33_69, 9.96_68], [9.07_12, -1.81_06, 7.03_80]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import shutil
import tempfile
import unittest
from transformers import ClapFeatureExtractor, ClapProcessor, RobertaTokenizer, RobertaTokenizerFast
from transformers.testing_utils import require_sentencepiece, require_torchaudio
from .test_feature_extraction_clap import floats_list
@require_torchaudio
@require_sentencepiece
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Union[str, Any] ) -> Any:
__lowerCAmelCase = 'laion/clap-htsat-unfused'
__lowerCAmelCase = tempfile.mkdtemp()
def lowercase ( self : Optional[int] , **lowerCAmelCase_ : Union[str, Any] ) -> Optional[int]:
return RobertaTokenizer.from_pretrained(self.checkpoint , **lowerCAmelCase_ )
def lowercase ( self : str , **lowerCAmelCase_ : Optional[int] ) -> List[str]:
return ClapFeatureExtractor.from_pretrained(self.checkpoint , **lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
shutil.rmtree(self.tmpdirname )
def lowercase ( self : Union[str, Any] ) -> Dict:
__lowerCAmelCase = self.get_tokenizer()
__lowerCAmelCase = self.get_feature_extractor()
__lowerCAmelCase = ClapProcessor(tokenizer=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ )
processor.save_pretrained(self.tmpdirname )
__lowerCAmelCase = ClapProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() )
self.assertIsInstance(processor.tokenizer , lowerCAmelCase_ )
self.assertEqual(processor.feature_extractor.to_json_string() , feature_extractor.to_json_string() )
self.assertIsInstance(processor.feature_extractor , lowerCAmelCase_ )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = ClapProcessor(tokenizer=self.get_tokenizer() , feature_extractor=self.get_feature_extractor() )
processor.save_pretrained(self.tmpdirname )
__lowerCAmelCase = self.get_tokenizer(bos_token='(BOS)' , eos_token='(EOS)' )
__lowerCAmelCase = self.get_feature_extractor(do_normalize=lowerCAmelCase_ , padding_value=1.0 )
__lowerCAmelCase = ClapProcessor.from_pretrained(
self.tmpdirname , bos_token='(BOS)' , eos_token='(EOS)' , do_normalize=lowerCAmelCase_ , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , lowerCAmelCase_ )
self.assertEqual(processor.feature_extractor.to_json_string() , feature_extractor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.feature_extractor , lowerCAmelCase_ )
def lowercase ( self : Any ) -> Dict:
__lowerCAmelCase = self.get_feature_extractor()
__lowerCAmelCase = self.get_tokenizer()
__lowerCAmelCase = ClapProcessor(tokenizer=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ )
__lowerCAmelCase = floats_list((3, 1_0_0_0) )
__lowerCAmelCase = feature_extractor(lowerCAmelCase_ , return_tensors='np' )
__lowerCAmelCase = processor(audios=lowerCAmelCase_ , 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 lowercase ( self : Dict ) -> Optional[int]:
__lowerCAmelCase = self.get_feature_extractor()
__lowerCAmelCase = self.get_tokenizer()
__lowerCAmelCase = ClapProcessor(tokenizer=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ )
__lowerCAmelCase = 'This is a test string'
__lowerCAmelCase = processor(text=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer(lowerCAmelCase_ )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def lowercase ( self : Union[str, Any] ) -> str:
__lowerCAmelCase = self.get_feature_extractor()
__lowerCAmelCase = self.get_tokenizer()
__lowerCAmelCase = ClapProcessor(tokenizer=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ )
__lowerCAmelCase = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
__lowerCAmelCase = processor.batch_decode(lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.batch_decode(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : int ) -> Dict:
__lowerCAmelCase = self.get_feature_extractor()
__lowerCAmelCase = self.get_tokenizer()
__lowerCAmelCase = ClapProcessor(tokenizer=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ )
self.assertListEqual(
processor.model_input_names[2:] , feature_extractor.model_input_names , msg='`processor` and `feature_extractor` model input names do not match' , )
| 53 |
import argparse
import json
import os
import re
import shutil
import torch
from transformers import BioGptConfig, BioGptForCausalLM
from transformers.models.biogpt.tokenization_biogpt import VOCAB_FILES_NAMES
from transformers.tokenization_utils_base import TOKENIZER_CONFIG_FILE
from transformers.utils import WEIGHTS_NAME, logging
logging.set_verbosity_warning()
_snake_case : Union[str, Any] = 2
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , *, # begin keyword-only arguments
lowerCAmelCase_ : str="<s>" , lowerCAmelCase_ : Dict="<pad>" , lowerCAmelCase_ : Any="</s>" , lowerCAmelCase_ : List[str]="<unk>" , lowerCAmelCase_ : Optional[Any]=None , ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = bos, unk, pad, eos
__lowerCAmelCase = []
__lowerCAmelCase = []
__lowerCAmelCase = {}
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
if extra_special_symbols:
for s in extra_special_symbols:
self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = len(self.symbols )
def __eq__( self : Dict , lowerCAmelCase_ : Dict ) -> str:
return self.indices == other.indices
def __getitem__( self : List[Any] , lowerCAmelCase_ : int ) -> Union[str, Any]:
if idx < len(self.symbols ):
return self.symbols[idx]
return self.unk_word
def __len__( self : Tuple ) -> List[Any]:
return len(self.symbols )
def __contains__( self : Optional[Any] , lowerCAmelCase_ : Dict ) -> Optional[int]:
return sym in self.indices
@classmethod
def lowercase ( cls : Dict , lowerCAmelCase_ : str ) -> str:
__lowerCAmelCase = cls()
d.add_from_file(lowerCAmelCase_ )
return d
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int=1 , lowerCAmelCase_ : Any=False ) -> Optional[Any]:
if word in self.indices and not overwrite:
__lowerCAmelCase = self.indices[word]
__lowerCAmelCase = self.count[idx] + n
return idx
else:
__lowerCAmelCase = len(self.symbols )
__lowerCAmelCase = idx
self.symbols.append(lowerCAmelCase_ )
self.count.append(lowerCAmelCase_ )
return idx
def lowercase ( self : str , lowerCAmelCase_ : Union[str, Any] ) -> Dict:
return 0
def lowercase ( self : Tuple , lowerCAmelCase_ : Union[str, Any] ) -> int:
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
try:
with open(lowerCAmelCase_ , 'r' , encoding='utf-8' ) as fd:
self.add_from_file(lowerCAmelCase_ )
except FileNotFoundError as fnfe:
raise fnfe
except UnicodeError:
raise Exception('Incorrect encoding detected in {}, please rebuild the dataset'.format(lowerCAmelCase_ ) )
return
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = self._load_meta(lowerCAmelCase_ )
for line in lines[indices_start_line:]:
try:
__lowerCAmelCase , __lowerCAmelCase = line.rstrip().rsplit(' ' , 1 )
if field == "#fairseq:overwrite":
__lowerCAmelCase = True
__lowerCAmelCase , __lowerCAmelCase = line.rsplit(' ' , 1 )
else:
__lowerCAmelCase = False
__lowerCAmelCase = int(lowerCAmelCase_ )
__lowerCAmelCase = line
if word in self and not overwrite:
raise RuntimeError(
'Duplicate word found when loading Dictionary: \'{}\'. '
'Duplicate words can overwrite earlier ones by adding the '
'#fairseq:overwrite flag at the end of the corresponding row '
'in the dictionary file. If using the Camembert model, please '
'download an updated copy of the model file.'.format(lowerCAmelCase_ ) )
self.add_symbol(lowerCAmelCase_ , n=lowerCAmelCase_ , overwrite=lowerCAmelCase_ )
except ValueError:
raise ValueError('Incorrect dictionary format, expected \'<token> <cnt> [flags]\'' )
def a_ ( lowerCAmelCase_ : List[str] ):
# (1) remove word breaking symbol, (2) add word ending symbol where the word is not broken up,
# e.g.: d = {'le@@': 5, 'tt@@': 6, 'er': 7} => {'le': 5, 'tt': 6, 'er</w>': 7}
__lowerCAmelCase = dict((re.sub(R'@@$', '', lowerCAmelCase_ ), v) if k.endswith('@@' ) else (re.sub(R'$', '</w>', lowerCAmelCase_ ), v) for k, v in d.items() )
__lowerCAmelCase = '<s> <pad> </s> <unk>'.split()
# restore the special tokens
for k in keep_keys:
del da[F"""{k}</w>"""]
__lowerCAmelCase = d[k] # restore
return da
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str] ):
# prep
if not os.path.exists(lowerCAmelCase_ ):
raise ValueError(F"""path {biogpt_checkpoint_path} does not exist!""" )
os.makedirs(lowerCAmelCase_, exist_ok=lowerCAmelCase_ )
print(F"""Writing results to {pytorch_dump_folder_path}""" )
# handle various types of models
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'checkpoint.pt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {checkpoint_file} does not exist!""" )
__lowerCAmelCase = torch.load(lowerCAmelCase_, map_location='cpu' )
__lowerCAmelCase = chkpt['cfg']['model']
# dicts
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'dict.txt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {dict_file} does not exist!""" )
__lowerCAmelCase = Dictionary.load(lowerCAmelCase_ )
__lowerCAmelCase = rewrite_dict_keys(src_dict.indices )
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['vocab_file'] )
print(F"""Generating {src_vocab_file} of {src_vocab_size} records""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# merges_file (bpecodes)
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'bpecodes' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {bpecodes_file} does not exist!""" )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['merges_file'] )
shutil.copyfile(lowerCAmelCase_, lowerCAmelCase_ )
# model config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'config.json' )
__lowerCAmelCase = {
'activation_dropout': args['activation_dropout'],
'architectures': ['BioGptForCausalLM'],
'attention_probs_dropout_prob': args['attention_dropout'],
'bos_token_id': 0,
'eos_token_id': 2,
'hidden_act': args['activation_fn'],
'hidden_dropout_prob': args['dropout'],
'hidden_size': args['decoder_embed_dim'],
'initializer_range': 0.02,
'intermediate_size': args['decoder_ffn_embed_dim'],
'layer_norm_eps': 1E-12,
'layerdrop': args['decoder_layerdrop'],
'max_position_embeddings': args['max_target_positions'],
'model_type': 'biogpt',
'num_attention_heads': args['decoder_attention_heads'],
'num_hidden_layers': args['decoder_layers'],
'pad_token_id': 1,
'scale_embedding': not args['no_scale_embedding'],
'tie_word_embeddings': args['share_decoder_input_output_embed'],
'vocab_size': src_vocab_size,
}
# good hparam defaults to start with
print(F"""Generating {biogpt_model_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# tokenizer config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = {
'bos_token': '<s>',
'eos_token': '</s>',
'model_max_length': 1024,
'pad_token': '<pad>',
'special_tokens_map_file': None,
'tokenizer_class': 'BioGptTokenizer',
'unk_token': '<unk>',
}
print(F"""Generating {biogpt_tokenizer_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# model
__lowerCAmelCase = chkpt['model']
# remove unneeded keys
__lowerCAmelCase = [
'decoder.version',
]
for k in ignore_keys:
model_state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = list(model_state_dict.keys() )
for layer_name in layer_names:
if layer_name.endswith('output_projection.weight' ):
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
else:
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
__lowerCAmelCase = BioGptConfig.from_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = BioGptForCausalLM(lowerCAmelCase_ )
# check that it loads ok
model_new.load_state_dict(lowerCAmelCase_ )
# save
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
print(F"""Generating {pytorch_weights_dump_path}""" )
torch.save(lowerCAmelCase_, lowerCAmelCase_ )
print('Conversion is done!' )
if __name__ == "__main__":
_snake_case : str = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--biogpt_checkpoint_path',
default=None,
type=str,
required=True,
help=(
'Path to the official PyTorch checkpoint file which is expected to reside in the dump dir with dicts,'
' bpecodes, etc.'
),
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_biogpt_checkpoint_to_pytorch(args.biogpt_checkpoint_path, args.pytorch_dump_folder_path)
| 53 | 1 |
import unittest
from knapsack import greedy_knapsack as kp
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Union[str, Any] ) -> List[str]:
__lowerCAmelCase = [1_0, 2_0, 3_0, 4_0, 5_0, 6_0]
__lowerCAmelCase = [2, 4, 6, 8, 1_0, 1_2]
__lowerCAmelCase = 1_0_0
self.assertEqual(kp.calc_profit(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) , 2_1_0 )
def lowercase ( self : Tuple ) -> Dict:
self.assertRaisesRegex(lowerCAmelCase_ , 'max_weight must greater than zero.' )
def lowercase ( self : str ) -> Tuple:
self.assertRaisesRegex(lowerCAmelCase_ , 'Weight can not be negative.' )
def lowercase ( self : int ) -> Optional[Any]:
self.assertRaisesRegex(lowerCAmelCase_ , 'Profit can not be negative.' )
def lowercase ( self : int ) -> Tuple:
self.assertRaisesRegex(lowerCAmelCase_ , 'max_weight must greater than zero.' )
def lowercase ( self : Optional[Any] ) -> int:
self.assertRaisesRegex(
lowerCAmelCase_ , 'The length of profit and weight must be same.' )
if __name__ == "__main__":
unittest.main()
| 53 |
from typing import Tuple, Union
from ...modeling_outputs import BackboneOutput
from ...modeling_utils import PreTrainedModel
from ...utils import is_timm_available, is_torch_available, requires_backends
from ...utils.backbone_utils import BackboneMixin
from .configuration_timm_backbone import TimmBackboneConfig
if is_timm_available():
import timm
if is_torch_available():
from torch import Tensor
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = """pixel_values"""
a_ = False
a_ = TimmBackboneConfig
def __init__( self : Tuple , lowerCAmelCase_ : Any , **lowerCAmelCase_ : Optional[int] ) -> Optional[Any]:
requires_backends(self , 'timm' )
super().__init__(lowerCAmelCase_ )
__lowerCAmelCase = config
if config.backbone is None:
raise ValueError('backbone is not set in the config. Please set it to a timm model name.' )
if config.backbone not in timm.list_models():
raise ValueError(f"""backbone {config.backbone} is not supported by timm.""" )
if hasattr(lowerCAmelCase_ , 'out_features' ) and config.out_features is not None:
raise ValueError('out_features is not supported by TimmBackbone. Please use out_indices instead.' )
__lowerCAmelCase = getattr(lowerCAmelCase_ , 'use_pretrained_backbone' , lowerCAmelCase_ )
if pretrained is None:
raise ValueError('use_pretrained_backbone is not set in the config. Please set it to True or False.' )
# We just take the final layer by default. This matches the default for the transformers models.
__lowerCAmelCase = config.out_indices if getattr(lowerCAmelCase_ , 'out_indices' , lowerCAmelCase_ ) is not None else (-1,)
__lowerCAmelCase = timm.create_model(
config.backbone , pretrained=lowerCAmelCase_ , features_only=config.features_only , in_chans=config.num_channels , out_indices=lowerCAmelCase_ , **lowerCAmelCase_ , )
# These are used to control the output of the model when called. If output_hidden_states is True, then
# return_layers is modified to include all layers.
__lowerCAmelCase = self._backbone.return_layers
__lowerCAmelCase = {layer['module']: str(lowerCAmelCase_ ) for i, layer in enumerate(self._backbone.feature_info.info )}
super()._init_backbone(lowerCAmelCase_ )
@classmethod
def lowercase ( cls : int , lowerCAmelCase_ : Dict , *lowerCAmelCase_ : Dict , **lowerCAmelCase_ : Union[str, Any] ) -> Optional[int]:
requires_backends(cls , ['vision', 'timm'] )
from ...models.timm_backbone import TimmBackboneConfig
__lowerCAmelCase = kwargs.pop('config' , TimmBackboneConfig() )
__lowerCAmelCase = kwargs.pop('use_timm_backbone' , lowerCAmelCase_ )
if not use_timm:
raise ValueError('use_timm_backbone must be True for timm backbones' )
__lowerCAmelCase = kwargs.pop('num_channels' , config.num_channels )
__lowerCAmelCase = kwargs.pop('features_only' , config.features_only )
__lowerCAmelCase = kwargs.pop('use_pretrained_backbone' , config.use_pretrained_backbone )
__lowerCAmelCase = kwargs.pop('out_indices' , config.out_indices )
__lowerCAmelCase = TimmBackboneConfig(
backbone=lowerCAmelCase_ , num_channels=lowerCAmelCase_ , features_only=lowerCAmelCase_ , use_pretrained_backbone=lowerCAmelCase_ , out_indices=lowerCAmelCase_ , )
return super()._from_config(lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Tuple , lowerCAmelCase_ : int ) -> Dict:
pass
def lowercase ( self : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : Any=None , **lowerCAmelCase_ : Dict ) -> Union[BackboneOutput, Tuple[Tensor, ...]]:
__lowerCAmelCase = return_dict if return_dict is not None else self.config.use_return_dict
__lowerCAmelCase = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__lowerCAmelCase = output_attentions if output_attentions is not None else self.config.output_attentions
if output_attentions:
raise ValueError('Cannot output attentions for timm backbones at the moment' )
if output_hidden_states:
# We modify the return layers to include all the stages of the backbone
__lowerCAmelCase = self._all_layers
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = self._return_layers
__lowerCAmelCase = tuple(hidden_states[i] for i in self.out_indices )
else:
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = tuple(lowerCAmelCase_ )
__lowerCAmelCase = tuple(lowerCAmelCase_ ) if hidden_states is not None else None
if not return_dict:
__lowerCAmelCase = (feature_maps,)
if output_hidden_states:
__lowerCAmelCase = output + (hidden_states,)
return output
return BackboneOutput(feature_maps=lowerCAmelCase_ , hidden_states=lowerCAmelCase_ , attentions=lowerCAmelCase_ )
| 53 | 1 |
import qiskit
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : int ):
__lowerCAmelCase = qiskit.Aer.get_backend('aer_simulator' )
# Create a Quantum Circuit acting on the q register
__lowerCAmelCase = qiskit.QuantumCircuit(lowerCAmelCase_, lowerCAmelCase_ )
# Map the quantum measurement to the classical bits
circuit.measure([0], [0] )
# Execute the circuit on the simulator
__lowerCAmelCase = qiskit.execute(lowerCAmelCase_, lowerCAmelCase_, shots=1000 )
# Return the histogram data of the results of the experiment.
return job.result().get_counts(lowerCAmelCase_ )
if __name__ == "__main__":
print(F"""Total count for various states are: {single_qubit_measure(1, 1)}""")
| 53 |
from __future__ import annotations
def a_ ( lowerCAmelCase_ : list[float] ):
if len(lowerCAmelCase_ ) < 2:
raise ValueError('Monogons and Digons are not polygons in the Euclidean space' )
if any(i <= 0 for i in nums ):
raise ValueError('All values must be greater than 0' )
__lowerCAmelCase = nums.copy()
copy_nums.sort()
return copy_nums[-1] < sum(copy_nums[:-1] )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 | 1 |
from __future__ import annotations
import math
from collections.abc import Callable
def a_ ( lowerCAmelCase_ : Callable[[int | float], int | float], lowerCAmelCase_ : int | float, lowerCAmelCase_ : int | float, lowerCAmelCase_ : int = 100, ):
__lowerCAmelCase = x_start
__lowerCAmelCase = fnc(lowerCAmelCase_ )
__lowerCAmelCase = 0.0
for _ in range(lowerCAmelCase_ ):
# Approximates curve as a sequence of linear lines and sums their length
__lowerCAmelCase = (x_end - x_start) / steps + xa
__lowerCAmelCase = fnc(lowerCAmelCase_ )
length += math.hypot(xa - xa, fxa - fxa )
# Increment step
__lowerCAmelCase = xa
__lowerCAmelCase = fxa
return length
if __name__ == "__main__":
def a_ ( lowerCAmelCase_ : Optional[Any] ):
return math.sin(10 * x )
print('f(x) = sin(10 * x)')
print('The length of the curve from x = -10 to x = 10 is:')
_snake_case : Union[str, Any] = 10
while i <= 100000:
print(F"""With {i} steps: {line_length(f, -10, 10, i)}""")
i *= 10
| 53 |
import inspect
import unittest
from transformers import RegNetConfig, is_flax_available
from transformers.testing_utils import require_flax, slow
from transformers.utils import cached_property, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor
if is_flax_available():
import jax
import jax.numpy as jnp
from transformers.models.regnet.modeling_flax_regnet import FlaxRegNetForImageClassification, FlaxRegNetModel
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int]=3 , lowerCAmelCase_ : Dict=3_2 , lowerCAmelCase_ : Tuple=3 , lowerCAmelCase_ : Union[str, Any]=1_0 , lowerCAmelCase_ : List[str]=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Optional[int]=[1, 1, 2, 1] , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : Tuple="relu" , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : Optional[int]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = embeddings_size
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = hidden_act
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = len(lowerCAmelCase_ )
def lowercase ( self : Optional[int] ) -> List[Any]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : Tuple ) -> List[Any]:
return RegNetConfig(
num_channels=self.num_channels , embeddings_size=self.embeddings_size , hidden_sizes=self.hidden_sizes , depths=self.depths , hidden_act=self.hidden_act , num_labels=self.num_labels , image_size=self.image_size , )
def lowercase ( self : List[str] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : List[str] ) -> str:
__lowerCAmelCase = FlaxRegNetModel(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
# Output shape (b, c, h, w)
self.parent.assertEqual(
result.last_hidden_state.shape , (self.batch_size, self.hidden_sizes[-1], self.image_size // 3_2, self.image_size // 3_2) , )
def lowercase ( self : str , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : int ) -> Tuple:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = FlaxRegNetForImageClassification(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_flax
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (FlaxRegNetModel, FlaxRegNetForImageClassification) if is_flax_available() else ()
a_ = False
a_ = False
a_ = False
def lowercase ( self : Dict ) -> None:
__lowerCAmelCase = FlaxRegNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : int ) -> Optional[int]:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : str ) -> Union[str, Any]:
return
def lowercase ( self : Dict ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCAmelCase_ )
@unittest.skip(reason='RegNet does not use inputs_embeds' )
def lowercase ( self : Union[str, Any] ) -> Any:
pass
@unittest.skip(reason='RegNet does not support input and output embeddings' )
def lowercase ( self : Tuple ) -> Tuple:
pass
def lowercase ( self : Optional[Any] ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.__call__ )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
def check_hidden_states_output(lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : str ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
with self.subTest(model_class.__name__ ):
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model_class(lowerCAmelCase_ )
@jax.jit
def model_jitted(lowerCAmelCase_ : Optional[int] , **lowerCAmelCase_ : Dict ):
return model(pixel_values=lowerCAmelCase_ , **lowerCAmelCase_ )
with self.subTest('JIT Enabled' ):
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
with self.subTest('JIT Disabled' ):
with jax.disable_jit():
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
self.assertEqual(len(lowerCAmelCase_ ) , len(lowerCAmelCase_ ) )
for jitted_output, output in zip(lowerCAmelCase_ , lowerCAmelCase_ ):
self.assertEqual(jitted_output.shape , output.shape )
def a_ ( ):
__lowerCAmelCase = Image.open('./tests/fixtures/tests_samples/COCO/000000039769.png' )
return image
@require_flax
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
return AutoImageProcessor.from_pretrained('facebook/regnet-y-040' ) if is_vision_available() else None
@slow
def lowercase ( self : Optional[Any] ) -> Union[str, Any]:
__lowerCAmelCase = FlaxRegNetForImageClassification.from_pretrained('facebook/regnet-y-040' )
__lowerCAmelCase = self.default_image_processor
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = image_processor(images=lowerCAmelCase_ , return_tensors='np' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
# verify the logits
__lowerCAmelCase = (1, 1_0_0_0)
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = jnp.array([-0.41_80, -1.50_51, -3.48_36] )
self.assertTrue(jnp.allclose(outputs.logits[0, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import os
import time
from dataclasses import dataclass, field
from enum import Enum
from typing import Dict, List, Optional, Union
import torch
from filelock import FileLock
from torch.utils.data import Dataset
from ...models.auto.modeling_auto import MODEL_FOR_QUESTION_ANSWERING_MAPPING
from ...tokenization_utils import PreTrainedTokenizer
from ...utils import logging
from ..processors.squad import SquadFeatures, SquadVaProcessor, SquadVaProcessor, squad_convert_examples_to_features
_snake_case : str = logging.get_logger(__name__)
_snake_case : Tuple = list(MODEL_FOR_QUESTION_ANSWERING_MAPPING.keys())
_snake_case : Union[str, Any] = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Model type selected in the list: """ + """, """.join(_UpperCamelCase )} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The input data dir. Should contain the .json files for the SQuAD task."""} )
a_ = field(
default=1_28 , metadata={
"""help""": (
"""The maximum total input sequence length after tokenization. Sequences longer """
"""than this will be truncated, sequences shorter will be padded."""
)
} , )
a_ = field(
default=1_28 , metadata={"""help""": """When splitting up a long document into chunks, how much stride to take between chunks."""} , )
a_ = field(
default=64 , metadata={
"""help""": (
"""The maximum number of tokens for the question. Questions longer than this will """
"""be truncated to this length."""
)
} , )
a_ = field(
default=30 , metadata={
"""help""": (
"""The maximum length of an answer that can be generated. This is needed because the start """
"""and end predictions are not conditioned on one another."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached training and evaluation sets"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """If true, the SQuAD examples contain some that do not have an answer."""} )
a_ = field(
default=0.0 , metadata={"""help""": """If null_score - best_non_null is greater than the threshold predict null."""} )
a_ = field(
default=20 , metadata={"""help""": """If null_score - best_non_null is greater than the threshold predict null."""} )
a_ = field(
default=0 , metadata={
"""help""": (
"""language id of input for language-specific xlm models (see"""
""" tokenization_xlm.PRETRAINED_INIT_CONFIGURATION)"""
)
} , )
a_ = field(default=1 , metadata={"""help""": """multiple threads for converting example to features"""} )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = """train"""
a_ = """dev"""
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = 42
a_ = 42
a_ = 42
a_ = 42
def __init__( self : int , lowerCAmelCase_ : SquadDataTrainingArguments , lowerCAmelCase_ : PreTrainedTokenizer , lowerCAmelCase_ : Optional[int] = None , lowerCAmelCase_ : Union[str, Split] = Split.train , lowerCAmelCase_ : Optional[bool] = False , lowerCAmelCase_ : Optional[str] = None , lowerCAmelCase_ : Optional[str] = "pt" , ) -> Tuple:
__lowerCAmelCase = args
__lowerCAmelCase = is_language_sensitive
__lowerCAmelCase = SquadVaProcessor() if args.version_2_with_negative else SquadVaProcessor()
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
try:
__lowerCAmelCase = Split[mode]
except KeyError:
raise KeyError('mode is not a valid split name' )
__lowerCAmelCase = mode
# Load data features from cache or dataset file
__lowerCAmelCase = 'v2' if args.version_2_with_negative else 'v1'
__lowerCAmelCase = os.path.join(
cache_dir if cache_dir is not None else args.data_dir , f"""cached_{mode.value}_{tokenizer.__class__.__name__}_{args.max_seq_length}_{version_tag}""" , )
# Make sure only the first process in distributed training processes the dataset,
# and the others will use the cache.
__lowerCAmelCase = cached_features_file + '.lock'
with FileLock(lowerCAmelCase_ ):
if os.path.exists(lowerCAmelCase_ ) and not args.overwrite_cache:
__lowerCAmelCase = time.time()
__lowerCAmelCase = torch.load(lowerCAmelCase_ )
# Legacy cache files have only features, while new cache files
# will have dataset and examples also.
__lowerCAmelCase = self.old_features['features']
__lowerCAmelCase = self.old_features.get('dataset' , lowerCAmelCase_ )
__lowerCAmelCase = self.old_features.get('examples' , lowerCAmelCase_ )
logger.info(
f"""Loading features from cached file {cached_features_file} [took %.3f s]""" , time.time() - start )
if self.dataset is None or self.examples is None:
logger.warning(
f"""Deleting cached file {cached_features_file} will allow dataset and examples to be cached in"""
' future run' )
else:
if mode == Split.dev:
__lowerCAmelCase = self.processor.get_dev_examples(args.data_dir )
else:
__lowerCAmelCase = self.processor.get_train_examples(args.data_dir )
__lowerCAmelCase , __lowerCAmelCase = squad_convert_examples_to_features(
examples=self.examples , tokenizer=lowerCAmelCase_ , max_seq_length=args.max_seq_length , doc_stride=args.doc_stride , max_query_length=args.max_query_length , is_training=mode == Split.train , threads=args.threads , return_dataset=lowerCAmelCase_ , )
__lowerCAmelCase = time.time()
torch.save(
{'features': self.features, 'dataset': self.dataset, 'examples': self.examples} , lowerCAmelCase_ , )
# ^ This seems to take a lot of time so I want to investigate why and how we can improve.
logger.info(
f"""Saving features into cached file {cached_features_file} [took {time.time() - start:.3f} s]""" )
def __len__( self : Tuple ) -> List[str]:
return len(self.features )
def __getitem__( self : Dict , lowerCAmelCase_ : Optional[int] ) -> Dict[str, torch.Tensor]:
# Convert to Tensors and build dataset
__lowerCAmelCase = self.features[i]
__lowerCAmelCase = torch.tensor(feature.input_ids , dtype=torch.long )
__lowerCAmelCase = torch.tensor(feature.attention_mask , dtype=torch.long )
__lowerCAmelCase = torch.tensor(feature.token_type_ids , dtype=torch.long )
__lowerCAmelCase = torch.tensor(feature.cls_index , dtype=torch.long )
__lowerCAmelCase = torch.tensor(feature.p_mask , dtype=torch.float )
__lowerCAmelCase = torch.tensor(feature.is_impossible , dtype=torch.float )
__lowerCAmelCase = {
'input_ids': input_ids,
'attention_mask': attention_mask,
'token_type_ids': token_type_ids,
}
if self.args.model_type in ["xlm", "roberta", "distilbert", "camembert"]:
del inputs["token_type_ids"]
if self.args.model_type in ["xlnet", "xlm"]:
inputs.update({'cls_index': cls_index, 'p_mask': p_mask} )
if self.args.version_2_with_negative:
inputs.update({'is_impossible': is_impossible} )
if self.is_language_sensitive:
inputs.update({'langs': (torch.ones(input_ids.shape , dtype=torch.intaa ) * self.args.lang_id)} )
if self.mode == Split.train:
__lowerCAmelCase = torch.tensor(feature.start_position , dtype=torch.long )
__lowerCAmelCase = torch.tensor(feature.end_position , dtype=torch.long )
inputs.update({'start_positions': start_positions, 'end_positions': end_positions} )
return inputs
| 53 |
import json
import logging
import math
import os
import sys
from dataclasses import dataclass, field
from typing import Optional
from datasets import Dataset, load_dataset
import transformers
from transformers import (
CONFIG_MAPPING,
MODEL_FOR_MASKED_LM_MAPPING,
AutoConfig,
AutoModelForMaskedLM,
AutoTokenizer,
DataCollatorForWholeWordMask,
HfArgumentParser,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
_snake_case : Optional[int] = logging.getLogger(__name__)
_snake_case : Dict = list(MODEL_FOR_MASKED_LM_MAPPING.keys())
_snake_case : List[Any] = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """If training from scratch, pass a model type from the list: """ + """, """.join(_UpperCamelCase )} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Override some existing default config settings when a model is trained from scratch. Example: """
"""n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained config name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained tokenizer name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."""} , )
a_ = field(
default="""main""" , metadata={"""help""": """The specific model version to use (can be a branch name, tag name or commit id)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Will use the token generated when running `huggingface-cli login` (necessary to use this script """
"""with private models)."""
)
} , )
def lowercase ( self : List[Any] ) -> List[Any]:
if self.config_overrides is not None and (self.config_name is not None or self.model_name_or_path is not None):
raise ValueError(
'--config_overrides can\'t be used in combination with --config_name or --model_name_or_path' )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The name of the dataset to use (via the datasets library)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(default=_UpperCamelCase , metadata={"""help""": """The input training data file (a text file)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input evaluation data file to evaluate the perplexity on (a text file)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input train ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input validation ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached training and evaluation sets"""} )
a_ = field(
default=5 , metadata={
"""help""": """The percentage of the train set used as validation set in case there's no validation split"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The maximum total input sequence length after tokenization. Sequences longer """
"""than this will be truncated. Default to the max input length of the model."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=0.15 , metadata={"""help""": """Ratio of tokens to mask for masked language modeling loss"""} )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Whether to pad all samples to `max_seq_length`. """
"""If False, will pad the samples dynamically when batching to the maximum length in the batch."""
)
} , )
def lowercase ( self : int ) -> int:
if self.train_file is not None:
__lowerCAmelCase = self.train_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`train_file` should be a csv, a json or a txt file."
if self.validation_file is not None:
__lowerCAmelCase = self.validation_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`validation_file` should be a csv, a json or a txt file."
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : Union[str, Any] ):
with open(lowerCAmelCase_, 'r', encoding='utf-8' ) as f:
__lowerCAmelCase = [json.loads(lowerCAmelCase_ ) for line in f.read().splitlines() if (len(lowerCAmelCase_ ) > 0 and not line.isspace())]
assert len(lowerCAmelCase_ ) == len(lowerCAmelCase_ )
__lowerCAmelCase = {c: dataset[c] for c in dataset.column_names}
__lowerCAmelCase = refs
return Dataset.from_dict(lowerCAmelCase_ )
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
__lowerCAmelCase = load_dataset(data_args.dataset_name, data_args.dataset_config_name )
if "validation" not in datasets.keys():
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[:{data_args.validation_split_percentage}%]""", )
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[{data_args.validation_split_percentage}%:]""", )
else:
__lowerCAmelCase = {}
if data_args.train_file is not None:
__lowerCAmelCase = data_args.train_file
if data_args.validation_file is not None:
__lowerCAmelCase = data_args.validation_file
__lowerCAmelCase = data_args.train_file.split('.' )[-1]
if extension == "txt":
__lowerCAmelCase = 'text'
__lowerCAmelCase = load_dataset(lowerCAmelCase_, data_files=lowerCAmelCase_ )
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.config_name:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.config_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
__lowerCAmelCase = CONFIG_MAPPING[model_args.model_type]()
logger.warning('You are instantiating a new config instance from scratch.' )
if model_args.config_overrides is not None:
logger.info(F"""Overriding config: {model_args.config_overrides}""" )
config.update_from_string(model_args.config_overrides )
logger.info(F"""New config: {config}""" )
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'use_fast': model_args.use_fast_tokenizer,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.tokenizer_name:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.tokenizer_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
raise ValueError(
'You are instantiating a new tokenizer from scratch. This is not supported by this script.'
'You can do it from another script, save it, and load it from here, using --tokenizer_name.' )
if model_args.model_name_or_path:
__lowerCAmelCase = AutoModelForMaskedLM.from_pretrained(
model_args.model_name_or_path, from_tf=bool('.ckpt' in model_args.model_name_or_path ), config=lowerCAmelCase_, cache_dir=model_args.cache_dir, revision=model_args.model_revision, use_auth_token=True if model_args.use_auth_token else None, )
else:
logger.info('Training new model from scratch' )
__lowerCAmelCase = AutoModelForMaskedLM.from_config(lowerCAmelCase_ )
model.resize_token_embeddings(len(lowerCAmelCase_ ) )
# Preprocessing the datasets.
# First we tokenize all the texts.
if training_args.do_train:
__lowerCAmelCase = datasets['train'].column_names
else:
__lowerCAmelCase = datasets['validation'].column_names
__lowerCAmelCase = 'text' if 'text' in column_names else column_names[0]
__lowerCAmelCase = 'max_length' if data_args.pad_to_max_length else False
def tokenize_function(lowerCAmelCase_ : str ):
# Remove empty lines
__lowerCAmelCase = [line for line in examples['text'] if len(lowerCAmelCase_ ) > 0 and not line.isspace()]
return tokenizer(examples['text'], padding=lowerCAmelCase_, truncation=lowerCAmelCase_, max_length=data_args.max_seq_length )
__lowerCAmelCase = datasets.map(
lowerCAmelCase_, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, remove_columns=[text_column_name], load_from_cache_file=not data_args.overwrite_cache, )
# Add the chinese references if provided
if data_args.train_ref_file is not None:
__lowerCAmelCase = add_chinese_references(tokenized_datasets['train'], data_args.train_ref_file )
if data_args.validation_ref_file is not None:
__lowerCAmelCase = add_chinese_references(
tokenized_datasets['validation'], data_args.validation_ref_file )
# If we have ref files, need to avoid it removed by trainer
__lowerCAmelCase = data_args.train_ref_file or data_args.validation_ref_file
if has_ref:
__lowerCAmelCase = False
# Data collator
# This one will take care of randomly masking the tokens.
__lowerCAmelCase = DataCollatorForWholeWordMask(tokenizer=lowerCAmelCase_, mlm_probability=data_args.mlm_probability )
# Initialize our Trainer
__lowerCAmelCase = Trainer(
model=lowerCAmelCase_, args=lowerCAmelCase_, train_dataset=tokenized_datasets['train'] if training_args.do_train else None, eval_dataset=tokenized_datasets['validation'] if training_args.do_eval else None, tokenizer=lowerCAmelCase_, data_collator=lowerCAmelCase_, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif model_args.model_name_or_path is not None and os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model() # Saves the tokenizer too for easy upload
__lowerCAmelCase = os.path.join(training_args.output_dir, 'train_results.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Train results *****' )
for key, value in sorted(train_result.metrics.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
# Need to save the state, since Trainer.save_model saves only the tokenizer with the model
trainer.state.save_to_json(os.path.join(training_args.output_dir, 'trainer_state.json' ) )
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = math.exp(eval_output['eval_loss'] )
__lowerCAmelCase = perplexity
__lowerCAmelCase = os.path.join(training_args.output_dir, 'eval_results_mlm_wwm.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Eval results *****' )
for key, value in sorted(results.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
return results
def a_ ( lowerCAmelCase_ : Tuple ):
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()
| 53 | 1 |
import warnings
from typing import Dict
import numpy as np
from ..utils import ExplicitEnum, add_end_docstrings, is_tf_available, is_torch_available
from .base import PIPELINE_INIT_ARGS, GenericTensor, Pipeline
if is_tf_available():
from ..models.auto.modeling_tf_auto import TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
if is_torch_available():
from ..models.auto.modeling_auto import MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
def a_ ( lowerCAmelCase_ : int ):
return 1.0 / (1.0 + np.exp(-_outputs ))
def a_ ( lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = np.max(_outputs, axis=-1, keepdims=lowerCAmelCase_ )
__lowerCAmelCase = np.exp(_outputs - maxes )
return shifted_exp / shifted_exp.sum(axis=-1, keepdims=lowerCAmelCase_ )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = """sigmoid"""
a_ = """softmax"""
a_ = """none"""
@add_end_docstrings(
_UpperCamelCase , R"""
return_all_scores (`bool`, *optional*, defaults to `False`):
Whether to return all prediction scores or just the one of the predicted class.
function_to_apply (`str`, *optional*, defaults to `\"default\"`):
The function to apply to the model outputs in order to retrieve the scores. Accepts four different values:
- `\"default\"`: if the model has a single label, will apply the sigmoid function on the output. If the model
has several labels, will apply the softmax function on the output.
- `\"sigmoid\"`: Applies the sigmoid function on the output.
- `\"softmax\"`: Applies the softmax function on the output.
- `\"none\"`: Does not apply any function on the output.
""" , )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = False
a_ = ClassificationFunction.NONE
def __init__( self : str , **lowerCAmelCase_ : Any ) -> Any:
super().__init__(**lowerCAmelCase_ )
self.check_model_type(
TF_MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING
if self.framework == 'tf'
else MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING )
def lowercase ( self : Any , lowerCAmelCase_ : str=None , lowerCAmelCase_ : Union[str, Any]=None , lowerCAmelCase_ : Any="" , **lowerCAmelCase_ : List[Any] ) -> Union[str, Any]:
# Using "" as default argument because we're going to use `top_k=None` in user code to declare
# "No top_k"
__lowerCAmelCase = tokenizer_kwargs
__lowerCAmelCase = {}
if hasattr(self.model.config , 'return_all_scores' ) and return_all_scores is None:
__lowerCAmelCase = self.model.config.return_all_scores
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ) or top_k is None:
__lowerCAmelCase = top_k
__lowerCAmelCase = False
elif return_all_scores is not None:
warnings.warn(
'`return_all_scores` is now deprecated, if want a similar functionality use `top_k=None` instead of'
' `return_all_scores=True` or `top_k=1` instead of `return_all_scores=False`.' , lowerCAmelCase_ , )
if return_all_scores:
__lowerCAmelCase = None
else:
__lowerCAmelCase = 1
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
__lowerCAmelCase = ClassificationFunction[function_to_apply.upper()]
if function_to_apply is not None:
__lowerCAmelCase = function_to_apply
return preprocess_params, {}, postprocess_params
def __call__( self : Union[str, Any] , *lowerCAmelCase_ : Tuple , **lowerCAmelCase_ : Any ) -> List[Any]:
__lowerCAmelCase = super().__call__(*lowerCAmelCase_ , **lowerCAmelCase_ )
# TODO try and retrieve it in a nicer way from _sanitize_parameters.
__lowerCAmelCase = 'top_k' not in kwargs
if isinstance(args[0] , lowerCAmelCase_ ) and _legacy:
# This pipeline is odd, and return a list when single item is run
return [result]
else:
return result
def lowercase ( self : List[str] , lowerCAmelCase_ : Any , **lowerCAmelCase_ : Optional[Any] ) -> Dict[str, GenericTensor]:
__lowerCAmelCase = self.framework
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
return self.tokenizer(**lowerCAmelCase_ , return_tensors=lowerCAmelCase_ , **lowerCAmelCase_ )
elif isinstance(lowerCAmelCase_ , lowerCAmelCase_ ) and len(lowerCAmelCase_ ) == 1 and isinstance(inputs[0] , lowerCAmelCase_ ) and len(inputs[0] ) == 2:
# It used to be valid to use a list of list of list for text pairs, keeping this path for BC
return self.tokenizer(
text=inputs[0][0] , text_pair=inputs[0][1] , return_tensors=lowerCAmelCase_ , **lowerCAmelCase_ )
elif isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
# This is likely an invalid usage of the pipeline attempting to pass text pairs.
raise ValueError(
'The pipeline received invalid inputs, if you are trying to send text pairs, you can try to send a'
' dictionary `{"text": "My text", "text_pair": "My pair"}` in order to send a text pair.' )
return self.tokenizer(lowerCAmelCase_ , return_tensors=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Any ) -> Union[str, Any]:
return self.model(**lowerCAmelCase_ )
def lowercase ( self : List[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple=None , lowerCAmelCase_ : Optional[Any]=1 , lowerCAmelCase_ : List[str]=True ) -> Optional[Any]:
# `_legacy` is used to determine if we're running the naked pipeline and in backward
# compatibility mode, or if running the pipeline with `pipeline(..., top_k=1)` we're running
# the more natural result containing the list.
# Default value before `set_parameters`
if function_to_apply is None:
if self.model.config.problem_type == "multi_label_classification" or self.model.config.num_labels == 1:
__lowerCAmelCase = ClassificationFunction.SIGMOID
elif self.model.config.problem_type == "single_label_classification" or self.model.config.num_labels > 1:
__lowerCAmelCase = ClassificationFunction.SOFTMAX
elif hasattr(self.model.config , 'function_to_apply' ) and function_to_apply is None:
__lowerCAmelCase = self.model.config.function_to_apply
else:
__lowerCAmelCase = ClassificationFunction.NONE
__lowerCAmelCase = model_outputs['logits'][0]
__lowerCAmelCase = outputs.numpy()
if function_to_apply == ClassificationFunction.SIGMOID:
__lowerCAmelCase = sigmoid(lowerCAmelCase_ )
elif function_to_apply == ClassificationFunction.SOFTMAX:
__lowerCAmelCase = softmax(lowerCAmelCase_ )
elif function_to_apply == ClassificationFunction.NONE:
__lowerCAmelCase = outputs
else:
raise ValueError(f"""Unrecognized `function_to_apply` argument: {function_to_apply}""" )
if top_k == 1 and _legacy:
return {"label": self.model.config.idalabel[scores.argmax().item()], "score": scores.max().item()}
__lowerCAmelCase = [
{'label': self.model.config.idalabel[i], 'score': score.item()} for i, score in enumerate(lowerCAmelCase_ )
]
if not _legacy:
dict_scores.sort(key=lambda lowerCAmelCase_ : x["score"] , reverse=lowerCAmelCase_ )
if top_k is not None:
__lowerCAmelCase = dict_scores[:top_k]
return dict_scores
| 53 |
def a_ ( lowerCAmelCase_ : int = 200_0000 ):
__lowerCAmelCase = [0 for i in range(n + 1 )]
__lowerCAmelCase = 1
__lowerCAmelCase = 1
for i in range(2, int(n**0.5 ) + 1 ):
if primality_list[i] == 0:
for j in range(i * i, n + 1, lowerCAmelCase_ ):
__lowerCAmelCase = 1
__lowerCAmelCase = 0
for i in range(lowerCAmelCase_ ):
if primality_list[i] == 0:
sum_of_primes += i
return sum_of_primes
if __name__ == "__main__":
print(F"""{solution() = }""")
| 53 | 1 |
import argparse
import shlex
import runhouse as rh
if __name__ == "__main__":
# Refer to https://runhouse-docs.readthedocs-hosted.com/en/latest/api/python/cluster.html#hardware-setup for cloud access
# setup instructions, if using on-demand hardware
# If user passes --user <user> --host <host> --key_path <key_path> <example> <args>, fill them in as BYO cluster
# If user passes --instance <instance> --provider <provider> <example> <args>, fill them in as on-demand cluster
# Throw an error if user passes both BYO and on-demand cluster args
# Otherwise, use default values
_snake_case : Optional[Any] = argparse.ArgumentParser()
parser.add_argument('--user', type=str, default='ubuntu')
parser.add_argument('--host', type=str, default='localhost')
parser.add_argument('--key_path', type=str, default=None)
parser.add_argument('--instance', type=str, default='V100:1')
parser.add_argument('--provider', type=str, default='cheapest')
parser.add_argument('--use_spot', type=bool, default=False)
parser.add_argument('--example', type=str, default='pytorch/text-generation/run_generation.py')
_snake_case , _snake_case : str = parser.parse_known_args()
if args.host != "localhost":
if args.instance != "V100:1" or args.provider != "cheapest":
raise ValueError('Cannot specify both BYO and on-demand cluster args')
_snake_case : int = rh.cluster(
name='rh-cluster', ips=[args.host], ssh_creds={'ssh_user': args.user, 'ssh_private_key': args.key_path}
)
else:
_snake_case : Optional[Any] = rh.cluster(
name='rh-cluster', instance_type=args.instance, provider=args.provider, use_spot=args.use_spot
)
_snake_case : Optional[Any] = args.example.rsplit('/', 1)[0]
# Set up remote environment
cluster.install_packages(['pip:./']) # Installs transformers from local source
# Note transformers is copied into the home directory on the remote machine, so we can install from there
cluster.run([F"""pip install -r transformers/examples/{example_dir}/requirements.txt"""])
cluster.run(['pip install torch --upgrade --extra-index-url https://download.pytorch.org/whl/cu117'])
# Run example. You can bypass the CLI wrapper and paste your own code here.
cluster.run([F"""python transformers/examples/{args.example} {" ".join(shlex.quote(arg) for arg in unknown)}"""])
# Alternatively, we can just import and run a training function (especially if there's no wrapper CLI):
# from my_script... import train
# reqs = ['pip:./', 'torch', 'datasets', 'accelerate', 'evaluate', 'tqdm', 'scipy', 'scikit-learn', 'tensorboard']
# launch_train_gpu = rh.function(fn=train,
# system=gpu,
# reqs=reqs,
# name='train_bert_glue')
#
# We can pass in arguments just like we would to a function:
# launch_train_gpu(num_epochs = 3, lr = 2e-5, seed = 42, batch_size = 16
# stream_logs=True)
| 53 |
import json
import logging
import os
import sys
from pathlib import Path
import finetune_rag
from transformers.file_utils import is_apex_available
from transformers.testing_utils import (
TestCasePlus,
execute_subprocess_async,
require_ray,
require_torch_gpu,
require_torch_multi_gpu,
)
logging.basicConfig(level=logging.DEBUG)
_snake_case : Tuple = logging.getLogger()
_snake_case : Any = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Any , lowerCAmelCase_ : Dict ) -> Optional[int]:
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = {'source': 'What is love ?', 'target': 'life'}
__lowerCAmelCase = {'train': 1_2, 'val': 2, 'test': 2}
for split in ["train", "test", "val"]:
for field in ["source", "target"]:
__lowerCAmelCase = '\n'.join([contents[field]] * n_lines[split] )
with open(os.path.join(lowerCAmelCase_ , f"""{split}.{field}""" ) , 'w' ) as f:
f.write(lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : int , lowerCAmelCase_ : str = "pytorch" ) -> List[str]:
__lowerCAmelCase = self.get_auto_remove_tmp_dir()
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'output' )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'data' )
self._create_dummy_data(data_dir=lowerCAmelCase_ )
__lowerCAmelCase = f"""
--data_dir {data_dir} \
--output_dir {output_dir} \
--model_name_or_path facebook/rag-sequence-base \
--model_type rag_sequence \
--do_train \
--do_predict \
--n_val -1 \
--val_check_interval 1.0 \
--train_batch_size 2 \
--eval_batch_size 1 \
--max_source_length 25 \
--max_target_length 25 \
--val_max_target_length 25 \
--test_max_target_length 25 \
--label_smoothing 0.1 \
--dropout 0.1 \
--attention_dropout 0.1 \
--weight_decay 0.001 \
--adam_epsilon 1e-08 \
--max_grad_norm 0.1 \
--lr_scheduler polynomial \
--learning_rate 3e-04 \
--num_train_epochs 1 \
--warmup_steps 4 \
--gradient_accumulation_steps 1 \
--distributed-port 8787 \
--use_dummy_dataset 1 \
--distributed_retriever {distributed_retriever} \
""".split()
if gpus > 0:
testargs.append(f"""--gpus={gpus}""" )
if is_apex_available():
testargs.append('--fp16' )
else:
testargs.append('--gpus=0' )
testargs.append('--distributed_backend=ddp_cpu' )
testargs.append('--num_processes=2' )
__lowerCAmelCase = [sys.executable, str(Path(finetune_rag.__file__ ).resolve() )] + testargs
execute_subprocess_async(lowerCAmelCase_ , env=self.get_env() )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'metrics.json' )
with open(lowerCAmelCase_ ) as f:
__lowerCAmelCase = json.load(lowerCAmelCase_ )
return result
@require_torch_gpu
def lowercase ( self : str ) -> int:
__lowerCAmelCase = self._run_finetune(gpus=1 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
def lowercase ( self : List[str] ) -> Dict:
__lowerCAmelCase = self._run_finetune(gpus=2 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_gpu
@require_ray
def lowercase ( self : int ) -> Tuple:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
@require_ray
def lowercase ( self : List[Any] ) -> str:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
| 53 | 1 |
import unittest
from diffusers.pipelines.pipeline_utils import is_safetensors_compatible
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Union[str, Any] ) -> List[str]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Any:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
# Removed: 'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> int:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
# Removed: 'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> str:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[str]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Union[str, Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
# Removed: 'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : List[str] ) -> List[Any]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : Optional[Any] ) -> Optional[Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
# 'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
| 53 |
import copy
import inspect
import unittest
from transformers import AutoBackbone
from transformers.configuration_utils import PretrainedConfig
from transformers.testing_utils import require_timm, require_torch, torch_device
from transformers.utils.import_utils import is_torch_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor
if is_torch_available():
import torch
from transformers import TimmBackbone, TimmBackboneConfig
from ...test_pipeline_mixin import PipelineTesterMixin
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : Optional[int]=None , lowerCAmelCase_ : Union[str, Any]="resnet50" , lowerCAmelCase_ : str=3 , lowerCAmelCase_ : List[str]=3_2 , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : Dict=True , lowerCAmelCase_ : Optional[Any]=True , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = out_indices if out_indices is not None else [4]
__lowerCAmelCase = stage_names
__lowerCAmelCase = out_features
__lowerCAmelCase = backbone
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = use_pretrained_backbone
__lowerCAmelCase = is_training
def lowercase ( self : List[str] ) -> Any:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : List[Any] ) -> Union[str, Any]:
return TimmBackboneConfig(
image_size=self.image_size , num_channels=self.num_channels , out_features=self.out_features , out_indices=self.out_indices , stage_names=self.stage_names , use_pretrained_backbone=self.use_pretrained_backbone , backbone=self.backbone , )
def lowercase ( self : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ) -> int:
__lowerCAmelCase = TimmBackbone(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.feature_map[-1].shape , (self.batch_size, model.channels[-1], 1_4, 1_4) , )
def lowercase ( self : List[str] ) -> str:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
@require_timm
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (TimmBackbone,) if is_torch_available() else ()
a_ = {"""feature-extraction""": TimmBackbone} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Tuple ) -> int:
__lowerCAmelCase = TimmBackboneModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : Dict ) -> List[str]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Union[str, Any] ) -> Optional[int]:
__lowerCAmelCase = 'resnet18'
__lowerCAmelCase = 'microsoft/resnet-18'
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(len(timm_model.stage_names ) , len(transformers_model.stage_names ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
# Out indices are set to the last layer by default. For timm models, we don't know
# the number of layers in advance, so we set it to (-1,), whereas for transformers
# models, we set it to [len(stage_names) - 1] (kept for backward compatibility).
self.assertEqual(timm_model.out_indices , (-1,) )
self.assertEqual(transformers_model.out_indices , [len(timm_model.stage_names ) - 1] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ , out_indices=[1, 2, 3] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , out_indices=[1, 2, 3] )
self.assertEqual(timm_model.out_indices , transformers_model.out_indices )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
@unittest.skip('TimmBackbone doesn\'t support feed forward chunking' )
def lowercase ( self : List[str] ) -> Tuple:
pass
@unittest.skip('TimmBackbone doesn\'t have num_hidden_layers attribute' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('TimmBackbone initialization is managed on the timm side' )
def lowercase ( self : str ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Any ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Optional[Any]:
pass
@unittest.skip('TimmBackbone model cannot be created without specifying a backbone checkpoint' )
def lowercase ( self : Dict ) -> Any:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Any ) -> Optional[int]:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : Union[str, Any] ) -> Tuple:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : List[str] ) -> Optional[int]:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Tuple ) -> List[str]:
pass
@unittest.skip('TimmBackbone doesn\'t have hidden size info in its configuration.' )
def lowercase ( self : int ) -> Optional[int]:
pass
@unittest.skip('TimmBackbone doesn\'t support output_attentions.' )
def lowercase ( self : Union[str, Any] ) -> str:
pass
@unittest.skip('Safetensors is not supported by timm.' )
def lowercase ( self : Dict ) -> str:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : List[str] ) -> Optional[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = True
__lowerCAmelCase = self.has_attentions
# no need to test all models as different heads yield the same functionality
__lowerCAmelCase = self.all_model_classes[0]
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs[0][-1]
# Encoder-/Decoder-only models
__lowerCAmelCase = outputs.hidden_states[0]
hidden_states.retain_grad()
if self.has_attentions:
__lowerCAmelCase = outputs.attentions[0]
attentions.retain_grad()
output.flatten()[0].backward(retain_graph=lowerCAmelCase_ )
self.assertIsNotNone(hidden_states.grad )
if self.has_attentions:
self.assertIsNotNone(attentions.grad )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , len(config.out_indices ) )
self.assertEqual(len(model.channels ) , len(config.out_indices ) )
# Check output of last stage is taken if out_features=None, out_indices=None
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , 1 )
self.assertEqual(len(model.channels ) , 1 )
# Check backbone can be initialized with fresh weights
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = False
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
| 53 | 1 |
import os
import textwrap
import pyarrow as pa
import pytest
from datasets import ClassLabel, Features, Image
from datasets.packaged_modules.csv.csv import Csv
from ..utils import require_pil
@pytest.fixture
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = tmp_path / 'file.csv'
__lowerCAmelCase = textwrap.dedent(
'\\n header1,header2\n 1,2\n 10,20\n ' )
with open(lowerCAmelCase_, 'w' ) as f:
f.write(lowerCAmelCase_ )
return str(lowerCAmelCase_ )
@pytest.fixture
def a_ ( lowerCAmelCase_ : Dict ):
__lowerCAmelCase = tmp_path / 'malformed_file.csv'
__lowerCAmelCase = textwrap.dedent(
'\\n header1,header2\n 1,2\n 10,20,\n ' )
with open(lowerCAmelCase_, 'w' ) as f:
f.write(lowerCAmelCase_ )
return str(lowerCAmelCase_ )
@pytest.fixture
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : str ):
__lowerCAmelCase = tmp_path / 'csv_with_image.csv'
__lowerCAmelCase = textwrap.dedent(
F"""\
image
{image_file}
""" )
with open(lowerCAmelCase_, 'w' ) as f:
f.write(lowerCAmelCase_ )
return str(lowerCAmelCase_ )
@pytest.fixture
def a_ ( lowerCAmelCase_ : Optional[int] ):
__lowerCAmelCase = tmp_path / 'csv_with_label.csv'
__lowerCAmelCase = textwrap.dedent(
'\\n label\n good\n bad\n good\n ' )
with open(lowerCAmelCase_, 'w' ) as f:
f.write(lowerCAmelCase_ )
return str(lowerCAmelCase_ )
@pytest.fixture
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = tmp_path / 'csv_with_int_list.csv'
__lowerCAmelCase = textwrap.dedent(
'\\n int_list\n 1 2 3\n 4 5 6\n 7 8 9\n ' )
with open(lowerCAmelCase_, 'w' ) as f:
f.write(lowerCAmelCase_ )
return str(lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : List[Any], lowerCAmelCase_ : Any ):
__lowerCAmelCase = Csv()
__lowerCAmelCase = csv._generate_tables([[csv_file, malformed_csv_file]] )
with pytest.raises(lowerCAmelCase_, match='Error tokenizing data' ):
for _ in generator:
pass
assert any(
record.levelname == 'ERROR'
and 'Failed to read file' in record.message
and os.path.basename(lowerCAmelCase_ ) in record.message
for record in caplog.records )
@require_pil
def a_ ( lowerCAmelCase_ : List[str] ):
with open(lowerCAmelCase_, encoding='utf-8' ) as f:
__lowerCAmelCase = f.read().splitlines()[1]
__lowerCAmelCase = Csv(encoding='utf-8', features=Features({'image': Image()} ) )
__lowerCAmelCase = csv._generate_tables([[csv_file_with_image]] )
__lowerCAmelCase = pa.concat_tables([table for _, table in generator] )
assert pa_table.schema.field('image' ).type == Image()()
__lowerCAmelCase = pa_table.to_pydict()['image']
assert generated_content == [{"path": image_file, "bytes": None}]
def a_ ( lowerCAmelCase_ : Dict ):
with open(lowerCAmelCase_, encoding='utf-8' ) as f:
__lowerCAmelCase = f.read().splitlines()[1:]
__lowerCAmelCase = Csv(encoding='utf-8', features=Features({'label': ClassLabel(names=['good', 'bad'] )} ) )
__lowerCAmelCase = csv._generate_tables([[csv_file_with_label]] )
__lowerCAmelCase = pa.concat_tables([table for _, table in generator] )
assert pa_table.schema.field('label' ).type == ClassLabel(names=['good', 'bad'] )()
__lowerCAmelCase = pa_table.to_pydict()['label']
assert generated_content == [ClassLabel(names=['good', 'bad'] ).straint(lowerCAmelCase_ ) for label in labels]
def a_ ( lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = Csv(encoding='utf-8', sep=',', converters={'int_list': lambda lowerCAmelCase_ : [int(lowerCAmelCase_ ) for i in x.split()]} )
__lowerCAmelCase = csv._generate_tables([[csv_file_with_int_list]] )
__lowerCAmelCase = pa.concat_tables([table for _, table in generator] )
assert pa.types.is_list(pa_table.schema.field('int_list' ).type )
__lowerCAmelCase = pa_table.to_pydict()['int_list']
assert generated_content == [[1, 2, 3], [4, 5, 6], [7, 8, 9]]
| 53 |
# Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import os
import platform
import numpy as np
import psutil
import torch
from accelerate import __version__ as version
from accelerate.commands.config import default_config_file, load_config_from_file
from ..utils import is_npu_available, is_xpu_available
def a_ ( lowerCAmelCase_ : str=None ):
if subparsers is not None:
__lowerCAmelCase = subparsers.add_parser('env' )
else:
__lowerCAmelCase = argparse.ArgumentParser('Accelerate env command' )
parser.add_argument(
'--config_file', default=lowerCAmelCase_, help='The config file to use for the default values in the launching script.' )
if subparsers is not None:
parser.set_defaults(func=lowerCAmelCase_ )
return parser
def a_ ( lowerCAmelCase_ : Optional[int] ):
__lowerCAmelCase = torch.__version__
__lowerCAmelCase = torch.cuda.is_available()
__lowerCAmelCase = is_xpu_available()
__lowerCAmelCase = is_npu_available()
__lowerCAmelCase = 'Not found'
# Get the default from the config file.
if args.config_file is not None or os.path.isfile(lowerCAmelCase_ ):
__lowerCAmelCase = load_config_from_file(args.config_file ).to_dict()
__lowerCAmelCase = {
'`Accelerate` version': version,
'Platform': platform.platform(),
'Python version': platform.python_version(),
'Numpy version': np.__version__,
'PyTorch version (GPU?)': F"""{pt_version} ({pt_cuda_available})""",
'PyTorch XPU available': str(lowerCAmelCase_ ),
'PyTorch NPU available': str(lowerCAmelCase_ ),
'System RAM': F"""{psutil.virtual_memory().total / 1024 ** 3:.2f} GB""",
}
if pt_cuda_available:
__lowerCAmelCase = torch.cuda.get_device_name()
print('\nCopy-and-paste the text below in your GitHub issue\n' )
print('\n'.join([F"""- {prop}: {val}""" for prop, val in info.items()] ) )
print('- `Accelerate` default config:' if args.config_file is None else '- `Accelerate` config passed:' )
__lowerCAmelCase = (
'\n'.join([F"""\t- {prop}: {val}""" for prop, val in accelerate_config.items()] )
if isinstance(lowerCAmelCase_, lowerCAmelCase_ )
else F"""\t{accelerate_config}"""
)
print(lowerCAmelCase_ )
__lowerCAmelCase = accelerate_config
return info
def a_ ( ):
__lowerCAmelCase = env_command_parser()
__lowerCAmelCase = parser.parse_args()
env_command(lowerCAmelCase_ )
return 0
if __name__ == "__main__":
raise SystemExit(main())
| 53 | 1 |
import dataclasses
import json
import sys
import types
from argparse import ArgumentDefaultsHelpFormatter, ArgumentParser, ArgumentTypeError
from copy import copy
from enum import Enum
from inspect import isclass
from pathlib import Path
from typing import Any, Callable, Dict, Iterable, List, Literal, NewType, Optional, Tuple, Union, get_type_hints
import yaml
_snake_case : List[str] = NewType('DataClass', Any)
_snake_case : str = NewType('DataClassType', Any)
def a_ ( lowerCAmelCase_ : Any ):
if isinstance(lowerCAmelCase_, lowerCAmelCase_ ):
return v
if v.lower() in ("yes", "true", "t", "y", "1"):
return True
elif v.lower() in ("no", "false", "f", "n", "0"):
return False
else:
raise ArgumentTypeError(
F"""Truthy value expected: got {v} but expected one of yes/no, true/false, t/f, y/n, 1/0 (case insensitive).""" )
def a_ ( lowerCAmelCase_ : list ):
__lowerCAmelCase = {str(lowerCAmelCase_ ): choice for choice in choices}
return lambda lowerCAmelCase_ : str_to_choice.get(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( *,
lowerCAmelCase_ : Union[str, List[str]] = None, lowerCAmelCase_ : str = None, lowerCAmelCase_ : Any = dataclasses.MISSING, lowerCAmelCase_ : Callable[[], Any] = dataclasses.MISSING, lowerCAmelCase_ : dict = None, **lowerCAmelCase_ : str, ):
if metadata is None:
# Important, don't use as default param in function signature because dict is mutable and shared across function calls
__lowerCAmelCase = {}
if aliases is not None:
__lowerCAmelCase = aliases
if help is not None:
__lowerCAmelCase = help
return dataclasses.field(metadata=lowerCAmelCase_, default=lowerCAmelCase_, default_factory=lowerCAmelCase_, **lowerCAmelCase_ )
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = 42
def __init__( self : Union[str, Any] , lowerCAmelCase_ : Union[DataClassType, Iterable[DataClassType]] , **lowerCAmelCase_ : Optional[Any] ) -> List[str]:
# To make the default appear when using --help
if "formatter_class" not in kwargs:
__lowerCAmelCase = ArgumentDefaultsHelpFormatter
super().__init__(**lowerCAmelCase_ )
if dataclasses.is_dataclass(lowerCAmelCase_ ):
__lowerCAmelCase = [dataclass_types]
__lowerCAmelCase = list(lowerCAmelCase_ )
for dtype in self.dataclass_types:
self._add_dataclass_arguments(lowerCAmelCase_ )
@staticmethod
def lowercase ( lowerCAmelCase_ : ArgumentParser , lowerCAmelCase_ : dataclasses.Field ) -> List[Any]:
__lowerCAmelCase = f"""--{field.name}"""
__lowerCAmelCase = field.metadata.copy()
# field.metadata is not used at all by Data Classes,
# it is provided as a third-party extension mechanism.
if isinstance(field.type , lowerCAmelCase_ ):
raise RuntimeError(
'Unresolved type detected, which should have been done with the help of '
'`typing.get_type_hints` method by default' )
__lowerCAmelCase = kwargs.pop('aliases' , [] )
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
__lowerCAmelCase = [aliases]
__lowerCAmelCase = getattr(field.type , '__origin__' , field.type )
if origin_type is Union or (hasattr(lowerCAmelCase_ , 'UnionType' ) and isinstance(lowerCAmelCase_ , types.UnionType )):
if str not in field.type.__args__ and (
len(field.type.__args__ ) != 2 or type(lowerCAmelCase_ ) not in field.type.__args__
):
raise ValueError(
'Only `Union[X, NoneType]` (i.e., `Optional[X]`) is allowed for `Union` because'
' the argument parser only supports one type per argument.'
f""" Problem encountered in field '{field.name}'.""" )
if type(lowerCAmelCase_ ) not in field.type.__args__:
# filter `str` in Union
__lowerCAmelCase = field.type.__args__[0] if field.type.__args__[1] == str else field.type.__args__[1]
__lowerCAmelCase = getattr(field.type , '__origin__' , field.type )
elif bool not in field.type.__args__:
# filter `NoneType` in Union (except for `Union[bool, NoneType]`)
__lowerCAmelCase = (
field.type.__args__[0] if isinstance(lowerCAmelCase_ , field.type.__args__[1] ) else field.type.__args__[1]
)
__lowerCAmelCase = getattr(field.type , '__origin__' , field.type )
# A variable to store kwargs for a boolean field, if needed
# so that we can init a `no_*` complement argument (see below)
__lowerCAmelCase = {}
if origin_type is Literal or (isinstance(field.type , lowerCAmelCase_ ) and issubclass(field.type , lowerCAmelCase_ )):
if origin_type is Literal:
__lowerCAmelCase = field.type.__args__
else:
__lowerCAmelCase = [x.value for x in field.type]
__lowerCAmelCase = make_choice_type_function(kwargs['choices'] )
if field.default is not dataclasses.MISSING:
__lowerCAmelCase = field.default
else:
__lowerCAmelCase = True
elif field.type is bool or field.type == Optional[bool]:
# Copy the currect kwargs to use to instantiate a `no_*` complement argument below.
# We do not initialize it here because the `no_*` alternative must be instantiated after the real argument
__lowerCAmelCase = copy(lowerCAmelCase_ )
# Hack because type=bool in argparse does not behave as we want.
__lowerCAmelCase = string_to_bool
if field.type is bool or (field.default is not None and field.default is not dataclasses.MISSING):
# Default value is False if we have no default when of type bool.
__lowerCAmelCase = False if field.default is dataclasses.MISSING else field.default
# This is the value that will get picked if we don't include --field_name in any way
__lowerCAmelCase = default
# This tells argparse we accept 0 or 1 value after --field_name
__lowerCAmelCase = '?'
# This is the value that will get picked if we do --field_name (without value)
__lowerCAmelCase = True
elif isclass(lowerCAmelCase_ ) and issubclass(lowerCAmelCase_ , lowerCAmelCase_ ):
__lowerCAmelCase = field.type.__args__[0]
__lowerCAmelCase = '+'
if field.default_factory is not dataclasses.MISSING:
__lowerCAmelCase = field.default_factory()
elif field.default is dataclasses.MISSING:
__lowerCAmelCase = True
else:
__lowerCAmelCase = field.type
if field.default is not dataclasses.MISSING:
__lowerCAmelCase = field.default
elif field.default_factory is not dataclasses.MISSING:
__lowerCAmelCase = field.default_factory()
else:
__lowerCAmelCase = True
parser.add_argument(lowerCAmelCase_ , *lowerCAmelCase_ , **lowerCAmelCase_ )
# Add a complement `no_*` argument for a boolean field AFTER the initial field has already been added.
# Order is important for arguments with the same destination!
# We use a copy of earlier kwargs because the original kwargs have changed a lot before reaching down
# here and we do not need those changes/additional keys.
if field.default is True and (field.type is bool or field.type == Optional[bool]):
__lowerCAmelCase = False
parser.add_argument(f"""--no_{field.name}""" , action='store_false' , dest=field.name , **lowerCAmelCase_ )
def lowercase ( self : int , lowerCAmelCase_ : DataClassType ) -> List[Any]:
if hasattr(lowerCAmelCase_ , '_argument_group_name' ):
__lowerCAmelCase = self.add_argument_group(dtype._argument_group_name )
else:
__lowerCAmelCase = self
try:
__lowerCAmelCase = get_type_hints(lowerCAmelCase_ )
except NameError:
raise RuntimeError(
f"""Type resolution failed for {dtype}. Try declaring the class in global scope or """
'removing line of `from __future__ import annotations` which opts in Postponed '
'Evaluation of Annotations (PEP 563)' )
except TypeError as ex:
# Remove this block when we drop Python 3.9 support
if sys.version_info[:2] < (3, 1_0) and "unsupported operand type(s) for |" in str(lowerCAmelCase_ ):
__lowerCAmelCase = '.'.join(map(lowerCAmelCase_ , sys.version_info[:3] ) )
raise RuntimeError(
f"""Type resolution failed for {dtype} on Python {python_version}. Try removing """
'line of `from __future__ import annotations` which opts in union types as '
'`X | Y` (PEP 604) via Postponed Evaluation of Annotations (PEP 563). To '
'support Python versions that lower than 3.10, you need to use '
'`typing.Union[X, Y]` instead of `X | Y` and `typing.Optional[X]` instead of '
'`X | None`.' ) from ex
raise
for field in dataclasses.fields(lowerCAmelCase_ ):
if not field.init:
continue
__lowerCAmelCase = type_hints[field.name]
self._parse_dataclass_field(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any]=None , lowerCAmelCase_ : Optional[Any]=False , lowerCAmelCase_ : Dict=True , lowerCAmelCase_ : Any=None , lowerCAmelCase_ : str=None , ) -> Tuple[DataClass, ...]:
if args_file_flag or args_filename or (look_for_args_file and len(sys.argv )):
__lowerCAmelCase = []
if args_filename:
args_files.append(Path(lowerCAmelCase_ ) )
elif look_for_args_file and len(sys.argv ):
args_files.append(Path(sys.argv[0] ).with_suffix('.args' ) )
# args files specified via command line flag should overwrite default args files so we add them last
if args_file_flag:
# Create special parser just to extract the args_file_flag values
__lowerCAmelCase = ArgumentParser()
args_file_parser.add_argument(lowerCAmelCase_ , type=lowerCAmelCase_ , action='append' )
# Use only remaining args for further parsing (remove the args_file_flag)
__lowerCAmelCase , __lowerCAmelCase = args_file_parser.parse_known_args(args=lowerCAmelCase_ )
__lowerCAmelCase = vars(lowerCAmelCase_ ).get(args_file_flag.lstrip('-' ) , lowerCAmelCase_ )
if cmd_args_file_paths:
args_files.extend([Path(lowerCAmelCase_ ) for p in cmd_args_file_paths] )
__lowerCAmelCase = []
for args_file in args_files:
if args_file.exists():
file_args += args_file.read_text().split()
# in case of duplicate arguments the last one has precedence
# args specified via the command line should overwrite args from files, so we add them last
__lowerCAmelCase = file_args + args if args is not None else file_args + sys.argv[1:]
__lowerCAmelCase , __lowerCAmelCase = self.parse_known_args(args=lowerCAmelCase_ )
__lowerCAmelCase = []
for dtype in self.dataclass_types:
__lowerCAmelCase = {f.name for f in dataclasses.fields(lowerCAmelCase_ ) if f.init}
__lowerCAmelCase = {k: v for k, v in vars(lowerCAmelCase_ ).items() if k in keys}
for k in keys:
delattr(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = dtype(**lowerCAmelCase_ )
outputs.append(lowerCAmelCase_ )
if len(namespace.__dict__ ) > 0:
# additional namespace.
outputs.append(lowerCAmelCase_ )
if return_remaining_strings:
return (*outputs, remaining_args)
else:
if remaining_args:
raise ValueError(f"""Some specified arguments are not used by the HfArgumentParser: {remaining_args}""" )
return (*outputs,)
def lowercase ( self : Dict , lowerCAmelCase_ : Dict[str, Any] , lowerCAmelCase_ : bool = False ) -> Tuple[DataClass, ...]:
__lowerCAmelCase = set(args.keys() )
__lowerCAmelCase = []
for dtype in self.dataclass_types:
__lowerCAmelCase = {f.name for f in dataclasses.fields(lowerCAmelCase_ ) if f.init}
__lowerCAmelCase = {k: v for k, v in args.items() if k in keys}
unused_keys.difference_update(inputs.keys() )
__lowerCAmelCase = dtype(**lowerCAmelCase_ )
outputs.append(lowerCAmelCase_ )
if not allow_extra_keys and unused_keys:
raise ValueError(f"""Some keys are not used by the HfArgumentParser: {sorted(lowerCAmelCase_ )}""" )
return tuple(lowerCAmelCase_ )
def lowercase ( self : List[str] , lowerCAmelCase_ : str , lowerCAmelCase_ : bool = False ) -> Tuple[DataClass, ...]:
with open(Path(lowerCAmelCase_ ) , encoding='utf-8' ) as open_json_file:
__lowerCAmelCase = json.loads(open_json_file.read() )
__lowerCAmelCase = self.parse_dict(lowerCAmelCase_ , allow_extra_keys=lowerCAmelCase_ )
return tuple(lowerCAmelCase_ )
def lowercase ( self : Optional[int] , lowerCAmelCase_ : str , lowerCAmelCase_ : bool = False ) -> Tuple[DataClass, ...]:
__lowerCAmelCase = self.parse_dict(yaml.safe_load(Path(lowerCAmelCase_ ).read_text() ) , allow_extra_keys=lowerCAmelCase_ )
return tuple(lowerCAmelCase_ )
| 53 |
import importlib
import sys
from argparse import REMAINDER, ArgumentParser
from pathlib import Path
import torch_xla.distributed.xla_multiprocessing as xmp
def a_ ( ):
__lowerCAmelCase = ArgumentParser(
description=(
'PyTorch TPU distributed training launch '
'helper utility that will spawn up '
'multiple distributed processes'
) )
# Optional arguments for the launch helper
parser.add_argument('--num_cores', type=lowerCAmelCase_, default=1, help='Number of TPU cores to use (1 or 8).' )
# positional
parser.add_argument(
'training_script', type=lowerCAmelCase_, help=(
'The full path to the single TPU training '
'program/script to be launched in parallel, '
'followed by all the arguments for the '
'training script'
), )
# rest from the training program
parser.add_argument('training_script_args', nargs=lowerCAmelCase_ )
return parser.parse_args()
def a_ ( ):
__lowerCAmelCase = parse_args()
# Import training_script as a module.
__lowerCAmelCase = Path(args.training_script )
sys.path.append(str(script_fpath.parent.resolve() ) )
__lowerCAmelCase = script_fpath.stem
__lowerCAmelCase = importlib.import_module(lowerCAmelCase_ )
# Patch sys.argv
__lowerCAmelCase = [args.training_script] + args.training_script_args + ['--tpu_num_cores', str(args.num_cores )]
xmp.spawn(mod._mp_fn, args=(), nprocs=args.num_cores )
if __name__ == "__main__":
main()
| 53 | 1 |
import csv
from collections import defaultdict
from dataclasses import dataclass, field
from typing import List, Optional
import matplotlib.pyplot as plt
import numpy as np
from matplotlib.ticker import ScalarFormatter
from transformers import HfArgumentParser
def a_ ( lowerCAmelCase_ : str=None, lowerCAmelCase_ : Any=None ):
return field(default_factory=lambda: default, metadata=lowerCAmelCase_ )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
metadata={"""help""": """The csv file to plot."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to plot along batch size or sequence length. Defaults to sequence length."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether the csv file has time results or memory results. Defaults to memory results."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Disable logarithmic scale when plotting"""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": """Whether the csv file has training results or inference results. Defaults to inference results."""
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Filename under which the plot will be saved. If unused no plot is saved."""} , )
a_ = list_field(
default=_UpperCamelCase , metadata={"""help""": """List of model names that are used instead of the ones in the csv file."""} )
def a_ ( lowerCAmelCase_ : Tuple ):
try:
int(lowerCAmelCase_ )
return True
except ValueError:
return False
def a_ ( lowerCAmelCase_ : Dict ):
try:
float(lowerCAmelCase_ )
return True
except ValueError:
return False
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Union[str, Any] , lowerCAmelCase_ : List[Any] ) -> Union[str, Any]:
__lowerCAmelCase = args
__lowerCAmelCase = defaultdict(lambda: {"bsz": [], "seq_len": [], "result": {}} )
with open(self.args.csv_file , newline='' ) as csv_file:
__lowerCAmelCase = csv.DictReader(lowerCAmelCase_ )
for row in reader:
__lowerCAmelCase = row['model']
self.result_dict[model_name]["bsz"].append(int(row['batch_size'] ) )
self.result_dict[model_name]["seq_len"].append(int(row['sequence_length'] ) )
if can_convert_to_int(row['result'] ):
# value is not None
__lowerCAmelCase = int(row['result'] )
elif can_convert_to_float(row['result'] ):
# value is not None
__lowerCAmelCase = float(row['result'] )
def lowercase ( self : List[Any] ) -> Tuple:
__lowerCAmelCase , __lowerCAmelCase = plt.subplots()
__lowerCAmelCase = 'Time usage' if self.args.is_time else 'Memory usage'
__lowerCAmelCase = title_str + ' for training' if self.args.is_train else title_str + ' for inference'
if not self.args.no_log_scale:
# set logarithm scales
ax.set_xscale('log' )
ax.set_yscale('log' )
for axis in [ax.xaxis, ax.yaxis]:
axis.set_major_formatter(ScalarFormatter() )
for model_name_idx, model_name in enumerate(self.result_dict.keys() ):
__lowerCAmelCase = sorted(set(self.result_dict[model_name]['bsz'] ) )
__lowerCAmelCase = sorted(set(self.result_dict[model_name]['seq_len'] ) )
__lowerCAmelCase = self.result_dict[model_name]['result']
((__lowerCAmelCase) , (__lowerCAmelCase)) = (
(batch_sizes, sequence_lengths) if self.args.plot_along_batch else (sequence_lengths, batch_sizes)
)
__lowerCAmelCase = (
model_name if self.args.short_model_names is None else self.args.short_model_names[model_name_idx]
)
for inner_loop_value in inner_loop_array:
if self.args.plot_along_batch:
__lowerCAmelCase = np.asarray(
[results[(x, inner_loop_value)] for x in x_axis_array if (x, inner_loop_value) in results] , dtype=lowerCAmelCase_ , )
else:
__lowerCAmelCase = np.asarray(
[results[(inner_loop_value, x)] for x in x_axis_array if (inner_loop_value, x) in results] , dtype=np.floataa , )
((__lowerCAmelCase) , (__lowerCAmelCase)) = (
('batch_size', 'len') if self.args.plot_along_batch else ('in #tokens', 'bsz')
)
__lowerCAmelCase = np.asarray(lowerCAmelCase_ , lowerCAmelCase_ )[: len(lowerCAmelCase_ )]
plt.scatter(
lowerCAmelCase_ , lowerCAmelCase_ , label=f"""{label_model_name} - {inner_loop_label}: {inner_loop_value}""" )
plt.plot(lowerCAmelCase_ , lowerCAmelCase_ , '--' )
title_str += f""" {label_model_name} vs."""
__lowerCAmelCase = title_str[:-4]
__lowerCAmelCase = 'Time in s' if self.args.is_time else 'Memory in MB'
# plot
plt.title(lowerCAmelCase_ )
plt.xlabel(lowerCAmelCase_ )
plt.ylabel(lowerCAmelCase_ )
plt.legend()
if self.args.figure_png_file is not None:
plt.savefig(self.args.figure_png_file )
else:
plt.show()
def a_ ( ):
__lowerCAmelCase = HfArgumentParser(lowerCAmelCase_ )
__lowerCAmelCase = parser.parse_args_into_dataclasses()[0]
__lowerCAmelCase = Plot(args=lowerCAmelCase_ )
plot.plot()
if __name__ == "__main__":
main()
| 53 |
import inspect
import unittest
from huggingface_hub import hf_hub_download
from transformers import ConvNextConfig, UperNetConfig
from transformers.testing_utils import require_torch, require_torch_multi_gpu, require_vision, slow, torch_device
from transformers.utils import is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import UperNetForSemanticSegmentation
from transformers.models.upernet.modeling_upernet import UPERNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Dict=1_3 , lowerCAmelCase_ : str=3_2 , lowerCAmelCase_ : Optional[Any]=3 , lowerCAmelCase_ : Any=4 , lowerCAmelCase_ : str=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Tuple=[2, 2, 3, 2] , lowerCAmelCase_ : str=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[int]=3_7 , lowerCAmelCase_ : Dict="gelu" , lowerCAmelCase_ : List[Any]=1_0 , lowerCAmelCase_ : str=0.02 , lowerCAmelCase_ : Dict=["stage2", "stage3", "stage4"] , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : List[Any]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = num_stages
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = out_features
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = num_stages
def lowercase ( self : Dict ) -> List[str]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = self.get_config()
return config, pixel_values, labels
def lowercase ( self : List[str] ) -> Union[str, Any]:
return ConvNextConfig(
num_channels=self.num_channels , num_stages=self.num_stages , hidden_sizes=self.hidden_sizes , depths=self.depths , is_training=self.is_training , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , out_features=self.out_features , )
def lowercase ( self : Dict ) -> List[str]:
return UperNetConfig(
backbone_config=self.get_backbone_config() , hidden_size=5_1_2 , pool_scales=[1, 2, 3, 6] , use_auxiliary_head=lowerCAmelCase_ , auxiliary_loss_weight=0.4 , auxiliary_in_channels=4_0 , auxiliary_channels=2_5_6 , auxiliary_num_convs=1 , auxiliary_concat_input=lowerCAmelCase_ , loss_ignore_index=2_5_5 , num_labels=self.num_labels , )
def lowercase ( self : List[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : int ) -> Optional[Any]:
__lowerCAmelCase = UperNetForSemanticSegmentation(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.logits.shape , (self.batch_size, self.num_labels, self.image_size, self.image_size) )
def lowercase ( self : Union[str, Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (UperNetForSemanticSegmentation,) if is_torch_available() else ()
a_ = {"""image-segmentation""": UperNetForSemanticSegmentation} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = UperNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Tuple ) -> Union[str, Any]:
return
def lowercase ( self : Optional[int] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_semantic_segmentation(*lowerCAmelCase_ )
@unittest.skip(reason='UperNet does not use inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Dict:
pass
@unittest.skip(reason='UperNet does not support input and output embeddings' )
def lowercase ( self : Optional[Any] ) -> Dict:
pass
@unittest.skip(reason='UperNet does not have a base model' )
def lowercase ( self : Optional[int] ) -> List[Any]:
pass
@unittest.skip(reason='UperNet does not have a base model' )
def lowercase ( self : str ) -> Dict:
pass
@require_torch_multi_gpu
@unittest.skip(reason='UperNet has some layers using `add_module` which doesn\'t work well with `nn.DataParallel`' )
def lowercase ( self : Optional[Any] ) -> Optional[int]:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : Tuple ) -> List[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Tuple:
def check_hidden_states_output(lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
# ConvNext's feature maps are of shape (batch_size, num_channels, height, width)
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [self.model_tester.image_size // 4, self.model_tester.image_size // 4] , )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Any ) -> Tuple:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = _config_zero_init(lowerCAmelCase_ )
__lowerCAmelCase = _config_zero_init(configs_no_init.backbone_config )
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(config=lowerCAmelCase_ )
for name, param in model.named_parameters():
if param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() , [0.0, 1.0] , msg=f"""Parameter {name} of model {model_class} seems not properly initialized""" , )
@unittest.skip(reason='UperNet does not have tied weights' )
def lowercase ( self : Any ) -> int:
pass
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in UPERNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
def a_ ( ):
__lowerCAmelCase = hf_hub_download(
repo_id='hf-internal-testing/fixtures_ade20k', repo_type='dataset', filename='ADE_val_00000001.jpg' )
__lowerCAmelCase = Image.open(lowerCAmelCase_ ).convert('RGB' )
return image
@require_torch
@require_vision
@slow
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Dict ) -> Union[str, Any]:
__lowerCAmelCase = AutoImageProcessor.from_pretrained('openmmlab/upernet-swin-tiny' )
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained('openmmlab/upernet-swin-tiny' ).to(lowerCAmelCase_ )
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = torch.Size((1, model.config.num_labels, 5_1_2, 5_1_2) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-7.59_58, -7.59_58, -7.43_02], [-7.59_58, -7.59_58, -7.43_02], [-7.47_97, -7.47_97, -7.30_68]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, 0, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = AutoImageProcessor.from_pretrained('openmmlab/upernet-convnext-tiny' )
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained('openmmlab/upernet-convnext-tiny' ).to(lowerCAmelCase_ )
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = torch.Size((1, model.config.num_labels, 5_1_2, 5_1_2) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-8.81_10, -8.81_10, -8.65_21], [-8.81_10, -8.81_10, -8.65_21], [-8.77_46, -8.77_46, -8.61_30]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, 0, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Dict=2 , lowerCAmelCase_ : Optional[int]=8 , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[Any]=9_9 , lowerCAmelCase_ : List[Any]=1_6 , lowerCAmelCase_ : int=5 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : str=3_6 , lowerCAmelCase_ : Optional[int]="gelu" , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : Optional[int]=0.0 , lowerCAmelCase_ : str=5_1_2 , lowerCAmelCase_ : List[str]=1_6 , lowerCAmelCase_ : str=2 , lowerCAmelCase_ : Tuple=0.02 , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : List[Any]=4 , lowerCAmelCase_ : List[str]=None , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_input_mask
__lowerCAmelCase = use_token_type_ids
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = type_vocab_size
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = num_labels
__lowerCAmelCase = num_choices
__lowerCAmelCase = scope
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = None
if self.use_input_mask:
__lowerCAmelCase = random_attention_mask([self.batch_size, self.seq_length] )
__lowerCAmelCase = None
if self.use_token_type_ids:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
__lowerCAmelCase = ids_tensor([self.batch_size] , self.num_choices )
__lowerCAmelCase = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowercase ( self : Any ) -> Union[str, Any]:
return MraConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , )
def lowercase ( self : Dict ) -> List[Any]:
__lowerCAmelCase = self.get_config()
__lowerCAmelCase = 3_0_0
return config
def lowercase ( self : Optional[int] ) -> Union[str, Any]:
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = self.prepare_config_and_inputs()
__lowerCAmelCase = True
__lowerCAmelCase = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
encoder_hidden_states,
encoder_attention_mask,
)
def lowercase ( self : Optional[int] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , ) -> Tuple:
__lowerCAmelCase = True
__lowerCAmelCase = MraModel(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , )
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , )
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Dict ) -> str:
__lowerCAmelCase = MraForQuestionAnswering(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , start_positions=lowerCAmelCase_ , end_positions=lowerCAmelCase_ , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowercase ( self : int , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : int , lowerCAmelCase_ : Dict ) -> Any:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForTokenClassification(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowercase ( self : int , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] ) -> List[Any]:
__lowerCAmelCase = self.num_choices
__lowerCAmelCase = MraForMultipleChoice(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowercase ( self : Tuple ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'input_ids': input_ids, 'token_type_ids': token_type_ids, 'attention_mask': input_mask}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
a_ = False
a_ = False
a_ = False
a_ = False
a_ = ()
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = MraModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : Tuple ) -> List[str]:
self.config_tester.run_common_tests()
def lowercase ( self : Optional[int] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
__lowerCAmelCase = type
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Any ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*lowerCAmelCase_ )
def lowercase ( self : List[str] ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*lowerCAmelCase_ )
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = MraModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
@unittest.skip(reason='MRA does not output attentions' )
def lowercase ( self : Optional[int] ) -> Tuple:
return
@require_torch
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@slow
def lowercase ( self : Optional[Any] ) -> List[str]:
__lowerCAmelCase = MraModel.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = torch.Size((1, 2_5_6, 7_6_8) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[-0.01_40, 0.08_30, -0.03_81], [0.15_46, 0.14_02, 0.02_20], [0.11_62, 0.08_51, 0.01_65]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : int ) -> Optional[int]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 2_5_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[9.25_95, -3.60_38, 11.88_19], [9.38_69, -3.26_93, 11.09_56], [11.85_24, -3.49_38, 13.12_10]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : Any ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-4096-8-d3' )
__lowerCAmelCase = torch.arange(4_0_9_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 4_0_9_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[5.47_89, -2.35_64, 7.50_64], [7.90_67, -1.33_69, 9.96_68], [9.07_12, -1.81_06, 7.03_80]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 |
import pytest
from datasets import Dataset, DatasetDict, Features, NamedSplit, Value
from datasets.io.text import TextDatasetReader
from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : Optional[Any] ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ )
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('keep_in_memory', [False, True] )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : int ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_, keep_in_memory=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize(
'features', [
None,
{'text': 'string'},
{'text': 'int32'},
{'text': 'float32'},
], )
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Dict, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = features.copy() if features else default_expected_features
__lowerCAmelCase = (
Features({feature: Value(lowerCAmelCase_ ) for feature, dtype in features.items()} ) if features is not None else None
)
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, features=lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize('split', [None, NamedSplit('train' ), 'train', 'test'] )
def a_ ( lowerCAmelCase_ : List[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Any ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_, split=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
assert dataset.split == split if split else "train"
@pytest.mark.parametrize('path_type', [str, list] )
def a_ ( lowerCAmelCase_ : Dict, lowerCAmelCase_ : Any, lowerCAmelCase_ : Dict ):
if issubclass(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = text_path
elif issubclass(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = [text_path]
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : int, lowerCAmelCase_ : Tuple=("train",) ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ )
for split in splits:
__lowerCAmelCase = dataset_dict[split]
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('keep_in_memory', [False, True] )
def a_ ( lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : Dict ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__lowerCAmelCase = TextDatasetReader({'train': text_path}, cache_dir=lowerCAmelCase_, keep_in_memory=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize(
'features', [
None,
{'text': 'string'},
{'text': 'int32'},
{'text': 'float32'},
], )
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[Any] ):
__lowerCAmelCase = tmp_path / 'cache'
# CSV file loses col_1 string dtype information: default now is "int64" instead of "string"
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = features.copy() if features else default_expected_features
__lowerCAmelCase = (
Features({feature: Value(lowerCAmelCase_ ) for feature, dtype in features.items()} ) if features is not None else None
)
__lowerCAmelCase = TextDatasetReader({'train': text_path}, features=lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize('split', [None, NamedSplit('train' ), 'train', 'test'] )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : Optional[int] ):
if split:
__lowerCAmelCase = {split: text_path}
else:
__lowerCAmelCase = 'train'
__lowerCAmelCase = {'train': text_path, 'test': text_path}
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_, splits=list(path.keys() ) )
assert all(dataset[split].split == split for split in path.keys() )
| 53 | 1 |
import os
import unittest
from transformers import BatchEncoding
from transformers.models.bert.tokenization_bert import (
BasicTokenizer,
WordpieceTokenizer,
_is_control,
_is_punctuation,
_is_whitespace,
)
from transformers.models.prophetnet.tokenization_prophetnet import VOCAB_FILES_NAMES, ProphetNetTokenizer
from transformers.testing_utils import require_torch, slow
from ...test_tokenization_common import TokenizerTesterMixin
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = ProphetNetTokenizer
a_ = False
def lowercase ( self : Union[str, Any] ) -> List[Any]:
super().setUp()
__lowerCAmelCase = [
'[UNK]',
'[CLS]',
'[SEP]',
'[PAD]',
'[MASK]',
'want',
'##want',
'##ed',
'wa',
'un',
'runn',
'##ing',
',',
'low',
'lowest',
]
__lowerCAmelCase = 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 lowercase ( self : Tuple , lowerCAmelCase_ : str ) -> List[str]:
__lowerCAmelCase = 'UNwant\u00E9d,running'
__lowerCAmelCase = 'unwanted, running'
return input_text, output_text
def lowercase ( self : str ) -> Optional[Any]:
__lowerCAmelCase = self.tokenizer_class(self.vocab_file )
__lowerCAmelCase = tokenizer.tokenize('UNwant\u00E9d,running' )
self.assertListEqual(lowerCAmelCase_ , ['un', '##want', '##ed', ',', 'runn', '##ing'] )
self.assertListEqual(tokenizer.convert_tokens_to_ids(lowerCAmelCase_ ) , [9, 6, 7, 1_2, 1_0, 1_1] )
def lowercase ( self : Optional[Any] ) -> Any:
__lowerCAmelCase = BasicTokenizer()
self.assertListEqual(tokenizer.tokenize('ah\u535A\u63A8zz' ) , ['ah', '\u535A', '\u63A8', 'zz'] )
def lowercase ( self : List[Any] ) -> List[Any]:
__lowerCAmelCase = BasicTokenizer(do_lower_case=lowerCAmelCase_ )
self.assertListEqual(
tokenizer.tokenize(' \tHeLLo!how \n Are yoU? ' ) , ['hello', '!', 'how', 'are', 'you', '?'] )
self.assertListEqual(tokenizer.tokenize('H\u00E9llo' ) , ['hello'] )
def lowercase ( self : Tuple ) -> Any:
__lowerCAmelCase = BasicTokenizer(do_lower_case=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ )
self.assertListEqual(
tokenizer.tokenize(' \tHäLLo!how \n Are yoU? ' ) , ['hällo', '!', 'how', 'are', 'you', '?'] )
self.assertListEqual(tokenizer.tokenize('H\u00E9llo' ) , ['h\u00E9llo'] )
def lowercase ( self : Tuple ) -> Dict:
__lowerCAmelCase = BasicTokenizer(do_lower_case=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ )
self.assertListEqual(
tokenizer.tokenize(' \tHäLLo!how \n Are yoU? ' ) , ['hallo', '!', 'how', 'are', 'you', '?'] )
self.assertListEqual(tokenizer.tokenize('H\u00E9llo' ) , ['hello'] )
def lowercase ( self : Union[str, Any] ) -> Dict:
__lowerCAmelCase = BasicTokenizer(do_lower_case=lowerCAmelCase_ )
self.assertListEqual(
tokenizer.tokenize(' \tHäLLo!how \n Are yoU? ' ) , ['hallo', '!', 'how', 'are', 'you', '?'] )
self.assertListEqual(tokenizer.tokenize('H\u00E9llo' ) , ['hello'] )
def lowercase ( self : Optional[Any] ) -> int:
__lowerCAmelCase = BasicTokenizer(do_lower_case=lowerCAmelCase_ )
self.assertListEqual(
tokenizer.tokenize(' \tHeLLo!how \n Are yoU? ' ) , ['HeLLo', '!', 'how', 'Are', 'yoU', '?'] )
def lowercase ( self : Optional[int] ) -> List[Any]:
__lowerCAmelCase = BasicTokenizer(do_lower_case=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ )
self.assertListEqual(
tokenizer.tokenize(' \tHäLLo!how \n Are yoU? ' ) , ['HäLLo', '!', 'how', 'Are', 'yoU', '?'] )
def lowercase ( self : Tuple ) -> List[Any]:
__lowerCAmelCase = BasicTokenizer(do_lower_case=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ )
self.assertListEqual(
tokenizer.tokenize(' \tHäLLo!how \n Are yoU? ' ) , ['HaLLo', '!', 'how', 'Are', 'yoU', '?'] )
def lowercase ( self : str ) -> Optional[int]:
__lowerCAmelCase = BasicTokenizer(do_lower_case=lowerCAmelCase_ , never_split=['[UNK]'] )
self.assertListEqual(
tokenizer.tokenize(' \tHeLLo!how \n Are yoU? [UNK]' ) , ['HeLLo', '!', 'how', 'Are', 'yoU', '?', '[UNK]'] )
def lowercase ( self : List[str] ) -> Tuple:
__lowerCAmelCase = ['[UNK]', '[CLS]', '[SEP]', 'want', '##want', '##ed', 'wa', 'un', 'runn', '##ing']
__lowerCAmelCase = {}
for i, token in enumerate(lowerCAmelCase_ ):
__lowerCAmelCase = i
__lowerCAmelCase = WordpieceTokenizer(vocab=lowerCAmelCase_ , unk_token='[UNK]' )
self.assertListEqual(tokenizer.tokenize('' ) , [] )
self.assertListEqual(tokenizer.tokenize('unwanted running' ) , ['un', '##want', '##ed', 'runn', '##ing'] )
self.assertListEqual(tokenizer.tokenize('unwantedX running' ) , ['[UNK]', 'runn', '##ing'] )
@require_torch
def lowercase ( self : str ) -> Any:
__lowerCAmelCase = self.tokenizer_class.from_pretrained('microsoft/prophetnet-large-uncased' )
__lowerCAmelCase = ['A long paragraph for summarization.', 'Another paragraph for summarization.']
__lowerCAmelCase = [1_0_3_7, 2_1_4_6, 2_0_4_2_3, 2_0_0_5, 7_6_8_0, 7_8_4_9, 3_9_8_9, 1_0_1_2, 1_0_2]
__lowerCAmelCase = tokenizer(lowerCAmelCase_ , padding=lowerCAmelCase_ , return_tensors='pt' )
self.assertIsInstance(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = list(batch.input_ids.numpy()[0] )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual((2, 9) , batch.input_ids.shape )
self.assertEqual((2, 9) , batch.attention_mask.shape )
def lowercase ( self : Dict ) -> int:
self.assertTrue(_is_whitespace(' ' ) )
self.assertTrue(_is_whitespace('\t' ) )
self.assertTrue(_is_whitespace('\r' ) )
self.assertTrue(_is_whitespace('\n' ) )
self.assertTrue(_is_whitespace('\u00A0' ) )
self.assertFalse(_is_whitespace('A' ) )
self.assertFalse(_is_whitespace('-' ) )
def lowercase ( self : Optional[int] ) -> Optional[Any]:
self.assertTrue(_is_control('\u0005' ) )
self.assertFalse(_is_control('A' ) )
self.assertFalse(_is_control(' ' ) )
self.assertFalse(_is_control('\t' ) )
self.assertFalse(_is_control('\r' ) )
def lowercase ( self : int ) -> Any:
self.assertTrue(_is_punctuation('-' ) )
self.assertTrue(_is_punctuation('$' ) )
self.assertTrue(_is_punctuation('`' ) )
self.assertTrue(_is_punctuation('.' ) )
self.assertFalse(_is_punctuation('A' ) )
self.assertFalse(_is_punctuation(' ' ) )
@slow
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = self.tokenizer_class.from_pretrained('microsoft/prophetnet-large-uncased' )
__lowerCAmelCase = tokenizer.encode('sequence builders' , add_special_tokens=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.encode('multi-sequence build' , add_special_tokens=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.build_inputs_with_special_tokens(lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.build_inputs_with_special_tokens(lowerCAmelCase_ , lowerCAmelCase_ )
assert encoded_sentence == text + [1_0_2]
assert encoded_pair == text + [1_0_2] + text_a + [1_0_2]
| 53 |
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 ViTConfig, ViTForImageClassification, ViTImageProcessor, ViTModel
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Union[str, Any] = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : int=False ):
__lowerCAmelCase = []
for i in range(config.num_hidden_layers ):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((F"""blocks.{i}.norm1.weight""", F"""vit.encoder.layer.{i}.layernorm_before.weight""") )
rename_keys.append((F"""blocks.{i}.norm1.bias""", F"""vit.encoder.layer.{i}.layernorm_before.bias""") )
rename_keys.append((F"""blocks.{i}.attn.proj.weight""", F"""vit.encoder.layer.{i}.attention.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.attn.proj.bias""", F"""vit.encoder.layer.{i}.attention.output.dense.bias""") )
rename_keys.append((F"""blocks.{i}.norm2.weight""", F"""vit.encoder.layer.{i}.layernorm_after.weight""") )
rename_keys.append((F"""blocks.{i}.norm2.bias""", F"""vit.encoder.layer.{i}.layernorm_after.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.weight""", F"""vit.encoder.layer.{i}.intermediate.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.bias""", F"""vit.encoder.layer.{i}.intermediate.dense.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.weight""", F"""vit.encoder.layer.{i}.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.bias""", F"""vit.encoder.layer.{i}.output.dense.bias""") )
# projection layer + position embeddings
rename_keys.extend(
[
('cls_token', 'vit.embeddings.cls_token'),
('patch_embed.proj.weight', 'vit.embeddings.patch_embeddings.projection.weight'),
('patch_embed.proj.bias', 'vit.embeddings.patch_embeddings.projection.bias'),
('pos_embed', 'vit.embeddings.position_embeddings'),
] )
if base_model:
# layernorm + pooler
rename_keys.extend(
[
('norm.weight', 'layernorm.weight'),
('norm.bias', 'layernorm.bias'),
] )
# if just the base model, we should remove "vit" from all keys that start with "vit"
__lowerCAmelCase = [(pair[0], pair[1][4:]) if pair[1].startswith('vit' ) else pair for pair in rename_keys]
else:
# layernorm + classification head
rename_keys.extend(
[
('norm.weight', 'vit.layernorm.weight'),
('norm.bias', 'vit.layernorm.bias'),
('head.weight', 'classifier.weight'),
('head.bias', 'classifier.bias'),
] )
return rename_keys
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Optional[int]=False ):
for i in range(config.num_hidden_layers ):
if base_model:
__lowerCAmelCase = ''
else:
__lowerCAmelCase = 'vit.'
# read in weights + bias of input projection layer (in timm, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[
: config.hidden_size, :
]
__lowerCAmelCase = in_proj_bias[: config.hidden_size]
__lowerCAmelCase = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
__lowerCAmelCase = in_proj_bias[
config.hidden_size : config.hidden_size * 2
]
__lowerCAmelCase = in_proj_weight[
-config.hidden_size :, :
]
__lowerCAmelCase = in_proj_bias[-config.hidden_size :]
def a_ ( lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = ['head.weight', 'head.bias']
for k in ignore_keys:
state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Optional[Any]=True ):
__lowerCAmelCase = ViTConfig()
# patch_size
if model_name[-1] == "8":
__lowerCAmelCase = 8
# set labels if required
if not base_model:
__lowerCAmelCase = 1000
__lowerCAmelCase = 'huggingface/label-files'
__lowerCAmelCase = 'imagenet-1k-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
__lowerCAmelCase = idalabel
__lowerCAmelCase = {v: k for k, v in idalabel.items()}
# size of the architecture
if model_name in ["dino_vits8", "dino_vits16"]:
__lowerCAmelCase = 384
__lowerCAmelCase = 1536
__lowerCAmelCase = 12
__lowerCAmelCase = 6
# load original model from torch hub
__lowerCAmelCase = torch.hub.load('facebookresearch/dino:main', lowerCAmelCase_ )
original_model.eval()
# load state_dict of original model, remove and rename some keys
__lowerCAmelCase = original_model.state_dict()
if base_model:
remove_classification_head_(lowerCAmelCase_ )
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_, base_model=lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_q_k_v(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# load HuggingFace model
if base_model:
__lowerCAmelCase = ViTModel(lowerCAmelCase_, add_pooling_layer=lowerCAmelCase_ ).eval()
else:
__lowerCAmelCase = ViTForImageClassification(lowerCAmelCase_ ).eval()
model.load_state_dict(lowerCAmelCase_ )
# Check outputs on an image, prepared by ViTImageProcessor
__lowerCAmelCase = ViTImageProcessor()
__lowerCAmelCase = image_processor(images=prepare_img(), return_tensors='pt' )
__lowerCAmelCase = encoding['pixel_values']
__lowerCAmelCase = model(lowerCAmelCase_ )
if base_model:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert torch.allclose(lowerCAmelCase_, outputs.last_hidden_state[:, 0, :], atol=1E-1 )
else:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert logits.shape == outputs.logits.shape
assert torch.allclose(lowerCAmelCase_, outputs.logits, atol=1E-3 )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
print(F"""Saving model {model_name} to {pytorch_dump_folder_path}""" )
model.save_pretrained(lowerCAmelCase_ )
print(F"""Saving image processor to {pytorch_dump_folder_path}""" )
image_processor.save_pretrained(lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : Dict = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='dino_vitb16',
type=str,
help='Name of the model trained with DINO you\'d like to convert.',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--base_model',
action='store_true',
help='Whether to only convert the base model (no projection head weights).',
)
parser.set_defaults(base_model=True)
_snake_case : List[Any] = parser.parse_args()
convert_vit_checkpoint(args.model_name, args.pytorch_dump_folder_path, args.base_model)
| 53 | 1 |
import unittest
from datasets import load_dataset
from transformers import BloomTokenizerFast
from transformers.testing_utils import require_tokenizers
from ...test_tokenization_common import TokenizerTesterMixin
@require_tokenizers
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = None
a_ = BloomTokenizerFast
a_ = BloomTokenizerFast
a_ = True
a_ = False
a_ = """tokenizer_file"""
a_ = {"""bos_token""": """<s>""", """eos_token""": """</s>""", """unk_token""": """<unk>""", """pad_token""": """<pad>"""}
def lowercase ( self : List[Any] ) -> List[str]:
super().setUp()
__lowerCAmelCase = BloomTokenizerFast.from_pretrained('bigscience/tokenizer' )
tokenizer.save_pretrained(self.tmpdirname )
def lowercase ( self : Dict , **lowerCAmelCase_ : Optional[Any] ) -> List[Any]:
kwargs.update(self.special_tokens_map )
return BloomTokenizerFast.from_pretrained(self.tmpdirname , **lowerCAmelCase_ )
def lowercase ( self : str ) -> Dict:
__lowerCAmelCase = self.get_rust_tokenizer()
__lowerCAmelCase = ['The quick brown fox</s>', 'jumps over the lazy dog</s>']
__lowerCAmelCase = [[2_1_7_5, 2_3_7_1_4, 7_3_1_7_3, 1_4_4_2_5_2, 2], [7_7, 1_3_2_6_1_9, 3_4_7_8, 3_6_8, 1_0_9_5_8_6, 3_5_4_3_3, 2]]
__lowerCAmelCase = tokenizer.batch_encode_plus(lowerCAmelCase_ )['input_ids']
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.batch_decode(lowerCAmelCase_ )
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : List[Any]=6 ) -> int:
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(f"""{tokenizer.__class__.__name__} ({pretrained_name})""" ):
__lowerCAmelCase = self.rust_tokenizer_class.from_pretrained(lowerCAmelCase_ , **lowerCAmelCase_ )
# tokenizer_r.pad_token = None # Hotfixing padding = None
# Simple input
__lowerCAmelCase = 'This is a simple input'
__lowerCAmelCase = ['This is a simple input 1', 'This is a simple input 2']
__lowerCAmelCase = ('This is a simple input', 'This is a pair')
__lowerCAmelCase = [
('This is a simple input 1', 'This is a simple input 2'),
('This is a simple pair 1', 'This is a simple pair 2'),
]
# Simple input tests
try:
tokenizer_r.encode(lowerCAmelCase_ , max_length=lowerCAmelCase_ )
tokenizer_r.encode_plus(lowerCAmelCase_ , max_length=lowerCAmelCase_ )
tokenizer_r.batch_encode_plus(lowerCAmelCase_ , max_length=lowerCAmelCase_ )
tokenizer_r.encode(lowerCAmelCase_ , max_length=lowerCAmelCase_ )
tokenizer_r.batch_encode_plus(lowerCAmelCase_ , max_length=lowerCAmelCase_ )
except ValueError:
self.fail('Bloom Tokenizer should be able to deal with padding' )
__lowerCAmelCase = None # Hotfixing padding = None
self.assertRaises(lowerCAmelCase_ , tokenizer_r.encode , lowerCAmelCase_ , max_length=lowerCAmelCase_ , padding='max_length' )
# Simple input
self.assertRaises(lowerCAmelCase_ , tokenizer_r.encode_plus , lowerCAmelCase_ , max_length=lowerCAmelCase_ , padding='max_length' )
# Simple input
self.assertRaises(
lowerCAmelCase_ , tokenizer_r.batch_encode_plus , lowerCAmelCase_ , max_length=lowerCAmelCase_ , padding='max_length' , )
# Pair input
self.assertRaises(lowerCAmelCase_ , tokenizer_r.encode , lowerCAmelCase_ , max_length=lowerCAmelCase_ , padding='max_length' )
# Pair input
self.assertRaises(lowerCAmelCase_ , tokenizer_r.encode_plus , lowerCAmelCase_ , max_length=lowerCAmelCase_ , padding='max_length' )
# Pair input
self.assertRaises(
lowerCAmelCase_ , tokenizer_r.batch_encode_plus , lowerCAmelCase_ , max_length=lowerCAmelCase_ , padding='max_length' , )
def lowercase ( self : List[str] ) -> Optional[int]:
__lowerCAmelCase = self.get_rust_tokenizer()
__lowerCAmelCase = load_dataset('xnli' , 'all_languages' , split='test' , streaming=lowerCAmelCase_ )
__lowerCAmelCase = next(iter(lowerCAmelCase_ ) )['premise'] # pick up one data
__lowerCAmelCase = list(sample_data.values() )
__lowerCAmelCase = list(map(tokenizer.encode , lowerCAmelCase_ ) )
__lowerCAmelCase = [tokenizer.decode(lowerCAmelCase_ , clean_up_tokenization_spaces=lowerCAmelCase_ ) for x in output_tokens]
self.assertListEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
# The test has to be overriden because BLOOM uses ALiBi positional embeddings that does not have
# any sequence length constraints. This test of the parent class will fail since it relies on the
# maximum sequence length of the positoonal embeddings.
self.assertGreaterEqual(len(self.tokenizer_class.pretrained_vocab_files_map ) , 1 )
self.assertGreaterEqual(len(list(self.tokenizer_class.pretrained_vocab_files_map.values() )[0] ) , 1 )
| 53 |
import unittest
from diffusers.pipelines.pipeline_utils import is_safetensors_compatible
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Union[str, Any] ) -> List[str]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Any:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
# Removed: 'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> int:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
# Removed: 'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> str:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[str]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Union[str, Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
# Removed: 'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : List[str] ) -> List[Any]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : Optional[Any] ) -> Optional[Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
# 'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
| 53 | 1 |
def a_ ( lowerCAmelCase_ : list[int] ):
__lowerCAmelCase = len(lowerCAmelCase_ )
for i in range(lowerCAmelCase_ ):
for j in range(i + 1, lowerCAmelCase_ ):
if numbers[j] < numbers[i]:
__lowerCAmelCase , __lowerCAmelCase = numbers[j], numbers[i]
return numbers
if __name__ == "__main__":
_snake_case : Any = input('Enter numbers separated by a comma:\n').strip()
_snake_case : Dict = [int(item) for item in user_input.split(',')]
print(exchange_sort(unsorted))
| 53 |
import math
def a_ ( lowerCAmelCase_ : list, lowerCAmelCase_ : int ):
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = int(math.floor(math.sqrt(lowerCAmelCase_ ) ) )
__lowerCAmelCase = 0
while arr[min(lowerCAmelCase_, lowerCAmelCase_ ) - 1] < x:
__lowerCAmelCase = step
step += int(math.floor(math.sqrt(lowerCAmelCase_ ) ) )
if prev >= n:
return -1
while arr[prev] < x:
__lowerCAmelCase = prev + 1
if prev == min(lowerCAmelCase_, lowerCAmelCase_ ):
return -1
if arr[prev] == x:
return prev
return -1
if __name__ == "__main__":
_snake_case : List[str] = input('Enter numbers separated by a comma:\n').strip()
_snake_case : Optional[Any] = [int(item) for item in user_input.split(',')]
_snake_case : List[str] = int(input('Enter the number to be searched:\n'))
_snake_case : Optional[int] = jump_search(arr, x)
if res == -1:
print('Number not found!')
else:
print(F"""Number {x} is at index {res}""")
| 53 | 1 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
is_vision_available,
)
_snake_case : int = {
'configuration_clip': [
'CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP',
'CLIPConfig',
'CLIPOnnxConfig',
'CLIPTextConfig',
'CLIPVisionConfig',
],
'processing_clip': ['CLIPProcessor'],
'tokenization_clip': ['CLIPTokenizer'],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : Optional[int] = ['CLIPTokenizerFast']
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : Optional[int] = ['CLIPFeatureExtractor']
_snake_case : Optional[Any] = ['CLIPImageProcessor']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : str = [
'CLIP_PRETRAINED_MODEL_ARCHIVE_LIST',
'CLIPModel',
'CLIPPreTrainedModel',
'CLIPTextModel',
'CLIPTextModelWithProjection',
'CLIPVisionModel',
'CLIPVisionModelWithProjection',
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : List[Any] = [
'TF_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST',
'TFCLIPModel',
'TFCLIPPreTrainedModel',
'TFCLIPTextModel',
'TFCLIPVisionModel',
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : Optional[int] = [
'FlaxCLIPModel',
'FlaxCLIPPreTrainedModel',
'FlaxCLIPTextModel',
'FlaxCLIPTextPreTrainedModel',
'FlaxCLIPVisionModel',
'FlaxCLIPVisionPreTrainedModel',
]
if TYPE_CHECKING:
from .configuration_clip import (
CLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
CLIPConfig,
CLIPOnnxConfig,
CLIPTextConfig,
CLIPVisionConfig,
)
from .processing_clip import CLIPProcessor
from .tokenization_clip import CLIPTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_clip_fast import CLIPTokenizerFast
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .feature_extraction_clip import CLIPFeatureExtractor
from .image_processing_clip import CLIPImageProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_clip import (
CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
CLIPModel,
CLIPPreTrainedModel,
CLIPTextModel,
CLIPTextModelWithProjection,
CLIPVisionModel,
CLIPVisionModelWithProjection,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_clip import (
TF_CLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
TFCLIPModel,
TFCLIPPreTrainedModel,
TFCLIPTextModel,
TFCLIPVisionModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_clip import (
FlaxCLIPModel,
FlaxCLIPPreTrainedModel,
FlaxCLIPTextModel,
FlaxCLIPTextPreTrainedModel,
FlaxCLIPVisionModel,
FlaxCLIPVisionPreTrainedModel,
)
else:
import sys
_snake_case : Dict = _LazyModule(__name__, globals()['__file__'], _import_structure, module_spec=__spec__)
| 53 |
import argparse
import torch
from transformers import RemBertConfig, RemBertModel, load_tf_weights_in_rembert
from transformers.utils import logging
logging.set_verbosity_info()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[Any], lowerCAmelCase_ : str ):
# Initialise PyTorch model
__lowerCAmelCase = RemBertConfig.from_json_file(lowerCAmelCase_ )
print('Building PyTorch model from configuration: {}'.format(str(lowerCAmelCase_ ) ) )
__lowerCAmelCase = RemBertModel(lowerCAmelCase_ )
# Load weights from tf checkpoint
load_tf_weights_in_rembert(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# Save pytorch-model
print('Save PyTorch model to {}'.format(lowerCAmelCase_ ) )
torch.save(model.state_dict(), lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--tf_checkpoint_path', default=None, type=str, required=True, help='Path to the TensorFlow checkpoint path.'
)
parser.add_argument(
'--rembert_config_file',
default=None,
type=str,
required=True,
help=(
'The config json file corresponding to the pre-trained RemBERT model. \n'
'This specifies the model architecture.'
),
)
parser.add_argument(
'--pytorch_dump_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_rembert_tf_checkpoint_to_pytorch(args.tf_checkpoint_path, args.rembert_config_file, args.pytorch_dump_path)
| 53 | 1 |
from dataclasses import dataclass
from typing import Dict, Optional, Tuple, Union
import torch
import torch.nn as nn
from ..configuration_utils import ConfigMixin, register_to_config
from ..utils import BaseOutput, apply_forward_hook
from .attention_processor import AttentionProcessor, AttnProcessor
from .modeling_utils import ModelMixin
from .vae import Decoder, DecoderOutput, DiagonalGaussianDistribution, Encoder
@dataclass
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = 42
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = True
@register_to_config
def __init__( self : Tuple , lowerCAmelCase_ : int = 3 , lowerCAmelCase_ : int = 3 , lowerCAmelCase_ : Tuple[str] = ("DownEncoderBlock2D",) , lowerCAmelCase_ : Tuple[str] = ("UpDecoderBlock2D",) , lowerCAmelCase_ : Tuple[int] = (6_4,) , lowerCAmelCase_ : int = 1 , lowerCAmelCase_ : str = "silu" , lowerCAmelCase_ : int = 4 , lowerCAmelCase_ : int = 3_2 , lowerCAmelCase_ : int = 3_2 , lowerCAmelCase_ : float = 0.1_82_15 , ) -> Union[str, Any]:
super().__init__()
# pass init params to Encoder
__lowerCAmelCase = Encoder(
in_channels=lowerCAmelCase_ , out_channels=lowerCAmelCase_ , down_block_types=lowerCAmelCase_ , block_out_channels=lowerCAmelCase_ , layers_per_block=lowerCAmelCase_ , act_fn=lowerCAmelCase_ , norm_num_groups=lowerCAmelCase_ , double_z=lowerCAmelCase_ , )
# pass init params to Decoder
__lowerCAmelCase = Decoder(
in_channels=lowerCAmelCase_ , out_channels=lowerCAmelCase_ , up_block_types=lowerCAmelCase_ , block_out_channels=lowerCAmelCase_ , layers_per_block=lowerCAmelCase_ , norm_num_groups=lowerCAmelCase_ , act_fn=lowerCAmelCase_ , )
__lowerCAmelCase = nn.Convad(2 * latent_channels , 2 * latent_channels , 1 )
__lowerCAmelCase = nn.Convad(lowerCAmelCase_ , lowerCAmelCase_ , 1 )
__lowerCAmelCase = False
__lowerCAmelCase = False
# only relevant if vae tiling is enabled
__lowerCAmelCase = self.config.sample_size
__lowerCAmelCase = (
self.config.sample_size[0]
if isinstance(self.config.sample_size , (list, tuple) )
else self.config.sample_size
)
__lowerCAmelCase = int(sample_size / (2 ** (len(self.config.block_out_channels ) - 1)) )
__lowerCAmelCase = 0.25
def lowercase ( self : Dict , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Union[str, Any]=False ) -> int:
if isinstance(lowerCAmelCase_ , (Encoder, Decoder) ):
__lowerCAmelCase = value
def lowercase ( self : Dict , lowerCAmelCase_ : bool = True ) -> Optional[int]:
__lowerCAmelCase = use_tiling
def lowercase ( self : Optional[Any] ) -> Any:
self.enable_tiling(lowerCAmelCase_ )
def lowercase ( self : List[str] ) -> Optional[int]:
__lowerCAmelCase = True
def lowercase ( self : Dict ) -> Optional[int]:
__lowerCAmelCase = False
@property
# Copied from diffusers.models.unet_2d_condition.UNet2DConditionModel.attn_processors
def lowercase ( self : List[str] ) -> Dict[str, AttentionProcessor]:
__lowerCAmelCase = {}
def fn_recursive_add_processors(lowerCAmelCase_ : str , lowerCAmelCase_ : torch.nn.Module , lowerCAmelCase_ : Dict[str, AttentionProcessor] ):
if hasattr(lowerCAmelCase_ , 'set_processor' ):
__lowerCAmelCase = module.processor
for sub_name, child in module.named_children():
fn_recursive_add_processors(f"""{name}.{sub_name}""" , lowerCAmelCase_ , lowerCAmelCase_ )
return processors
for name, module in self.named_children():
fn_recursive_add_processors(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
return processors
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Union[AttentionProcessor, Dict[str, AttentionProcessor]] ) -> str:
__lowerCAmelCase = len(self.attn_processors.keys() )
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ) and len(lowerCAmelCase_ ) != count:
raise ValueError(
f"""A dict of processors was passed, but the number of processors {len(lowerCAmelCase_ )} does not match the"""
f""" number of attention layers: {count}. Please make sure to pass {count} processor classes.""" )
def fn_recursive_attn_processor(lowerCAmelCase_ : str , lowerCAmelCase_ : torch.nn.Module , lowerCAmelCase_ : Optional[int] ):
if hasattr(lowerCAmelCase_ , 'set_processor' ):
if not isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
module.set_processor(lowerCAmelCase_ )
else:
module.set_processor(processor.pop(f"""{name}.processor""" ) )
for sub_name, child in module.named_children():
fn_recursive_attn_processor(f"""{name}.{sub_name}""" , lowerCAmelCase_ , lowerCAmelCase_ )
for name, module in self.named_children():
fn_recursive_attn_processor(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
self.set_attn_processor(AttnProcessor() )
@apply_forward_hook
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = True ) -> AutoencoderKLOutput:
if self.use_tiling and (x.shape[-1] > self.tile_sample_min_size or x.shape[-2] > self.tile_sample_min_size):
return self.tiled_encode(lowerCAmelCase_ , return_dict=lowerCAmelCase_ )
if self.use_slicing and x.shape[0] > 1:
__lowerCAmelCase = [self.encoder(lowerCAmelCase_ ) for x_slice in x.split(1 )]
__lowerCAmelCase = torch.cat(lowerCAmelCase_ )
else:
__lowerCAmelCase = self.encoder(lowerCAmelCase_ )
__lowerCAmelCase = self.quant_conv(lowerCAmelCase_ )
__lowerCAmelCase = DiagonalGaussianDistribution(lowerCAmelCase_ )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=lowerCAmelCase_ )
def lowercase ( self : str , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = True ) -> Union[DecoderOutput, torch.FloatTensor]:
if self.use_tiling and (z.shape[-1] > self.tile_latent_min_size or z.shape[-2] > self.tile_latent_min_size):
return self.tiled_decode(lowerCAmelCase_ , return_dict=lowerCAmelCase_ )
__lowerCAmelCase = self.post_quant_conv(lowerCAmelCase_ )
__lowerCAmelCase = self.decoder(lowerCAmelCase_ )
if not return_dict:
return (dec,)
return DecoderOutput(sample=lowerCAmelCase_ )
@apply_forward_hook
def lowercase ( self : str , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = True ) -> Union[DecoderOutput, torch.FloatTensor]:
if self.use_slicing and z.shape[0] > 1:
__lowerCAmelCase = [self._decode(lowerCAmelCase_ ).sample for z_slice in z.split(1 )]
__lowerCAmelCase = torch.cat(lowerCAmelCase_ )
else:
__lowerCAmelCase = self._decode(lowerCAmelCase_ ).sample
if not return_dict:
return (decoded,)
return DecoderOutput(sample=lowerCAmelCase_ )
def lowercase ( self : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[int] ) -> Any:
__lowerCAmelCase = min(a.shape[2] , b.shape[2] , lowerCAmelCase_ )
for y in range(lowerCAmelCase_ ):
__lowerCAmelCase = a[:, :, -blend_extent + y, :] * (1 - y / blend_extent) + b[:, :, y, :] * (y / blend_extent)
return b
def lowercase ( self : List[Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple ) -> Tuple:
__lowerCAmelCase = min(a.shape[3] , b.shape[3] , lowerCAmelCase_ )
for x in range(lowerCAmelCase_ ):
__lowerCAmelCase = a[:, :, :, -blend_extent + x] * (1 - x / blend_extent) + b[:, :, :, x] * (x / blend_extent)
return b
def lowercase ( self : int , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = True ) -> AutoencoderKLOutput:
__lowerCAmelCase = int(self.tile_sample_min_size * (1 - self.tile_overlap_factor) )
__lowerCAmelCase = int(self.tile_latent_min_size * self.tile_overlap_factor )
__lowerCAmelCase = self.tile_latent_min_size - blend_extent
# Split the image into 512x512 tiles and encode them separately.
__lowerCAmelCase = []
for i in range(0 , x.shape[2] , lowerCAmelCase_ ):
__lowerCAmelCase = []
for j in range(0 , x.shape[3] , lowerCAmelCase_ ):
__lowerCAmelCase = x[:, :, i : i + self.tile_sample_min_size, j : j + self.tile_sample_min_size]
__lowerCAmelCase = self.encoder(lowerCAmelCase_ )
__lowerCAmelCase = self.quant_conv(lowerCAmelCase_ )
row.append(lowerCAmelCase_ )
rows.append(lowerCAmelCase_ )
__lowerCAmelCase = []
for i, row in enumerate(lowerCAmelCase_ ):
__lowerCAmelCase = []
for j, tile in enumerate(lowerCAmelCase_ ):
# blend the above tile and the left tile
# to the current tile and add the current tile to the result row
if i > 0:
__lowerCAmelCase = self.blend_v(rows[i - 1][j] , lowerCAmelCase_ , lowerCAmelCase_ )
if j > 0:
__lowerCAmelCase = self.blend_h(row[j - 1] , lowerCAmelCase_ , lowerCAmelCase_ )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(lowerCAmelCase_ , dim=3 ) )
__lowerCAmelCase = torch.cat(lowerCAmelCase_ , dim=2 )
__lowerCAmelCase = DiagonalGaussianDistribution(lowerCAmelCase_ )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=lowerCAmelCase_ )
def lowercase ( self : List[Any] , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = True ) -> Union[DecoderOutput, torch.FloatTensor]:
__lowerCAmelCase = int(self.tile_latent_min_size * (1 - self.tile_overlap_factor) )
__lowerCAmelCase = int(self.tile_sample_min_size * self.tile_overlap_factor )
__lowerCAmelCase = self.tile_sample_min_size - blend_extent
# Split z into overlapping 64x64 tiles and decode them separately.
# The tiles have an overlap to avoid seams between tiles.
__lowerCAmelCase = []
for i in range(0 , z.shape[2] , lowerCAmelCase_ ):
__lowerCAmelCase = []
for j in range(0 , z.shape[3] , lowerCAmelCase_ ):
__lowerCAmelCase = z[:, :, i : i + self.tile_latent_min_size, j : j + self.tile_latent_min_size]
__lowerCAmelCase = self.post_quant_conv(lowerCAmelCase_ )
__lowerCAmelCase = self.decoder(lowerCAmelCase_ )
row.append(lowerCAmelCase_ )
rows.append(lowerCAmelCase_ )
__lowerCAmelCase = []
for i, row in enumerate(lowerCAmelCase_ ):
__lowerCAmelCase = []
for j, tile in enumerate(lowerCAmelCase_ ):
# blend the above tile and the left tile
# to the current tile and add the current tile to the result row
if i > 0:
__lowerCAmelCase = self.blend_v(rows[i - 1][j] , lowerCAmelCase_ , lowerCAmelCase_ )
if j > 0:
__lowerCAmelCase = self.blend_h(row[j - 1] , lowerCAmelCase_ , lowerCAmelCase_ )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(lowerCAmelCase_ , dim=3 ) )
__lowerCAmelCase = torch.cat(lowerCAmelCase_ , dim=2 )
if not return_dict:
return (dec,)
return DecoderOutput(sample=lowerCAmelCase_ )
def lowercase ( self : Tuple , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = False , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Optional[torch.Generator] = None , ) -> Union[DecoderOutput, torch.FloatTensor]:
__lowerCAmelCase = sample
__lowerCAmelCase = self.encode(lowerCAmelCase_ ).latent_dist
if sample_posterior:
__lowerCAmelCase = posterior.sample(generator=lowerCAmelCase_ )
else:
__lowerCAmelCase = posterior.mode()
__lowerCAmelCase = self.decode(lowerCAmelCase_ ).sample
if not return_dict:
return (dec,)
return DecoderOutput(sample=lowerCAmelCase_ )
| 53 |
import argparse
import json
import pickle
from pathlib import Path
import requests
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import MaskFormerConfig, MaskFormerForInstanceSegmentation, MaskFormerImageProcessor, SwinConfig
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Any = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = SwinConfig.from_pretrained(
'microsoft/swin-tiny-patch4-window7-224', out_features=['stage1', 'stage2', 'stage3', 'stage4'] )
__lowerCAmelCase = MaskFormerConfig(backbone_config=lowerCAmelCase_ )
__lowerCAmelCase = 'huggingface/label-files'
if "ade20k-full" in model_name:
# this should be ok
__lowerCAmelCase = 847
__lowerCAmelCase = 'maskformer-ade20k-full-id2label.json'
elif "ade" in model_name:
# this should be ok
__lowerCAmelCase = 150
__lowerCAmelCase = 'ade20k-id2label.json'
elif "coco-stuff" in model_name:
# this should be ok
__lowerCAmelCase = 171
__lowerCAmelCase = 'maskformer-coco-stuff-id2label.json'
elif "coco" in model_name:
# TODO
__lowerCAmelCase = 133
__lowerCAmelCase = 'coco-panoptic-id2label.json'
elif "cityscapes" in model_name:
# this should be ok
__lowerCAmelCase = 19
__lowerCAmelCase = 'cityscapes-id2label.json'
elif "vistas" in model_name:
# this should be ok
__lowerCAmelCase = 65
__lowerCAmelCase = 'mapillary-vistas-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
return config
def a_ ( lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = []
# stem
# fmt: off
rename_keys.append(('backbone.patch_embed.proj.weight', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.weight') )
rename_keys.append(('backbone.patch_embed.proj.bias', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.bias') )
rename_keys.append(('backbone.patch_embed.norm.weight', 'model.pixel_level_module.encoder.model.embeddings.norm.weight') )
rename_keys.append(('backbone.patch_embed.norm.bias', 'model.pixel_level_module.encoder.model.embeddings.norm.bias') )
# stages
for i in range(len(config.backbone_config.depths ) ):
for j in range(config.backbone_config.depths[i] ):
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_bias_table""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_bias_table""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_index""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_index""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.bias""") )
if i < 3:
rename_keys.append((F"""backbone.layers.{i}.downsample.reduction.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.reduction.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.bias""") )
rename_keys.append((F"""backbone.norm{i}.weight""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.weight""") )
rename_keys.append((F"""backbone.norm{i}.bias""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.bias""") )
# FPN
rename_keys.append(('sem_seg_head.layer_4.weight', 'model.pixel_level_module.decoder.fpn.stem.0.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.weight', 'model.pixel_level_module.decoder.fpn.stem.1.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.bias', 'model.pixel_level_module.decoder.fpn.stem.1.bias') )
for source_index, target_index in zip(range(3, 0, -1 ), range(0, 3 ) ):
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.0.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.bias""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.0.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.bias""") )
rename_keys.append(('sem_seg_head.mask_features.weight', 'model.pixel_level_module.decoder.mask_projection.weight') )
rename_keys.append(('sem_seg_head.mask_features.bias', 'model.pixel_level_module.decoder.mask_projection.bias') )
# Transformer decoder
for idx in range(config.decoder_config.decoder_layers ):
# self-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.bias""") )
# cross-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.bias""") )
# MLP 1
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc1.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc1.bias""") )
# MLP 2
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc2.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc2.bias""") )
# layernorm 1 (self-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.bias""") )
# layernorm 2 (cross-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.bias""") )
# layernorm 3 (final layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.weight""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.bias""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.bias""") )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.weight', 'model.transformer_module.decoder.layernorm.weight') )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.bias', 'model.transformer_module.decoder.layernorm.bias') )
# heads on top
rename_keys.append(('sem_seg_head.predictor.query_embed.weight', 'model.transformer_module.queries_embedder.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.weight', 'model.transformer_module.input_projection.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.bias', 'model.transformer_module.input_projection.bias') )
rename_keys.append(('sem_seg_head.predictor.class_embed.weight', 'class_predictor.weight') )
rename_keys.append(('sem_seg_head.predictor.class_embed.bias', 'class_predictor.bias') )
for i in range(3 ):
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.weight""", F"""mask_embedder.{i}.0.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.bias""", F"""mask_embedder.{i}.0.bias""") )
# fmt: on
return rename_keys
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int ):
__lowerCAmelCase = [int(backbone_config.embed_dim * 2**i ) for i in range(len(backbone_config.depths ) )]
for i in range(len(backbone_config.depths ) ):
__lowerCAmelCase = num_features[i]
for j in range(backbone_config.depths[i] ):
# fmt: off
# read in weights + bias of input projection layer (in original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[:dim, :]
__lowerCAmelCase = in_proj_bias[: dim]
__lowerCAmelCase = in_proj_weight[
dim : dim * 2, :
]
__lowerCAmelCase = in_proj_bias[
dim : dim * 2
]
__lowerCAmelCase = in_proj_weight[
-dim :, :
]
__lowerCAmelCase = in_proj_bias[-dim :]
# fmt: on
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : Dict ):
# fmt: off
__lowerCAmelCase = config.decoder_config.hidden_size
for idx in range(config.decoder_config.decoder_layers ):
# read in weights + bias of self-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# read in weights + bias of cross-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# fmt: on
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : bool = False ):
__lowerCAmelCase = get_maskformer_config(lowerCAmelCase_ )
# load original state_dict
with open(lowerCAmelCase_, 'rb' ) as f:
__lowerCAmelCase = pickle.load(lowerCAmelCase_ )
__lowerCAmelCase = data['model']
# for name, param in state_dict.items():
# print(name, param.shape)
# rename keys
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_swin_q_k_v(lowerCAmelCase_, config.backbone_config )
read_in_decoder_q_k_v(lowerCAmelCase_, lowerCAmelCase_ )
# update to torch tensors
for key, value in state_dict.items():
__lowerCAmelCase = torch.from_numpy(lowerCAmelCase_ )
# load 🤗 model
__lowerCAmelCase = MaskFormerForInstanceSegmentation(lowerCAmelCase_ )
model.eval()
for name, param in model.named_parameters():
print(lowerCAmelCase_, param.shape )
__lowerCAmelCase , __lowerCAmelCase = model.load_state_dict(lowerCAmelCase_, strict=lowerCAmelCase_ )
assert missing_keys == [
"model.pixel_level_module.encoder.model.layernorm.weight",
"model.pixel_level_module.encoder.model.layernorm.bias",
]
assert len(lowerCAmelCase_ ) == 0, F"""Unexpected keys: {unexpected_keys}"""
# verify results
__lowerCAmelCase = prepare_img()
if "vistas" in model_name:
__lowerCAmelCase = 65
elif "cityscapes" in model_name:
__lowerCAmelCase = 6_5535
else:
__lowerCAmelCase = 255
__lowerCAmelCase = True if 'ade' in model_name else False
__lowerCAmelCase = MaskFormerImageProcessor(ignore_index=lowerCAmelCase_, reduce_labels=lowerCAmelCase_ )
__lowerCAmelCase = image_processor(lowerCAmelCase_, return_tensors='pt' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
print('Logits:', outputs.class_queries_logits[0, :3, :3] )
if model_name == "maskformer-swin-tiny-ade":
__lowerCAmelCase = torch.tensor(
[[3.6353, -4.4770, -2.6065], [0.5081, -4.2394, -3.5343], [2.1909, -5.0353, -1.9323]] )
assert torch.allclose(outputs.class_queries_logits[0, :3, :3], lowerCAmelCase_, atol=1E-4 )
print('Looks ok!' )
if pytorch_dump_folder_path is not None:
print(F"""Saving model and image processor to {pytorch_dump_folder_path}""" )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
model.save_pretrained(lowerCAmelCase_ )
image_processor.save_pretrained(lowerCAmelCase_ )
if push_to_hub:
print('Pushing model and image processor to the hub...' )
model.push_to_hub(F"""nielsr/{model_name}""" )
image_processor.push_to_hub(F"""nielsr/{model_name}""" )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='maskformer-swin-tiny-ade',
type=str,
help=('Name of the MaskFormer model you\'d like to convert',),
)
parser.add_argument(
'--checkpoint_path',
default='/Users/nielsrogge/Documents/MaskFormer_checkpoints/MaskFormer-Swin-tiny-ADE20k/model.pkl',
type=str,
help='Path to the original state dict (.pth file).',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--push_to_hub', action='store_true', help='Whether or not to push the converted model to the 🤗 hub.'
)
_snake_case : List[str] = parser.parse_args()
convert_maskformer_checkpoint(
args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub
)
| 53 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
_snake_case : Optional[int] = logging.get_logger(__name__)
_snake_case : Union[str, Any] = {
'MIT/ast-finetuned-audioset-10-10-0.4593': (
'https://huggingface.co/MIT/ast-finetuned-audioset-10-10-0.4593/resolve/main/config.json'
),
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = """audio-spectrogram-transformer"""
def __init__( self : Union[str, Any] , lowerCAmelCase_ : int=7_6_8 , lowerCAmelCase_ : List[Any]=1_2 , lowerCAmelCase_ : int=1_2 , lowerCAmelCase_ : Union[str, Any]=3_0_7_2 , lowerCAmelCase_ : Union[str, Any]="gelu" , lowerCAmelCase_ : Dict=0.0 , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : List[Any]=0.02 , lowerCAmelCase_ : Optional[Any]=1e-12 , lowerCAmelCase_ : Union[str, Any]=1_6 , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : str=1_0 , lowerCAmelCase_ : List[Any]=1_0 , lowerCAmelCase_ : Dict=1_0_2_4 , lowerCAmelCase_ : Union[str, Any]=1_2_8 , **lowerCAmelCase_ : str , ) -> Any:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = initializer_range
__lowerCAmelCase = layer_norm_eps
__lowerCAmelCase = patch_size
__lowerCAmelCase = qkv_bias
__lowerCAmelCase = frequency_stride
__lowerCAmelCase = time_stride
__lowerCAmelCase = max_length
__lowerCAmelCase = num_mel_bins
| 53 |
import json
import logging
import os
import re
import sys
from dataclasses import dataclass, field
from typing import Any, Dict, List, Optional, Union
import datasets
import numpy as np
import torch
import torchaudio
from packaging import version
from torch import nn
import transformers
from transformers import (
HfArgumentParser,
Trainer,
TrainingArguments,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaForCTC,
WavaVecaProcessor,
is_apex_available,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
if is_apex_available():
from apex import amp
if version.parse(version.parse(torch.__version__).base_version) >= version.parse('1.6'):
_snake_case : List[Any] = True
from torch.cuda.amp import autocast
_snake_case : Dict = logging.getLogger(__name__)
def a_ ( lowerCAmelCase_ : str=None, lowerCAmelCase_ : str=None ):
return field(default_factory=lambda: default, metadata=lowerCAmelCase_ )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
metadata={"""help""": """Path to pretrained model or model identifier from huggingface.co/models"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to freeze the feature extractor layers of the model."""} )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout ratio for the attention probabilities."""} )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout ratio for activations inside the fully connected layer."""} )
a_ = field(
default=0.1 , metadata={
"""help""": """The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler."""
} , )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout probabilitiy for all 1D convolutional layers in feature extractor."""} , )
a_ = field(
default=0.05 , metadata={
"""help""": (
"""Propability of each feature vector along the time axis to be chosen as the start of the vector"""
"""span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"""
"""vectors will be masked along the time axis. This is only relevant if ``apply_spec_augment is True``."""
)
} , )
a_ = field(default=0.0 , metadata={"""help""": """The LayerDrop probability."""} )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(
default="""train+validation""" , metadata={
"""help""": """The name of the training data set split to use (via the datasets library). Defaults to 'train'"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached preprocessed datasets or not."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""For debugging purposes or quicker training, truncate the number of training examples to this """
"""value if set."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""For debugging purposes or quicker training, truncate the number of validation examples to this """
"""value if set."""
)
} , )
a_ = list_field(
default=[""",""", """?""", """.""", """!""", """-""", """;""", """:""", """\"\"""", """%""", """'""", """\"""", """�"""] , metadata={"""help""": """A list of characters to remove from the transcripts."""} , )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = 42
a_ = True
a_ = None
a_ = None
a_ = None
a_ = None
def __call__( self : int , lowerCAmelCase_ : List[Dict[str, Union[List[int], torch.Tensor]]] ) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# different padding methods
__lowerCAmelCase = [{'input_values': feature['input_values']} for feature in features]
__lowerCAmelCase = [{'input_ids': feature['labels']} for feature in features]
__lowerCAmelCase = self.processor.pad(
lowerCAmelCase_ , padding=self.padding , max_length=self.max_length , pad_to_multiple_of=self.pad_to_multiple_of , return_tensors='pt' , )
__lowerCAmelCase = self.processor.pad(
labels=lowerCAmelCase_ , padding=self.padding , max_length=self.max_length_labels , pad_to_multiple_of=self.pad_to_multiple_of_labels , return_tensors='pt' , )
# replace padding with -100 to ignore loss correctly
__lowerCAmelCase = labels_batch['input_ids'].masked_fill(labels_batch.attention_mask.ne(1 ) , -1_0_0 )
__lowerCAmelCase = labels
return batch
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Tuple , lowerCAmelCase_ : nn.Module , lowerCAmelCase_ : Dict[str, Union[torch.Tensor, Any]] ) -> torch.Tensor:
model.train()
__lowerCAmelCase = self._prepare_inputs(lowerCAmelCase_ )
if self.use_amp:
with autocast():
__lowerCAmelCase = self.compute_loss(lowerCAmelCase_ , lowerCAmelCase_ )
else:
__lowerCAmelCase = self.compute_loss(lowerCAmelCase_ , lowerCAmelCase_ )
if self.args.n_gpu > 1:
if model.module.config.ctc_loss_reduction == "mean":
__lowerCAmelCase = loss.mean()
elif model.module.config.ctc_loss_reduction == "sum":
__lowerCAmelCase = loss.sum() / (inputs['labels'] >= 0).sum()
else:
raise ValueError(f"""{model.config.ctc_loss_reduction} is not valid. Choose one of ['mean', 'sum']""" )
if self.args.gradient_accumulation_steps > 1:
__lowerCAmelCase = loss / self.args.gradient_accumulation_steps
if self.use_amp:
self.scaler.scale(lowerCAmelCase_ ).backward()
elif self.use_apex:
with amp.scale_loss(lowerCAmelCase_ , self.optimizer ) as scaled_loss:
scaled_loss.backward()
elif self.deepspeed:
self.deepspeed.backward(lowerCAmelCase_ )
else:
loss.backward()
return loss.detach()
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets:
__lowerCAmelCase = datasets.load_dataset(
'common_voice', data_args.dataset_config_name, split=data_args.train_split_name )
__lowerCAmelCase = datasets.load_dataset('common_voice', data_args.dataset_config_name, split='test' )
# Create and save tokenizer
__lowerCAmelCase = F"""[{"".join(data_args.chars_to_ignore )}]"""
def remove_special_characters(lowerCAmelCase_ : Any ):
__lowerCAmelCase = re.sub(lowerCAmelCase_, '', batch['sentence'] ).lower() + ' '
return batch
__lowerCAmelCase = train_dataset.map(lowerCAmelCase_, remove_columns=['sentence'] )
__lowerCAmelCase = eval_dataset.map(lowerCAmelCase_, remove_columns=['sentence'] )
def extract_all_chars(lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = ' '.join(batch['text'] )
__lowerCAmelCase = list(set(lowerCAmelCase_ ) )
return {"vocab": [vocab], "all_text": [all_text]}
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, batched=lowerCAmelCase_, batch_size=-1, keep_in_memory=lowerCAmelCase_, remove_columns=train_dataset.column_names, )
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, batched=lowerCAmelCase_, batch_size=-1, keep_in_memory=lowerCAmelCase_, remove_columns=eval_dataset.column_names, )
__lowerCAmelCase = list(set(vocab_train['vocab'][0] ) | set(vocab_test['vocab'][0] ) )
__lowerCAmelCase = {v: k for k, v in enumerate(lowerCAmelCase_ )}
__lowerCAmelCase = vocab_dict[' ']
del vocab_dict[" "]
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = len(lowerCAmelCase_ )
with open('vocab.json', 'w' ) as vocab_file:
json.dump(lowerCAmelCase_, lowerCAmelCase_ )
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = WavaVecaCTCTokenizer(
'vocab.json', unk_token='[UNK]', pad_token='[PAD]', word_delimiter_token='|', )
__lowerCAmelCase = WavaVecaFeatureExtractor(
feature_size=1, sampling_rate=1_6000, padding_value=0.0, do_normalize=lowerCAmelCase_, return_attention_mask=lowerCAmelCase_ )
__lowerCAmelCase = WavaVecaProcessor(feature_extractor=lowerCAmelCase_, tokenizer=lowerCAmelCase_ )
__lowerCAmelCase = WavaVecaForCTC.from_pretrained(
model_args.model_name_or_path, cache_dir=model_args.cache_dir, activation_dropout=model_args.activation_dropout, attention_dropout=model_args.attention_dropout, hidden_dropout=model_args.hidden_dropout, feat_proj_dropout=model_args.feat_proj_dropout, mask_time_prob=model_args.mask_time_prob, gradient_checkpointing=training_args.gradient_checkpointing, layerdrop=model_args.layerdrop, ctc_loss_reduction='mean', pad_token_id=processor.tokenizer.pad_token_id, vocab_size=len(processor.tokenizer ), )
if data_args.max_train_samples is not None:
__lowerCAmelCase = min(len(lowerCAmelCase_ ), data_args.max_train_samples )
__lowerCAmelCase = train_dataset.select(range(lowerCAmelCase_ ) )
if data_args.max_val_samples is not None:
__lowerCAmelCase = eval_dataset.select(range(data_args.max_val_samples ) )
__lowerCAmelCase = torchaudio.transforms.Resample(4_8000, 1_6000 )
# Preprocessing the datasets.
# We need to read the aduio files as arrays and tokenize the targets.
def speech_file_to_array_fn(lowerCAmelCase_ : int ):
__lowerCAmelCase , __lowerCAmelCase = torchaudio.load(batch['path'] )
__lowerCAmelCase = resampler(lowerCAmelCase_ ).squeeze().numpy()
__lowerCAmelCase = 1_6000
__lowerCAmelCase = batch['text']
return batch
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, remove_columns=train_dataset.column_names, num_proc=data_args.preprocessing_num_workers, )
__lowerCAmelCase = eval_dataset.map(
lowerCAmelCase_, remove_columns=eval_dataset.column_names, num_proc=data_args.preprocessing_num_workers, )
def prepare_dataset(lowerCAmelCase_ : Union[str, Any] ):
# check that all files have the correct sampling rate
assert (
len(set(batch['sampling_rate'] ) ) == 1
), F"""Make sure all inputs have the same sampling rate of {processor.feature_extractor.sampling_rate}."""
__lowerCAmelCase = processor(
audio=batch['speech'], text=batch['target_text'], sampling_rate=batch['sampling_rate'][0] )
batch.update(lowerCAmelCase_ )
return batch
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, remove_columns=train_dataset.column_names, batch_size=training_args.per_device_train_batch_size, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, )
__lowerCAmelCase = eval_dataset.map(
lowerCAmelCase_, remove_columns=eval_dataset.column_names, batch_size=training_args.per_device_train_batch_size, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, )
# Metric
__lowerCAmelCase = datasets.load_metric('wer' )
def compute_metrics(lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = pred.predictions
__lowerCAmelCase = np.argmax(lowerCAmelCase_, axis=-1 )
__lowerCAmelCase = processor.tokenizer.pad_token_id
__lowerCAmelCase = processor.batch_decode(lowerCAmelCase_ )
# we do not want to group tokens when computing the metrics
__lowerCAmelCase = processor.batch_decode(pred.label_ids, group_tokens=lowerCAmelCase_ )
__lowerCAmelCase = wer_metric.compute(predictions=lowerCAmelCase_, references=lowerCAmelCase_ )
return {"wer": wer}
if model_args.freeze_feature_extractor:
model.freeze_feature_extractor()
# Data collator
__lowerCAmelCase = DataCollatorCTCWithPadding(processor=lowerCAmelCase_, padding=lowerCAmelCase_ )
# Initialize our Trainer
__lowerCAmelCase = CTCTrainer(
model=lowerCAmelCase_, data_collator=lowerCAmelCase_, args=lowerCAmelCase_, compute_metrics=lowerCAmelCase_, train_dataset=train_dataset if training_args.do_train else None, eval_dataset=eval_dataset if training_args.do_eval else None, tokenizer=processor.feature_extractor, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
# Save the feature_extractor and the tokenizer
if is_main_process(training_args.local_rank ):
processor.save_pretrained(training_args.output_dir )
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model()
__lowerCAmelCase = train_result.metrics
__lowerCAmelCase = (
data_args.max_train_samples if data_args.max_train_samples is not None else len(lowerCAmelCase_ )
)
__lowerCAmelCase = min(lowerCAmelCase_, len(lowerCAmelCase_ ) )
trainer.log_metrics('train', lowerCAmelCase_ )
trainer.save_metrics('train', lowerCAmelCase_ )
trainer.save_state()
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = data_args.max_val_samples if data_args.max_val_samples is not None else len(lowerCAmelCase_ )
__lowerCAmelCase = min(lowerCAmelCase_, len(lowerCAmelCase_ ) )
trainer.log_metrics('eval', lowerCAmelCase_ )
trainer.save_metrics('eval', lowerCAmelCase_ )
return results
if __name__ == "__main__":
main()
| 53 | 1 |
from typing import Dict, Iterable, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_DEFAULT_MEAN,
IMAGENET_DEFAULT_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, logging
_snake_case : List[Any] = logging.get_logger(__name__)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = ["""pixel_values"""]
def __init__( self : Optional[int] , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Union[int, float] = 1 / 2_5_5 , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_MEAN , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_STD , **lowerCAmelCase_ : Any , ) -> None:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = size if size is not None else {'shortest_edge': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else {'height': 2_2_4, 'width': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = do_resize
__lowerCAmelCase = size
__lowerCAmelCase = resample
__lowerCAmelCase = do_center_crop
__lowerCAmelCase = crop_size
__lowerCAmelCase = do_rescale
__lowerCAmelCase = rescale_factor
__lowerCAmelCase = do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else IMAGENET_DEFAULT_MEAN
__lowerCAmelCase = image_std if image_std is not None else IMAGENET_DEFAULT_STD
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : Optional[int] , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
# size_dict is a dict with either keys "height" and "width" or "shortest_edge"
if "shortest_edge" in size:
__lowerCAmelCase = int((2_5_6 / 2_2_4) * size['shortest_edge'] )
__lowerCAmelCase = get_resize_output_image_size(lowerCAmelCase_ , size=lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = {'height': output_size[0], 'width': output_size[1]}
if "height" not in size_dict or "width" not in size_dict:
raise ValueError(
f"""Size dict must have keys 'height' and 'width' or 'shortest_edge'. Got {size_dict.keys()}""" )
return resize(
lowerCAmelCase_ , size=(size_dict['height'], size_dict['width']) , resample=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : str , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : str , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ )
if "height" not in size or "width" not in size:
raise ValueError(f"""Size dict must have keys 'height' and 'width'. Got {size.keys()}""" )
return center_crop(lowerCAmelCase_ , size=(size['height'], size['width']) , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[int, float] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : int , ) -> np.ndarray:
return rescale(lowerCAmelCase_ , scale=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : int , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : List[str] , ) -> np.ndarray:
return normalize(lowerCAmelCase_ , mean=lowerCAmelCase_ , std=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : ImageInput , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : PILImageResampling = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[float] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[TensorType] = None , lowerCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **lowerCAmelCase_ : str , ) -> BatchFeature:
__lowerCAmelCase = do_resize if do_resize is not None else self.do_resize
__lowerCAmelCase = resample if resample is not None else self.resample
__lowerCAmelCase = do_center_crop if do_center_crop is not None else self.do_center_crop
__lowerCAmelCase = do_rescale if do_rescale is not None else self.do_rescale
__lowerCAmelCase = rescale_factor if rescale_factor is not None else self.rescale_factor
__lowerCAmelCase = do_normalize if do_normalize is not None else self.do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else self.image_mean
__lowerCAmelCase = image_std if image_std is not None else self.image_std
__lowerCAmelCase = size if size is not None else self.size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else self.crop_size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = make_list_of_images(lowerCAmelCase_ )
if not valid_images(lowerCAmelCase_ ):
raise ValueError(
'Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '
'torch.Tensor, tf.Tensor or jax.ndarray.' )
if do_resize and size is None:
raise ValueError('Size must be specified if do_resize is True.' )
if do_center_crop and crop_size is None:
raise ValueError('Crop size must be specified if do_center_crop is True.' )
if do_rescale and rescale_factor is None:
raise ValueError('Rescale factor must be specified if do_rescale is True.' )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError('Image mean and std must be specified if do_normalize is True.' )
# All transformations expect numpy arrays.
__lowerCAmelCase = [to_numpy_array(lowerCAmelCase_ ) for image in images]
if do_resize:
__lowerCAmelCase = [self.resize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_center_crop:
__lowerCAmelCase = [self.center_crop(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_rescale:
__lowerCAmelCase = [self.rescale(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_normalize:
__lowerCAmelCase = [self.normalize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = [to_channel_dimension_format(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = {'pixel_values': images}
return BatchFeature(data=lowerCAmelCase_ , tensor_type=lowerCAmelCase_ )
| 53 |
import json
from typing import List, Optional, Tuple
from tokenizers import normalizers
from ....tokenization_utils_fast import PreTrainedTokenizerFast
from ....utils import logging
from .tokenization_retribert import RetriBertTokenizer
_snake_case : Any = logging.get_logger(__name__)
_snake_case : int = {'vocab_file': 'vocab.txt', 'tokenizer_file': 'tokenizer.json'}
_snake_case : Optional[Any] = {
'vocab_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/vocab.txt'
),
},
'tokenizer_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/tokenizer.json'
),
},
}
_snake_case : str = {
'yjernite/retribert-base-uncased': 512,
}
_snake_case : Optional[int] = {
'yjernite/retribert-base-uncased': {'do_lower_case': True},
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = PRETRAINED_VOCAB_FILES_MAP
a_ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
a_ = PRETRAINED_INIT_CONFIGURATION
a_ = RetriBertTokenizer
a_ = ["""input_ids""", """attention_mask"""]
def __init__( self : Dict , lowerCAmelCase_ : Optional[Any]=None , lowerCAmelCase_ : Tuple=None , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : str="[UNK]" , lowerCAmelCase_ : Optional[Any]="[SEP]" , lowerCAmelCase_ : List[str]="[PAD]" , lowerCAmelCase_ : Optional[int]="[CLS]" , lowerCAmelCase_ : List[Any]="[MASK]" , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : List[str]=None , **lowerCAmelCase_ : List[Any] , ) -> Dict:
super().__init__(
lowerCAmelCase_ , tokenizer_file=lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , unk_token=lowerCAmelCase_ , sep_token=lowerCAmelCase_ , pad_token=lowerCAmelCase_ , cls_token=lowerCAmelCase_ , mask_token=lowerCAmelCase_ , tokenize_chinese_chars=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ , **lowerCAmelCase_ , )
__lowerCAmelCase = json.loads(self.backend_tokenizer.normalizer.__getstate__() )
if (
normalizer_state.get('lowercase' , lowerCAmelCase_ ) != do_lower_case
or normalizer_state.get('strip_accents' , lowerCAmelCase_ ) != strip_accents
or normalizer_state.get('handle_chinese_chars' , lowerCAmelCase_ ) != tokenize_chinese_chars
):
__lowerCAmelCase = getattr(lowerCAmelCase_ , normalizer_state.pop('type' ) )
__lowerCAmelCase = do_lower_case
__lowerCAmelCase = strip_accents
__lowerCAmelCase = tokenize_chinese_chars
__lowerCAmelCase = normalizer_class(**lowerCAmelCase_ )
__lowerCAmelCase = do_lower_case
def lowercase ( self : str , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Optional[int]=None ) -> Optional[int]:
__lowerCAmelCase = [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 lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[int] , lowerCAmelCase_ : Optional[List[int]] = None ) -> List[int]:
__lowerCAmelCase = [self.sep_token_id]
__lowerCAmelCase = [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 lowercase ( self : int , lowerCAmelCase_ : str , lowerCAmelCase_ : Optional[str] = None ) -> Tuple[str]:
__lowerCAmelCase = self._tokenizer.model.save(lowerCAmelCase_ , name=lowerCAmelCase_ )
return tuple(lowerCAmelCase_ )
| 53 | 1 |
import copy
import inspect
import unittest
from transformers import PretrainedConfig, SwiftFormerConfig
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 torch import nn
from transformers import SwiftFormerForImageClassification, SwiftFormerModel
from transformers.models.swiftformer.modeling_swiftformer import SWIFTFORMER_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import ViTImageProcessor
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any=1_3 , lowerCAmelCase_ : Tuple=3 , lowerCAmelCase_ : int=True , lowerCAmelCase_ : Tuple=True , lowerCAmelCase_ : Optional[Any]=0.1 , lowerCAmelCase_ : Any=0.1 , lowerCAmelCase_ : Union[str, Any]=2_2_4 , lowerCAmelCase_ : List[str]=1_0_0_0 , lowerCAmelCase_ : Union[str, Any]=[3, 3, 6, 4] , lowerCAmelCase_ : List[str]=[4_8, 5_6, 1_1_2, 2_2_0] , ) -> Any:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = num_labels
__lowerCAmelCase = image_size
__lowerCAmelCase = layer_depths
__lowerCAmelCase = embed_dims
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.num_labels )
__lowerCAmelCase = self.get_config()
return config, pixel_values, labels
def lowercase ( self : Optional[Any] ) -> Optional[int]:
return SwiftFormerConfig(
depths=self.layer_depths , embed_dims=self.embed_dims , mlp_ratio=4 , downsamples=[True, True, True, True] , hidden_act='gelu' , num_labels=self.num_labels , down_patch_size=3 , down_stride=2 , down_pad=1 , drop_rate=0.0 , drop_path_rate=0.0 , use_layer_scale=lowerCAmelCase_ , layer_scale_init_value=1e-5 , )
def lowercase ( self : Optional[int] , lowerCAmelCase_ : int , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Tuple ) -> int:
__lowerCAmelCase = SwiftFormerModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.embed_dims[-1], 7, 7) )
def lowercase ( self : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Optional[int] ) -> Optional[int]:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = SwiftFormerForImageClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
__lowerCAmelCase = SwiftFormerForImageClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : str ) -> Any:
((__lowerCAmelCase) , (__lowerCAmelCase) , (__lowerCAmelCase)) = self.prepare_config_and_inputs()
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (SwiftFormerModel, SwiftFormerForImageClassification) if is_torch_available() else ()
a_ = (
{"""feature-extraction""": SwiftFormerModel, """image-classification""": SwiftFormerForImageClassification}
if is_torch_available()
else {}
)
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Optional[int] ) -> List[Any]:
__lowerCAmelCase = SwiftFormerModelTester(self )
__lowerCAmelCase = ConfigTester(
self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ , hidden_size=3_7 , num_attention_heads=1_2 , num_hidden_layers=1_2 , )
def lowercase ( self : Optional[Any] ) -> Optional[int]:
self.config_tester.run_common_tests()
@unittest.skip(reason='SwiftFormer does not use inputs_embeds' )
def lowercase ( self : List[str] ) -> Tuple:
pass
def lowercase ( self : int ) -> Dict:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCAmelCase_ , nn.Linear ) )
def lowercase ( self : str ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Dict:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCAmelCase_ )
@slow
def lowercase ( self : Union[str, Any] ) -> int:
for model_name in SWIFTFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = SwiftFormerModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
@unittest.skip(reason='SwiftFormer does not output attentions' )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
pass
def lowercase ( self : Any ) -> int:
def check_hidden_states_output(lowerCAmelCase_ : Dict , lowerCAmelCase_ : str , lowerCAmelCase_ : List[Any] ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.hidden_states
__lowerCAmelCase = 8
self.assertEqual(len(lowerCAmelCase_ ) , lowerCAmelCase_ ) # TODO
# SwiftFormer's feature maps are of shape (batch_size, embed_dims, height, width)
# with the width and height being successively divided by 2, after every 2 blocks
for i in range(len(lowerCAmelCase_ ) ):
self.assertEqual(
hidden_states[i].shape , torch.Size(
[
self.model_tester.batch_size,
self.model_tester.embed_dims[i // 2],
(self.model_tester.image_size // 4) // 2 ** (i // 2),
(self.model_tester.image_size // 4) // 2 ** (i // 2),
] ) , )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : List[str] ) -> List[str]:
def _config_zero_init(lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
for key in configs_no_init.__dict__.keys():
if "_range" in key or "_std" in key or "initializer_factor" in key or "layer_scale" in key:
setattr(lowerCAmelCase_ , lowerCAmelCase_ , 1e-10 )
if isinstance(getattr(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) , lowerCAmelCase_ ):
__lowerCAmelCase = _config_zero_init(getattr(lowerCAmelCase_ , lowerCAmelCase_ ) )
setattr(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
return configs_no_init
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = _config_zero_init(lowerCAmelCase_ )
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(config=lowerCAmelCase_ )
for name, param in model.named_parameters():
if param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9) / 1e9).round().item() , [0.0, 1.0] , msg=f"""Parameter {name} of model {model_class} seems not properly initialized""" , )
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : Any ) -> List[str]:
pass
def a_ ( ):
__lowerCAmelCase = Image.open('./tests/fixtures/tests_samples/COCO/000000039769.png' )
return image
@require_torch
@require_vision
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def lowercase ( self : Dict ) -> Tuple:
return ViTImageProcessor.from_pretrained('MBZUAI/swiftformer-xs' ) if is_vision_available() else None
@slow
def lowercase ( self : str ) -> List[str]:
__lowerCAmelCase = SwiftFormerForImageClassification.from_pretrained('MBZUAI/swiftformer-xs' ).to(lowerCAmelCase_ )
__lowerCAmelCase = self.default_image_processor
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = image_processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
# forward pass
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
# verify the logits
__lowerCAmelCase = torch.Size((1, 1_0_0_0) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor([[-2.17_03e00, 2.11_07e00, -2.08_11e00]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 |
import numpy as np
from cva import COLOR_BGR2GRAY, cvtColor, imread
from numpy import array, uinta
from PIL import Image
from digital_image_processing import change_contrast as cc
from digital_image_processing import convert_to_negative as cn
from digital_image_processing import sepia as sp
from digital_image_processing.dithering import burkes as bs
from digital_image_processing.edge_detection import canny
from digital_image_processing.filters import convolve as conv
from digital_image_processing.filters import gaussian_filter as gg
from digital_image_processing.filters import local_binary_pattern as lbp
from digital_image_processing.filters import median_filter as med
from digital_image_processing.filters import sobel_filter as sob
from digital_image_processing.resize import resize as rs
_snake_case : Union[str, Any] = imread(R'digital_image_processing/image_data/lena_small.jpg')
_snake_case : Optional[int] = cvtColor(img, COLOR_BGR2GRAY)
def a_ ( ):
__lowerCAmelCase = cn.convert_to_negative(lowerCAmelCase_ )
# assert negative_img array for at least one True
assert negative_img.any()
def a_ ( ):
with Image.open('digital_image_processing/image_data/lena_small.jpg' ) as img:
# Work around assertion for response
assert str(cc.change_contrast(lowerCAmelCase_, 110 ) ).startswith(
'<PIL.Image.Image image mode=RGB size=100x100 at' )
def a_ ( ):
__lowerCAmelCase = canny.gen_gaussian_kernel(9, sigma=1.4 )
# Assert ambiguous array
assert resp.all()
def a_ ( ):
__lowerCAmelCase = imread('digital_image_processing/image_data/lena_small.jpg', 0 )
# assert ambiguous array for all == True
assert canny_img.all()
__lowerCAmelCase = canny.canny(lowerCAmelCase_ )
# assert canny array for at least one True
assert canny_array.any()
def a_ ( ):
assert gg.gaussian_filter(lowerCAmelCase_, 5, sigma=0.9 ).all()
def a_ ( ):
# laplace diagonals
__lowerCAmelCase = array([[0.25, 0.5, 0.25], [0.5, -3, 0.5], [0.25, 0.5, 0.25]] )
__lowerCAmelCase = conv.img_convolve(lowerCAmelCase_, lowerCAmelCase_ ).astype(lowerCAmelCase_ )
assert res.any()
def a_ ( ):
assert med.median_filter(lowerCAmelCase_, 3 ).any()
def a_ ( ):
__lowerCAmelCase , __lowerCAmelCase = sob.sobel_filter(lowerCAmelCase_ )
assert grad.any() and theta.any()
def a_ ( ):
__lowerCAmelCase = sp.make_sepia(lowerCAmelCase_, 20 )
assert sepia.all()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg" ):
__lowerCAmelCase = bs.Burkes(imread(lowerCAmelCase_, 1 ), 120 )
burkes.process()
assert burkes.output_img.any()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg", ):
__lowerCAmelCase = rs.NearestNeighbour(imread(lowerCAmelCase_, 1 ), 400, 200 )
nn.process()
assert nn.output.any()
def a_ ( ):
__lowerCAmelCase = 'digital_image_processing/image_data/lena.jpg'
# Reading the image and converting it to grayscale.
__lowerCAmelCase = imread(lowerCAmelCase_, 0 )
# Test for get_neighbors_pixel function() return not None
__lowerCAmelCase = 0
__lowerCAmelCase = 0
__lowerCAmelCase = image[x_coordinate][y_coordinate]
__lowerCAmelCase = lbp.get_neighbors_pixel(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert neighbors_pixels is not None
# Test for local_binary_pattern function()
# Create a numpy array as the same height and width of read image
__lowerCAmelCase = np.zeros((image.shape[0], image.shape[1]) )
# Iterating through the image and calculating the local binary pattern value
# for each pixel.
for i in range(0, image.shape[0] ):
for j in range(0, image.shape[1] ):
__lowerCAmelCase = lbp.local_binary_value(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert lbp_image.any()
| 53 | 1 |
import argparse
from collections import defaultdict
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : str, lowerCAmelCase_ : int, lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = F"""{file}_{class_name}_{test_name}"""
done_test[_id] += 1
with open(lowerCAmelCase_, 'r' ) as f:
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = F"""class {class_name}("""
__lowerCAmelCase = F"""{4 * " "}def {test_name}("""
__lowerCAmelCase = F"""{8 * " "}{correct_line.split()[0]}"""
__lowerCAmelCase = F"""{16 * " "}{correct_line.split()[0]}"""
__lowerCAmelCase = False
__lowerCAmelCase = False
__lowerCAmelCase = False
__lowerCAmelCase = False
__lowerCAmelCase = 0
__lowerCAmelCase = 0
__lowerCAmelCase = []
for line in lines:
if line.startswith(lowerCAmelCase_ ):
__lowerCAmelCase = True
elif in_class and line.startswith(lowerCAmelCase_ ):
__lowerCAmelCase = True
elif in_class and in_func and (line.startswith(lowerCAmelCase_ ) or line.startswith(lowerCAmelCase_ )):
__lowerCAmelCase = len(line.split(correct_line.split()[0] )[0] )
count += 1
if count == done_test[_id]:
__lowerCAmelCase = True
if in_class and in_func and in_line:
if ")" not in line:
continue
else:
__lowerCAmelCase = True
if in_class and in_func and in_line and insert_line:
new_lines.append(F"""{spaces * " "}{correct_line}""" )
__lowerCAmelCase = __lowerCAmelCase = __lowerCAmelCase = __lowerCAmelCase = False
else:
new_lines.append(lowerCAmelCase_ )
with open(lowerCAmelCase_, 'w' ) as f:
for line in new_lines:
f.write(lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Any=None ):
if fail is not None:
with open(lowerCAmelCase_, 'r' ) as f:
__lowerCAmelCase = {l.strip() for l in f.readlines()}
else:
__lowerCAmelCase = None
with open(lowerCAmelCase_, 'r' ) as f:
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = defaultdict(lowerCAmelCase_ )
for line in correct_lines:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = line.split(';' )
if test_failures is None or "::".join([file, class_name, test_name] ) in test_failures:
overwrite_file(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
parser.add_argument('--correct_filename', help='filename of tests with expected result')
parser.add_argument('--fail_filename', help='filename of test failures', type=str, default=None)
_snake_case : str = parser.parse_args()
main(args.correct_filename, args.fail_filename)
| 53 |
from typing import Dict, Iterable, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_DEFAULT_MEAN,
IMAGENET_DEFAULT_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, logging
_snake_case : List[Any] = logging.get_logger(__name__)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = ["""pixel_values"""]
def __init__( self : Optional[int] , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Union[int, float] = 1 / 2_5_5 , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_MEAN , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_STD , **lowerCAmelCase_ : Any , ) -> None:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = size if size is not None else {'shortest_edge': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else {'height': 2_2_4, 'width': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = do_resize
__lowerCAmelCase = size
__lowerCAmelCase = resample
__lowerCAmelCase = do_center_crop
__lowerCAmelCase = crop_size
__lowerCAmelCase = do_rescale
__lowerCAmelCase = rescale_factor
__lowerCAmelCase = do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else IMAGENET_DEFAULT_MEAN
__lowerCAmelCase = image_std if image_std is not None else IMAGENET_DEFAULT_STD
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : Optional[int] , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
# size_dict is a dict with either keys "height" and "width" or "shortest_edge"
if "shortest_edge" in size:
__lowerCAmelCase = int((2_5_6 / 2_2_4) * size['shortest_edge'] )
__lowerCAmelCase = get_resize_output_image_size(lowerCAmelCase_ , size=lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = {'height': output_size[0], 'width': output_size[1]}
if "height" not in size_dict or "width" not in size_dict:
raise ValueError(
f"""Size dict must have keys 'height' and 'width' or 'shortest_edge'. Got {size_dict.keys()}""" )
return resize(
lowerCAmelCase_ , size=(size_dict['height'], size_dict['width']) , resample=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : str , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : str , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ )
if "height" not in size or "width" not in size:
raise ValueError(f"""Size dict must have keys 'height' and 'width'. Got {size.keys()}""" )
return center_crop(lowerCAmelCase_ , size=(size['height'], size['width']) , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[int, float] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : int , ) -> np.ndarray:
return rescale(lowerCAmelCase_ , scale=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : int , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : List[str] , ) -> np.ndarray:
return normalize(lowerCAmelCase_ , mean=lowerCAmelCase_ , std=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : ImageInput , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : PILImageResampling = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[float] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[TensorType] = None , lowerCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **lowerCAmelCase_ : str , ) -> BatchFeature:
__lowerCAmelCase = do_resize if do_resize is not None else self.do_resize
__lowerCAmelCase = resample if resample is not None else self.resample
__lowerCAmelCase = do_center_crop if do_center_crop is not None else self.do_center_crop
__lowerCAmelCase = do_rescale if do_rescale is not None else self.do_rescale
__lowerCAmelCase = rescale_factor if rescale_factor is not None else self.rescale_factor
__lowerCAmelCase = do_normalize if do_normalize is not None else self.do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else self.image_mean
__lowerCAmelCase = image_std if image_std is not None else self.image_std
__lowerCAmelCase = size if size is not None else self.size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else self.crop_size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = make_list_of_images(lowerCAmelCase_ )
if not valid_images(lowerCAmelCase_ ):
raise ValueError(
'Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '
'torch.Tensor, tf.Tensor or jax.ndarray.' )
if do_resize and size is None:
raise ValueError('Size must be specified if do_resize is True.' )
if do_center_crop and crop_size is None:
raise ValueError('Crop size must be specified if do_center_crop is True.' )
if do_rescale and rescale_factor is None:
raise ValueError('Rescale factor must be specified if do_rescale is True.' )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError('Image mean and std must be specified if do_normalize is True.' )
# All transformations expect numpy arrays.
__lowerCAmelCase = [to_numpy_array(lowerCAmelCase_ ) for image in images]
if do_resize:
__lowerCAmelCase = [self.resize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_center_crop:
__lowerCAmelCase = [self.center_crop(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_rescale:
__lowerCAmelCase = [self.rescale(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_normalize:
__lowerCAmelCase = [self.normalize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = [to_channel_dimension_format(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = {'pixel_values': images}
return BatchFeature(data=lowerCAmelCase_ , tensor_type=lowerCAmelCase_ )
| 53 | 1 |
import logging
import os
from dataclasses import dataclass, field
from functools import partial
from pathlib import Path
from tempfile import TemporaryDirectory
from typing import List, Optional
import faiss
import torch
from datasets import Features, Sequence, Value, load_dataset
from transformers import DPRContextEncoder, DPRContextEncoderTokenizerFast, HfArgumentParser
_snake_case : List[Any] = logging.getLogger(__name__)
torch.set_grad_enabled(False)
_snake_case : Any = 'cuda' if torch.cuda.is_available() else 'cpu'
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : Dict=100, lowerCAmelCase_ : Optional[Any]=" " ):
__lowerCAmelCase = text.split(lowerCAmelCase_ )
return [character.join(text[i : i + n] ).strip() for i in range(0, len(lowerCAmelCase_ ), lowerCAmelCase_ )]
def a_ ( lowerCAmelCase_ : dict ):
__lowerCAmelCase , __lowerCAmelCase = [], []
for title, text in zip(documents['title'], documents['text'] ):
if text is not None:
for passage in split_text(lowerCAmelCase_ ):
titles.append(title if title is not None else '' )
texts.append(lowerCAmelCase_ )
return {"title": titles, "text": texts}
def a_ ( lowerCAmelCase_ : dict, lowerCAmelCase_ : DPRContextEncoder, lowerCAmelCase_ : DPRContextEncoderTokenizerFast ):
__lowerCAmelCase = ctx_tokenizer(
documents['title'], documents['text'], truncation=lowerCAmelCase_, padding='longest', return_tensors='pt' )['input_ids']
__lowerCAmelCase = ctx_encoder(input_ids.to(device=lowerCAmelCase_ ), return_dict=lowerCAmelCase_ ).pooler_output
return {"embeddings": embeddings.detach().cpu().numpy()}
def a_ ( lowerCAmelCase_ : "RagExampleArguments", lowerCAmelCase_ : "ProcessingArguments", lowerCAmelCase_ : "IndexHnswArguments", ):
######################################
logger.info('Step 1 - Create the dataset' )
######################################
# The dataset needed for RAG must have three columns:
# - title (string): title of the document
# - text (string): text of a passage of the document
# - embeddings (array of dimension d): DPR representation of the passage
# Let's say you have documents in tab-separated csv files with columns "title" and "text"
assert os.path.isfile(rag_example_args.csv_path ), "Please provide a valid path to a csv file"
# You can load a Dataset object this way
__lowerCAmelCase = load_dataset(
'csv', data_files=[rag_example_args.csv_path], split='train', delimiter='\t', column_names=['title', 'text'] )
# More info about loading csv files in the documentation: https://huggingface.co/docs/datasets/loading_datasets.html?highlight=csv#csv-files
# Then split the documents into passages of 100 words
__lowerCAmelCase = dataset.map(lowerCAmelCase_, batched=lowerCAmelCase_, num_proc=processing_args.num_proc )
# And compute the embeddings
__lowerCAmelCase = DPRContextEncoder.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name ).to(device=lowerCAmelCase_ )
__lowerCAmelCase = DPRContextEncoderTokenizerFast.from_pretrained(rag_example_args.dpr_ctx_encoder_model_name )
__lowerCAmelCase = Features(
{'text': Value('string' ), 'title': Value('string' ), 'embeddings': Sequence(Value('float32' ) )} ) # optional, save as float32 instead of float64 to save space
__lowerCAmelCase = dataset.map(
partial(lowerCAmelCase_, ctx_encoder=lowerCAmelCase_, ctx_tokenizer=lowerCAmelCase_ ), batched=lowerCAmelCase_, batch_size=processing_args.batch_size, features=lowerCAmelCase_, )
# And finally save your dataset
__lowerCAmelCase = os.path.join(rag_example_args.output_dir, 'my_knowledge_dataset' )
dataset.save_to_disk(lowerCAmelCase_ )
# from datasets import load_from_disk
# dataset = load_from_disk(passages_path) # to reload the dataset
######################################
logger.info('Step 2 - Index the dataset' )
######################################
# Let's use the Faiss implementation of HNSW for fast approximate nearest neighbor search
__lowerCAmelCase = faiss.IndexHNSWFlat(index_hnsw_args.d, index_hnsw_args.m, faiss.METRIC_INNER_PRODUCT )
dataset.add_faiss_index('embeddings', custom_index=lowerCAmelCase_ )
# And save the index
__lowerCAmelCase = os.path.join(rag_example_args.output_dir, 'my_knowledge_dataset_hnsw_index.faiss' )
dataset.get_index('embeddings' ).save(lowerCAmelCase_ )
# dataset.load_faiss_index("embeddings", index_path) # to reload the index
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=str(Path(_UpperCamelCase ).parent / """test_run""" / """dummy-kb""" / """my_knowledge_dataset.csv""" ) , metadata={"""help""": """Path to a tab-separated csv file with columns 'title' and 'text'"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Question that is passed as input to RAG. Default is 'What does Moses' rod turn into ?'."""} , )
a_ = field(
default="""facebook/rag-sequence-nq""" , metadata={"""help""": """The RAG model to use. Either 'facebook/rag-sequence-nq' or 'facebook/rag-token-nq'"""} , )
a_ = field(
default="""facebook/dpr-ctx_encoder-multiset-base""" , metadata={
"""help""": (
"""The DPR context encoder model to use. Either 'facebook/dpr-ctx_encoder-single-nq-base' or"""
""" 'facebook/dpr-ctx_encoder-multiset-base'"""
)
} , )
a_ = field(
default=str(Path(_UpperCamelCase ).parent / """test_run""" / """dummy-kb""" ) , metadata={"""help""": """Path to a directory where the dataset passages and the index will be saved"""} , )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": """The number of processes to use to split the documents into passages. Default is single process."""
} , )
a_ = field(
default=16 , metadata={
"""help""": """The batch size to use when computing the passages embeddings using the DPR context encoder."""
} , )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=7_68 , metadata={"""help""": """The dimension of the embeddings to pass to the HNSW Faiss index."""} , )
a_ = field(
default=1_28 , metadata={
"""help""": (
"""The number of bi-directional links created for every new element during the HNSW index construction."""
)
} , )
if __name__ == "__main__":
logging.basicConfig(level=logging.WARNING)
logger.setLevel(logging.INFO)
_snake_case : Tuple = HfArgumentParser((RagExampleArguments, ProcessingArguments, IndexHnswArguments))
_snake_case , _snake_case , _snake_case : Union[str, Any] = parser.parse_args_into_dataclasses()
with TemporaryDirectory() as tmp_dir:
_snake_case : Tuple = rag_example_args.output_dir or tmp_dir
main(rag_example_args, processing_args, index_hnsw_args)
| 53 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Dict=2 , lowerCAmelCase_ : Optional[int]=8 , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[Any]=9_9 , lowerCAmelCase_ : List[Any]=1_6 , lowerCAmelCase_ : int=5 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : str=3_6 , lowerCAmelCase_ : Optional[int]="gelu" , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : Optional[int]=0.0 , lowerCAmelCase_ : str=5_1_2 , lowerCAmelCase_ : List[str]=1_6 , lowerCAmelCase_ : str=2 , lowerCAmelCase_ : Tuple=0.02 , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : List[Any]=4 , lowerCAmelCase_ : List[str]=None , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_input_mask
__lowerCAmelCase = use_token_type_ids
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = type_vocab_size
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = num_labels
__lowerCAmelCase = num_choices
__lowerCAmelCase = scope
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = None
if self.use_input_mask:
__lowerCAmelCase = random_attention_mask([self.batch_size, self.seq_length] )
__lowerCAmelCase = None
if self.use_token_type_ids:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
__lowerCAmelCase = ids_tensor([self.batch_size] , self.num_choices )
__lowerCAmelCase = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowercase ( self : Any ) -> Union[str, Any]:
return MraConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , )
def lowercase ( self : Dict ) -> List[Any]:
__lowerCAmelCase = self.get_config()
__lowerCAmelCase = 3_0_0
return config
def lowercase ( self : Optional[int] ) -> Union[str, Any]:
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = self.prepare_config_and_inputs()
__lowerCAmelCase = True
__lowerCAmelCase = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
encoder_hidden_states,
encoder_attention_mask,
)
def lowercase ( self : Optional[int] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , ) -> Tuple:
__lowerCAmelCase = True
__lowerCAmelCase = MraModel(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , )
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , )
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Dict ) -> str:
__lowerCAmelCase = MraForQuestionAnswering(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , start_positions=lowerCAmelCase_ , end_positions=lowerCAmelCase_ , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowercase ( self : int , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : int , lowerCAmelCase_ : Dict ) -> Any:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForTokenClassification(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowercase ( self : int , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] ) -> List[Any]:
__lowerCAmelCase = self.num_choices
__lowerCAmelCase = MraForMultipleChoice(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowercase ( self : Tuple ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'input_ids': input_ids, 'token_type_ids': token_type_ids, 'attention_mask': input_mask}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
a_ = False
a_ = False
a_ = False
a_ = False
a_ = ()
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = MraModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : Tuple ) -> List[str]:
self.config_tester.run_common_tests()
def lowercase ( self : Optional[int] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
__lowerCAmelCase = type
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Any ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*lowerCAmelCase_ )
def lowercase ( self : List[str] ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*lowerCAmelCase_ )
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = MraModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
@unittest.skip(reason='MRA does not output attentions' )
def lowercase ( self : Optional[int] ) -> Tuple:
return
@require_torch
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@slow
def lowercase ( self : Optional[Any] ) -> List[str]:
__lowerCAmelCase = MraModel.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = torch.Size((1, 2_5_6, 7_6_8) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[-0.01_40, 0.08_30, -0.03_81], [0.15_46, 0.14_02, 0.02_20], [0.11_62, 0.08_51, 0.01_65]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : int ) -> Optional[int]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 2_5_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[9.25_95, -3.60_38, 11.88_19], [9.38_69, -3.26_93, 11.09_56], [11.85_24, -3.49_38, 13.12_10]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : Any ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-4096-8-d3' )
__lowerCAmelCase = torch.arange(4_0_9_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 4_0_9_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[5.47_89, -2.35_64, 7.50_64], [7.90_67, -1.33_69, 9.96_68], [9.07_12, -1.81_06, 7.03_80]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = ''
for ch in key:
if ch == " " or ch not in key_no_dups and ch.isalpha():
key_no_dups += ch
return key_no_dups
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = [chr(i + 65 ) for i in range(26 )]
# Remove duplicate characters from key
__lowerCAmelCase = remove_duplicates(key.upper() )
__lowerCAmelCase = len(lowerCAmelCase_ )
# First fill cipher with key characters
__lowerCAmelCase = {alphabet[i]: char for i, char in enumerate(lowerCAmelCase_ )}
# Then map remaining characters in alphabet to
# the alphabet from the beginning
for i in range(len(lowerCAmelCase_ ), 26 ):
__lowerCAmelCase = alphabet[i - offset]
# Ensure we are not mapping letters to letters previously mapped
while char in key:
offset -= 1
__lowerCAmelCase = alphabet[i - offset]
__lowerCAmelCase = char
return cipher_alphabet
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : dict[str, str] ):
return "".join(cipher_map.get(lowerCAmelCase_, lowerCAmelCase_ ) for ch in message.upper() )
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : dict[str, str] ):
__lowerCAmelCase = {v: k for k, v in cipher_map.items()}
return "".join(rev_cipher_map.get(lowerCAmelCase_, lowerCAmelCase_ ) for ch in message.upper() )
def a_ ( ):
__lowerCAmelCase = input('Enter message to encode or decode: ' ).strip()
__lowerCAmelCase = input('Enter keyword: ' ).strip()
__lowerCAmelCase = input('Encipher or decipher? E/D:' ).strip()[0].lower()
try:
__lowerCAmelCase = {'e': encipher, 'd': decipher}[option]
except KeyError:
raise KeyError('invalid input option' )
__lowerCAmelCase = create_cipher_map(lowerCAmelCase_ )
print(func(lowerCAmelCase_, lowerCAmelCase_ ) )
if __name__ == "__main__":
import doctest
doctest.testmod()
main()
| 53 |
import argparse
import json
import os
import re
import shutil
import torch
from transformers import BioGptConfig, BioGptForCausalLM
from transformers.models.biogpt.tokenization_biogpt import VOCAB_FILES_NAMES
from transformers.tokenization_utils_base import TOKENIZER_CONFIG_FILE
from transformers.utils import WEIGHTS_NAME, logging
logging.set_verbosity_warning()
_snake_case : Union[str, Any] = 2
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , *, # begin keyword-only arguments
lowerCAmelCase_ : str="<s>" , lowerCAmelCase_ : Dict="<pad>" , lowerCAmelCase_ : Any="</s>" , lowerCAmelCase_ : List[str]="<unk>" , lowerCAmelCase_ : Optional[Any]=None , ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = bos, unk, pad, eos
__lowerCAmelCase = []
__lowerCAmelCase = []
__lowerCAmelCase = {}
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
if extra_special_symbols:
for s in extra_special_symbols:
self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = len(self.symbols )
def __eq__( self : Dict , lowerCAmelCase_ : Dict ) -> str:
return self.indices == other.indices
def __getitem__( self : List[Any] , lowerCAmelCase_ : int ) -> Union[str, Any]:
if idx < len(self.symbols ):
return self.symbols[idx]
return self.unk_word
def __len__( self : Tuple ) -> List[Any]:
return len(self.symbols )
def __contains__( self : Optional[Any] , lowerCAmelCase_ : Dict ) -> Optional[int]:
return sym in self.indices
@classmethod
def lowercase ( cls : Dict , lowerCAmelCase_ : str ) -> str:
__lowerCAmelCase = cls()
d.add_from_file(lowerCAmelCase_ )
return d
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int=1 , lowerCAmelCase_ : Any=False ) -> Optional[Any]:
if word in self.indices and not overwrite:
__lowerCAmelCase = self.indices[word]
__lowerCAmelCase = self.count[idx] + n
return idx
else:
__lowerCAmelCase = len(self.symbols )
__lowerCAmelCase = idx
self.symbols.append(lowerCAmelCase_ )
self.count.append(lowerCAmelCase_ )
return idx
def lowercase ( self : str , lowerCAmelCase_ : Union[str, Any] ) -> Dict:
return 0
def lowercase ( self : Tuple , lowerCAmelCase_ : Union[str, Any] ) -> int:
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
try:
with open(lowerCAmelCase_ , 'r' , encoding='utf-8' ) as fd:
self.add_from_file(lowerCAmelCase_ )
except FileNotFoundError as fnfe:
raise fnfe
except UnicodeError:
raise Exception('Incorrect encoding detected in {}, please rebuild the dataset'.format(lowerCAmelCase_ ) )
return
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = self._load_meta(lowerCAmelCase_ )
for line in lines[indices_start_line:]:
try:
__lowerCAmelCase , __lowerCAmelCase = line.rstrip().rsplit(' ' , 1 )
if field == "#fairseq:overwrite":
__lowerCAmelCase = True
__lowerCAmelCase , __lowerCAmelCase = line.rsplit(' ' , 1 )
else:
__lowerCAmelCase = False
__lowerCAmelCase = int(lowerCAmelCase_ )
__lowerCAmelCase = line
if word in self and not overwrite:
raise RuntimeError(
'Duplicate word found when loading Dictionary: \'{}\'. '
'Duplicate words can overwrite earlier ones by adding the '
'#fairseq:overwrite flag at the end of the corresponding row '
'in the dictionary file. If using the Camembert model, please '
'download an updated copy of the model file.'.format(lowerCAmelCase_ ) )
self.add_symbol(lowerCAmelCase_ , n=lowerCAmelCase_ , overwrite=lowerCAmelCase_ )
except ValueError:
raise ValueError('Incorrect dictionary format, expected \'<token> <cnt> [flags]\'' )
def a_ ( lowerCAmelCase_ : List[str] ):
# (1) remove word breaking symbol, (2) add word ending symbol where the word is not broken up,
# e.g.: d = {'le@@': 5, 'tt@@': 6, 'er': 7} => {'le': 5, 'tt': 6, 'er</w>': 7}
__lowerCAmelCase = dict((re.sub(R'@@$', '', lowerCAmelCase_ ), v) if k.endswith('@@' ) else (re.sub(R'$', '</w>', lowerCAmelCase_ ), v) for k, v in d.items() )
__lowerCAmelCase = '<s> <pad> </s> <unk>'.split()
# restore the special tokens
for k in keep_keys:
del da[F"""{k}</w>"""]
__lowerCAmelCase = d[k] # restore
return da
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str] ):
# prep
if not os.path.exists(lowerCAmelCase_ ):
raise ValueError(F"""path {biogpt_checkpoint_path} does not exist!""" )
os.makedirs(lowerCAmelCase_, exist_ok=lowerCAmelCase_ )
print(F"""Writing results to {pytorch_dump_folder_path}""" )
# handle various types of models
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'checkpoint.pt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {checkpoint_file} does not exist!""" )
__lowerCAmelCase = torch.load(lowerCAmelCase_, map_location='cpu' )
__lowerCAmelCase = chkpt['cfg']['model']
# dicts
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'dict.txt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {dict_file} does not exist!""" )
__lowerCAmelCase = Dictionary.load(lowerCAmelCase_ )
__lowerCAmelCase = rewrite_dict_keys(src_dict.indices )
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['vocab_file'] )
print(F"""Generating {src_vocab_file} of {src_vocab_size} records""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# merges_file (bpecodes)
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'bpecodes' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {bpecodes_file} does not exist!""" )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['merges_file'] )
shutil.copyfile(lowerCAmelCase_, lowerCAmelCase_ )
# model config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'config.json' )
__lowerCAmelCase = {
'activation_dropout': args['activation_dropout'],
'architectures': ['BioGptForCausalLM'],
'attention_probs_dropout_prob': args['attention_dropout'],
'bos_token_id': 0,
'eos_token_id': 2,
'hidden_act': args['activation_fn'],
'hidden_dropout_prob': args['dropout'],
'hidden_size': args['decoder_embed_dim'],
'initializer_range': 0.02,
'intermediate_size': args['decoder_ffn_embed_dim'],
'layer_norm_eps': 1E-12,
'layerdrop': args['decoder_layerdrop'],
'max_position_embeddings': args['max_target_positions'],
'model_type': 'biogpt',
'num_attention_heads': args['decoder_attention_heads'],
'num_hidden_layers': args['decoder_layers'],
'pad_token_id': 1,
'scale_embedding': not args['no_scale_embedding'],
'tie_word_embeddings': args['share_decoder_input_output_embed'],
'vocab_size': src_vocab_size,
}
# good hparam defaults to start with
print(F"""Generating {biogpt_model_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# tokenizer config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = {
'bos_token': '<s>',
'eos_token': '</s>',
'model_max_length': 1024,
'pad_token': '<pad>',
'special_tokens_map_file': None,
'tokenizer_class': 'BioGptTokenizer',
'unk_token': '<unk>',
}
print(F"""Generating {biogpt_tokenizer_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# model
__lowerCAmelCase = chkpt['model']
# remove unneeded keys
__lowerCAmelCase = [
'decoder.version',
]
for k in ignore_keys:
model_state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = list(model_state_dict.keys() )
for layer_name in layer_names:
if layer_name.endswith('output_projection.weight' ):
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
else:
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
__lowerCAmelCase = BioGptConfig.from_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = BioGptForCausalLM(lowerCAmelCase_ )
# check that it loads ok
model_new.load_state_dict(lowerCAmelCase_ )
# save
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
print(F"""Generating {pytorch_weights_dump_path}""" )
torch.save(lowerCAmelCase_, lowerCAmelCase_ )
print('Conversion is done!' )
if __name__ == "__main__":
_snake_case : str = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--biogpt_checkpoint_path',
default=None,
type=str,
required=True,
help=(
'Path to the official PyTorch checkpoint file which is expected to reside in the dump dir with dicts,'
' bpecodes, etc.'
),
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_biogpt_checkpoint_to_pytorch(args.biogpt_checkpoint_path, args.pytorch_dump_folder_path)
| 53 | 1 |
import tempfile
import unittest
import numpy as np
from diffusers import (
DDIMScheduler,
DPMSolverMultistepScheduler,
EulerAncestralDiscreteScheduler,
EulerDiscreteScheduler,
LMSDiscreteScheduler,
OnnxStableDiffusionPipeline,
PNDMScheduler,
)
from diffusers.utils.testing_utils import is_onnx_available, nightly, require_onnxruntime, require_torch_gpu
from ..test_pipelines_onnx_common import OnnxPipelineTesterMixin
if is_onnx_available():
import onnxruntime as ort
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = """hf-internal-testing/tiny-random-OnnxStableDiffusionPipeline"""
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Tuple=0 ) -> str:
__lowerCAmelCase = np.random.RandomState(lowerCAmelCase_ )
__lowerCAmelCase = {
'prompt': 'A painting of a squirrel eating a burger',
'generator': generator,
'num_inference_steps': 2,
'guidance_scale': 7.5,
'output_type': 'numpy',
}
return inputs
def lowercase ( self : Tuple ) -> List[str]:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(self.hub_checkpoint , provider='CPUExecutionProvider' )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = pipe(**lowerCAmelCase_ ).images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 1_2_8, 1_2_8, 3)
__lowerCAmelCase = np.array([0.6_50_72, 0.5_84_92, 0.4_82_19, 0.5_55_21, 0.5_31_80, 0.5_59_39, 0.5_06_97, 0.3_98_00, 0.4_64_55] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def lowercase ( self : Optional[Any] ) -> Optional[Any]:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(self.hub_checkpoint , provider='CPUExecutionProvider' )
__lowerCAmelCase = PNDMScheduler.from_config(pipe.scheduler.config , skip_prk_steps=lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = pipe(**lowerCAmelCase_ ).images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 1_2_8, 1_2_8, 3)
__lowerCAmelCase = np.array([0.6_58_63, 0.5_94_25, 0.4_93_26, 0.5_63_13, 0.5_38_75, 0.5_66_27, 0.5_10_65, 0.3_97_77, 0.4_63_30] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def lowercase ( self : Optional[int] ) -> Optional[Any]:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(self.hub_checkpoint , provider='CPUExecutionProvider' )
__lowerCAmelCase = LMSDiscreteScheduler.from_config(pipe.scheduler.config )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = pipe(**lowerCAmelCase_ ).images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 1_2_8, 1_2_8, 3)
__lowerCAmelCase = np.array([0.5_37_55, 0.6_07_86, 0.4_74_02, 0.4_94_88, 0.5_18_69, 0.4_98_19, 0.4_79_85, 0.3_89_57, 0.4_42_79] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def lowercase ( self : str ) -> Union[str, Any]:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(self.hub_checkpoint , provider='CPUExecutionProvider' )
__lowerCAmelCase = EulerDiscreteScheduler.from_config(pipe.scheduler.config )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = pipe(**lowerCAmelCase_ ).images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 1_2_8, 1_2_8, 3)
__lowerCAmelCase = np.array([0.5_37_55, 0.6_07_86, 0.4_74_02, 0.4_94_88, 0.5_18_69, 0.4_98_19, 0.4_79_85, 0.3_89_57, 0.4_42_79] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def lowercase ( self : Any ) -> int:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(self.hub_checkpoint , provider='CPUExecutionProvider' )
__lowerCAmelCase = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = pipe(**lowerCAmelCase_ ).images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 1_2_8, 1_2_8, 3)
__lowerCAmelCase = np.array([0.5_38_17, 0.6_08_12, 0.4_73_84, 0.4_95_30, 0.5_18_94, 0.4_98_14, 0.4_79_84, 0.3_89_58, 0.4_42_71] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def lowercase ( self : List[str] ) -> Optional[Any]:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(self.hub_checkpoint , provider='CPUExecutionProvider' )
__lowerCAmelCase = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = pipe(**lowerCAmelCase_ ).images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 1_2_8, 1_2_8, 3)
__lowerCAmelCase = np.array([0.5_38_95, 0.6_08_08, 0.4_79_33, 0.4_96_08, 0.5_18_86, 0.4_99_50, 0.4_80_53, 0.3_89_57, 0.4_42_00] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
def lowercase ( self : Optional[int] ) -> int:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(self.hub_checkpoint , provider='CPUExecutionProvider' )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = 3 * [inputs['prompt']]
# forward
__lowerCAmelCase = pipe(**lowerCAmelCase_ )
__lowerCAmelCase = output.images[0, -3:, -3:, -1]
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = 3 * [inputs.pop('prompt' )]
__lowerCAmelCase = pipe.tokenizer(
lowerCAmelCase_ , padding='max_length' , max_length=pipe.tokenizer.model_max_length , truncation=lowerCAmelCase_ , return_tensors='np' , )
__lowerCAmelCase = text_inputs['input_ids']
__lowerCAmelCase = pipe.text_encoder(input_ids=text_inputs.astype(np.intaa ) )[0]
__lowerCAmelCase = prompt_embeds
# forward
__lowerCAmelCase = pipe(**lowerCAmelCase_ )
__lowerCAmelCase = output.images[0, -3:, -3:, -1]
assert np.abs(image_slice_a.flatten() - image_slice_a.flatten() ).max() < 1e-4
def lowercase ( self : List[str] ) -> Optional[Any]:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(self.hub_checkpoint , provider='CPUExecutionProvider' )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = 3 * ['this is a negative prompt']
__lowerCAmelCase = negative_prompt
__lowerCAmelCase = 3 * [inputs['prompt']]
# forward
__lowerCAmelCase = pipe(**lowerCAmelCase_ )
__lowerCAmelCase = output.images[0, -3:, -3:, -1]
__lowerCAmelCase = self.get_dummy_inputs()
__lowerCAmelCase = 3 * [inputs.pop('prompt' )]
__lowerCAmelCase = []
for p in [prompt, negative_prompt]:
__lowerCAmelCase = pipe.tokenizer(
lowerCAmelCase_ , padding='max_length' , max_length=pipe.tokenizer.model_max_length , truncation=lowerCAmelCase_ , return_tensors='np' , )
__lowerCAmelCase = text_inputs['input_ids']
embeds.append(pipe.text_encoder(input_ids=text_inputs.astype(np.intaa ) )[0] )
__lowerCAmelCase , __lowerCAmelCase = embeds
# forward
__lowerCAmelCase = pipe(**lowerCAmelCase_ )
__lowerCAmelCase = output.images[0, -3:, -3:, -1]
assert np.abs(image_slice_a.flatten() - image_slice_a.flatten() ).max() < 1e-4
@nightly
@require_onnxruntime
@require_torch_gpu
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@property
def lowercase ( self : Any ) -> Optional[int]:
return (
"CUDAExecutionProvider",
{
"gpu_mem_limit": "15000000000", # 15GB
"arena_extend_strategy": "kSameAsRequested",
},
)
@property
def lowercase ( self : List[Any] ) -> str:
__lowerCAmelCase = ort.SessionOptions()
__lowerCAmelCase = False
return options
def lowercase ( self : int ) -> Dict:
# using the PNDM scheduler by default
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(
'CompVis/stable-diffusion-v1-4' , revision='onnx' , safety_checker=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ , provider=self.gpu_provider , sess_options=self.gpu_options , )
sd_pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = 'A painting of a squirrel eating a burger'
np.random.seed(0 )
__lowerCAmelCase = sd_pipe([prompt] , guidance_scale=6.0 , num_inference_steps=1_0 , output_type='np' )
__lowerCAmelCase = output.images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 5_1_2, 5_1_2, 3)
__lowerCAmelCase = np.array([0.04_52, 0.03_90, 0.00_87, 0.03_50, 0.06_17, 0.03_64, 0.05_44, 0.05_23, 0.07_20] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-3
def lowercase ( self : Optional[int] ) -> Tuple:
__lowerCAmelCase = DDIMScheduler.from_pretrained(
'runwayml/stable-diffusion-v1-5' , subfolder='scheduler' , revision='onnx' )
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(
'runwayml/stable-diffusion-v1-5' , revision='onnx' , scheduler=lowerCAmelCase_ , safety_checker=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ , provider=self.gpu_provider , sess_options=self.gpu_options , )
sd_pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = 'open neural network exchange'
__lowerCAmelCase = np.random.RandomState(0 )
__lowerCAmelCase = sd_pipe([prompt] , guidance_scale=7.5 , num_inference_steps=1_0 , generator=lowerCAmelCase_ , output_type='np' )
__lowerCAmelCase = output.images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 5_1_2, 5_1_2, 3)
__lowerCAmelCase = np.array([0.28_67, 0.19_74, 0.14_81, 0.72_94, 0.72_51, 0.66_67, 0.41_94, 0.56_42, 0.64_86] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-3
def lowercase ( self : List[str] ) -> int:
__lowerCAmelCase = LMSDiscreteScheduler.from_pretrained(
'runwayml/stable-diffusion-v1-5' , subfolder='scheduler' , revision='onnx' )
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(
'runwayml/stable-diffusion-v1-5' , revision='onnx' , scheduler=lowerCAmelCase_ , safety_checker=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ , provider=self.gpu_provider , sess_options=self.gpu_options , )
sd_pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = 'open neural network exchange'
__lowerCAmelCase = np.random.RandomState(0 )
__lowerCAmelCase = sd_pipe([prompt] , guidance_scale=7.5 , num_inference_steps=1_0 , generator=lowerCAmelCase_ , output_type='np' )
__lowerCAmelCase = output.images
__lowerCAmelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 5_1_2, 5_1_2, 3)
__lowerCAmelCase = np.array([0.23_06, 0.19_59, 0.15_93, 0.65_49, 0.63_94, 0.54_08, 0.50_65, 0.60_10, 0.61_61] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-3
def lowercase ( self : List[str] ) -> List[Any]:
__lowerCAmelCase = 0
def test_callback_fn(lowerCAmelCase_ : int , lowerCAmelCase_ : int , lowerCAmelCase_ : np.ndarray ) -> None:
__lowerCAmelCase = True
nonlocal number_of_steps
number_of_steps += 1
if step == 0:
assert latents.shape == (1, 4, 6_4, 6_4)
__lowerCAmelCase = latents[0, -3:, -3:, -1]
__lowerCAmelCase = np.array(
[-0.67_72, -0.38_35, -1.24_56, 0.19_05, -1.09_74, 0.69_67, -1.93_53, 0.01_78, 1.01_67] )
assert np.abs(latents_slice.flatten() - expected_slice ).max() < 1e-3
elif step == 5:
assert latents.shape == (1, 4, 6_4, 6_4)
__lowerCAmelCase = latents[0, -3:, -3:, -1]
__lowerCAmelCase = np.array(
[-0.33_51, 0.22_41, -0.18_37, -0.23_25, -0.65_77, 0.33_93, -0.02_41, 0.58_99, 1.38_75] )
assert np.abs(latents_slice.flatten() - expected_slice ).max() < 1e-3
__lowerCAmelCase = False
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(
'runwayml/stable-diffusion-v1-5' , revision='onnx' , safety_checker=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ , provider=self.gpu_provider , sess_options=self.gpu_options , )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = 'Andromeda galaxy in a bottle'
__lowerCAmelCase = np.random.RandomState(0 )
pipe(
prompt=lowerCAmelCase_ , num_inference_steps=5 , guidance_scale=7.5 , generator=lowerCAmelCase_ , callback=lowerCAmelCase_ , callback_steps=1 , )
assert test_callback_fn.has_been_called
assert number_of_steps == 6
def lowercase ( self : Optional[int] ) -> Optional[Any]:
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(
'runwayml/stable-diffusion-v1-5' , revision='onnx' , safety_checker=lowerCAmelCase_ , feature_extractor=lowerCAmelCase_ , provider=self.gpu_provider , sess_options=self.gpu_options , )
assert isinstance(lowerCAmelCase_ , lowerCAmelCase_ )
assert pipe.safety_checker is None
__lowerCAmelCase = pipe('example prompt' , num_inference_steps=2 ).images[0]
assert image is not None
# check that there's no error when saving a pipeline with one of the models being None
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = OnnxStableDiffusionPipeline.from_pretrained(lowerCAmelCase_ )
# sanity check that the pipeline still works
assert pipe.safety_checker is None
__lowerCAmelCase = pipe('example prompt' , num_inference_steps=2 ).images[0]
assert image is not None
| 53 |
from typing import Tuple, Union
from ...modeling_outputs import BackboneOutput
from ...modeling_utils import PreTrainedModel
from ...utils import is_timm_available, is_torch_available, requires_backends
from ...utils.backbone_utils import BackboneMixin
from .configuration_timm_backbone import TimmBackboneConfig
if is_timm_available():
import timm
if is_torch_available():
from torch import Tensor
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = """pixel_values"""
a_ = False
a_ = TimmBackboneConfig
def __init__( self : Tuple , lowerCAmelCase_ : Any , **lowerCAmelCase_ : Optional[int] ) -> Optional[Any]:
requires_backends(self , 'timm' )
super().__init__(lowerCAmelCase_ )
__lowerCAmelCase = config
if config.backbone is None:
raise ValueError('backbone is not set in the config. Please set it to a timm model name.' )
if config.backbone not in timm.list_models():
raise ValueError(f"""backbone {config.backbone} is not supported by timm.""" )
if hasattr(lowerCAmelCase_ , 'out_features' ) and config.out_features is not None:
raise ValueError('out_features is not supported by TimmBackbone. Please use out_indices instead.' )
__lowerCAmelCase = getattr(lowerCAmelCase_ , 'use_pretrained_backbone' , lowerCAmelCase_ )
if pretrained is None:
raise ValueError('use_pretrained_backbone is not set in the config. Please set it to True or False.' )
# We just take the final layer by default. This matches the default for the transformers models.
__lowerCAmelCase = config.out_indices if getattr(lowerCAmelCase_ , 'out_indices' , lowerCAmelCase_ ) is not None else (-1,)
__lowerCAmelCase = timm.create_model(
config.backbone , pretrained=lowerCAmelCase_ , features_only=config.features_only , in_chans=config.num_channels , out_indices=lowerCAmelCase_ , **lowerCAmelCase_ , )
# These are used to control the output of the model when called. If output_hidden_states is True, then
# return_layers is modified to include all layers.
__lowerCAmelCase = self._backbone.return_layers
__lowerCAmelCase = {layer['module']: str(lowerCAmelCase_ ) for i, layer in enumerate(self._backbone.feature_info.info )}
super()._init_backbone(lowerCAmelCase_ )
@classmethod
def lowercase ( cls : int , lowerCAmelCase_ : Dict , *lowerCAmelCase_ : Dict , **lowerCAmelCase_ : Union[str, Any] ) -> Optional[int]:
requires_backends(cls , ['vision', 'timm'] )
from ...models.timm_backbone import TimmBackboneConfig
__lowerCAmelCase = kwargs.pop('config' , TimmBackboneConfig() )
__lowerCAmelCase = kwargs.pop('use_timm_backbone' , lowerCAmelCase_ )
if not use_timm:
raise ValueError('use_timm_backbone must be True for timm backbones' )
__lowerCAmelCase = kwargs.pop('num_channels' , config.num_channels )
__lowerCAmelCase = kwargs.pop('features_only' , config.features_only )
__lowerCAmelCase = kwargs.pop('use_pretrained_backbone' , config.use_pretrained_backbone )
__lowerCAmelCase = kwargs.pop('out_indices' , config.out_indices )
__lowerCAmelCase = TimmBackboneConfig(
backbone=lowerCAmelCase_ , num_channels=lowerCAmelCase_ , features_only=lowerCAmelCase_ , use_pretrained_backbone=lowerCAmelCase_ , out_indices=lowerCAmelCase_ , )
return super()._from_config(lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Tuple , lowerCAmelCase_ : int ) -> Dict:
pass
def lowercase ( self : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : Any=None , **lowerCAmelCase_ : Dict ) -> Union[BackboneOutput, Tuple[Tensor, ...]]:
__lowerCAmelCase = return_dict if return_dict is not None else self.config.use_return_dict
__lowerCAmelCase = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__lowerCAmelCase = output_attentions if output_attentions is not None else self.config.output_attentions
if output_attentions:
raise ValueError('Cannot output attentions for timm backbones at the moment' )
if output_hidden_states:
# We modify the return layers to include all the stages of the backbone
__lowerCAmelCase = self._all_layers
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = self._return_layers
__lowerCAmelCase = tuple(hidden_states[i] for i in self.out_indices )
else:
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = tuple(lowerCAmelCase_ )
__lowerCAmelCase = tuple(lowerCAmelCase_ ) if hidden_states is not None else None
if not return_dict:
__lowerCAmelCase = (feature_maps,)
if output_hidden_states:
__lowerCAmelCase = output + (hidden_states,)
return output
return BackboneOutput(feature_maps=lowerCAmelCase_ , hidden_states=lowerCAmelCase_ , attentions=lowerCAmelCase_ )
| 53 | 1 |
import math
from collections import defaultdict
from typing import List, Optional, Tuple, Union
import numpy as np
import torch
from ..configuration_utils import ConfigMixin, register_to_config
from .scheduling_utils import KarrasDiffusionSchedulers, SchedulerMixin, SchedulerOutput
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : List[Any]=0.999, lowerCAmelCase_ : List[str]="cosine", ):
if alpha_transform_type == "cosine":
def alpha_bar_fn(lowerCAmelCase_ : int ):
return math.cos((t + 0.008) / 1.008 * math.pi / 2 ) ** 2
elif alpha_transform_type == "exp":
def alpha_bar_fn(lowerCAmelCase_ : str ):
return math.exp(t * -12.0 )
else:
raise ValueError(F"""Unsupported alpha_tranform_type: {alpha_transform_type}""" )
__lowerCAmelCase = []
for i in range(lowerCAmelCase_ ):
__lowerCAmelCase = i / num_diffusion_timesteps
__lowerCAmelCase = (i + 1) / num_diffusion_timesteps
betas.append(min(1 - alpha_bar_fn(lowerCAmelCase_ ) / alpha_bar_fn(lowerCAmelCase_ ), lowerCAmelCase_ ) )
return torch.tensor(lowerCAmelCase_, dtype=torch.floataa )
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = [e.name for e in KarrasDiffusionSchedulers]
a_ = 2
@register_to_config
def __init__( self : int , lowerCAmelCase_ : int = 1_0_0_0 , lowerCAmelCase_ : float = 0.0_00_85 , lowerCAmelCase_ : float = 0.0_12 , lowerCAmelCase_ : str = "linear" , lowerCAmelCase_ : Optional[Union[np.ndarray, List[float]]] = None , lowerCAmelCase_ : str = "epsilon" , lowerCAmelCase_ : Optional[bool] = False , lowerCAmelCase_ : Optional[bool] = False , lowerCAmelCase_ : float = 1.0 , lowerCAmelCase_ : str = "linspace" , lowerCAmelCase_ : int = 0 , ) -> List[Any]:
if trained_betas is not None:
__lowerCAmelCase = torch.tensor(lowerCAmelCase_ , dtype=torch.floataa )
elif beta_schedule == "linear":
__lowerCAmelCase = torch.linspace(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , dtype=torch.floataa )
elif beta_schedule == "scaled_linear":
# this schedule is very specific to the latent diffusion model.
__lowerCAmelCase = (
torch.linspace(beta_start**0.5 , beta_end**0.5 , lowerCAmelCase_ , dtype=torch.floataa ) ** 2
)
elif beta_schedule == "squaredcos_cap_v2":
# Glide cosine schedule
__lowerCAmelCase = betas_for_alpha_bar(lowerCAmelCase_ , alpha_transform_type='cosine' )
elif beta_schedule == "exp":
__lowerCAmelCase = betas_for_alpha_bar(lowerCAmelCase_ , alpha_transform_type='exp' )
else:
raise NotImplementedError(f"""{beta_schedule} does is not implemented for {self.__class__}""" )
__lowerCAmelCase = 1.0 - self.betas
__lowerCAmelCase = torch.cumprod(self.alphas , dim=0 )
# set all values
self.set_timesteps(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = use_karras_sigmas
def lowercase ( self : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[Any]=None ) -> str:
if schedule_timesteps is None:
__lowerCAmelCase = self.timesteps
__lowerCAmelCase = (schedule_timesteps == timestep).nonzero()
# The sigma index that is taken for the **very** first `step`
# is always the second index (or the last index if there is only 1)
# This way we can ensure we don't accidentally skip a sigma in
# case we start in the middle of the denoising schedule (e.g. for image-to-image)
if len(self._index_counter ) == 0:
__lowerCAmelCase = 1 if len(lowerCAmelCase_ ) > 1 else 0
else:
__lowerCAmelCase = timestep.cpu().item() if torch.is_tensor(lowerCAmelCase_ ) else timestep
__lowerCAmelCase = self._index_counter[timestep_int]
return indices[pos].item()
@property
def lowercase ( self : Any ) -> List[Any]:
# standard deviation of the initial noise distribution
if self.config.timestep_spacing in ["linspace", "trailing"]:
return self.sigmas.max()
return (self.sigmas.max() ** 2 + 1) ** 0.5
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : Union[float, torch.FloatTensor] , ) -> torch.FloatTensor:
__lowerCAmelCase = self.index_for_timestep(lowerCAmelCase_ )
__lowerCAmelCase = self.sigmas[step_index]
__lowerCAmelCase = sample / ((sigma**2 + 1) ** 0.5)
return sample
def lowercase ( self : List[Any] , lowerCAmelCase_ : int , lowerCAmelCase_ : Union[str, torch.device] = None , lowerCAmelCase_ : Optional[int] = None , ) -> Union[str, Any]:
__lowerCAmelCase = num_inference_steps
__lowerCAmelCase = num_train_timesteps or self.config.num_train_timesteps
# "linspace", "leading", "trailing" corresponds to annotation of Table 2. of https://arxiv.org/abs/2305.08891
if self.config.timestep_spacing == "linspace":
__lowerCAmelCase = np.linspace(0 , num_train_timesteps - 1 , lowerCAmelCase_ , dtype=lowerCAmelCase_ )[::-1].copy()
elif self.config.timestep_spacing == "leading":
__lowerCAmelCase = num_train_timesteps // self.num_inference_steps
# creates integer timesteps by multiplying by ratio
# casting to int to avoid issues when num_inference_step is power of 3
__lowerCAmelCase = (np.arange(0 , lowerCAmelCase_ ) * step_ratio).round()[::-1].copy().astype(lowerCAmelCase_ )
timesteps += self.config.steps_offset
elif self.config.timestep_spacing == "trailing":
__lowerCAmelCase = num_train_timesteps / self.num_inference_steps
# creates integer timesteps by multiplying by ratio
# casting to int to avoid issues when num_inference_step is power of 3
__lowerCAmelCase = (np.arange(lowerCAmelCase_ , 0 , -step_ratio )).round().copy().astype(lowerCAmelCase_ )
timesteps -= 1
else:
raise ValueError(
f"""{self.config.timestep_spacing} is not supported. Please make sure to choose one of 'linspace', 'leading' or 'trailing'.""" )
__lowerCAmelCase = np.array(((1 - self.alphas_cumprod) / self.alphas_cumprod) ** 0.5 )
__lowerCAmelCase = np.log(lowerCAmelCase_ )
__lowerCAmelCase = np.interp(lowerCAmelCase_ , np.arange(0 , len(lowerCAmelCase_ ) ) , lowerCAmelCase_ )
if self.config.use_karras_sigmas:
__lowerCAmelCase = self._convert_to_karras(in_sigmas=lowerCAmelCase_ , num_inference_steps=self.num_inference_steps )
__lowerCAmelCase = np.array([self._sigma_to_t(lowerCAmelCase_ , lowerCAmelCase_ ) for sigma in sigmas] )
__lowerCAmelCase = np.concatenate([sigmas, [0.0]] ).astype(np.floataa )
__lowerCAmelCase = torch.from_numpy(lowerCAmelCase_ ).to(device=lowerCAmelCase_ )
__lowerCAmelCase = torch.cat([sigmas[:1], sigmas[1:-1].repeat_interleave(2 ), sigmas[-1:]] )
__lowerCAmelCase = torch.from_numpy(lowerCAmelCase_ )
__lowerCAmelCase = torch.cat([timesteps[:1], timesteps[1:].repeat_interleave(2 )] )
if str(lowerCAmelCase_ ).startswith('mps' ):
# mps does not support float64
__lowerCAmelCase = timesteps.to(lowerCAmelCase_ , dtype=torch.floataa )
else:
__lowerCAmelCase = timesteps.to(device=lowerCAmelCase_ )
# empty dt and derivative
__lowerCAmelCase = None
__lowerCAmelCase = None
# for exp beta schedules, such as the one for `pipeline_shap_e.py`
# we need an index counter
__lowerCAmelCase = defaultdict(lowerCAmelCase_ )
def lowercase ( self : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : List[Any] ) -> int:
# get log sigma
__lowerCAmelCase = np.log(lowerCAmelCase_ )
# get distribution
__lowerCAmelCase = log_sigma - log_sigmas[:, np.newaxis]
# get sigmas range
__lowerCAmelCase = np.cumsum((dists >= 0) , axis=0 ).argmax(axis=0 ).clip(max=log_sigmas.shape[0] - 2 )
__lowerCAmelCase = low_idx + 1
__lowerCAmelCase = log_sigmas[low_idx]
__lowerCAmelCase = log_sigmas[high_idx]
# interpolate sigmas
__lowerCAmelCase = (low - log_sigma) / (low - high)
__lowerCAmelCase = np.clip(lowerCAmelCase_ , 0 , 1 )
# transform interpolation to time range
__lowerCAmelCase = (1 - w) * low_idx + w * high_idx
__lowerCAmelCase = t.reshape(sigma.shape )
return t
def lowercase ( self : Tuple , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : Optional[Any] ) -> torch.FloatTensor:
__lowerCAmelCase = in_sigmas[-1].item()
__lowerCAmelCase = in_sigmas[0].item()
__lowerCAmelCase = 7.0 # 7.0 is the value used in the paper
__lowerCAmelCase = np.linspace(0 , 1 , lowerCAmelCase_ )
__lowerCAmelCase = sigma_min ** (1 / rho)
__lowerCAmelCase = sigma_max ** (1 / rho)
__lowerCAmelCase = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
return sigmas
@property
def lowercase ( self : List[str] ) -> Any:
return self.dt is None
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[torch.FloatTensor, np.ndarray] , lowerCAmelCase_ : Union[float, torch.FloatTensor] , lowerCAmelCase_ : Union[torch.FloatTensor, np.ndarray] , lowerCAmelCase_ : bool = True , ) -> Union[SchedulerOutput, Tuple]:
__lowerCAmelCase = self.index_for_timestep(lowerCAmelCase_ )
# advance index counter by 1
__lowerCAmelCase = timestep.cpu().item() if torch.is_tensor(lowerCAmelCase_ ) else timestep
self._index_counter[timestep_int] += 1
if self.state_in_first_order:
__lowerCAmelCase = self.sigmas[step_index]
__lowerCAmelCase = self.sigmas[step_index + 1]
else:
# 2nd order / Heun's method
__lowerCAmelCase = self.sigmas[step_index - 1]
__lowerCAmelCase = self.sigmas[step_index]
# currently only gamma=0 is supported. This usually works best anyways.
# We can support gamma in the future but then need to scale the timestep before
# passing it to the model which requires a change in API
__lowerCAmelCase = 0
__lowerCAmelCase = sigma * (gamma + 1) # Note: sigma_hat == sigma for now
# 1. compute predicted original sample (x_0) from sigma-scaled predicted noise
if self.config.prediction_type == "epsilon":
__lowerCAmelCase = sigma_hat if self.state_in_first_order else sigma_next
__lowerCAmelCase = sample - sigma_input * model_output
elif self.config.prediction_type == "v_prediction":
__lowerCAmelCase = sigma_hat if self.state_in_first_order else sigma_next
__lowerCAmelCase = model_output * (-sigma_input / (sigma_input**2 + 1) ** 0.5) + (
sample / (sigma_input**2 + 1)
)
elif self.config.prediction_type == "sample":
__lowerCAmelCase = model_output
else:
raise ValueError(
f"""prediction_type given as {self.config.prediction_type} must be one of `epsilon`, or `v_prediction`""" )
if self.config.clip_sample:
__lowerCAmelCase = pred_original_sample.clamp(
-self.config.clip_sample_range , self.config.clip_sample_range )
if self.state_in_first_order:
# 2. Convert to an ODE derivative for 1st order
__lowerCAmelCase = (sample - pred_original_sample) / sigma_hat
# 3. delta timestep
__lowerCAmelCase = sigma_next - sigma_hat
# store for 2nd order step
__lowerCAmelCase = derivative
__lowerCAmelCase = dt
__lowerCAmelCase = sample
else:
# 2. 2nd order / Heun's method
__lowerCAmelCase = (sample - pred_original_sample) / sigma_next
__lowerCAmelCase = (self.prev_derivative + derivative) / 2
# 3. take prev timestep & sample
__lowerCAmelCase = self.dt
__lowerCAmelCase = self.sample
# free dt and derivative
# Note, this puts the scheduler in "first order mode"
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = sample + derivative * dt
if not return_dict:
return (prev_sample,)
return SchedulerOutput(prev_sample=lowerCAmelCase_ )
def lowercase ( self : List[Any] , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : torch.FloatTensor , ) -> torch.FloatTensor:
# Make sure sigmas and timesteps have the same device and dtype as original_samples
__lowerCAmelCase = self.sigmas.to(device=original_samples.device , dtype=original_samples.dtype )
if original_samples.device.type == "mps" and torch.is_floating_point(lowerCAmelCase_ ):
# mps does not support float64
__lowerCAmelCase = self.timesteps.to(original_samples.device , dtype=torch.floataa )
__lowerCAmelCase = timesteps.to(original_samples.device , dtype=torch.floataa )
else:
__lowerCAmelCase = self.timesteps.to(original_samples.device )
__lowerCAmelCase = timesteps.to(original_samples.device )
__lowerCAmelCase = [self.index_for_timestep(lowerCAmelCase_ , lowerCAmelCase_ ) for t in timesteps]
__lowerCAmelCase = sigmas[step_indices].flatten()
while len(sigma.shape ) < len(original_samples.shape ):
__lowerCAmelCase = sigma.unsqueeze(-1 )
__lowerCAmelCase = original_samples + noise * sigma
return noisy_samples
def __len__( self : Union[str, Any] ) -> Union[str, Any]:
return self.config.num_train_timesteps
| 53 |
from __future__ import annotations
def a_ ( lowerCAmelCase_ : list[float] ):
if len(lowerCAmelCase_ ) < 2:
raise ValueError('Monogons and Digons are not polygons in the Euclidean space' )
if any(i <= 0 for i in nums ):
raise ValueError('All values must be greater than 0' )
__lowerCAmelCase = nums.copy()
copy_nums.sort()
return copy_nums[-1] < sum(copy_nums[:-1] )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 | 1 |
from math import ceil
def a_ ( lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Dict ):
__lowerCAmelCase = list(range(0, lowerCAmelCase_ ) )
__lowerCAmelCase = [item for sublist in list(device_map.values() ) for item in sublist]
# Duplicate check
__lowerCAmelCase = []
for i in device_map_blocks:
if device_map_blocks.count(lowerCAmelCase_ ) > 1 and i not in duplicate_blocks:
duplicate_blocks.append(lowerCAmelCase_ )
# Missing blocks
__lowerCAmelCase = [i for i in blocks if i not in device_map_blocks]
__lowerCAmelCase = [i for i in device_map_blocks if i not in blocks]
if len(lowerCAmelCase_ ) != 0:
raise ValueError(
'Duplicate attention blocks specified in device_map. Attention blocks must be specified to one device.'
' These attention blocks were specified more than once: ' + str(lowerCAmelCase_ ) )
if len(lowerCAmelCase_ ) != 0:
raise ValueError(
'There are attention blocks for this model that are not specified in the device_map. Add these attention '
'blocks to a device on the device_map: ' + str(lowerCAmelCase_ ) )
if len(lowerCAmelCase_ ) != 0:
raise ValueError(
'The device_map contains more attention blocks than this model has. Remove these from the device_map:'
+ str(lowerCAmelCase_ ) )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = list(range(lowerCAmelCase_ ) )
__lowerCAmelCase = int(ceil(n_layers / len(lowerCAmelCase_ ) ) )
__lowerCAmelCase = [layers[i : i + n_blocks] for i in range(0, lowerCAmelCase_, lowerCAmelCase_ )]
return dict(zip(lowerCAmelCase_, lowerCAmelCase_ ) )
| 53 |
import inspect
import unittest
from transformers import RegNetConfig, is_flax_available
from transformers.testing_utils import require_flax, slow
from transformers.utils import cached_property, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor
if is_flax_available():
import jax
import jax.numpy as jnp
from transformers.models.regnet.modeling_flax_regnet import FlaxRegNetForImageClassification, FlaxRegNetModel
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int]=3 , lowerCAmelCase_ : Dict=3_2 , lowerCAmelCase_ : Tuple=3 , lowerCAmelCase_ : Union[str, Any]=1_0 , lowerCAmelCase_ : List[str]=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Optional[int]=[1, 1, 2, 1] , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : Tuple="relu" , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : Optional[int]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = embeddings_size
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = hidden_act
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = len(lowerCAmelCase_ )
def lowercase ( self : Optional[int] ) -> List[Any]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : Tuple ) -> List[Any]:
return RegNetConfig(
num_channels=self.num_channels , embeddings_size=self.embeddings_size , hidden_sizes=self.hidden_sizes , depths=self.depths , hidden_act=self.hidden_act , num_labels=self.num_labels , image_size=self.image_size , )
def lowercase ( self : List[str] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : List[str] ) -> str:
__lowerCAmelCase = FlaxRegNetModel(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
# Output shape (b, c, h, w)
self.parent.assertEqual(
result.last_hidden_state.shape , (self.batch_size, self.hidden_sizes[-1], self.image_size // 3_2, self.image_size // 3_2) , )
def lowercase ( self : str , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : int ) -> Tuple:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = FlaxRegNetForImageClassification(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_flax
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (FlaxRegNetModel, FlaxRegNetForImageClassification) if is_flax_available() else ()
a_ = False
a_ = False
a_ = False
def lowercase ( self : Dict ) -> None:
__lowerCAmelCase = FlaxRegNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : int ) -> Optional[int]:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : str ) -> Union[str, Any]:
return
def lowercase ( self : Dict ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCAmelCase_ )
@unittest.skip(reason='RegNet does not use inputs_embeds' )
def lowercase ( self : Union[str, Any] ) -> Any:
pass
@unittest.skip(reason='RegNet does not support input and output embeddings' )
def lowercase ( self : Tuple ) -> Tuple:
pass
def lowercase ( self : Optional[Any] ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.__call__ )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
def check_hidden_states_output(lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : str ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
with self.subTest(model_class.__name__ ):
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model_class(lowerCAmelCase_ )
@jax.jit
def model_jitted(lowerCAmelCase_ : Optional[int] , **lowerCAmelCase_ : Dict ):
return model(pixel_values=lowerCAmelCase_ , **lowerCAmelCase_ )
with self.subTest('JIT Enabled' ):
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
with self.subTest('JIT Disabled' ):
with jax.disable_jit():
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
self.assertEqual(len(lowerCAmelCase_ ) , len(lowerCAmelCase_ ) )
for jitted_output, output in zip(lowerCAmelCase_ , lowerCAmelCase_ ):
self.assertEqual(jitted_output.shape , output.shape )
def a_ ( ):
__lowerCAmelCase = Image.open('./tests/fixtures/tests_samples/COCO/000000039769.png' )
return image
@require_flax
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
return AutoImageProcessor.from_pretrained('facebook/regnet-y-040' ) if is_vision_available() else None
@slow
def lowercase ( self : Optional[Any] ) -> Union[str, Any]:
__lowerCAmelCase = FlaxRegNetForImageClassification.from_pretrained('facebook/regnet-y-040' )
__lowerCAmelCase = self.default_image_processor
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = image_processor(images=lowerCAmelCase_ , return_tensors='np' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
# verify the logits
__lowerCAmelCase = (1, 1_0_0_0)
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = jnp.array([-0.41_80, -1.50_51, -3.48_36] )
self.assertTrue(jnp.allclose(outputs.logits[0, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import argparse
import json
import pickle
from pathlib import Path
import requests
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import MaskFormerConfig, MaskFormerForInstanceSegmentation, MaskFormerImageProcessor, SwinConfig
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Any = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = SwinConfig.from_pretrained(
'microsoft/swin-tiny-patch4-window7-224', out_features=['stage1', 'stage2', 'stage3', 'stage4'] )
__lowerCAmelCase = MaskFormerConfig(backbone_config=lowerCAmelCase_ )
__lowerCAmelCase = 'huggingface/label-files'
if "ade20k-full" in model_name:
# this should be ok
__lowerCAmelCase = 847
__lowerCAmelCase = 'maskformer-ade20k-full-id2label.json'
elif "ade" in model_name:
# this should be ok
__lowerCAmelCase = 150
__lowerCAmelCase = 'ade20k-id2label.json'
elif "coco-stuff" in model_name:
# this should be ok
__lowerCAmelCase = 171
__lowerCAmelCase = 'maskformer-coco-stuff-id2label.json'
elif "coco" in model_name:
# TODO
__lowerCAmelCase = 133
__lowerCAmelCase = 'coco-panoptic-id2label.json'
elif "cityscapes" in model_name:
# this should be ok
__lowerCAmelCase = 19
__lowerCAmelCase = 'cityscapes-id2label.json'
elif "vistas" in model_name:
# this should be ok
__lowerCAmelCase = 65
__lowerCAmelCase = 'mapillary-vistas-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
return config
def a_ ( lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = []
# stem
# fmt: off
rename_keys.append(('backbone.patch_embed.proj.weight', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.weight') )
rename_keys.append(('backbone.patch_embed.proj.bias', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.bias') )
rename_keys.append(('backbone.patch_embed.norm.weight', 'model.pixel_level_module.encoder.model.embeddings.norm.weight') )
rename_keys.append(('backbone.patch_embed.norm.bias', 'model.pixel_level_module.encoder.model.embeddings.norm.bias') )
# stages
for i in range(len(config.backbone_config.depths ) ):
for j in range(config.backbone_config.depths[i] ):
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_bias_table""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_bias_table""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_index""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_index""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.bias""") )
if i < 3:
rename_keys.append((F"""backbone.layers.{i}.downsample.reduction.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.reduction.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.bias""") )
rename_keys.append((F"""backbone.norm{i}.weight""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.weight""") )
rename_keys.append((F"""backbone.norm{i}.bias""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.bias""") )
# FPN
rename_keys.append(('sem_seg_head.layer_4.weight', 'model.pixel_level_module.decoder.fpn.stem.0.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.weight', 'model.pixel_level_module.decoder.fpn.stem.1.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.bias', 'model.pixel_level_module.decoder.fpn.stem.1.bias') )
for source_index, target_index in zip(range(3, 0, -1 ), range(0, 3 ) ):
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.0.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.bias""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.0.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.bias""") )
rename_keys.append(('sem_seg_head.mask_features.weight', 'model.pixel_level_module.decoder.mask_projection.weight') )
rename_keys.append(('sem_seg_head.mask_features.bias', 'model.pixel_level_module.decoder.mask_projection.bias') )
# Transformer decoder
for idx in range(config.decoder_config.decoder_layers ):
# self-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.bias""") )
# cross-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.bias""") )
# MLP 1
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc1.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc1.bias""") )
# MLP 2
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc2.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc2.bias""") )
# layernorm 1 (self-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.bias""") )
# layernorm 2 (cross-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.bias""") )
# layernorm 3 (final layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.weight""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.bias""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.bias""") )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.weight', 'model.transformer_module.decoder.layernorm.weight') )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.bias', 'model.transformer_module.decoder.layernorm.bias') )
# heads on top
rename_keys.append(('sem_seg_head.predictor.query_embed.weight', 'model.transformer_module.queries_embedder.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.weight', 'model.transformer_module.input_projection.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.bias', 'model.transformer_module.input_projection.bias') )
rename_keys.append(('sem_seg_head.predictor.class_embed.weight', 'class_predictor.weight') )
rename_keys.append(('sem_seg_head.predictor.class_embed.bias', 'class_predictor.bias') )
for i in range(3 ):
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.weight""", F"""mask_embedder.{i}.0.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.bias""", F"""mask_embedder.{i}.0.bias""") )
# fmt: on
return rename_keys
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int ):
__lowerCAmelCase = [int(backbone_config.embed_dim * 2**i ) for i in range(len(backbone_config.depths ) )]
for i in range(len(backbone_config.depths ) ):
__lowerCAmelCase = num_features[i]
for j in range(backbone_config.depths[i] ):
# fmt: off
# read in weights + bias of input projection layer (in original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[:dim, :]
__lowerCAmelCase = in_proj_bias[: dim]
__lowerCAmelCase = in_proj_weight[
dim : dim * 2, :
]
__lowerCAmelCase = in_proj_bias[
dim : dim * 2
]
__lowerCAmelCase = in_proj_weight[
-dim :, :
]
__lowerCAmelCase = in_proj_bias[-dim :]
# fmt: on
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : Dict ):
# fmt: off
__lowerCAmelCase = config.decoder_config.hidden_size
for idx in range(config.decoder_config.decoder_layers ):
# read in weights + bias of self-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# read in weights + bias of cross-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# fmt: on
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : bool = False ):
__lowerCAmelCase = get_maskformer_config(lowerCAmelCase_ )
# load original state_dict
with open(lowerCAmelCase_, 'rb' ) as f:
__lowerCAmelCase = pickle.load(lowerCAmelCase_ )
__lowerCAmelCase = data['model']
# for name, param in state_dict.items():
# print(name, param.shape)
# rename keys
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_swin_q_k_v(lowerCAmelCase_, config.backbone_config )
read_in_decoder_q_k_v(lowerCAmelCase_, lowerCAmelCase_ )
# update to torch tensors
for key, value in state_dict.items():
__lowerCAmelCase = torch.from_numpy(lowerCAmelCase_ )
# load 🤗 model
__lowerCAmelCase = MaskFormerForInstanceSegmentation(lowerCAmelCase_ )
model.eval()
for name, param in model.named_parameters():
print(lowerCAmelCase_, param.shape )
__lowerCAmelCase , __lowerCAmelCase = model.load_state_dict(lowerCAmelCase_, strict=lowerCAmelCase_ )
assert missing_keys == [
"model.pixel_level_module.encoder.model.layernorm.weight",
"model.pixel_level_module.encoder.model.layernorm.bias",
]
assert len(lowerCAmelCase_ ) == 0, F"""Unexpected keys: {unexpected_keys}"""
# verify results
__lowerCAmelCase = prepare_img()
if "vistas" in model_name:
__lowerCAmelCase = 65
elif "cityscapes" in model_name:
__lowerCAmelCase = 6_5535
else:
__lowerCAmelCase = 255
__lowerCAmelCase = True if 'ade' in model_name else False
__lowerCAmelCase = MaskFormerImageProcessor(ignore_index=lowerCAmelCase_, reduce_labels=lowerCAmelCase_ )
__lowerCAmelCase = image_processor(lowerCAmelCase_, return_tensors='pt' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
print('Logits:', outputs.class_queries_logits[0, :3, :3] )
if model_name == "maskformer-swin-tiny-ade":
__lowerCAmelCase = torch.tensor(
[[3.6353, -4.4770, -2.6065], [0.5081, -4.2394, -3.5343], [2.1909, -5.0353, -1.9323]] )
assert torch.allclose(outputs.class_queries_logits[0, :3, :3], lowerCAmelCase_, atol=1E-4 )
print('Looks ok!' )
if pytorch_dump_folder_path is not None:
print(F"""Saving model and image processor to {pytorch_dump_folder_path}""" )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
model.save_pretrained(lowerCAmelCase_ )
image_processor.save_pretrained(lowerCAmelCase_ )
if push_to_hub:
print('Pushing model and image processor to the hub...' )
model.push_to_hub(F"""nielsr/{model_name}""" )
image_processor.push_to_hub(F"""nielsr/{model_name}""" )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='maskformer-swin-tiny-ade',
type=str,
help=('Name of the MaskFormer model you\'d like to convert',),
)
parser.add_argument(
'--checkpoint_path',
default='/Users/nielsrogge/Documents/MaskFormer_checkpoints/MaskFormer-Swin-tiny-ADE20k/model.pkl',
type=str,
help='Path to the original state dict (.pth file).',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--push_to_hub', action='store_true', help='Whether or not to push the converted model to the 🤗 hub.'
)
_snake_case : List[str] = parser.parse_args()
convert_maskformer_checkpoint(
args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub
)
| 53 |
import json
import logging
import math
import os
import sys
from dataclasses import dataclass, field
from typing import Optional
from datasets import Dataset, load_dataset
import transformers
from transformers import (
CONFIG_MAPPING,
MODEL_FOR_MASKED_LM_MAPPING,
AutoConfig,
AutoModelForMaskedLM,
AutoTokenizer,
DataCollatorForWholeWordMask,
HfArgumentParser,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
_snake_case : Optional[int] = logging.getLogger(__name__)
_snake_case : Dict = list(MODEL_FOR_MASKED_LM_MAPPING.keys())
_snake_case : List[Any] = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """If training from scratch, pass a model type from the list: """ + """, """.join(_UpperCamelCase )} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Override some existing default config settings when a model is trained from scratch. Example: """
"""n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained config name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained tokenizer name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."""} , )
a_ = field(
default="""main""" , metadata={"""help""": """The specific model version to use (can be a branch name, tag name or commit id)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Will use the token generated when running `huggingface-cli login` (necessary to use this script """
"""with private models)."""
)
} , )
def lowercase ( self : List[Any] ) -> List[Any]:
if self.config_overrides is not None and (self.config_name is not None or self.model_name_or_path is not None):
raise ValueError(
'--config_overrides can\'t be used in combination with --config_name or --model_name_or_path' )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The name of the dataset to use (via the datasets library)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(default=_UpperCamelCase , metadata={"""help""": """The input training data file (a text file)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input evaluation data file to evaluate the perplexity on (a text file)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input train ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input validation ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached training and evaluation sets"""} )
a_ = field(
default=5 , metadata={
"""help""": """The percentage of the train set used as validation set in case there's no validation split"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The maximum total input sequence length after tokenization. Sequences longer """
"""than this will be truncated. Default to the max input length of the model."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=0.15 , metadata={"""help""": """Ratio of tokens to mask for masked language modeling loss"""} )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Whether to pad all samples to `max_seq_length`. """
"""If False, will pad the samples dynamically when batching to the maximum length in the batch."""
)
} , )
def lowercase ( self : int ) -> int:
if self.train_file is not None:
__lowerCAmelCase = self.train_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`train_file` should be a csv, a json or a txt file."
if self.validation_file is not None:
__lowerCAmelCase = self.validation_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`validation_file` should be a csv, a json or a txt file."
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : Union[str, Any] ):
with open(lowerCAmelCase_, 'r', encoding='utf-8' ) as f:
__lowerCAmelCase = [json.loads(lowerCAmelCase_ ) for line in f.read().splitlines() if (len(lowerCAmelCase_ ) > 0 and not line.isspace())]
assert len(lowerCAmelCase_ ) == len(lowerCAmelCase_ )
__lowerCAmelCase = {c: dataset[c] for c in dataset.column_names}
__lowerCAmelCase = refs
return Dataset.from_dict(lowerCAmelCase_ )
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
__lowerCAmelCase = load_dataset(data_args.dataset_name, data_args.dataset_config_name )
if "validation" not in datasets.keys():
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[:{data_args.validation_split_percentage}%]""", )
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[{data_args.validation_split_percentage}%:]""", )
else:
__lowerCAmelCase = {}
if data_args.train_file is not None:
__lowerCAmelCase = data_args.train_file
if data_args.validation_file is not None:
__lowerCAmelCase = data_args.validation_file
__lowerCAmelCase = data_args.train_file.split('.' )[-1]
if extension == "txt":
__lowerCAmelCase = 'text'
__lowerCAmelCase = load_dataset(lowerCAmelCase_, data_files=lowerCAmelCase_ )
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.config_name:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.config_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
__lowerCAmelCase = CONFIG_MAPPING[model_args.model_type]()
logger.warning('You are instantiating a new config instance from scratch.' )
if model_args.config_overrides is not None:
logger.info(F"""Overriding config: {model_args.config_overrides}""" )
config.update_from_string(model_args.config_overrides )
logger.info(F"""New config: {config}""" )
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'use_fast': model_args.use_fast_tokenizer,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.tokenizer_name:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.tokenizer_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
raise ValueError(
'You are instantiating a new tokenizer from scratch. This is not supported by this script.'
'You can do it from another script, save it, and load it from here, using --tokenizer_name.' )
if model_args.model_name_or_path:
__lowerCAmelCase = AutoModelForMaskedLM.from_pretrained(
model_args.model_name_or_path, from_tf=bool('.ckpt' in model_args.model_name_or_path ), config=lowerCAmelCase_, cache_dir=model_args.cache_dir, revision=model_args.model_revision, use_auth_token=True if model_args.use_auth_token else None, )
else:
logger.info('Training new model from scratch' )
__lowerCAmelCase = AutoModelForMaskedLM.from_config(lowerCAmelCase_ )
model.resize_token_embeddings(len(lowerCAmelCase_ ) )
# Preprocessing the datasets.
# First we tokenize all the texts.
if training_args.do_train:
__lowerCAmelCase = datasets['train'].column_names
else:
__lowerCAmelCase = datasets['validation'].column_names
__lowerCAmelCase = 'text' if 'text' in column_names else column_names[0]
__lowerCAmelCase = 'max_length' if data_args.pad_to_max_length else False
def tokenize_function(lowerCAmelCase_ : str ):
# Remove empty lines
__lowerCAmelCase = [line for line in examples['text'] if len(lowerCAmelCase_ ) > 0 and not line.isspace()]
return tokenizer(examples['text'], padding=lowerCAmelCase_, truncation=lowerCAmelCase_, max_length=data_args.max_seq_length )
__lowerCAmelCase = datasets.map(
lowerCAmelCase_, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, remove_columns=[text_column_name], load_from_cache_file=not data_args.overwrite_cache, )
# Add the chinese references if provided
if data_args.train_ref_file is not None:
__lowerCAmelCase = add_chinese_references(tokenized_datasets['train'], data_args.train_ref_file )
if data_args.validation_ref_file is not None:
__lowerCAmelCase = add_chinese_references(
tokenized_datasets['validation'], data_args.validation_ref_file )
# If we have ref files, need to avoid it removed by trainer
__lowerCAmelCase = data_args.train_ref_file or data_args.validation_ref_file
if has_ref:
__lowerCAmelCase = False
# Data collator
# This one will take care of randomly masking the tokens.
__lowerCAmelCase = DataCollatorForWholeWordMask(tokenizer=lowerCAmelCase_, mlm_probability=data_args.mlm_probability )
# Initialize our Trainer
__lowerCAmelCase = Trainer(
model=lowerCAmelCase_, args=lowerCAmelCase_, train_dataset=tokenized_datasets['train'] if training_args.do_train else None, eval_dataset=tokenized_datasets['validation'] if training_args.do_eval else None, tokenizer=lowerCAmelCase_, data_collator=lowerCAmelCase_, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif model_args.model_name_or_path is not None and os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model() # Saves the tokenizer too for easy upload
__lowerCAmelCase = os.path.join(training_args.output_dir, 'train_results.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Train results *****' )
for key, value in sorted(train_result.metrics.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
# Need to save the state, since Trainer.save_model saves only the tokenizer with the model
trainer.state.save_to_json(os.path.join(training_args.output_dir, 'trainer_state.json' ) )
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = math.exp(eval_output['eval_loss'] )
__lowerCAmelCase = perplexity
__lowerCAmelCase = os.path.join(training_args.output_dir, 'eval_results_mlm_wwm.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Eval results *****' )
for key, value in sorted(results.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
return results
def a_ ( lowerCAmelCase_ : Tuple ):
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()
| 53 | 1 |
import unittest
from parameterized import parameterized
from transformers import LlamaConfig, is_torch_available, set_seed
from transformers.testing_utils import require_torch, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import LlamaForCausalLM, LlamaForSequenceClassification, LlamaModel, LlamaTokenizer
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Any , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Any=1_3 , lowerCAmelCase_ : List[Any]=7 , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : List[Any]=False , lowerCAmelCase_ : List[Any]=True , lowerCAmelCase_ : Any=9_9 , lowerCAmelCase_ : List[str]=3_2 , lowerCAmelCase_ : List[Any]=5 , lowerCAmelCase_ : List[Any]=4 , lowerCAmelCase_ : List[Any]=3_7 , lowerCAmelCase_ : Tuple="gelu" , lowerCAmelCase_ : Dict=0.1 , lowerCAmelCase_ : List[Any]=0.1 , lowerCAmelCase_ : Any=5_1_2 , lowerCAmelCase_ : List[Any]=1_6 , lowerCAmelCase_ : Tuple=2 , lowerCAmelCase_ : int=0.02 , lowerCAmelCase_ : List[str]=3 , lowerCAmelCase_ : List[Any]=4 , lowerCAmelCase_ : Dict=None , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_input_mask
__lowerCAmelCase = use_token_type_ids
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = type_vocab_size
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = num_labels
__lowerCAmelCase = num_choices
__lowerCAmelCase = scope
def lowercase ( self : List[Any] ) -> Dict:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = None
if self.use_input_mask:
__lowerCAmelCase = random_attention_mask([self.batch_size, self.seq_length] )
__lowerCAmelCase = None
if self.use_token_type_ids:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
__lowerCAmelCase = ids_tensor([self.batch_size] , self.num_choices )
__lowerCAmelCase = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowercase ( self : Union[str, Any] ) -> Any:
return LlamaConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , )
def lowercase ( self : List[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] ) -> Dict:
__lowerCAmelCase = LlamaModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Dict , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , ) -> int:
__lowerCAmelCase = True
__lowerCAmelCase = LlamaModel(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , )
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , )
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : int , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Optional[int] , ) -> Optional[Any]:
__lowerCAmelCase = LlamaForCausalLM(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowercase ( self : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : List[str] , ) -> Optional[int]:
__lowerCAmelCase = True
__lowerCAmelCase = True
__lowerCAmelCase = LlamaForCausalLM(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
# first forward pass
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , use_cache=lowerCAmelCase_ , )
__lowerCAmelCase = outputs.past_key_values
# create hypothetical multiple next token and extent to next_input_ids
__lowerCAmelCase = ids_tensor((self.batch_size, 3) , config.vocab_size )
__lowerCAmelCase = ids_tensor((self.batch_size, 3) , vocab_size=2 )
# append to next input_ids and
__lowerCAmelCase = torch.cat([input_ids, next_tokens] , dim=-1 )
__lowerCAmelCase = torch.cat([input_mask, next_mask] , dim=-1 )
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , output_hidden_states=lowerCAmelCase_ , )['hidden_states'][0]
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , past_key_values=lowerCAmelCase_ , output_hidden_states=lowerCAmelCase_ , )['hidden_states'][0]
# select random slice
__lowerCAmelCase = ids_tensor((1,) , output_from_past.shape[-1] ).item()
__lowerCAmelCase = output_from_no_past[:, -3:, random_slice_idx].detach()
__lowerCAmelCase = output_from_past[:, :, random_slice_idx].detach()
self.parent.assertTrue(output_from_past_slice.shape[1] == next_tokens.shape[1] )
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(lowerCAmelCase_ , lowerCAmelCase_ , atol=1e-3 ) )
def lowercase ( self : Optional[int] ) -> Union[str, Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'input_ids': input_ids, 'attention_mask': input_mask}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (LlamaModel, LlamaForCausalLM, LlamaForSequenceClassification) if is_torch_available() else ()
a_ = (LlamaForCausalLM,) if is_torch_available() else ()
a_ = (
{
"""feature-extraction""": LlamaModel,
"""text-classification""": LlamaForSequenceClassification,
"""text-generation""": LlamaForCausalLM,
"""zero-shot""": LlamaForSequenceClassification,
}
if is_torch_available()
else {}
)
a_ = False
a_ = False
def lowercase ( self : Union[str, Any] ) -> Optional[int]:
__lowerCAmelCase = LlamaModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : Any ) -> Optional[Any]:
self.config_tester.run_common_tests()
def lowercase ( self : Any ) -> List[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : int ) -> Dict:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
__lowerCAmelCase = type
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Optional[Any] ) -> Any:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = 3
__lowerCAmelCase = input_dict['input_ids']
__lowerCAmelCase = input_ids.ne(1 ).to(lowerCAmelCase_ )
__lowerCAmelCase = ids_tensor([self.model_tester.batch_size] , self.model_tester.type_sequence_label_size )
__lowerCAmelCase = LlamaForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.assertEqual(result.logits.shape , (self.model_tester.batch_size, self.model_tester.num_labels) )
def lowercase ( self : Optional[Any] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = 3
__lowerCAmelCase = 'single_label_classification'
__lowerCAmelCase = input_dict['input_ids']
__lowerCAmelCase = input_ids.ne(1 ).to(lowerCAmelCase_ )
__lowerCAmelCase = ids_tensor([self.model_tester.batch_size] , self.model_tester.type_sequence_label_size )
__lowerCAmelCase = LlamaForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.assertEqual(result.logits.shape , (self.model_tester.batch_size, self.model_tester.num_labels) )
def lowercase ( self : Tuple ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = 3
__lowerCAmelCase = 'multi_label_classification'
__lowerCAmelCase = input_dict['input_ids']
__lowerCAmelCase = input_ids.ne(1 ).to(lowerCAmelCase_ )
__lowerCAmelCase = ids_tensor(
[self.model_tester.batch_size, config.num_labels] , self.model_tester.type_sequence_label_size ).to(torch.float )
__lowerCAmelCase = LlamaForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.assertEqual(result.logits.shape , (self.model_tester.batch_size, self.model_tester.num_labels) )
@unittest.skip('LLaMA buffers include complex numbers, which breaks this test' )
def lowercase ( self : List[str] ) -> Optional[int]:
pass
@parameterized.expand([('linear',), ('dynamic',)] )
def lowercase ( self : Optional[int] , lowerCAmelCase_ : List[str] ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = ids_tensor([1, 1_0] , config.vocab_size )
__lowerCAmelCase = ids_tensor([1, int(config.max_position_embeddings * 1.5 )] , config.vocab_size )
set_seed(4_2 ) # Fixed seed at init time so the two models get the same random weights
__lowerCAmelCase = LlamaModel(lowerCAmelCase_ )
original_model.to(lowerCAmelCase_ )
original_model.eval()
__lowerCAmelCase = original_model(lowerCAmelCase_ ).last_hidden_state
__lowerCAmelCase = original_model(lowerCAmelCase_ ).last_hidden_state
set_seed(4_2 ) # Fixed seed at init time so the two models get the same random weights
__lowerCAmelCase = {'type': scaling_type, 'factor': 10.0}
__lowerCAmelCase = LlamaModel(lowerCAmelCase_ )
scaled_model.to(lowerCAmelCase_ )
scaled_model.eval()
__lowerCAmelCase = scaled_model(lowerCAmelCase_ ).last_hidden_state
__lowerCAmelCase = scaled_model(lowerCAmelCase_ ).last_hidden_state
# Dynamic scaling does not change the RoPE embeddings until it receives an input longer than the original
# maximum sequence length, so the outputs for the short input should match.
if scaling_type == "dynamic":
self.assertTrue(torch.allclose(lowerCAmelCase_ , lowerCAmelCase_ , atol=1e-5 ) )
else:
self.assertFalse(torch.allclose(lowerCAmelCase_ , lowerCAmelCase_ , atol=1e-5 ) )
# The output should be different for long inputs
self.assertFalse(torch.allclose(lowerCAmelCase_ , lowerCAmelCase_ , atol=1e-5 ) )
@require_torch
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@unittest.skip('Logits are not exactly the same, once we fix the instabalities somehow, will update!' )
@slow
def lowercase ( self : Any ) -> Optional[int]:
__lowerCAmelCase = [1, 3_0_6, 4_6_5_8, 2_7_8, 6_5_9_3, 3_1_0, 2_8_3_4, 3_3_8]
__lowerCAmelCase = LlamaForCausalLM.from_pretrained('meta-llama/Llama-2-7b-hf' , device_map='auto' )
__lowerCAmelCase = model(torch.tensor([input_ids] ) )
# Expected mean on dim = -1
__lowerCAmelCase = torch.tensor([[-6.65_50, -4.12_27, -4.98_59, -3.24_06, 0.82_62, -3.00_33, 1.29_64, -3.36_99]] )
torch.testing.assert_close(out.mean(-1 ) , lowerCAmelCase_ , atol=1e-2 , rtol=1e-2 )
# slicing logits[0, 0, 0:30]
# fmt: off
__lowerCAmelCase = torch.tensor([-12.82_81, -7.44_53, -0.46_39, -8.06_25, -7.25_00, -8.00_00, -6.48_83, -7.76_95, -7.84_38, -7.03_12, -6.21_88, -7.13_28, -1.84_96, 1.99_61, -8.62_50, -6.72_27, -12.82_81, -6.94_92, -7.07_42, -7.78_52, -7.58_20, -7.90_62, -6.93_75, -7.98_05, -8.34_38, -8.15_62, -8.04_69, -7.62_50, -7.74_22, -7.33_98,] )
# fmt: on
torch.testing.assert_close(out[0, 0, :3_0] , lowerCAmelCase_ , atol=1e-5 , rtol=1e-5 )
@unittest.skip('Logits are not exactly the same, once we fix the instabalities somehow, will update!' )
@slow
def lowercase ( self : List[str] ) -> List[str]:
__lowerCAmelCase = [1, 3_0_6, 4_6_5_8, 2_7_8, 6_5_9_3, 3_1_0, 2_8_3_4, 3_3_8]
__lowerCAmelCase = LlamaForCausalLM.from_pretrained('meta-llama/Llama-2-13b-hf' , device_map='auto' )
__lowerCAmelCase = model(torch.tensor(lowerCAmelCase_ ) )
# Expected mean on dim = -1
__lowerCAmelCase = torch.tensor([[-2.06_22, -1.27_94, -1.16_38, -0.97_88, -1.46_03, -1.02_38, -1.78_93, -1.44_11]] )
torch.testing.assert_close(out.mean(-1 ) , lowerCAmelCase_ , atol=1e-2 , rtol=1e-2 )
# slicing logits[0, 0, 0:30]
# fmt: off
__lowerCAmelCase = torch.tensor([-8.14_06, -8.05_47, 2.74_61, -1.23_44, -0.14_48, -1.82_62, -1.00_20, -1.81_54, -1.68_95, -1.85_16, -2.35_74, -0.92_77, 3.75_98, 6.57_42, -1.29_98, -0.11_77, -8.14_06, -2.96_88, -2.91_99, -3.16_99, -3.52_54, -2.35_55, -2.79_88, -3.41_41, -2.82_62, -4.51_95, -3.33_79, -3.31_64, -2.78_32, -3.02_73] )
# fmt: on
torch.testing.assert_close(out[0, 0, :3_0] , lowerCAmelCase_ , atol=1e-5 , rtol=1e-5 )
@unittest.skip('Logits are not exactly the same, once we fix the instabalities somehow, will update!' )
@slow
def lowercase ( self : int ) -> Tuple:
__lowerCAmelCase = [1, 3_0_6, 4_6_5_8, 2_7_8, 6_5_9_3, 3_1_0, 2_8_3_4, 3_3_8]
__lowerCAmelCase = LlamaForCausalLM.from_pretrained('meta-llama/Llama-2-13b-chat-hf' , device_map='auto' )
__lowerCAmelCase = model(torch.tensor(lowerCAmelCase_ ) )
# Expected mean on dim = -1
__lowerCAmelCase = torch.tensor([[-0.85_62, -1.85_20, -0.75_51, -0.41_62, -1.51_61, -1.20_38, -2.48_23, -2.32_54]] )
torch.testing.assert_close(out.mean(-1 ) , lowerCAmelCase_ , atol=1e-2 , rtol=1e-2 )
# slicing logits[0, 0, 0:30]
# fmt: off
__lowerCAmelCase = torch.tensor([-2.22_27, 4.88_28, 0.90_23, -0.45_78, -0.78_71, -0.10_33, -0.62_21, -0.57_86, -0.78_03, -1.06_74, -1.29_20, -0.15_70, 0.80_08, 2.07_23, -0.94_97, 0.27_71, -2.22_27, -0.76_12, -1.43_46, -1.20_61, -1.64_26, -0.30_00, -0.71_39, -1.19_34, -1.86_91, -1.69_73, -1.59_47, -1.27_05, -0.35_23, -0.55_13] )
# fmt: on
torch.testing.assert_close(out.mean(-1 ) , lowerCAmelCase_ , atol=1e-2 , rtol=1e-2 )
@unittest.skip(
'Logits are not exactly the same, once we fix the instabalities somehow, will update! Also it is gonna be a `too_slow` test' )
@slow
def lowercase ( self : Optional[Any] ) -> Any:
__lowerCAmelCase = [1, 3_0_6, 4_6_5_8, 2_7_8, 6_5_9_3, 3_1_0, 2_8_3_4, 3_3_8]
__lowerCAmelCase = LlamaForCausalLM.from_pretrained('meta-llama/Llama-2-70b-hf' , device_map='auto' )
__lowerCAmelCase = model(torch.tensor(lowerCAmelCase_ ) )
__lowerCAmelCase = torch.tensor(
[[-4.23_27, -3.33_60, -4.66_65, -4.76_31, -1.81_80, -3.41_70, -1.42_11, -3.18_10]] , dtype=torch.floataa )
torch.testing.assert_close(out.mean(-1 ) , lowerCAmelCase_ , atol=1e-2 , rtol=1e-2 )
# fmt: off
__lowerCAmelCase = torch.tensor([-9.49_22, -3.95_51, 1.79_98, -5.67_58, -5.10_55, -5.89_84, -4.83_20, -6.80_86, -6.53_91, -5.61_72, -5.58_20, -5.53_52, 1.78_81, 3.62_89, -6.51_17, -3.47_85, -9.50_00, -6.03_52, -6.81_25, -6.01_95, -6.68_36, -5.47_27, -6.28_12, -6.03_91, -7.33_98, -7.42_97, -7.48_44, -6.58_20, -5.87_89, -5.53_12] )
# fmt: on
torch.testing.assert_close(out[0, 0, :3_0] , lowerCAmelCase_ , atol=1e-5 , rtol=1e-5 )
@unittest.skip('Model is curently gated' )
@slow
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
__lowerCAmelCase = 'Simply put, the theory of relativity states that 1) the laws of physics are the same everywhere in the universe and 2) the passage of time and the length of objects can vary depending on the observer\'s frame of reference.\n\nThe first part of the theory, that the laws of physics are the same everywhere, is known as the "princi'
__lowerCAmelCase = 'Simply put, the theory of relativity states that '
__lowerCAmelCase = LlamaTokenizer.from_pretrained('meta-llama/Llama-2-13b-chat-hf' )
__lowerCAmelCase = tokenizer.encode(lowerCAmelCase_ , return_tensors='pt' )
__lowerCAmelCase = LlamaForCausalLM.from_pretrained(
'meta-llama/Llama-2-13b-chat-hf' , device_map='sequential' , use_safetensors=lowerCAmelCase_ )
# greedy generation outputs
__lowerCAmelCase = model.generate(lowerCAmelCase_ , max_new_tokens=6_4 , top_p=lowerCAmelCase_ , temperature=1 , do_sample=lowerCAmelCase_ )
__lowerCAmelCase = tokenizer.decode(generated_ids[0] , skip_special_tokens=lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
| 53 |
def a_ ( lowerCAmelCase_ : int = 200_0000 ):
__lowerCAmelCase = [0 for i in range(n + 1 )]
__lowerCAmelCase = 1
__lowerCAmelCase = 1
for i in range(2, int(n**0.5 ) + 1 ):
if primality_list[i] == 0:
for j in range(i * i, n + 1, lowerCAmelCase_ ):
__lowerCAmelCase = 1
__lowerCAmelCase = 0
for i in range(lowerCAmelCase_ ):
if primality_list[i] == 0:
sum_of_primes += i
return sum_of_primes
if __name__ == "__main__":
print(F"""{solution() = }""")
| 53 | 1 |
from datetime import datetime
import requests
from bsa import BeautifulSoup
if __name__ == "__main__":
_snake_case : List[Any] = input('Enter image url: ').strip()
print(F"""Downloading image from {url} ...""")
_snake_case : Union[str, Any] = BeautifulSoup(requests.get(url).content, 'html.parser')
# The image URL is in the content field of the first meta tag with property og:image
_snake_case : List[Any] = soup.find('meta', {'property': 'og:image'})['content']
_snake_case : str = requests.get(image_url).content
_snake_case : Any = F"""{datetime.now():%Y-%m-%d_%H:%M:%S}.jpg"""
with open(file_name, 'wb') as fp:
fp.write(image_data)
print(F"""Done. Image saved to disk as {file_name}.""")
| 53 |
import json
import logging
import os
import sys
from pathlib import Path
import finetune_rag
from transformers.file_utils import is_apex_available
from transformers.testing_utils import (
TestCasePlus,
execute_subprocess_async,
require_ray,
require_torch_gpu,
require_torch_multi_gpu,
)
logging.basicConfig(level=logging.DEBUG)
_snake_case : Tuple = logging.getLogger()
_snake_case : Any = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Any , lowerCAmelCase_ : Dict ) -> Optional[int]:
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = {'source': 'What is love ?', 'target': 'life'}
__lowerCAmelCase = {'train': 1_2, 'val': 2, 'test': 2}
for split in ["train", "test", "val"]:
for field in ["source", "target"]:
__lowerCAmelCase = '\n'.join([contents[field]] * n_lines[split] )
with open(os.path.join(lowerCAmelCase_ , f"""{split}.{field}""" ) , 'w' ) as f:
f.write(lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : int , lowerCAmelCase_ : str = "pytorch" ) -> List[str]:
__lowerCAmelCase = self.get_auto_remove_tmp_dir()
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'output' )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'data' )
self._create_dummy_data(data_dir=lowerCAmelCase_ )
__lowerCAmelCase = f"""
--data_dir {data_dir} \
--output_dir {output_dir} \
--model_name_or_path facebook/rag-sequence-base \
--model_type rag_sequence \
--do_train \
--do_predict \
--n_val -1 \
--val_check_interval 1.0 \
--train_batch_size 2 \
--eval_batch_size 1 \
--max_source_length 25 \
--max_target_length 25 \
--val_max_target_length 25 \
--test_max_target_length 25 \
--label_smoothing 0.1 \
--dropout 0.1 \
--attention_dropout 0.1 \
--weight_decay 0.001 \
--adam_epsilon 1e-08 \
--max_grad_norm 0.1 \
--lr_scheduler polynomial \
--learning_rate 3e-04 \
--num_train_epochs 1 \
--warmup_steps 4 \
--gradient_accumulation_steps 1 \
--distributed-port 8787 \
--use_dummy_dataset 1 \
--distributed_retriever {distributed_retriever} \
""".split()
if gpus > 0:
testargs.append(f"""--gpus={gpus}""" )
if is_apex_available():
testargs.append('--fp16' )
else:
testargs.append('--gpus=0' )
testargs.append('--distributed_backend=ddp_cpu' )
testargs.append('--num_processes=2' )
__lowerCAmelCase = [sys.executable, str(Path(finetune_rag.__file__ ).resolve() )] + testargs
execute_subprocess_async(lowerCAmelCase_ , env=self.get_env() )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'metrics.json' )
with open(lowerCAmelCase_ ) as f:
__lowerCAmelCase = json.load(lowerCAmelCase_ )
return result
@require_torch_gpu
def lowercase ( self : str ) -> int:
__lowerCAmelCase = self._run_finetune(gpus=1 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
def lowercase ( self : List[str] ) -> Dict:
__lowerCAmelCase = self._run_finetune(gpus=2 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_gpu
@require_ray
def lowercase ( self : int ) -> Tuple:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
@require_ray
def lowercase ( self : List[Any] ) -> str:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
| 53 | 1 |
from math import sqrt
def a_ ( lowerCAmelCase_ : int ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (
number >= 0
), "'number' must been an int and positive"
__lowerCAmelCase = True
# 0 and 1 are none primes.
if number <= 1:
__lowerCAmelCase = False
for divisor in range(2, int(round(sqrt(lowerCAmelCase_ ) ) ) + 1 ):
# if 'number' divisible by 'divisor' then sets 'status'
# of false and break up the loop.
if number % divisor == 0:
__lowerCAmelCase = False
break
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "'status' must been from type bool"
return status
def a_ ( lowerCAmelCase_ : int ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (n > 2), "'N' must been an int and > 2"
# beginList: contains all natural numbers from 2 up to N
__lowerCAmelCase = list(range(2, n + 1 ) )
__lowerCAmelCase = [] # this list will be returns.
# actual sieve of erathostenes
for i in range(len(lowerCAmelCase_ ) ):
for j in range(i + 1, len(lowerCAmelCase_ ) ):
if (begin_list[i] != 0) and (begin_list[j] % begin_list[i] == 0):
__lowerCAmelCase = 0
# filters actual prime numbers.
__lowerCAmelCase = [x for x in begin_list if x != 0]
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "'ans' must been from type list"
return ans
def a_ ( lowerCAmelCase_ : int ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (n > 2), "'N' must been an int and > 2"
__lowerCAmelCase = []
# 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(lowerCAmelCase_ ):
ans.append(lowerCAmelCase_ )
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "'ans' must been from type list"
return ans
def a_ ( lowerCAmelCase_ : Tuple ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and number >= 0, "'number' must been an int and >= 0"
__lowerCAmelCase = [] # this list will be returns of the function.
# potential prime number factors.
__lowerCAmelCase = 2
__lowerCAmelCase = number
if number == 0 or number == 1:
ans.append(lowerCAmelCase_ )
# if 'number' not prime then builds the prime factorization of 'number'
elif not is_prime(lowerCAmelCase_ ):
while quotient != 1:
if is_prime(lowerCAmelCase_ ) and (quotient % factor == 0):
ans.append(lowerCAmelCase_ )
quotient /= factor
else:
factor += 1
else:
ans.append(lowerCAmelCase_ )
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "'ans' must been from type list"
return ans
def a_ ( lowerCAmelCase_ : Dict ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (
number >= 0
), "'number' bust been an int and >= 0"
__lowerCAmelCase = 0
# prime factorization of 'number'
__lowerCAmelCase = prime_factorization(lowerCAmelCase_ )
__lowerCAmelCase = max(lowerCAmelCase_ )
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "'ans' must been from type int"
return ans
def a_ ( lowerCAmelCase_ : Tuple ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (
number >= 0
), "'number' bust been an int and >= 0"
__lowerCAmelCase = 0
# prime factorization of 'number'
__lowerCAmelCase = prime_factorization(lowerCAmelCase_ )
__lowerCAmelCase = min(lowerCAmelCase_ )
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "'ans' must been from type int"
return ans
def a_ ( lowerCAmelCase_ : Optional[Any] ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "'number' must been an int"
assert isinstance(number % 2 == 0, lowerCAmelCase_ ), "compare bust been from type bool"
return number % 2 == 0
def a_ ( lowerCAmelCase_ : str ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "'number' must been an int"
assert isinstance(number % 2 != 0, lowerCAmelCase_ ), "compare bust been from type bool"
return number % 2 != 0
def a_ ( lowerCAmelCase_ : Any ):
assert (
isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (number > 2) and is_even(lowerCAmelCase_ )
), "'number' must been an int, even and > 2"
__lowerCAmelCase = [] # this list will returned
# creates a list of prime numbers between 2 up to 'number'
__lowerCAmelCase = get_prime_numbers(lowerCAmelCase_ )
__lowerCAmelCase = len(lowerCAmelCase_ )
# run variable for while-loops.
__lowerCAmelCase = 0
__lowerCAmelCase = None
# exit variable. for break up the loops
__lowerCAmelCase = True
while i < len_pn and loop:
__lowerCAmelCase = i + 1
while j < len_pn and loop:
if prime_numbers[i] + prime_numbers[j] == number:
__lowerCAmelCase = False
ans.append(prime_numbers[i] )
ans.append(prime_numbers[j] )
j += 1
i += 1
# precondition
assert (
isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and (len(lowerCAmelCase_ ) == 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 a_ ( lowerCAmelCase_ : List[Any], lowerCAmelCase_ : int ):
assert (
isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and (numbera >= 0)
and (numbera >= 0)
), "'number1' and 'number2' must been positive integer."
__lowerCAmelCase = 0
while numbera != 0:
__lowerCAmelCase = numbera % numbera
__lowerCAmelCase = numbera
__lowerCAmelCase = rest
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (
numbera >= 0
), "'number' must been from type int and positive"
return numbera
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : List[Any] ):
assert (
isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and (numbera >= 1)
and (numbera >= 1)
), "'number1' and 'number2' must been positive integer."
__lowerCAmelCase = 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'
__lowerCAmelCase = prime_factorization(lowerCAmelCase_ )
__lowerCAmelCase = prime_factorization(lowerCAmelCase_ )
elif numbera == 1 or numbera == 1:
__lowerCAmelCase = []
__lowerCAmelCase = []
__lowerCAmelCase = max(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = 0
__lowerCAmelCase = 0
__lowerCAmelCase = [] # 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:
__lowerCAmelCase = prime_fac_a.count(lowerCAmelCase_ )
__lowerCAmelCase = prime_fac_a.count(lowerCAmelCase_ )
for _ in range(max(lowerCAmelCase_, lowerCAmelCase_ ) ):
ans *= n
else:
__lowerCAmelCase = prime_fac_a.count(lowerCAmelCase_ )
for _ in range(lowerCAmelCase_ ):
ans *= n
done.append(lowerCAmelCase_ )
# iterates through primeFac2
for n in prime_fac_a:
if n not in done:
__lowerCAmelCase = prime_fac_a.count(lowerCAmelCase_ )
for _ in range(lowerCAmelCase_ ):
ans *= n
done.append(lowerCAmelCase_ )
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (
ans >= 0
), "'ans' must been from type int and positive"
return ans
def a_ ( lowerCAmelCase_ : str ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (n >= 0), "'number' must been a positive int"
__lowerCAmelCase = 0
__lowerCAmelCase = 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(lowerCAmelCase_ ):
ans += 1
# precondition
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and is_prime(
lowerCAmelCase_ ), "'ans' must been a prime number and from type int"
return ans
def a_ ( lowerCAmelCase_ : Dict, lowerCAmelCase_ : str ):
assert (
is_prime(lowerCAmelCase_ ) and is_prime(lowerCAmelCase_ ) and (p_number_a < p_number_a)
), "The arguments must been prime numbers and 'pNumber1' < 'pNumber2'"
__lowerCAmelCase = p_number_a + 1 # jump to the next number
__lowerCAmelCase = [] # this list will be returns.
# if number is not prime then
# fetch the next prime number.
while not is_prime(lowerCAmelCase_ ):
number += 1
while number < p_number_a:
ans.append(lowerCAmelCase_ )
number += 1
# fetch the next prime number.
while not is_prime(lowerCAmelCase_ ):
number += 1
# precondition
assert (
isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and ans[0] != p_number_a
and ans[len(lowerCAmelCase_ ) - 1] != p_number_a
), "'ans' must been a list without the arguments"
# 'ans' contains not 'pNumber1' and 'pNumber2' !
return ans
def a_ ( lowerCAmelCase_ : Optional[Any] ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (n >= 1), "'n' must been int and >= 1"
__lowerCAmelCase = [] # will be returned.
for divisor in range(1, n + 1 ):
if n % divisor == 0:
ans.append(lowerCAmelCase_ )
# precondition
assert ans[0] == 1 and ans[len(lowerCAmelCase_ ) - 1] == n, "Error in function getDivisiors(...)"
return ans
def a_ ( lowerCAmelCase_ : int ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (
number > 1
), "'number' must been an int and >= 1"
__lowerCAmelCase = get_divisors(lowerCAmelCase_ )
# precondition
assert (
isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and (divisors[0] == 1)
and (divisors[len(lowerCAmelCase_ ) - 1] == number)
), "Error in help-function getDivisiors(...)"
# summed all divisors up to 'number' (exclusive), hence [:-1]
return sum(divisors[:-1] ) == number
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : Any ):
assert (
isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and isinstance(lowerCAmelCase_, lowerCAmelCase_ )
and (denominator != 0)
), "The arguments must been from type int and 'denominator' != 0"
# build the greatest common divisor of numerator and denominator.
__lowerCAmelCase = gcd(abs(lowerCAmelCase_ ), abs(lowerCAmelCase_ ) )
# precondition
assert (
isinstance(lowerCAmelCase_, lowerCAmelCase_ )
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 a_ ( lowerCAmelCase_ : Dict ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (n >= 0), "'n' must been a int and >= 0"
__lowerCAmelCase = 1 # this will be return.
for factor in range(1, n + 1 ):
ans *= factor
return ans
def a_ ( lowerCAmelCase_ : Dict ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ) and (n >= 0), "'n' must been an int and >= 0"
__lowerCAmelCase = 0
__lowerCAmelCase = 1
__lowerCAmelCase = 1 # this will be return
for _ in range(n - 1 ):
__lowerCAmelCase = ans
ans += fiba
__lowerCAmelCase = tmp
return ans
| 53 |
import copy
import inspect
import unittest
from transformers import AutoBackbone
from transformers.configuration_utils import PretrainedConfig
from transformers.testing_utils import require_timm, require_torch, torch_device
from transformers.utils.import_utils import is_torch_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor
if is_torch_available():
import torch
from transformers import TimmBackbone, TimmBackboneConfig
from ...test_pipeline_mixin import PipelineTesterMixin
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : Optional[int]=None , lowerCAmelCase_ : Union[str, Any]="resnet50" , lowerCAmelCase_ : str=3 , lowerCAmelCase_ : List[str]=3_2 , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : Dict=True , lowerCAmelCase_ : Optional[Any]=True , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = out_indices if out_indices is not None else [4]
__lowerCAmelCase = stage_names
__lowerCAmelCase = out_features
__lowerCAmelCase = backbone
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = use_pretrained_backbone
__lowerCAmelCase = is_training
def lowercase ( self : List[str] ) -> Any:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : List[Any] ) -> Union[str, Any]:
return TimmBackboneConfig(
image_size=self.image_size , num_channels=self.num_channels , out_features=self.out_features , out_indices=self.out_indices , stage_names=self.stage_names , use_pretrained_backbone=self.use_pretrained_backbone , backbone=self.backbone , )
def lowercase ( self : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ) -> int:
__lowerCAmelCase = TimmBackbone(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.feature_map[-1].shape , (self.batch_size, model.channels[-1], 1_4, 1_4) , )
def lowercase ( self : List[str] ) -> str:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
@require_timm
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (TimmBackbone,) if is_torch_available() else ()
a_ = {"""feature-extraction""": TimmBackbone} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Tuple ) -> int:
__lowerCAmelCase = TimmBackboneModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : Dict ) -> List[str]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Union[str, Any] ) -> Optional[int]:
__lowerCAmelCase = 'resnet18'
__lowerCAmelCase = 'microsoft/resnet-18'
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(len(timm_model.stage_names ) , len(transformers_model.stage_names ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
# Out indices are set to the last layer by default. For timm models, we don't know
# the number of layers in advance, so we set it to (-1,), whereas for transformers
# models, we set it to [len(stage_names) - 1] (kept for backward compatibility).
self.assertEqual(timm_model.out_indices , (-1,) )
self.assertEqual(transformers_model.out_indices , [len(timm_model.stage_names ) - 1] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ , out_indices=[1, 2, 3] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , out_indices=[1, 2, 3] )
self.assertEqual(timm_model.out_indices , transformers_model.out_indices )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
@unittest.skip('TimmBackbone doesn\'t support feed forward chunking' )
def lowercase ( self : List[str] ) -> Tuple:
pass
@unittest.skip('TimmBackbone doesn\'t have num_hidden_layers attribute' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('TimmBackbone initialization is managed on the timm side' )
def lowercase ( self : str ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Any ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Optional[Any]:
pass
@unittest.skip('TimmBackbone model cannot be created without specifying a backbone checkpoint' )
def lowercase ( self : Dict ) -> Any:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Any ) -> Optional[int]:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : Union[str, Any] ) -> Tuple:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : List[str] ) -> Optional[int]:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Tuple ) -> List[str]:
pass
@unittest.skip('TimmBackbone doesn\'t have hidden size info in its configuration.' )
def lowercase ( self : int ) -> Optional[int]:
pass
@unittest.skip('TimmBackbone doesn\'t support output_attentions.' )
def lowercase ( self : Union[str, Any] ) -> str:
pass
@unittest.skip('Safetensors is not supported by timm.' )
def lowercase ( self : Dict ) -> str:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : List[str] ) -> Optional[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = True
__lowerCAmelCase = self.has_attentions
# no need to test all models as different heads yield the same functionality
__lowerCAmelCase = self.all_model_classes[0]
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs[0][-1]
# Encoder-/Decoder-only models
__lowerCAmelCase = outputs.hidden_states[0]
hidden_states.retain_grad()
if self.has_attentions:
__lowerCAmelCase = outputs.attentions[0]
attentions.retain_grad()
output.flatten()[0].backward(retain_graph=lowerCAmelCase_ )
self.assertIsNotNone(hidden_states.grad )
if self.has_attentions:
self.assertIsNotNone(attentions.grad )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , len(config.out_indices ) )
self.assertEqual(len(model.channels ) , len(config.out_indices ) )
# Check output of last stage is taken if out_features=None, out_indices=None
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , 1 )
self.assertEqual(len(model.channels ) , 1 )
# Check backbone can be initialized with fresh weights
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = False
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
| 53 | 1 |
import random
import unittest
import torch
from diffusers import IFInpaintingPipeline
from diffusers.utils import floats_tensor
from diffusers.utils.import_utils import is_xformers_available
from diffusers.utils.testing_utils import skip_mps, torch_device
from ..pipeline_params import (
TEXT_GUIDED_IMAGE_INPAINTING_BATCH_PARAMS,
TEXT_GUIDED_IMAGE_INPAINTING_PARAMS,
)
from ..test_pipelines_common import PipelineTesterMixin
from . import IFPipelineTesterMixin
@skip_mps
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = IFInpaintingPipeline
a_ = TEXT_GUIDED_IMAGE_INPAINTING_PARAMS - {"""width""", """height"""}
a_ = TEXT_GUIDED_IMAGE_INPAINTING_BATCH_PARAMS
a_ = PipelineTesterMixin.required_optional_params - {"""latents"""}
def lowercase ( self : Union[str, Any] ) -> str:
return self._get_dummy_components()
def lowercase ( self : Dict , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Optional[Any]=0 ) -> Any:
if str(lowerCAmelCase_ ).startswith('mps' ):
__lowerCAmelCase = torch.manual_seed(lowerCAmelCase_ )
else:
__lowerCAmelCase = torch.Generator(device=lowerCAmelCase_ ).manual_seed(lowerCAmelCase_ )
__lowerCAmelCase = floats_tensor((1, 3, 3_2, 3_2) , rng=random.Random(lowerCAmelCase_ ) ).to(lowerCAmelCase_ )
__lowerCAmelCase = floats_tensor((1, 3, 3_2, 3_2) , rng=random.Random(lowerCAmelCase_ ) ).to(lowerCAmelCase_ )
__lowerCAmelCase = {
'prompt': 'A painting of a squirrel eating a burger',
'image': image,
'mask_image': mask_image,
'generator': generator,
'num_inference_steps': 2,
'output_type': 'numpy',
}
return inputs
@unittest.skipIf(
torch_device != 'cuda' or not is_xformers_available() , reason='XFormers attention is only available with CUDA and `xformers` installed' , )
def lowercase ( self : Any ) -> List[Any]:
self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=1e-3 )
def lowercase ( self : str ) -> Tuple:
self._test_save_load_optional_components()
@unittest.skipIf(torch_device != 'cuda' , reason='float16 requires CUDA' )
def lowercase ( self : str ) -> int:
# Due to non-determinism in save load of the hf-internal-testing/tiny-random-t5 text encoder
super().test_save_load_floataa(expected_max_diff=1e-1 )
def lowercase ( self : Tuple ) -> str:
self._test_attention_slicing_forward_pass(expected_max_diff=1e-2 )
def lowercase ( self : str ) -> Optional[Any]:
self._test_save_load_local()
def lowercase ( self : Optional[Any] ) -> List[Any]:
self._test_inference_batch_single_identical(
expected_max_diff=1e-2 , )
| 53 |
# Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import os
import platform
import numpy as np
import psutil
import torch
from accelerate import __version__ as version
from accelerate.commands.config import default_config_file, load_config_from_file
from ..utils import is_npu_available, is_xpu_available
def a_ ( lowerCAmelCase_ : str=None ):
if subparsers is not None:
__lowerCAmelCase = subparsers.add_parser('env' )
else:
__lowerCAmelCase = argparse.ArgumentParser('Accelerate env command' )
parser.add_argument(
'--config_file', default=lowerCAmelCase_, help='The config file to use for the default values in the launching script.' )
if subparsers is not None:
parser.set_defaults(func=lowerCAmelCase_ )
return parser
def a_ ( lowerCAmelCase_ : Optional[int] ):
__lowerCAmelCase = torch.__version__
__lowerCAmelCase = torch.cuda.is_available()
__lowerCAmelCase = is_xpu_available()
__lowerCAmelCase = is_npu_available()
__lowerCAmelCase = 'Not found'
# Get the default from the config file.
if args.config_file is not None or os.path.isfile(lowerCAmelCase_ ):
__lowerCAmelCase = load_config_from_file(args.config_file ).to_dict()
__lowerCAmelCase = {
'`Accelerate` version': version,
'Platform': platform.platform(),
'Python version': platform.python_version(),
'Numpy version': np.__version__,
'PyTorch version (GPU?)': F"""{pt_version} ({pt_cuda_available})""",
'PyTorch XPU available': str(lowerCAmelCase_ ),
'PyTorch NPU available': str(lowerCAmelCase_ ),
'System RAM': F"""{psutil.virtual_memory().total / 1024 ** 3:.2f} GB""",
}
if pt_cuda_available:
__lowerCAmelCase = torch.cuda.get_device_name()
print('\nCopy-and-paste the text below in your GitHub issue\n' )
print('\n'.join([F"""- {prop}: {val}""" for prop, val in info.items()] ) )
print('- `Accelerate` default config:' if args.config_file is None else '- `Accelerate` config passed:' )
__lowerCAmelCase = (
'\n'.join([F"""\t- {prop}: {val}""" for prop, val in accelerate_config.items()] )
if isinstance(lowerCAmelCase_, lowerCAmelCase_ )
else F"""\t{accelerate_config}"""
)
print(lowerCAmelCase_ )
__lowerCAmelCase = accelerate_config
return info
def a_ ( ):
__lowerCAmelCase = env_command_parser()
__lowerCAmelCase = parser.parse_args()
env_command(lowerCAmelCase_ )
return 0
if __name__ == "__main__":
raise SystemExit(main())
| 53 | 1 |
from typing import List, Optional, Tuple, Union
import PIL
import torch
from torchvision import transforms
from diffusers.pipeline_utils import DiffusionPipeline, ImagePipelineOutput
from diffusers.schedulers import DDIMScheduler
from diffusers.utils import randn_tensor
_snake_case : Dict = transforms.Compose(
[
transforms.Resize((256, 256)),
transforms.ToTensor(),
transforms.Normalize([0.5], [0.5]),
]
)
def a_ ( lowerCAmelCase_ : List[str] ):
if isinstance(lowerCAmelCase_, torch.Tensor ):
return image
elif isinstance(lowerCAmelCase_, PIL.Image.Image ):
__lowerCAmelCase = [image]
__lowerCAmelCase = [trans(img.convert('RGB' ) ) for img in image]
__lowerCAmelCase = torch.stack(lowerCAmelCase_ )
return image
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def __init__( self : str , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : List[str] ) -> Tuple:
super().__init__()
# make sure scheduler can always be converted to DDIM
__lowerCAmelCase = DDIMScheduler.from_config(scheduler.config )
self.register_modules(unet=lowerCAmelCase_ , scheduler=lowerCAmelCase_ )
def lowercase ( self : List[Any] , lowerCAmelCase_ : Tuple ) -> Union[str, Any]:
if strength < 0 or strength > 1:
raise ValueError(f"""The value of strength should in [0.0, 1.0] but is {strength}""" )
def lowercase ( self : Any , lowerCAmelCase_ : str , lowerCAmelCase_ : str , lowerCAmelCase_ : Optional[Any] ) -> List[str]:
# get the original timestep using init_timestep
__lowerCAmelCase = min(int(num_inference_steps * strength ) , lowerCAmelCase_ )
__lowerCAmelCase = max(num_inference_steps - init_timestep , 0 )
__lowerCAmelCase = self.scheduler.timesteps[t_start:]
return timesteps, num_inference_steps - t_start
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any , lowerCAmelCase_ : str , lowerCAmelCase_ : Optional[Any]=None ) -> Any:
if not isinstance(lowerCAmelCase_ , (torch.Tensor, PIL.Image.Image, list) ):
raise ValueError(
f"""`image` has to be of type `torch.Tensor`, `PIL.Image.Image` or list but is {type(lowerCAmelCase_ )}""" )
__lowerCAmelCase = image.to(device=lowerCAmelCase_ , dtype=lowerCAmelCase_ )
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ) and len(lowerCAmelCase_ ) != batch_size:
raise ValueError(
f"""You have passed a list of generators of length {len(lowerCAmelCase_ )}, but requested an effective batch"""
f""" size of {batch_size}. Make sure the batch size matches the length of the generators.""" )
__lowerCAmelCase = init_latents.shape
__lowerCAmelCase = randn_tensor(lowerCAmelCase_ , generator=lowerCAmelCase_ , device=lowerCAmelCase_ , dtype=lowerCAmelCase_ )
# get latents
print('add noise to latents at timestep' , lowerCAmelCase_ )
__lowerCAmelCase = self.scheduler.add_noise(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = init_latents
return latents
@torch.no_grad()
def __call__( self : int , lowerCAmelCase_ : Union[torch.FloatTensor, PIL.Image.Image] = None , lowerCAmelCase_ : float = 0.8 , lowerCAmelCase_ : int = 1 , lowerCAmelCase_ : Optional[Union[torch.Generator, List[torch.Generator]]] = None , lowerCAmelCase_ : float = 0.0 , lowerCAmelCase_ : int = 5_0 , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[str] = "pil" , lowerCAmelCase_ : bool = True , ) -> Union[ImagePipelineOutput, Tuple]:
self.check_inputs(lowerCAmelCase_ )
# 2. Preprocess image
__lowerCAmelCase = preprocess(lowerCAmelCase_ )
# 3. set timesteps
self.scheduler.set_timesteps(lowerCAmelCase_ , device=self.device )
__lowerCAmelCase , __lowerCAmelCase = self.get_timesteps(lowerCAmelCase_ , lowerCAmelCase_ , self.device )
__lowerCAmelCase = timesteps[:1].repeat(lowerCAmelCase_ )
# 4. Prepare latent variables
__lowerCAmelCase = self.prepare_latents(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , self.unet.dtype , self.device , lowerCAmelCase_ )
__lowerCAmelCase = latents
# 5. Denoising loop
for t in self.progress_bar(lowerCAmelCase_ ):
# 1. predict noise model_output
__lowerCAmelCase = self.unet(lowerCAmelCase_ , lowerCAmelCase_ ).sample
# 2. predict previous mean of image x_t-1 and add variance depending on eta
# eta corresponds to η in paper and should be between [0, 1]
# do x_t -> x_t-1
__lowerCAmelCase = self.scheduler.step(
lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , eta=lowerCAmelCase_ , use_clipped_model_output=lowerCAmelCase_ , generator=lowerCAmelCase_ , ).prev_sample
__lowerCAmelCase = (image / 2 + 0.5).clamp(0 , 1 )
__lowerCAmelCase = image.cpu().permute(0 , 2 , 3 , 1 ).numpy()
if output_type == "pil":
__lowerCAmelCase = self.numpy_to_pil(lowerCAmelCase_ )
if not return_dict:
return (image, latent_timestep.item())
return ImagePipelineOutput(images=lowerCAmelCase_ )
| 53 |
import importlib
import sys
from argparse import REMAINDER, ArgumentParser
from pathlib import Path
import torch_xla.distributed.xla_multiprocessing as xmp
def a_ ( ):
__lowerCAmelCase = ArgumentParser(
description=(
'PyTorch TPU distributed training launch '
'helper utility that will spawn up '
'multiple distributed processes'
) )
# Optional arguments for the launch helper
parser.add_argument('--num_cores', type=lowerCAmelCase_, default=1, help='Number of TPU cores to use (1 or 8).' )
# positional
parser.add_argument(
'training_script', type=lowerCAmelCase_, help=(
'The full path to the single TPU training '
'program/script to be launched in parallel, '
'followed by all the arguments for the '
'training script'
), )
# rest from the training program
parser.add_argument('training_script_args', nargs=lowerCAmelCase_ )
return parser.parse_args()
def a_ ( ):
__lowerCAmelCase = parse_args()
# Import training_script as a module.
__lowerCAmelCase = Path(args.training_script )
sys.path.append(str(script_fpath.parent.resolve() ) )
__lowerCAmelCase = script_fpath.stem
__lowerCAmelCase = importlib.import_module(lowerCAmelCase_ )
# Patch sys.argv
__lowerCAmelCase = [args.training_script] + args.training_script_args + ['--tpu_num_cores', str(args.num_cores )]
xmp.spawn(mod._mp_fn, args=(), nprocs=args.num_cores )
if __name__ == "__main__":
main()
| 53 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
from ...utils.backbone_utils import BackboneConfigMixin, get_aligned_output_features_output_indices
_snake_case : List[str] = logging.get_logger(__name__)
_snake_case : Optional[int] = {
'facebook/convnextv2-tiny-1k-224': 'https://huggingface.co/facebook/convnextv2-tiny-1k-224/resolve/main/config.json',
}
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = """convnextv2"""
def __init__( self : List[str] , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : Optional[Any]=4 , lowerCAmelCase_ : str=4 , lowerCAmelCase_ : Union[str, Any]=None , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : int="gelu" , lowerCAmelCase_ : Any=0.02 , lowerCAmelCase_ : Tuple=1e-12 , lowerCAmelCase_ : Dict=0.0 , lowerCAmelCase_ : Tuple=2_2_4 , lowerCAmelCase_ : Tuple=None , lowerCAmelCase_ : Any=None , **lowerCAmelCase_ : int , ) -> Union[str, Any]:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = num_channels
__lowerCAmelCase = patch_size
__lowerCAmelCase = num_stages
__lowerCAmelCase = [9_6, 1_9_2, 3_8_4, 7_6_8] if hidden_sizes is None else hidden_sizes
__lowerCAmelCase = [3, 3, 9, 3] if depths is None else depths
__lowerCAmelCase = hidden_act
__lowerCAmelCase = initializer_range
__lowerCAmelCase = layer_norm_eps
__lowerCAmelCase = drop_path_rate
__lowerCAmelCase = image_size
__lowerCAmelCase = ['stem'] + [f"""stage{idx}""" for idx in range(1 , len(self.depths ) + 1 )]
__lowerCAmelCase , __lowerCAmelCase = get_aligned_output_features_output_indices(
out_features=lowerCAmelCase_ , out_indices=lowerCAmelCase_ , stage_names=self.stage_names )
| 53 |
import inspect
import unittest
from huggingface_hub import hf_hub_download
from transformers import ConvNextConfig, UperNetConfig
from transformers.testing_utils import require_torch, require_torch_multi_gpu, require_vision, slow, torch_device
from transformers.utils import is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import UperNetForSemanticSegmentation
from transformers.models.upernet.modeling_upernet import UPERNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Dict=1_3 , lowerCAmelCase_ : str=3_2 , lowerCAmelCase_ : Optional[Any]=3 , lowerCAmelCase_ : Any=4 , lowerCAmelCase_ : str=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Tuple=[2, 2, 3, 2] , lowerCAmelCase_ : str=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[int]=3_7 , lowerCAmelCase_ : Dict="gelu" , lowerCAmelCase_ : List[Any]=1_0 , lowerCAmelCase_ : str=0.02 , lowerCAmelCase_ : Dict=["stage2", "stage3", "stage4"] , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : List[Any]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = num_stages
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = out_features
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = num_stages
def lowercase ( self : Dict ) -> List[str]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = self.get_config()
return config, pixel_values, labels
def lowercase ( self : List[str] ) -> Union[str, Any]:
return ConvNextConfig(
num_channels=self.num_channels , num_stages=self.num_stages , hidden_sizes=self.hidden_sizes , depths=self.depths , is_training=self.is_training , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , out_features=self.out_features , )
def lowercase ( self : Dict ) -> List[str]:
return UperNetConfig(
backbone_config=self.get_backbone_config() , hidden_size=5_1_2 , pool_scales=[1, 2, 3, 6] , use_auxiliary_head=lowerCAmelCase_ , auxiliary_loss_weight=0.4 , auxiliary_in_channels=4_0 , auxiliary_channels=2_5_6 , auxiliary_num_convs=1 , auxiliary_concat_input=lowerCAmelCase_ , loss_ignore_index=2_5_5 , num_labels=self.num_labels , )
def lowercase ( self : List[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : int ) -> Optional[Any]:
__lowerCAmelCase = UperNetForSemanticSegmentation(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.logits.shape , (self.batch_size, self.num_labels, self.image_size, self.image_size) )
def lowercase ( self : Union[str, Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (UperNetForSemanticSegmentation,) if is_torch_available() else ()
a_ = {"""image-segmentation""": UperNetForSemanticSegmentation} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = UperNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Tuple ) -> Union[str, Any]:
return
def lowercase ( self : Optional[int] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_semantic_segmentation(*lowerCAmelCase_ )
@unittest.skip(reason='UperNet does not use inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Dict:
pass
@unittest.skip(reason='UperNet does not support input and output embeddings' )
def lowercase ( self : Optional[Any] ) -> Dict:
pass
@unittest.skip(reason='UperNet does not have a base model' )
def lowercase ( self : Optional[int] ) -> List[Any]:
pass
@unittest.skip(reason='UperNet does not have a base model' )
def lowercase ( self : str ) -> Dict:
pass
@require_torch_multi_gpu
@unittest.skip(reason='UperNet has some layers using `add_module` which doesn\'t work well with `nn.DataParallel`' )
def lowercase ( self : Optional[Any] ) -> Optional[int]:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : Tuple ) -> List[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Tuple:
def check_hidden_states_output(lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
# ConvNext's feature maps are of shape (batch_size, num_channels, height, width)
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [self.model_tester.image_size // 4, self.model_tester.image_size // 4] , )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Any ) -> Tuple:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = _config_zero_init(lowerCAmelCase_ )
__lowerCAmelCase = _config_zero_init(configs_no_init.backbone_config )
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(config=lowerCAmelCase_ )
for name, param in model.named_parameters():
if param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() , [0.0, 1.0] , msg=f"""Parameter {name} of model {model_class} seems not properly initialized""" , )
@unittest.skip(reason='UperNet does not have tied weights' )
def lowercase ( self : Any ) -> int:
pass
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in UPERNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
def a_ ( ):
__lowerCAmelCase = hf_hub_download(
repo_id='hf-internal-testing/fixtures_ade20k', repo_type='dataset', filename='ADE_val_00000001.jpg' )
__lowerCAmelCase = Image.open(lowerCAmelCase_ ).convert('RGB' )
return image
@require_torch
@require_vision
@slow
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Dict ) -> Union[str, Any]:
__lowerCAmelCase = AutoImageProcessor.from_pretrained('openmmlab/upernet-swin-tiny' )
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained('openmmlab/upernet-swin-tiny' ).to(lowerCAmelCase_ )
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = torch.Size((1, model.config.num_labels, 5_1_2, 5_1_2) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-7.59_58, -7.59_58, -7.43_02], [-7.59_58, -7.59_58, -7.43_02], [-7.47_97, -7.47_97, -7.30_68]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, 0, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = AutoImageProcessor.from_pretrained('openmmlab/upernet-convnext-tiny' )
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained('openmmlab/upernet-convnext-tiny' ).to(lowerCAmelCase_ )
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = torch.Size((1, model.config.num_labels, 5_1_2, 5_1_2) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-8.81_10, -8.81_10, -8.65_21], [-8.81_10, -8.81_10, -8.65_21], [-8.77_46, -8.77_46, -8.61_30]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, 0, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_tokenizers_available, is_torch_available
_snake_case : Tuple = {
'configuration_bloom': ['BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP', 'BloomConfig', 'BloomOnnxConfig'],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : List[str] = ['BloomTokenizerFast']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : Optional[int] = [
'BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST',
'BloomForCausalLM',
'BloomModel',
'BloomPreTrainedModel',
'BloomForSequenceClassification',
'BloomForTokenClassification',
'BloomForQuestionAnswering',
]
if TYPE_CHECKING:
from .configuration_bloom import BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP, BloomConfig, BloomOnnxConfig
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_bloom_fast import BloomTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_bloom import (
BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST,
BloomForCausalLM,
BloomForQuestionAnswering,
BloomForSequenceClassification,
BloomForTokenClassification,
BloomModel,
BloomPreTrainedModel,
)
else:
import sys
_snake_case : Dict = _LazyModule(__name__, globals()['__file__'], _import_structure, module_spec=__spec__)
| 53 |
import pytest
from datasets import Dataset, DatasetDict, Features, NamedSplit, Value
from datasets.io.text import TextDatasetReader
from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : Optional[Any] ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ )
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('keep_in_memory', [False, True] )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : int ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_, keep_in_memory=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize(
'features', [
None,
{'text': 'string'},
{'text': 'int32'},
{'text': 'float32'},
], )
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Dict, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = features.copy() if features else default_expected_features
__lowerCAmelCase = (
Features({feature: Value(lowerCAmelCase_ ) for feature, dtype in features.items()} ) if features is not None else None
)
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, features=lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize('split', [None, NamedSplit('train' ), 'train', 'test'] )
def a_ ( lowerCAmelCase_ : List[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Any ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_, split=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
assert dataset.split == split if split else "train"
@pytest.mark.parametrize('path_type', [str, list] )
def a_ ( lowerCAmelCase_ : Dict, lowerCAmelCase_ : Any, lowerCAmelCase_ : Dict ):
if issubclass(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = text_path
elif issubclass(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = [text_path]
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : int, lowerCAmelCase_ : Tuple=("train",) ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ )
for split in splits:
__lowerCAmelCase = dataset_dict[split]
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('keep_in_memory', [False, True] )
def a_ ( lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : Dict ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__lowerCAmelCase = TextDatasetReader({'train': text_path}, cache_dir=lowerCAmelCase_, keep_in_memory=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize(
'features', [
None,
{'text': 'string'},
{'text': 'int32'},
{'text': 'float32'},
], )
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[Any] ):
__lowerCAmelCase = tmp_path / 'cache'
# CSV file loses col_1 string dtype information: default now is "int64" instead of "string"
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = features.copy() if features else default_expected_features
__lowerCAmelCase = (
Features({feature: Value(lowerCAmelCase_ ) for feature, dtype in features.items()} ) if features is not None else None
)
__lowerCAmelCase = TextDatasetReader({'train': text_path}, features=lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize('split', [None, NamedSplit('train' ), 'train', 'test'] )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : Optional[int] ):
if split:
__lowerCAmelCase = {split: text_path}
else:
__lowerCAmelCase = 'train'
__lowerCAmelCase = {'train': text_path, 'test': text_path}
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_, splits=list(path.keys() ) )
assert all(dataset[split].split == split for split in path.keys() )
| 53 | 1 |
def a_ ( lowerCAmelCase_ : float, lowerCAmelCase_ : float, lowerCAmelCase_ : float, lowerCAmelCase_ : float, lowerCAmelCase_ : float, ):
__lowerCAmelCase = [redshift, radiation_density, matter_density, dark_energy]
if any(p < 0 for p in parameters ):
raise ValueError('All input parameters must be positive' )
if any(p > 1 for p in parameters[1:4] ):
raise ValueError('Relative densities cannot be greater than one' )
else:
__lowerCAmelCase = 1 - (matter_density + radiation_density + dark_energy)
__lowerCAmelCase = (
radiation_density * (redshift + 1) ** 4
+ matter_density * (redshift + 1) ** 3
+ curvature * (redshift + 1) ** 2
+ dark_energy
)
__lowerCAmelCase = hubble_constant * e_a ** (1 / 2)
return hubble
if __name__ == "__main__":
import doctest
# run doctest
doctest.testmod()
# demo LCDM approximation
_snake_case : Dict = 0.3
print(
hubble_parameter(
hubble_constant=68.3,
radiation_density=1e-4,
matter_density=matter_density,
dark_energy=1 - matter_density,
redshift=0,
)
)
| 53 |
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 ViTConfig, ViTForImageClassification, ViTImageProcessor, ViTModel
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Union[str, Any] = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : int=False ):
__lowerCAmelCase = []
for i in range(config.num_hidden_layers ):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((F"""blocks.{i}.norm1.weight""", F"""vit.encoder.layer.{i}.layernorm_before.weight""") )
rename_keys.append((F"""blocks.{i}.norm1.bias""", F"""vit.encoder.layer.{i}.layernorm_before.bias""") )
rename_keys.append((F"""blocks.{i}.attn.proj.weight""", F"""vit.encoder.layer.{i}.attention.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.attn.proj.bias""", F"""vit.encoder.layer.{i}.attention.output.dense.bias""") )
rename_keys.append((F"""blocks.{i}.norm2.weight""", F"""vit.encoder.layer.{i}.layernorm_after.weight""") )
rename_keys.append((F"""blocks.{i}.norm2.bias""", F"""vit.encoder.layer.{i}.layernorm_after.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.weight""", F"""vit.encoder.layer.{i}.intermediate.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.bias""", F"""vit.encoder.layer.{i}.intermediate.dense.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.weight""", F"""vit.encoder.layer.{i}.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.bias""", F"""vit.encoder.layer.{i}.output.dense.bias""") )
# projection layer + position embeddings
rename_keys.extend(
[
('cls_token', 'vit.embeddings.cls_token'),
('patch_embed.proj.weight', 'vit.embeddings.patch_embeddings.projection.weight'),
('patch_embed.proj.bias', 'vit.embeddings.patch_embeddings.projection.bias'),
('pos_embed', 'vit.embeddings.position_embeddings'),
] )
if base_model:
# layernorm + pooler
rename_keys.extend(
[
('norm.weight', 'layernorm.weight'),
('norm.bias', 'layernorm.bias'),
] )
# if just the base model, we should remove "vit" from all keys that start with "vit"
__lowerCAmelCase = [(pair[0], pair[1][4:]) if pair[1].startswith('vit' ) else pair for pair in rename_keys]
else:
# layernorm + classification head
rename_keys.extend(
[
('norm.weight', 'vit.layernorm.weight'),
('norm.bias', 'vit.layernorm.bias'),
('head.weight', 'classifier.weight'),
('head.bias', 'classifier.bias'),
] )
return rename_keys
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Optional[int]=False ):
for i in range(config.num_hidden_layers ):
if base_model:
__lowerCAmelCase = ''
else:
__lowerCAmelCase = 'vit.'
# read in weights + bias of input projection layer (in timm, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[
: config.hidden_size, :
]
__lowerCAmelCase = in_proj_bias[: config.hidden_size]
__lowerCAmelCase = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
__lowerCAmelCase = in_proj_bias[
config.hidden_size : config.hidden_size * 2
]
__lowerCAmelCase = in_proj_weight[
-config.hidden_size :, :
]
__lowerCAmelCase = in_proj_bias[-config.hidden_size :]
def a_ ( lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = ['head.weight', 'head.bias']
for k in ignore_keys:
state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Optional[Any]=True ):
__lowerCAmelCase = ViTConfig()
# patch_size
if model_name[-1] == "8":
__lowerCAmelCase = 8
# set labels if required
if not base_model:
__lowerCAmelCase = 1000
__lowerCAmelCase = 'huggingface/label-files'
__lowerCAmelCase = 'imagenet-1k-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
__lowerCAmelCase = idalabel
__lowerCAmelCase = {v: k for k, v in idalabel.items()}
# size of the architecture
if model_name in ["dino_vits8", "dino_vits16"]:
__lowerCAmelCase = 384
__lowerCAmelCase = 1536
__lowerCAmelCase = 12
__lowerCAmelCase = 6
# load original model from torch hub
__lowerCAmelCase = torch.hub.load('facebookresearch/dino:main', lowerCAmelCase_ )
original_model.eval()
# load state_dict of original model, remove and rename some keys
__lowerCAmelCase = original_model.state_dict()
if base_model:
remove_classification_head_(lowerCAmelCase_ )
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_, base_model=lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_q_k_v(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# load HuggingFace model
if base_model:
__lowerCAmelCase = ViTModel(lowerCAmelCase_, add_pooling_layer=lowerCAmelCase_ ).eval()
else:
__lowerCAmelCase = ViTForImageClassification(lowerCAmelCase_ ).eval()
model.load_state_dict(lowerCAmelCase_ )
# Check outputs on an image, prepared by ViTImageProcessor
__lowerCAmelCase = ViTImageProcessor()
__lowerCAmelCase = image_processor(images=prepare_img(), return_tensors='pt' )
__lowerCAmelCase = encoding['pixel_values']
__lowerCAmelCase = model(lowerCAmelCase_ )
if base_model:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert torch.allclose(lowerCAmelCase_, outputs.last_hidden_state[:, 0, :], atol=1E-1 )
else:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert logits.shape == outputs.logits.shape
assert torch.allclose(lowerCAmelCase_, outputs.logits, atol=1E-3 )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
print(F"""Saving model {model_name} to {pytorch_dump_folder_path}""" )
model.save_pretrained(lowerCAmelCase_ )
print(F"""Saving image processor to {pytorch_dump_folder_path}""" )
image_processor.save_pretrained(lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : Dict = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='dino_vitb16',
type=str,
help='Name of the model trained with DINO you\'d like to convert.',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--base_model',
action='store_true',
help='Whether to only convert the base model (no projection head weights).',
)
parser.set_defaults(base_model=True)
_snake_case : List[Any] = parser.parse_args()
convert_vit_checkpoint(args.model_name, args.pytorch_dump_folder_path, args.base_model)
| 53 | 1 |
import importlib
import inspect
import json
import os
import re
import shutil
import sys
from pathlib import Path
from typing import Dict, Optional, Union
from urllib import request
from huggingface_hub import HfFolder, cached_download, hf_hub_download, model_info
from packaging import version
from .. import __version__
from . import DIFFUSERS_DYNAMIC_MODULE_NAME, HF_MODULES_CACHE, logging
_snake_case : Optional[Any] = (
'https://raw.githubusercontent.com/huggingface/diffusers/{revision}/examples/community/{pipeline}.py'
)
_snake_case : Optional[int] = logging.get_logger(__name__) # pylint: disable=invalid-name
def a_ ( ):
__lowerCAmelCase = 'https://pypi.org/pypi/diffusers/json'
__lowerCAmelCase = json.loads(request.urlopen(lowerCAmelCase_ ).read() )['releases'].keys()
return sorted(lowerCAmelCase_, key=lambda lowerCAmelCase_ : version.Version(lowerCAmelCase_ ) )
def a_ ( ):
# This function has already been executed if HF_MODULES_CACHE already is in the Python path.
if HF_MODULES_CACHE in sys.path:
return
sys.path.append(lowerCAmelCase_ )
os.makedirs(lowerCAmelCase_, exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = Path(lowerCAmelCase_ ) / '__init__.py'
if not init_path.exists():
init_path.touch()
def a_ ( lowerCAmelCase_ : Union[str, os.PathLike] ):
init_hf_modules()
__lowerCAmelCase = Path(lowerCAmelCase_ ) / name
# If the parent module does not exist yet, recursively create it.
if not dynamic_module_path.parent.exists():
create_dynamic_module(dynamic_module_path.parent )
os.makedirs(lowerCAmelCase_, exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = dynamic_module_path / '__init__.py'
if not init_path.exists():
init_path.touch()
def a_ ( lowerCAmelCase_ : List[Any] ):
with open(lowerCAmelCase_, 'r', encoding='utf-8' ) as f:
__lowerCAmelCase = f.read()
# Imports of the form `import .xxx`
__lowerCAmelCase = re.findall('^\s*import\s+\.(\S+)\s*$', lowerCAmelCase_, flags=re.MULTILINE )
# Imports of the form `from .xxx import yyy`
relative_imports += re.findall('^\s*from\s+\.(\S+)\s+import', lowerCAmelCase_, flags=re.MULTILINE )
# Unique-ify
return list(set(lowerCAmelCase_ ) )
def a_ ( lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = False
__lowerCAmelCase = [module_file]
__lowerCAmelCase = []
# Let's recurse through all relative imports
while not no_change:
__lowerCAmelCase = []
for f in files_to_check:
new_imports.extend(get_relative_imports(lowerCAmelCase_ ) )
__lowerCAmelCase = Path(lowerCAmelCase_ ).parent
__lowerCAmelCase = [str(module_path / m ) for m in new_imports]
__lowerCAmelCase = [f for f in new_import_files if f not in all_relative_imports]
__lowerCAmelCase = [F"""{f}.py""" for f in new_import_files]
__lowerCAmelCase = len(lowerCAmelCase_ ) == 0
all_relative_imports.extend(lowerCAmelCase_ )
return all_relative_imports
def a_ ( lowerCAmelCase_ : List[str] ):
with open(lowerCAmelCase_, 'r', encoding='utf-8' ) as f:
__lowerCAmelCase = f.read()
# Imports of the form `import xxx`
__lowerCAmelCase = re.findall('^\s*import\s+(\S+)\s*$', lowerCAmelCase_, flags=re.MULTILINE )
# Imports of the form `from xxx import yyy`
imports += re.findall('^\s*from\s+(\S+)\s+import', lowerCAmelCase_, flags=re.MULTILINE )
# Only keep the top-level module
__lowerCAmelCase = [imp.split('.' )[0] for imp in imports if not imp.startswith('.' )]
# Unique-ify and test we got them all
__lowerCAmelCase = list(set(lowerCAmelCase_ ) )
__lowerCAmelCase = []
for imp in imports:
try:
importlib.import_module(lowerCAmelCase_ )
except ImportError:
missing_packages.append(lowerCAmelCase_ )
if len(lowerCAmelCase_ ) > 0:
raise ImportError(
'This modeling file requires the following packages that were not found in your environment: '
F"""{", ".join(lowerCAmelCase_ )}. Run `pip install {" ".join(lowerCAmelCase_ )}`""" )
return get_relative_imports(lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : List[Any], lowerCAmelCase_ : Dict ):
__lowerCAmelCase = module_path.replace(os.path.sep, '.' )
__lowerCAmelCase = importlib.import_module(lowerCAmelCase_ )
if class_name is None:
return find_pipeline_class(lowerCAmelCase_ )
return getattr(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Optional[int] ):
from ..pipelines import DiffusionPipeline
__lowerCAmelCase = dict(inspect.getmembers(lowerCAmelCase_, inspect.isclass ) )
__lowerCAmelCase = None
for cls_name, cls in cls_members.items():
if (
cls_name != DiffusionPipeline.__name__
and issubclass(cls, lowerCAmelCase_ )
and cls.__module__.split('.' )[0] != "diffusers"
):
if pipeline_class is not None:
raise ValueError(
F"""Multiple classes that inherit from {DiffusionPipeline.__name__} have been found:"""
F""" {pipeline_class.__name__}, and {cls_name}. Please make sure to define only one in"""
F""" {loaded_module}.""" )
__lowerCAmelCase = cls
return pipeline_class
def a_ ( lowerCAmelCase_ : Union[str, os.PathLike], lowerCAmelCase_ : str, lowerCAmelCase_ : Optional[Union[str, os.PathLike]] = None, lowerCAmelCase_ : bool = False, lowerCAmelCase_ : bool = False, lowerCAmelCase_ : Optional[Dict[str, str]] = None, lowerCAmelCase_ : Optional[Union[bool, str]] = None, lowerCAmelCase_ : Optional[str] = None, lowerCAmelCase_ : bool = False, ):
__lowerCAmelCase = str(lowerCAmelCase_ )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
if os.path.isfile(lowerCAmelCase_ ):
__lowerCAmelCase = module_file_or_url
__lowerCAmelCase = 'local'
elif pretrained_model_name_or_path.count('/' ) == 0:
__lowerCAmelCase = get_diffusers_versions()
# cut ".dev0"
__lowerCAmelCase = 'v' + '.'.join(__version__.split('.' )[:3] )
# retrieve github version that matches
if revision is None:
__lowerCAmelCase = latest_version if latest_version[1:] in available_versions else 'main'
logger.info(F"""Defaulting to latest_version: {revision}.""" )
elif revision in available_versions:
__lowerCAmelCase = F"""v{revision}"""
elif revision == "main":
__lowerCAmelCase = revision
else:
raise ValueError(
F"""`custom_revision`: {revision} does not exist. Please make sure to choose one of"""
F""" {", ".join(available_versions + ["main"] )}.""" )
# community pipeline on GitHub
__lowerCAmelCase = COMMUNITY_PIPELINES_URL.format(revision=lowerCAmelCase_, pipeline=lowerCAmelCase_ )
try:
__lowerCAmelCase = cached_download(
lowerCAmelCase_, cache_dir=lowerCAmelCase_, force_download=lowerCAmelCase_, proxies=lowerCAmelCase_, resume_download=lowerCAmelCase_, local_files_only=lowerCAmelCase_, use_auth_token=lowerCAmelCase_, )
__lowerCAmelCase = 'git'
__lowerCAmelCase = pretrained_model_name_or_path + '.py'
except EnvironmentError:
logger.error(F"""Could not locate the {module_file} inside {pretrained_model_name_or_path}.""" )
raise
else:
try:
# Load from URL or cache if already cached
__lowerCAmelCase = hf_hub_download(
lowerCAmelCase_, lowerCAmelCase_, cache_dir=lowerCAmelCase_, force_download=lowerCAmelCase_, proxies=lowerCAmelCase_, resume_download=lowerCAmelCase_, local_files_only=lowerCAmelCase_, use_auth_token=lowerCAmelCase_, )
__lowerCAmelCase = os.path.join('local', '--'.join(pretrained_model_name_or_path.split('/' ) ) )
except EnvironmentError:
logger.error(F"""Could not locate the {module_file} inside {pretrained_model_name_or_path}.""" )
raise
# Check we have all the requirements in our environment
__lowerCAmelCase = check_imports(lowerCAmelCase_ )
# Now we move the module inside our cached dynamic modules.
__lowerCAmelCase = DIFFUSERS_DYNAMIC_MODULE_NAME + os.path.sep + submodule
create_dynamic_module(lowerCAmelCase_ )
__lowerCAmelCase = Path(lowerCAmelCase_ ) / full_submodule
if submodule == "local" or submodule == "git":
# We always copy local files (we could hash the file to see if there was a change, and give them the name of
# that hash, to only copy when there is a modification but it seems overkill for now).
# The only reason we do the copy is to avoid putting too many folders in sys.path.
shutil.copy(lowerCAmelCase_, submodule_path / module_file )
for module_needed in modules_needed:
__lowerCAmelCase = F"""{module_needed}.py"""
shutil.copy(os.path.join(lowerCAmelCase_, lowerCAmelCase_ ), submodule_path / module_needed )
else:
# Get the commit hash
# TODO: we will get this info in the etag soon, so retrieve it from there and not here.
if isinstance(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = use_auth_token
elif use_auth_token is True:
__lowerCAmelCase = HfFolder.get_token()
else:
__lowerCAmelCase = None
__lowerCAmelCase = model_info(lowerCAmelCase_, revision=lowerCAmelCase_, token=lowerCAmelCase_ ).sha
# The module file will end up being placed in a subfolder with the git hash of the repo. This way we get the
# benefit of versioning.
__lowerCAmelCase = submodule_path / commit_hash
__lowerCAmelCase = full_submodule + os.path.sep + commit_hash
create_dynamic_module(lowerCAmelCase_ )
if not (submodule_path / module_file).exists():
shutil.copy(lowerCAmelCase_, submodule_path / module_file )
# Make sure we also have every file with relative
for module_needed in modules_needed:
if not (submodule_path / module_needed).exists():
get_cached_module_file(
lowerCAmelCase_, F"""{module_needed}.py""", cache_dir=lowerCAmelCase_, force_download=lowerCAmelCase_, resume_download=lowerCAmelCase_, proxies=lowerCAmelCase_, use_auth_token=lowerCAmelCase_, revision=lowerCAmelCase_, local_files_only=lowerCAmelCase_, )
return os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Union[str, os.PathLike], lowerCAmelCase_ : str, lowerCAmelCase_ : Optional[str] = None, lowerCAmelCase_ : Optional[Union[str, os.PathLike]] = None, lowerCAmelCase_ : bool = False, lowerCAmelCase_ : bool = False, lowerCAmelCase_ : Optional[Dict[str, str]] = None, lowerCAmelCase_ : Optional[Union[bool, str]] = None, lowerCAmelCase_ : Optional[str] = None, lowerCAmelCase_ : bool = False, **lowerCAmelCase_ : List[Any], ):
__lowerCAmelCase = get_cached_module_file(
lowerCAmelCase_, lowerCAmelCase_, cache_dir=lowerCAmelCase_, force_download=lowerCAmelCase_, resume_download=lowerCAmelCase_, proxies=lowerCAmelCase_, use_auth_token=lowerCAmelCase_, revision=lowerCAmelCase_, local_files_only=lowerCAmelCase_, )
return get_class_in_module(lowerCAmelCase_, final_module.replace('.py', '' ) )
| 53 |
import unittest
from diffusers.pipelines.pipeline_utils import is_safetensors_compatible
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Union[str, Any] ) -> List[str]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Any:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
# Removed: 'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> int:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
# Removed: 'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> str:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[str]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Union[str, Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
# Removed: 'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : List[str] ) -> List[Any]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : Optional[Any] ) -> Optional[Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
# 'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
| 53 | 1 |
import torch
def a_ ( ):
if torch.cuda.is_available():
__lowerCAmelCase = torch.cuda.device_count()
else:
__lowerCAmelCase = 0
print(F"""Successfully ran on {num_gpus} GPUs""" )
if __name__ == "__main__":
main()
| 53 |
import math
def a_ ( lowerCAmelCase_ : list, lowerCAmelCase_ : int ):
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = int(math.floor(math.sqrt(lowerCAmelCase_ ) ) )
__lowerCAmelCase = 0
while arr[min(lowerCAmelCase_, lowerCAmelCase_ ) - 1] < x:
__lowerCAmelCase = step
step += int(math.floor(math.sqrt(lowerCAmelCase_ ) ) )
if prev >= n:
return -1
while arr[prev] < x:
__lowerCAmelCase = prev + 1
if prev == min(lowerCAmelCase_, lowerCAmelCase_ ):
return -1
if arr[prev] == x:
return prev
return -1
if __name__ == "__main__":
_snake_case : List[str] = input('Enter numbers separated by a comma:\n').strip()
_snake_case : Optional[Any] = [int(item) for item in user_input.split(',')]
_snake_case : List[str] = int(input('Enter the number to be searched:\n'))
_snake_case : Optional[int] = jump_search(arr, x)
if res == -1:
print('Number not found!')
else:
print(F"""Number {x} is at index {res}""")
| 53 | 1 |
from __future__ import annotations
import unittest
from transformers import AutoTokenizer, PegasusConfig, is_tf_available
from transformers.testing_utils import require_sentencepiece, require_tf, require_tokenizers, slow
from transformers.utils import cached_property
from ...test_configuration_common import ConfigTester
from ...test_modeling_tf_common import TFModelTesterMixin, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_tf_available():
import tensorflow as tf
from transformers import TFAutoModelForSeqaSeqLM, TFPegasusForConditionalGeneration, TFPegasusModel
@require_tf
class _UpperCAmelCase :
"""simple docstring"""
a_ = PegasusConfig
a_ = {}
a_ = """gelu"""
def __init__( self : Tuple , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[int]=1_3 , lowerCAmelCase_ : List[Any]=7 , lowerCAmelCase_ : Optional[int]=True , lowerCAmelCase_ : List[str]=False , lowerCAmelCase_ : Union[str, Any]=9_9 , lowerCAmelCase_ : Dict=3_2 , lowerCAmelCase_ : Tuple=2 , lowerCAmelCase_ : int=4 , lowerCAmelCase_ : List[str]=3_7 , lowerCAmelCase_ : Tuple=0.1 , lowerCAmelCase_ : Optional[int]=0.1 , lowerCAmelCase_ : Optional[int]=4_0 , lowerCAmelCase_ : Dict=2 , lowerCAmelCase_ : Dict=1 , lowerCAmelCase_ : int=0 , ) -> Optional[int]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = eos_token_id
__lowerCAmelCase = pad_token_id
__lowerCAmelCase = bos_token_id
def lowercase ( self : Optional[int] ) -> Any:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length - 1] , self.vocab_size )
__lowerCAmelCase = tf.expand_dims(tf.constant([self.eos_token_id] * self.batch_size ) , 1 )
__lowerCAmelCase = tf.concat([input_ids, eos_tensor] , axis=1 )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = self.config_cls(
vocab_size=self.vocab_size , d_model=self.hidden_size , encoder_layers=self.num_hidden_layers , decoder_layers=self.num_hidden_layers , encoder_attention_heads=self.num_attention_heads , decoder_attention_heads=self.num_attention_heads , encoder_ffn_dim=self.intermediate_size , decoder_ffn_dim=self.intermediate_size , dropout=self.hidden_dropout_prob , attention_dropout=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , eos_token_ids=[2] , bos_token_id=self.bos_token_id , pad_token_id=self.pad_token_id , decoder_start_token_id=self.pad_token_id , **self.config_updates , )
__lowerCAmelCase = prepare_pegasus_inputs_dict(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
return config, inputs_dict
def lowercase ( self : List[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Any ) -> Tuple:
__lowerCAmelCase = TFPegasusModel(config=lowerCAmelCase_ ).get_decoder()
__lowerCAmelCase = inputs_dict['input_ids']
__lowerCAmelCase = input_ids[:1, :]
__lowerCAmelCase = inputs_dict['attention_mask'][:1, :]
__lowerCAmelCase = inputs_dict['head_mask']
__lowerCAmelCase = 1
# first forward pass
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , head_mask=lowerCAmelCase_ , use_cache=lowerCAmelCase_ )
__lowerCAmelCase , __lowerCAmelCase = outputs.to_tuple()
# create hypothetical next token and extent to next_input_ids
__lowerCAmelCase = ids_tensor((self.batch_size, 3) , config.vocab_size )
__lowerCAmelCase = tf.cast(ids_tensor((self.batch_size, 3) , 2 ) , tf.inta )
# append to next input_ids and
__lowerCAmelCase = tf.concat([input_ids, next_tokens] , axis=-1 )
__lowerCAmelCase = tf.concat([attention_mask, next_attn_mask] , axis=-1 )
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ )[0]
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , past_key_values=lowerCAmelCase_ )[0]
self.parent.assertEqual(next_tokens.shape[1] , output_from_past.shape[1] )
# select random slice
__lowerCAmelCase = int(ids_tensor((1,) , output_from_past.shape[-1] ) )
__lowerCAmelCase = output_from_no_past[:, -3:, random_slice_idx]
__lowerCAmelCase = output_from_past[:, :, random_slice_idx]
# test that outputs are equal for slice
tf.debugging.assert_near(lowerCAmelCase_ , lowerCAmelCase_ , rtol=1e-3 )
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : int, lowerCAmelCase_ : List[str], lowerCAmelCase_ : Optional[int]=None, lowerCAmelCase_ : Optional[int]=None, lowerCAmelCase_ : List[Any]=None, lowerCAmelCase_ : str=None, lowerCAmelCase_ : Union[str, Any]=None, ):
if attention_mask is None:
__lowerCAmelCase = tf.cast(tf.math.not_equal(lowerCAmelCase_, config.pad_token_id ), tf.inta )
if decoder_attention_mask is None:
__lowerCAmelCase = tf.concat(
[
tf.ones(decoder_input_ids[:, :1].shape, dtype=tf.inta ),
tf.cast(tf.math.not_equal(decoder_input_ids[:, 1:], config.pad_token_id ), tf.inta ),
], axis=-1, )
if head_mask is None:
__lowerCAmelCase = tf.ones((config.encoder_layers, config.encoder_attention_heads) )
if decoder_head_mask is None:
__lowerCAmelCase = tf.ones((config.decoder_layers, config.decoder_attention_heads) )
if cross_attn_head_mask is None:
__lowerCAmelCase = tf.ones((config.decoder_layers, config.decoder_attention_heads) )
return {
"input_ids": input_ids,
"decoder_input_ids": decoder_input_ids,
"attention_mask": attention_mask,
"decoder_attention_mask": decoder_attention_mask,
"head_mask": head_mask,
"decoder_head_mask": decoder_head_mask,
"cross_attn_head_mask": cross_attn_head_mask,
}
@require_tf
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (TFPegasusForConditionalGeneration, TFPegasusModel) if is_tf_available() else ()
a_ = (TFPegasusForConditionalGeneration,) if is_tf_available() else ()
a_ = (
{
"""conversational""": TFPegasusForConditionalGeneration,
"""feature-extraction""": TFPegasusModel,
"""summarization""": TFPegasusForConditionalGeneration,
"""text2text-generation""": TFPegasusForConditionalGeneration,
"""translation""": TFPegasusForConditionalGeneration,
}
if is_tf_available()
else {}
)
a_ = True
a_ = False
a_ = False
def lowercase ( self : Tuple ) -> Tuple:
__lowerCAmelCase = TFPegasusModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> Optional[Any]:
self.config_tester.run_common_tests()
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
self.model_tester.check_decoder_model_past_large_inputs(*lowerCAmelCase_ )
@require_sentencepiece
@require_tokenizers
@require_tf
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
a_ = [
""" PG&E stated it scheduled the blackouts in response to forecasts for high winds amid dry conditions. The aim is to reduce the risk of wildfires. Nearly 800 thousand customers were scheduled to be affected by the shutoffs which were expected to last through at least midday tomorrow.""",
""" The London trio are up for best UK act and best album, as well as getting two nominations in the best song category.\"We got told like this morning 'Oh I think you're nominated'\", said Dappy.\"And I was like 'Oh yeah, which one?' And now we've got nominated for four awards. I mean, wow!\"Bandmate Fazer added: \"We thought it's best of us to come down and mingle with everyone and say hello to the cameras. And now we find we've got four nominations.\"The band have two shots at the best song prize, getting the nod for their Tynchy Stryder collaboration Number One, and single Strong Again.Their album Uncle B will also go up against records by the likes of Beyonce and Kanye West.N-Dubz picked up the best newcomer Mobo in 2007, but female member Tulisa said they wouldn't be too disappointed if they didn't win this time around.\"At the end of the day we're grateful to be where we are in our careers.\"If it don't happen then it don't happen - live to fight another day and keep on making albums and hits for the fans.\"Dappy also revealed they could be performing live several times on the night.The group will be doing Number One and also a possible rendition of the War Child single, I Got Soul.The charity song is a re-working of The Killers' All These Things That I've Done and is set to feature artists like Chipmunk, Ironik and Pixie Lott.This year's Mobos will be held outside of London for the first time, in Glasgow on 30 September.N-Dubz said they were looking forward to performing for their Scottish fans and boasted about their recent shows north of the border.\"We just done Edinburgh the other day,\" said Dappy.\"We smashed up an N-Dubz show over there. We done Aberdeen about three or four months ago - we smashed up that show over there! Everywhere we go we smash it up!\" """,
]
a_ = [
"""California's largest electricity provider has cut power to hundreds of thousands of customers in an effort to"""
""" reduce the risk of wildfires.""",
"""N-Dubz have revealed they\'re \"grateful\" to have been nominated for four Mobo Awards.""",
] # differs slightly from pytorch, likely due to numerical differences in linear layers
a_ = """google/pegasus-xsum"""
@cached_property
def lowercase ( self : Any ) -> List[str]:
return AutoTokenizer.from_pretrained(self.model_name )
@cached_property
def lowercase ( self : Union[str, Any] ) -> List[Any]:
__lowerCAmelCase = TFAutoModelForSeqaSeqLM.from_pretrained(self.model_name )
return model
def lowercase ( self : str , **lowerCAmelCase_ : List[str] ) -> Optional[int]:
__lowerCAmelCase = self.translate_src_text(**lowerCAmelCase_ )
assert self.expected_text == generated_words
def lowercase ( self : Optional[int] , **lowerCAmelCase_ : str ) -> Dict:
__lowerCAmelCase = self.tokenizer(self.src_text , **lowerCAmelCase_ , padding=lowerCAmelCase_ , return_tensors='tf' )
__lowerCAmelCase = self.model.generate(
model_inputs.input_ids , attention_mask=model_inputs.attention_mask , num_beams=2 , use_cache=lowerCAmelCase_ , )
__lowerCAmelCase = self.tokenizer.batch_decode(generated_ids.numpy() , skip_special_tokens=lowerCAmelCase_ )
return generated_words
@slow
def lowercase ( self : Dict ) -> List[str]:
self._assert_generated_batch_equal_expected()
| 53 |
import argparse
import torch
from transformers import RemBertConfig, RemBertModel, load_tf_weights_in_rembert
from transformers.utils import logging
logging.set_verbosity_info()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[Any], lowerCAmelCase_ : str ):
# Initialise PyTorch model
__lowerCAmelCase = RemBertConfig.from_json_file(lowerCAmelCase_ )
print('Building PyTorch model from configuration: {}'.format(str(lowerCAmelCase_ ) ) )
__lowerCAmelCase = RemBertModel(lowerCAmelCase_ )
# Load weights from tf checkpoint
load_tf_weights_in_rembert(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# Save pytorch-model
print('Save PyTorch model to {}'.format(lowerCAmelCase_ ) )
torch.save(model.state_dict(), lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--tf_checkpoint_path', default=None, type=str, required=True, help='Path to the TensorFlow checkpoint path.'
)
parser.add_argument(
'--rembert_config_file',
default=None,
type=str,
required=True,
help=(
'The config json file corresponding to the pre-trained RemBERT model. \n'
'This specifies the model architecture.'
),
)
parser.add_argument(
'--pytorch_dump_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_rembert_tf_checkpoint_to_pytorch(args.tf_checkpoint_path, args.rembert_config_file, args.pytorch_dump_path)
| 53 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
_snake_case : str = logging.get_logger(__name__)
_snake_case : Any = {
'facebook/s2t-wav2vec2-large-en-de': (
'https://huggingface.co/facebook/s2t-wav2vec2-large-en-de/resolve/main/config.json'
),
# See all Speech2Text models at https://huggingface.co/models?filter=speech2text2
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = """speech_to_text_2"""
a_ = ["""past_key_values"""]
a_ = {"""num_attention_heads""": """decoder_attention_heads""", """hidden_size""": """d_model"""}
def __init__( self : Tuple , lowerCAmelCase_ : str=1_0_0_0_0 , lowerCAmelCase_ : Optional[int]=6 , lowerCAmelCase_ : Union[str, Any]=2_0_4_8 , lowerCAmelCase_ : int=4 , lowerCAmelCase_ : Optional[int]=0.0 , lowerCAmelCase_ : Tuple=True , lowerCAmelCase_ : Union[str, Any]="relu" , lowerCAmelCase_ : int=2_5_6 , lowerCAmelCase_ : Dict=0.1 , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : Dict=0.02 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Optional[Any]=1 , lowerCAmelCase_ : Any=0 , lowerCAmelCase_ : Any=2 , lowerCAmelCase_ : str=1_0_2_4 , **lowerCAmelCase_ : Optional[Any] , ) -> Optional[Any]:
__lowerCAmelCase = vocab_size
__lowerCAmelCase = d_model
__lowerCAmelCase = decoder_ffn_dim
__lowerCAmelCase = decoder_layers
__lowerCAmelCase = decoder_attention_heads
__lowerCAmelCase = dropout
__lowerCAmelCase = attention_dropout
__lowerCAmelCase = activation_dropout
__lowerCAmelCase = activation_function
__lowerCAmelCase = init_std
__lowerCAmelCase = decoder_layerdrop
__lowerCAmelCase = use_cache
__lowerCAmelCase = decoder_layers
__lowerCAmelCase = scale_embedding # scale factor will be sqrt(d_model) if True
__lowerCAmelCase = max_target_positions
super().__init__(
pad_token_id=lowerCAmelCase_ , bos_token_id=lowerCAmelCase_ , eos_token_id=lowerCAmelCase_ , decoder_start_token_id=lowerCAmelCase_ , **lowerCAmelCase_ , )
| 53 |
import argparse
import json
import pickle
from pathlib import Path
import requests
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import MaskFormerConfig, MaskFormerForInstanceSegmentation, MaskFormerImageProcessor, SwinConfig
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Any = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = SwinConfig.from_pretrained(
'microsoft/swin-tiny-patch4-window7-224', out_features=['stage1', 'stage2', 'stage3', 'stage4'] )
__lowerCAmelCase = MaskFormerConfig(backbone_config=lowerCAmelCase_ )
__lowerCAmelCase = 'huggingface/label-files'
if "ade20k-full" in model_name:
# this should be ok
__lowerCAmelCase = 847
__lowerCAmelCase = 'maskformer-ade20k-full-id2label.json'
elif "ade" in model_name:
# this should be ok
__lowerCAmelCase = 150
__lowerCAmelCase = 'ade20k-id2label.json'
elif "coco-stuff" in model_name:
# this should be ok
__lowerCAmelCase = 171
__lowerCAmelCase = 'maskformer-coco-stuff-id2label.json'
elif "coco" in model_name:
# TODO
__lowerCAmelCase = 133
__lowerCAmelCase = 'coco-panoptic-id2label.json'
elif "cityscapes" in model_name:
# this should be ok
__lowerCAmelCase = 19
__lowerCAmelCase = 'cityscapes-id2label.json'
elif "vistas" in model_name:
# this should be ok
__lowerCAmelCase = 65
__lowerCAmelCase = 'mapillary-vistas-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
return config
def a_ ( lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = []
# stem
# fmt: off
rename_keys.append(('backbone.patch_embed.proj.weight', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.weight') )
rename_keys.append(('backbone.patch_embed.proj.bias', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.bias') )
rename_keys.append(('backbone.patch_embed.norm.weight', 'model.pixel_level_module.encoder.model.embeddings.norm.weight') )
rename_keys.append(('backbone.patch_embed.norm.bias', 'model.pixel_level_module.encoder.model.embeddings.norm.bias') )
# stages
for i in range(len(config.backbone_config.depths ) ):
for j in range(config.backbone_config.depths[i] ):
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_bias_table""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_bias_table""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_index""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_index""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.bias""") )
if i < 3:
rename_keys.append((F"""backbone.layers.{i}.downsample.reduction.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.reduction.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.bias""") )
rename_keys.append((F"""backbone.norm{i}.weight""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.weight""") )
rename_keys.append((F"""backbone.norm{i}.bias""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.bias""") )
# FPN
rename_keys.append(('sem_seg_head.layer_4.weight', 'model.pixel_level_module.decoder.fpn.stem.0.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.weight', 'model.pixel_level_module.decoder.fpn.stem.1.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.bias', 'model.pixel_level_module.decoder.fpn.stem.1.bias') )
for source_index, target_index in zip(range(3, 0, -1 ), range(0, 3 ) ):
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.0.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.bias""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.0.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.bias""") )
rename_keys.append(('sem_seg_head.mask_features.weight', 'model.pixel_level_module.decoder.mask_projection.weight') )
rename_keys.append(('sem_seg_head.mask_features.bias', 'model.pixel_level_module.decoder.mask_projection.bias') )
# Transformer decoder
for idx in range(config.decoder_config.decoder_layers ):
# self-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.bias""") )
# cross-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.bias""") )
# MLP 1
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc1.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc1.bias""") )
# MLP 2
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc2.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc2.bias""") )
# layernorm 1 (self-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.bias""") )
# layernorm 2 (cross-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.bias""") )
# layernorm 3 (final layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.weight""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.bias""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.bias""") )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.weight', 'model.transformer_module.decoder.layernorm.weight') )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.bias', 'model.transformer_module.decoder.layernorm.bias') )
# heads on top
rename_keys.append(('sem_seg_head.predictor.query_embed.weight', 'model.transformer_module.queries_embedder.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.weight', 'model.transformer_module.input_projection.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.bias', 'model.transformer_module.input_projection.bias') )
rename_keys.append(('sem_seg_head.predictor.class_embed.weight', 'class_predictor.weight') )
rename_keys.append(('sem_seg_head.predictor.class_embed.bias', 'class_predictor.bias') )
for i in range(3 ):
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.weight""", F"""mask_embedder.{i}.0.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.bias""", F"""mask_embedder.{i}.0.bias""") )
# fmt: on
return rename_keys
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int ):
__lowerCAmelCase = [int(backbone_config.embed_dim * 2**i ) for i in range(len(backbone_config.depths ) )]
for i in range(len(backbone_config.depths ) ):
__lowerCAmelCase = num_features[i]
for j in range(backbone_config.depths[i] ):
# fmt: off
# read in weights + bias of input projection layer (in original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[:dim, :]
__lowerCAmelCase = in_proj_bias[: dim]
__lowerCAmelCase = in_proj_weight[
dim : dim * 2, :
]
__lowerCAmelCase = in_proj_bias[
dim : dim * 2
]
__lowerCAmelCase = in_proj_weight[
-dim :, :
]
__lowerCAmelCase = in_proj_bias[-dim :]
# fmt: on
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : Dict ):
# fmt: off
__lowerCAmelCase = config.decoder_config.hidden_size
for idx in range(config.decoder_config.decoder_layers ):
# read in weights + bias of self-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# read in weights + bias of cross-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# fmt: on
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : bool = False ):
__lowerCAmelCase = get_maskformer_config(lowerCAmelCase_ )
# load original state_dict
with open(lowerCAmelCase_, 'rb' ) as f:
__lowerCAmelCase = pickle.load(lowerCAmelCase_ )
__lowerCAmelCase = data['model']
# for name, param in state_dict.items():
# print(name, param.shape)
# rename keys
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_swin_q_k_v(lowerCAmelCase_, config.backbone_config )
read_in_decoder_q_k_v(lowerCAmelCase_, lowerCAmelCase_ )
# update to torch tensors
for key, value in state_dict.items():
__lowerCAmelCase = torch.from_numpy(lowerCAmelCase_ )
# load 🤗 model
__lowerCAmelCase = MaskFormerForInstanceSegmentation(lowerCAmelCase_ )
model.eval()
for name, param in model.named_parameters():
print(lowerCAmelCase_, param.shape )
__lowerCAmelCase , __lowerCAmelCase = model.load_state_dict(lowerCAmelCase_, strict=lowerCAmelCase_ )
assert missing_keys == [
"model.pixel_level_module.encoder.model.layernorm.weight",
"model.pixel_level_module.encoder.model.layernorm.bias",
]
assert len(lowerCAmelCase_ ) == 0, F"""Unexpected keys: {unexpected_keys}"""
# verify results
__lowerCAmelCase = prepare_img()
if "vistas" in model_name:
__lowerCAmelCase = 65
elif "cityscapes" in model_name:
__lowerCAmelCase = 6_5535
else:
__lowerCAmelCase = 255
__lowerCAmelCase = True if 'ade' in model_name else False
__lowerCAmelCase = MaskFormerImageProcessor(ignore_index=lowerCAmelCase_, reduce_labels=lowerCAmelCase_ )
__lowerCAmelCase = image_processor(lowerCAmelCase_, return_tensors='pt' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
print('Logits:', outputs.class_queries_logits[0, :3, :3] )
if model_name == "maskformer-swin-tiny-ade":
__lowerCAmelCase = torch.tensor(
[[3.6353, -4.4770, -2.6065], [0.5081, -4.2394, -3.5343], [2.1909, -5.0353, -1.9323]] )
assert torch.allclose(outputs.class_queries_logits[0, :3, :3], lowerCAmelCase_, atol=1E-4 )
print('Looks ok!' )
if pytorch_dump_folder_path is not None:
print(F"""Saving model and image processor to {pytorch_dump_folder_path}""" )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
model.save_pretrained(lowerCAmelCase_ )
image_processor.save_pretrained(lowerCAmelCase_ )
if push_to_hub:
print('Pushing model and image processor to the hub...' )
model.push_to_hub(F"""nielsr/{model_name}""" )
image_processor.push_to_hub(F"""nielsr/{model_name}""" )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='maskformer-swin-tiny-ade',
type=str,
help=('Name of the MaskFormer model you\'d like to convert',),
)
parser.add_argument(
'--checkpoint_path',
default='/Users/nielsrogge/Documents/MaskFormer_checkpoints/MaskFormer-Swin-tiny-ADE20k/model.pkl',
type=str,
help='Path to the original state dict (.pth file).',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--push_to_hub', action='store_true', help='Whether or not to push the converted model to the 🤗 hub.'
)
_snake_case : List[str] = parser.parse_args()
convert_maskformer_checkpoint(
args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub
)
| 53 | 1 |
def a_ ( lowerCAmelCase_ : float, lowerCAmelCase_ : float ):
if density <= 0:
raise ValueError('Impossible fluid density' )
if bulk_modulus <= 0:
raise ValueError('Impossible bulk modulus' )
return (bulk_modulus / density) ** 0.5
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 |
import json
import logging
import os
import re
import sys
from dataclasses import dataclass, field
from typing import Any, Dict, List, Optional, Union
import datasets
import numpy as np
import torch
import torchaudio
from packaging import version
from torch import nn
import transformers
from transformers import (
HfArgumentParser,
Trainer,
TrainingArguments,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaForCTC,
WavaVecaProcessor,
is_apex_available,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
if is_apex_available():
from apex import amp
if version.parse(version.parse(torch.__version__).base_version) >= version.parse('1.6'):
_snake_case : List[Any] = True
from torch.cuda.amp import autocast
_snake_case : Dict = logging.getLogger(__name__)
def a_ ( lowerCAmelCase_ : str=None, lowerCAmelCase_ : str=None ):
return field(default_factory=lambda: default, metadata=lowerCAmelCase_ )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
metadata={"""help""": """Path to pretrained model or model identifier from huggingface.co/models"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to freeze the feature extractor layers of the model."""} )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout ratio for the attention probabilities."""} )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout ratio for activations inside the fully connected layer."""} )
a_ = field(
default=0.1 , metadata={
"""help""": """The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler."""
} , )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout probabilitiy for all 1D convolutional layers in feature extractor."""} , )
a_ = field(
default=0.05 , metadata={
"""help""": (
"""Propability of each feature vector along the time axis to be chosen as the start of the vector"""
"""span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"""
"""vectors will be masked along the time axis. This is only relevant if ``apply_spec_augment is True``."""
)
} , )
a_ = field(default=0.0 , metadata={"""help""": """The LayerDrop probability."""} )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(
default="""train+validation""" , metadata={
"""help""": """The name of the training data set split to use (via the datasets library). Defaults to 'train'"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached preprocessed datasets or not."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""For debugging purposes or quicker training, truncate the number of training examples to this """
"""value if set."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""For debugging purposes or quicker training, truncate the number of validation examples to this """
"""value if set."""
)
} , )
a_ = list_field(
default=[""",""", """?""", """.""", """!""", """-""", """;""", """:""", """\"\"""", """%""", """'""", """\"""", """�"""] , metadata={"""help""": """A list of characters to remove from the transcripts."""} , )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = 42
a_ = True
a_ = None
a_ = None
a_ = None
a_ = None
def __call__( self : int , lowerCAmelCase_ : List[Dict[str, Union[List[int], torch.Tensor]]] ) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# different padding methods
__lowerCAmelCase = [{'input_values': feature['input_values']} for feature in features]
__lowerCAmelCase = [{'input_ids': feature['labels']} for feature in features]
__lowerCAmelCase = self.processor.pad(
lowerCAmelCase_ , padding=self.padding , max_length=self.max_length , pad_to_multiple_of=self.pad_to_multiple_of , return_tensors='pt' , )
__lowerCAmelCase = self.processor.pad(
labels=lowerCAmelCase_ , padding=self.padding , max_length=self.max_length_labels , pad_to_multiple_of=self.pad_to_multiple_of_labels , return_tensors='pt' , )
# replace padding with -100 to ignore loss correctly
__lowerCAmelCase = labels_batch['input_ids'].masked_fill(labels_batch.attention_mask.ne(1 ) , -1_0_0 )
__lowerCAmelCase = labels
return batch
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Tuple , lowerCAmelCase_ : nn.Module , lowerCAmelCase_ : Dict[str, Union[torch.Tensor, Any]] ) -> torch.Tensor:
model.train()
__lowerCAmelCase = self._prepare_inputs(lowerCAmelCase_ )
if self.use_amp:
with autocast():
__lowerCAmelCase = self.compute_loss(lowerCAmelCase_ , lowerCAmelCase_ )
else:
__lowerCAmelCase = self.compute_loss(lowerCAmelCase_ , lowerCAmelCase_ )
if self.args.n_gpu > 1:
if model.module.config.ctc_loss_reduction == "mean":
__lowerCAmelCase = loss.mean()
elif model.module.config.ctc_loss_reduction == "sum":
__lowerCAmelCase = loss.sum() / (inputs['labels'] >= 0).sum()
else:
raise ValueError(f"""{model.config.ctc_loss_reduction} is not valid. Choose one of ['mean', 'sum']""" )
if self.args.gradient_accumulation_steps > 1:
__lowerCAmelCase = loss / self.args.gradient_accumulation_steps
if self.use_amp:
self.scaler.scale(lowerCAmelCase_ ).backward()
elif self.use_apex:
with amp.scale_loss(lowerCAmelCase_ , self.optimizer ) as scaled_loss:
scaled_loss.backward()
elif self.deepspeed:
self.deepspeed.backward(lowerCAmelCase_ )
else:
loss.backward()
return loss.detach()
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets:
__lowerCAmelCase = datasets.load_dataset(
'common_voice', data_args.dataset_config_name, split=data_args.train_split_name )
__lowerCAmelCase = datasets.load_dataset('common_voice', data_args.dataset_config_name, split='test' )
# Create and save tokenizer
__lowerCAmelCase = F"""[{"".join(data_args.chars_to_ignore )}]"""
def remove_special_characters(lowerCAmelCase_ : Any ):
__lowerCAmelCase = re.sub(lowerCAmelCase_, '', batch['sentence'] ).lower() + ' '
return batch
__lowerCAmelCase = train_dataset.map(lowerCAmelCase_, remove_columns=['sentence'] )
__lowerCAmelCase = eval_dataset.map(lowerCAmelCase_, remove_columns=['sentence'] )
def extract_all_chars(lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = ' '.join(batch['text'] )
__lowerCAmelCase = list(set(lowerCAmelCase_ ) )
return {"vocab": [vocab], "all_text": [all_text]}
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, batched=lowerCAmelCase_, batch_size=-1, keep_in_memory=lowerCAmelCase_, remove_columns=train_dataset.column_names, )
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, batched=lowerCAmelCase_, batch_size=-1, keep_in_memory=lowerCAmelCase_, remove_columns=eval_dataset.column_names, )
__lowerCAmelCase = list(set(vocab_train['vocab'][0] ) | set(vocab_test['vocab'][0] ) )
__lowerCAmelCase = {v: k for k, v in enumerate(lowerCAmelCase_ )}
__lowerCAmelCase = vocab_dict[' ']
del vocab_dict[" "]
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = len(lowerCAmelCase_ )
with open('vocab.json', 'w' ) as vocab_file:
json.dump(lowerCAmelCase_, lowerCAmelCase_ )
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = WavaVecaCTCTokenizer(
'vocab.json', unk_token='[UNK]', pad_token='[PAD]', word_delimiter_token='|', )
__lowerCAmelCase = WavaVecaFeatureExtractor(
feature_size=1, sampling_rate=1_6000, padding_value=0.0, do_normalize=lowerCAmelCase_, return_attention_mask=lowerCAmelCase_ )
__lowerCAmelCase = WavaVecaProcessor(feature_extractor=lowerCAmelCase_, tokenizer=lowerCAmelCase_ )
__lowerCAmelCase = WavaVecaForCTC.from_pretrained(
model_args.model_name_or_path, cache_dir=model_args.cache_dir, activation_dropout=model_args.activation_dropout, attention_dropout=model_args.attention_dropout, hidden_dropout=model_args.hidden_dropout, feat_proj_dropout=model_args.feat_proj_dropout, mask_time_prob=model_args.mask_time_prob, gradient_checkpointing=training_args.gradient_checkpointing, layerdrop=model_args.layerdrop, ctc_loss_reduction='mean', pad_token_id=processor.tokenizer.pad_token_id, vocab_size=len(processor.tokenizer ), )
if data_args.max_train_samples is not None:
__lowerCAmelCase = min(len(lowerCAmelCase_ ), data_args.max_train_samples )
__lowerCAmelCase = train_dataset.select(range(lowerCAmelCase_ ) )
if data_args.max_val_samples is not None:
__lowerCAmelCase = eval_dataset.select(range(data_args.max_val_samples ) )
__lowerCAmelCase = torchaudio.transforms.Resample(4_8000, 1_6000 )
# Preprocessing the datasets.
# We need to read the aduio files as arrays and tokenize the targets.
def speech_file_to_array_fn(lowerCAmelCase_ : int ):
__lowerCAmelCase , __lowerCAmelCase = torchaudio.load(batch['path'] )
__lowerCAmelCase = resampler(lowerCAmelCase_ ).squeeze().numpy()
__lowerCAmelCase = 1_6000
__lowerCAmelCase = batch['text']
return batch
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, remove_columns=train_dataset.column_names, num_proc=data_args.preprocessing_num_workers, )
__lowerCAmelCase = eval_dataset.map(
lowerCAmelCase_, remove_columns=eval_dataset.column_names, num_proc=data_args.preprocessing_num_workers, )
def prepare_dataset(lowerCAmelCase_ : Union[str, Any] ):
# check that all files have the correct sampling rate
assert (
len(set(batch['sampling_rate'] ) ) == 1
), F"""Make sure all inputs have the same sampling rate of {processor.feature_extractor.sampling_rate}."""
__lowerCAmelCase = processor(
audio=batch['speech'], text=batch['target_text'], sampling_rate=batch['sampling_rate'][0] )
batch.update(lowerCAmelCase_ )
return batch
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, remove_columns=train_dataset.column_names, batch_size=training_args.per_device_train_batch_size, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, )
__lowerCAmelCase = eval_dataset.map(
lowerCAmelCase_, remove_columns=eval_dataset.column_names, batch_size=training_args.per_device_train_batch_size, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, )
# Metric
__lowerCAmelCase = datasets.load_metric('wer' )
def compute_metrics(lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = pred.predictions
__lowerCAmelCase = np.argmax(lowerCAmelCase_, axis=-1 )
__lowerCAmelCase = processor.tokenizer.pad_token_id
__lowerCAmelCase = processor.batch_decode(lowerCAmelCase_ )
# we do not want to group tokens when computing the metrics
__lowerCAmelCase = processor.batch_decode(pred.label_ids, group_tokens=lowerCAmelCase_ )
__lowerCAmelCase = wer_metric.compute(predictions=lowerCAmelCase_, references=lowerCAmelCase_ )
return {"wer": wer}
if model_args.freeze_feature_extractor:
model.freeze_feature_extractor()
# Data collator
__lowerCAmelCase = DataCollatorCTCWithPadding(processor=lowerCAmelCase_, padding=lowerCAmelCase_ )
# Initialize our Trainer
__lowerCAmelCase = CTCTrainer(
model=lowerCAmelCase_, data_collator=lowerCAmelCase_, args=lowerCAmelCase_, compute_metrics=lowerCAmelCase_, train_dataset=train_dataset if training_args.do_train else None, eval_dataset=eval_dataset if training_args.do_eval else None, tokenizer=processor.feature_extractor, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
# Save the feature_extractor and the tokenizer
if is_main_process(training_args.local_rank ):
processor.save_pretrained(training_args.output_dir )
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model()
__lowerCAmelCase = train_result.metrics
__lowerCAmelCase = (
data_args.max_train_samples if data_args.max_train_samples is not None else len(lowerCAmelCase_ )
)
__lowerCAmelCase = min(lowerCAmelCase_, len(lowerCAmelCase_ ) )
trainer.log_metrics('train', lowerCAmelCase_ )
trainer.save_metrics('train', lowerCAmelCase_ )
trainer.save_state()
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = data_args.max_val_samples if data_args.max_val_samples is not None else len(lowerCAmelCase_ )
__lowerCAmelCase = min(lowerCAmelCase_, len(lowerCAmelCase_ ) )
trainer.log_metrics('eval', lowerCAmelCase_ )
trainer.save_metrics('eval', lowerCAmelCase_ )
return results
if __name__ == "__main__":
main()
| 53 | 1 |
from __future__ import annotations
import numpy as np
def a_ ( lowerCAmelCase_ : list[float] ):
return np.maximum(0, lowerCAmelCase_ )
if __name__ == "__main__":
print(np.array(relu([-1, 0, 5]))) # --> [0, 0, 5]
| 53 |
import json
from typing import List, Optional, Tuple
from tokenizers import normalizers
from ....tokenization_utils_fast import PreTrainedTokenizerFast
from ....utils import logging
from .tokenization_retribert import RetriBertTokenizer
_snake_case : Any = logging.get_logger(__name__)
_snake_case : int = {'vocab_file': 'vocab.txt', 'tokenizer_file': 'tokenizer.json'}
_snake_case : Optional[Any] = {
'vocab_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/vocab.txt'
),
},
'tokenizer_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/tokenizer.json'
),
},
}
_snake_case : str = {
'yjernite/retribert-base-uncased': 512,
}
_snake_case : Optional[int] = {
'yjernite/retribert-base-uncased': {'do_lower_case': True},
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = PRETRAINED_VOCAB_FILES_MAP
a_ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
a_ = PRETRAINED_INIT_CONFIGURATION
a_ = RetriBertTokenizer
a_ = ["""input_ids""", """attention_mask"""]
def __init__( self : Dict , lowerCAmelCase_ : Optional[Any]=None , lowerCAmelCase_ : Tuple=None , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : str="[UNK]" , lowerCAmelCase_ : Optional[Any]="[SEP]" , lowerCAmelCase_ : List[str]="[PAD]" , lowerCAmelCase_ : Optional[int]="[CLS]" , lowerCAmelCase_ : List[Any]="[MASK]" , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : List[str]=None , **lowerCAmelCase_ : List[Any] , ) -> Dict:
super().__init__(
lowerCAmelCase_ , tokenizer_file=lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , unk_token=lowerCAmelCase_ , sep_token=lowerCAmelCase_ , pad_token=lowerCAmelCase_ , cls_token=lowerCAmelCase_ , mask_token=lowerCAmelCase_ , tokenize_chinese_chars=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ , **lowerCAmelCase_ , )
__lowerCAmelCase = json.loads(self.backend_tokenizer.normalizer.__getstate__() )
if (
normalizer_state.get('lowercase' , lowerCAmelCase_ ) != do_lower_case
or normalizer_state.get('strip_accents' , lowerCAmelCase_ ) != strip_accents
or normalizer_state.get('handle_chinese_chars' , lowerCAmelCase_ ) != tokenize_chinese_chars
):
__lowerCAmelCase = getattr(lowerCAmelCase_ , normalizer_state.pop('type' ) )
__lowerCAmelCase = do_lower_case
__lowerCAmelCase = strip_accents
__lowerCAmelCase = tokenize_chinese_chars
__lowerCAmelCase = normalizer_class(**lowerCAmelCase_ )
__lowerCAmelCase = do_lower_case
def lowercase ( self : str , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Optional[int]=None ) -> Optional[int]:
__lowerCAmelCase = [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 lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[int] , lowerCAmelCase_ : Optional[List[int]] = None ) -> List[int]:
__lowerCAmelCase = [self.sep_token_id]
__lowerCAmelCase = [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 lowercase ( self : int , lowerCAmelCase_ : str , lowerCAmelCase_ : Optional[str] = None ) -> Tuple[str]:
__lowerCAmelCase = self._tokenizer.model.save(lowerCAmelCase_ , name=lowerCAmelCase_ )
return tuple(lowerCAmelCase_ )
| 53 | 1 |
from math import isclose, sqrt
def a_ ( lowerCAmelCase_ : float, lowerCAmelCase_ : float, lowerCAmelCase_ : float ):
__lowerCAmelCase = point_y / 4 / point_x
__lowerCAmelCase = 2 * normal_gradient / (1 + normal_gradient * normal_gradient)
__lowerCAmelCase = (1 - normal_gradient * normal_gradient) / (
1 + normal_gradient * normal_gradient
)
__lowerCAmelCase = (sa - ca * incoming_gradient) / (ca + sa * incoming_gradient)
# to find the next point, solve the simultaeneous equations:
# y^2 + 4x^2 = 100
# y - b = m * (x - a)
# ==> A x^2 + B x + C = 0
__lowerCAmelCase = outgoing_gradient**2 + 4
__lowerCAmelCase = 2 * outgoing_gradient * (point_y - outgoing_gradient * point_x)
__lowerCAmelCase = (point_y - outgoing_gradient * point_x) ** 2 - 100
__lowerCAmelCase = (
-linear_term - sqrt(linear_term**2 - 4 * quadratic_term * constant_term )
) / (2 * quadratic_term)
__lowerCAmelCase = (
-linear_term + sqrt(linear_term**2 - 4 * quadratic_term * constant_term )
) / (2 * quadratic_term)
# two solutions, one of which is our input point
__lowerCAmelCase = x_minus if isclose(lowerCAmelCase_, lowerCAmelCase_ ) else x_plus
__lowerCAmelCase = point_y + outgoing_gradient * (next_x - point_x)
return next_x, next_y, outgoing_gradient
def a_ ( lowerCAmelCase_ : float = 1.4, lowerCAmelCase_ : float = -9.6 ):
__lowerCAmelCase = 0
__lowerCAmelCase = first_x_coord
__lowerCAmelCase = first_y_coord
__lowerCAmelCase = (10.1 - point_y) / (0.0 - point_x)
while not (-0.01 <= point_x <= 0.01 and point_y > 0):
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = next_point(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
num_reflections += 1
return num_reflections
if __name__ == "__main__":
print(F"""{solution() = }""")
| 53 |
import numpy as np
from cva import COLOR_BGR2GRAY, cvtColor, imread
from numpy import array, uinta
from PIL import Image
from digital_image_processing import change_contrast as cc
from digital_image_processing import convert_to_negative as cn
from digital_image_processing import sepia as sp
from digital_image_processing.dithering import burkes as bs
from digital_image_processing.edge_detection import canny
from digital_image_processing.filters import convolve as conv
from digital_image_processing.filters import gaussian_filter as gg
from digital_image_processing.filters import local_binary_pattern as lbp
from digital_image_processing.filters import median_filter as med
from digital_image_processing.filters import sobel_filter as sob
from digital_image_processing.resize import resize as rs
_snake_case : Union[str, Any] = imread(R'digital_image_processing/image_data/lena_small.jpg')
_snake_case : Optional[int] = cvtColor(img, COLOR_BGR2GRAY)
def a_ ( ):
__lowerCAmelCase = cn.convert_to_negative(lowerCAmelCase_ )
# assert negative_img array for at least one True
assert negative_img.any()
def a_ ( ):
with Image.open('digital_image_processing/image_data/lena_small.jpg' ) as img:
# Work around assertion for response
assert str(cc.change_contrast(lowerCAmelCase_, 110 ) ).startswith(
'<PIL.Image.Image image mode=RGB size=100x100 at' )
def a_ ( ):
__lowerCAmelCase = canny.gen_gaussian_kernel(9, sigma=1.4 )
# Assert ambiguous array
assert resp.all()
def a_ ( ):
__lowerCAmelCase = imread('digital_image_processing/image_data/lena_small.jpg', 0 )
# assert ambiguous array for all == True
assert canny_img.all()
__lowerCAmelCase = canny.canny(lowerCAmelCase_ )
# assert canny array for at least one True
assert canny_array.any()
def a_ ( ):
assert gg.gaussian_filter(lowerCAmelCase_, 5, sigma=0.9 ).all()
def a_ ( ):
# laplace diagonals
__lowerCAmelCase = array([[0.25, 0.5, 0.25], [0.5, -3, 0.5], [0.25, 0.5, 0.25]] )
__lowerCAmelCase = conv.img_convolve(lowerCAmelCase_, lowerCAmelCase_ ).astype(lowerCAmelCase_ )
assert res.any()
def a_ ( ):
assert med.median_filter(lowerCAmelCase_, 3 ).any()
def a_ ( ):
__lowerCAmelCase , __lowerCAmelCase = sob.sobel_filter(lowerCAmelCase_ )
assert grad.any() and theta.any()
def a_ ( ):
__lowerCAmelCase = sp.make_sepia(lowerCAmelCase_, 20 )
assert sepia.all()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg" ):
__lowerCAmelCase = bs.Burkes(imread(lowerCAmelCase_, 1 ), 120 )
burkes.process()
assert burkes.output_img.any()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg", ):
__lowerCAmelCase = rs.NearestNeighbour(imread(lowerCAmelCase_, 1 ), 400, 200 )
nn.process()
assert nn.output.any()
def a_ ( ):
__lowerCAmelCase = 'digital_image_processing/image_data/lena.jpg'
# Reading the image and converting it to grayscale.
__lowerCAmelCase = imread(lowerCAmelCase_, 0 )
# Test for get_neighbors_pixel function() return not None
__lowerCAmelCase = 0
__lowerCAmelCase = 0
__lowerCAmelCase = image[x_coordinate][y_coordinate]
__lowerCAmelCase = lbp.get_neighbors_pixel(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert neighbors_pixels is not None
# Test for local_binary_pattern function()
# Create a numpy array as the same height and width of read image
__lowerCAmelCase = np.zeros((image.shape[0], image.shape[1]) )
# Iterating through the image and calculating the local binary pattern value
# for each pixel.
for i in range(0, image.shape[0] ):
for j in range(0, image.shape[1] ):
__lowerCAmelCase = lbp.local_binary_value(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert lbp_image.any()
| 53 | 1 |
import logging
import numpy as np
import pytest
from scipy.linalg import eigh
logging.basicConfig(level=logging.INFO, format='%(message)s')
def a_ ( lowerCAmelCase_ : np.ndarray ):
return input_array.reshape((input_array.size, 1) )
def a_ ( lowerCAmelCase_ : np.ndarray, lowerCAmelCase_ : np.ndarray, lowerCAmelCase_ : int ):
__lowerCAmelCase = np.nan
for i in range(lowerCAmelCase_ ):
__lowerCAmelCase = features[:, labels == i]
__lowerCAmelCase = data.mean(1 )
# Centralize the data of class i
__lowerCAmelCase = data - column_reshape(lowerCAmelCase_ )
if i > 0:
# If covariance_sum is not None
covariance_sum += np.dot(lowerCAmelCase_, centered_data.T )
else:
# If covariance_sum is np.nan (i.e. first loop)
__lowerCAmelCase = np.dot(lowerCAmelCase_, centered_data.T )
return covariance_sum / features.shape[1]
def a_ ( lowerCAmelCase_ : np.ndarray, lowerCAmelCase_ : np.ndarray, lowerCAmelCase_ : int ):
__lowerCAmelCase = features.mean(1 )
__lowerCAmelCase = np.nan
for i in range(lowerCAmelCase_ ):
__lowerCAmelCase = features[:, labels == i]
__lowerCAmelCase = data.shape[1]
__lowerCAmelCase = data.mean(1 )
if i > 0:
# If covariance_sum is not None
covariance_sum += device_data * np.dot(
column_reshape(lowerCAmelCase_ ) - column_reshape(lowerCAmelCase_ ), (column_reshape(lowerCAmelCase_ ) - column_reshape(lowerCAmelCase_ )).T, )
else:
# If covariance_sum is np.nan (i.e. first loop)
__lowerCAmelCase = device_data * np.dot(
column_reshape(lowerCAmelCase_ ) - column_reshape(lowerCAmelCase_ ), (column_reshape(lowerCAmelCase_ ) - column_reshape(lowerCAmelCase_ )).T, )
return covariance_sum / features.shape[1]
def a_ ( lowerCAmelCase_ : np.ndarray, lowerCAmelCase_ : int ):
# Check if the features have been loaded
if features.any():
__lowerCAmelCase = features.mean(1 )
# Center the dataset
__lowerCAmelCase = features - np.reshape(lowerCAmelCase_, (data_mean.size, 1) )
__lowerCAmelCase = np.dot(lowerCAmelCase_, centered_data.T ) / features.shape[1]
__lowerCAmelCase , __lowerCAmelCase = np.linalg.eigh(lowerCAmelCase_ )
# Take all the columns in the reverse order (-1), and then takes only the first
__lowerCAmelCase = eigenvectors[:, ::-1][:, 0:dimensions]
# Project the database on the new space
__lowerCAmelCase = np.dot(filtered_eigenvectors.T, lowerCAmelCase_ )
logging.info('Principal Component Analysis computed' )
return projected_data
else:
logging.basicConfig(level=logging.ERROR, format='%(message)s', force=lowerCAmelCase_ )
logging.error('Dataset empty' )
raise AssertionError
def a_ ( lowerCAmelCase_ : np.ndarray, lowerCAmelCase_ : np.ndarray, lowerCAmelCase_ : int, lowerCAmelCase_ : int ):
assert classes > dimensions
# Check if features have been already loaded
if features.any:
__lowerCAmelCase , __lowerCAmelCase = eigh(
covariance_between_classes(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ ), covariance_within_classes(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ ), )
__lowerCAmelCase = eigenvectors[:, ::-1][:, :dimensions]
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = np.linalg.svd(lowerCAmelCase_ )
__lowerCAmelCase = svd_matrix[:, 0:dimensions]
__lowerCAmelCase = np.dot(filtered_svd_matrix.T, lowerCAmelCase_ )
logging.info('Linear Discriminant Analysis computed' )
return projected_data
else:
logging.basicConfig(level=logging.ERROR, format='%(message)s', force=lowerCAmelCase_ )
logging.error('Dataset empty' )
raise AssertionError
def a_ ( ):
# Create dummy dataset with 2 classes and 3 features
__lowerCAmelCase = np.array([[1, 2, 3, 4, 5], [2, 3, 4, 5, 6], [3, 4, 5, 6, 7]] )
__lowerCAmelCase = np.array([0, 0, 0, 1, 1] )
__lowerCAmelCase = 2
__lowerCAmelCase = 2
# Assert that the function raises an AssertionError if dimensions > classes
with pytest.raises(lowerCAmelCase_ ) as error_info:
__lowerCAmelCase = linear_discriminant_analysis(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
if isinstance(lowerCAmelCase_, np.ndarray ):
raise AssertionError(
'Did not raise AssertionError for dimensions > classes' )
assert error_info.type is AssertionError
def a_ ( ):
__lowerCAmelCase = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]] )
__lowerCAmelCase = 2
__lowerCAmelCase = np.array([[6.9282_0323, 8.6602_5404, 10.3923_0485], [3.0, 3.0, 3.0]] )
with pytest.raises(lowerCAmelCase_ ) as error_info:
__lowerCAmelCase = principal_component_analysis(lowerCAmelCase_, lowerCAmelCase_ )
if not np.allclose(lowerCAmelCase_, lowerCAmelCase_ ):
raise AssertionError
assert error_info.type is AssertionError
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 |
from typing import Dict, Iterable, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_DEFAULT_MEAN,
IMAGENET_DEFAULT_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, logging
_snake_case : List[Any] = logging.get_logger(__name__)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = ["""pixel_values"""]
def __init__( self : Optional[int] , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Union[int, float] = 1 / 2_5_5 , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_MEAN , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_STD , **lowerCAmelCase_ : Any , ) -> None:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = size if size is not None else {'shortest_edge': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else {'height': 2_2_4, 'width': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = do_resize
__lowerCAmelCase = size
__lowerCAmelCase = resample
__lowerCAmelCase = do_center_crop
__lowerCAmelCase = crop_size
__lowerCAmelCase = do_rescale
__lowerCAmelCase = rescale_factor
__lowerCAmelCase = do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else IMAGENET_DEFAULT_MEAN
__lowerCAmelCase = image_std if image_std is not None else IMAGENET_DEFAULT_STD
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : Optional[int] , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
# size_dict is a dict with either keys "height" and "width" or "shortest_edge"
if "shortest_edge" in size:
__lowerCAmelCase = int((2_5_6 / 2_2_4) * size['shortest_edge'] )
__lowerCAmelCase = get_resize_output_image_size(lowerCAmelCase_ , size=lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = {'height': output_size[0], 'width': output_size[1]}
if "height" not in size_dict or "width" not in size_dict:
raise ValueError(
f"""Size dict must have keys 'height' and 'width' or 'shortest_edge'. Got {size_dict.keys()}""" )
return resize(
lowerCAmelCase_ , size=(size_dict['height'], size_dict['width']) , resample=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : str , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : str , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ )
if "height" not in size or "width" not in size:
raise ValueError(f"""Size dict must have keys 'height' and 'width'. Got {size.keys()}""" )
return center_crop(lowerCAmelCase_ , size=(size['height'], size['width']) , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[int, float] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : int , ) -> np.ndarray:
return rescale(lowerCAmelCase_ , scale=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : int , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : List[str] , ) -> np.ndarray:
return normalize(lowerCAmelCase_ , mean=lowerCAmelCase_ , std=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : ImageInput , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : PILImageResampling = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[float] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[TensorType] = None , lowerCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **lowerCAmelCase_ : str , ) -> BatchFeature:
__lowerCAmelCase = do_resize if do_resize is not None else self.do_resize
__lowerCAmelCase = resample if resample is not None else self.resample
__lowerCAmelCase = do_center_crop if do_center_crop is not None else self.do_center_crop
__lowerCAmelCase = do_rescale if do_rescale is not None else self.do_rescale
__lowerCAmelCase = rescale_factor if rescale_factor is not None else self.rescale_factor
__lowerCAmelCase = do_normalize if do_normalize is not None else self.do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else self.image_mean
__lowerCAmelCase = image_std if image_std is not None else self.image_std
__lowerCAmelCase = size if size is not None else self.size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else self.crop_size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = make_list_of_images(lowerCAmelCase_ )
if not valid_images(lowerCAmelCase_ ):
raise ValueError(
'Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '
'torch.Tensor, tf.Tensor or jax.ndarray.' )
if do_resize and size is None:
raise ValueError('Size must be specified if do_resize is True.' )
if do_center_crop and crop_size is None:
raise ValueError('Crop size must be specified if do_center_crop is True.' )
if do_rescale and rescale_factor is None:
raise ValueError('Rescale factor must be specified if do_rescale is True.' )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError('Image mean and std must be specified if do_normalize is True.' )
# All transformations expect numpy arrays.
__lowerCAmelCase = [to_numpy_array(lowerCAmelCase_ ) for image in images]
if do_resize:
__lowerCAmelCase = [self.resize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_center_crop:
__lowerCAmelCase = [self.center_crop(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_rescale:
__lowerCAmelCase = [self.rescale(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_normalize:
__lowerCAmelCase = [self.normalize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = [to_channel_dimension_format(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = {'pixel_values': images}
return BatchFeature(data=lowerCAmelCase_ , tensor_type=lowerCAmelCase_ )
| 53 | 1 |
import pytest
import datasets
# Import fixture modules as plugins
_snake_case : List[str] = ['tests.fixtures.files', 'tests.fixtures.hub', 'tests.fixtures.fsspec']
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : List[str] ):
# Mark tests as "unit" by default if not marked as "integration" (or already marked as "unit")
for item in items:
if any(marker in item.keywords for marker in ['integration', 'unit'] ):
continue
item.add_marker(pytest.mark.unit )
def a_ ( lowerCAmelCase_ : int ):
config.addinivalue_line('markers', 'torchaudio_latest: mark test to run with torchaudio>=0.12' )
@pytest.fixture(autouse=lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : Dict ):
# test_hf_cache_home = tmp_path_factory.mktemp("cache") # TODO: why a cache dir per test function does not work?
__lowerCAmelCase = tmp_path_factory.getbasetemp() / 'cache'
__lowerCAmelCase = test_hf_cache_home / 'datasets'
__lowerCAmelCase = test_hf_cache_home / 'metrics'
__lowerCAmelCase = test_hf_cache_home / 'modules'
monkeypatch.setattr('datasets.config.HF_DATASETS_CACHE', str(lowerCAmelCase_ ) )
monkeypatch.setattr('datasets.config.HF_METRICS_CACHE', str(lowerCAmelCase_ ) )
monkeypatch.setattr('datasets.config.HF_MODULES_CACHE', str(lowerCAmelCase_ ) )
__lowerCAmelCase = test_hf_datasets_cache / 'downloads'
monkeypatch.setattr('datasets.config.DOWNLOADED_DATASETS_PATH', str(lowerCAmelCase_ ) )
__lowerCAmelCase = test_hf_datasets_cache / 'downloads' / 'extracted'
monkeypatch.setattr('datasets.config.EXTRACTED_DATASETS_PATH', str(lowerCAmelCase_ ) )
@pytest.fixture(autouse=lowerCAmelCase_, scope='session' )
def a_ ( ):
datasets.disable_progress_bar()
@pytest.fixture(autouse=lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : str ):
# don't take tests into account when counting downloads
monkeypatch.setattr('datasets.config.HF_UPDATE_DOWNLOAD_COUNTS', lowerCAmelCase_ )
@pytest.fixture
def a_ ( lowerCAmelCase_ : List[str] ):
# Required to suppress RemovedIn20Warning when feature(s) are not compatible with SQLAlchemy 2.0
# To be removed once SQLAlchemy 2.0 supported
monkeypatch.setattr('sqlalchemy.util.deprecations.SILENCE_UBER_WARNING', lowerCAmelCase_ )
| 53 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Dict=2 , lowerCAmelCase_ : Optional[int]=8 , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[Any]=9_9 , lowerCAmelCase_ : List[Any]=1_6 , lowerCAmelCase_ : int=5 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : str=3_6 , lowerCAmelCase_ : Optional[int]="gelu" , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : Optional[int]=0.0 , lowerCAmelCase_ : str=5_1_2 , lowerCAmelCase_ : List[str]=1_6 , lowerCAmelCase_ : str=2 , lowerCAmelCase_ : Tuple=0.02 , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : List[Any]=4 , lowerCAmelCase_ : List[str]=None , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_input_mask
__lowerCAmelCase = use_token_type_ids
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = type_vocab_size
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = num_labels
__lowerCAmelCase = num_choices
__lowerCAmelCase = scope
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = None
if self.use_input_mask:
__lowerCAmelCase = random_attention_mask([self.batch_size, self.seq_length] )
__lowerCAmelCase = None
if self.use_token_type_ids:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
__lowerCAmelCase = ids_tensor([self.batch_size] , self.num_choices )
__lowerCAmelCase = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowercase ( self : Any ) -> Union[str, Any]:
return MraConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , )
def lowercase ( self : Dict ) -> List[Any]:
__lowerCAmelCase = self.get_config()
__lowerCAmelCase = 3_0_0
return config
def lowercase ( self : Optional[int] ) -> Union[str, Any]:
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = self.prepare_config_and_inputs()
__lowerCAmelCase = True
__lowerCAmelCase = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
encoder_hidden_states,
encoder_attention_mask,
)
def lowercase ( self : Optional[int] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , ) -> Tuple:
__lowerCAmelCase = True
__lowerCAmelCase = MraModel(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , )
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , )
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Dict ) -> str:
__lowerCAmelCase = MraForQuestionAnswering(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , start_positions=lowerCAmelCase_ , end_positions=lowerCAmelCase_ , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowercase ( self : int , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : int , lowerCAmelCase_ : Dict ) -> Any:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForTokenClassification(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowercase ( self : int , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] ) -> List[Any]:
__lowerCAmelCase = self.num_choices
__lowerCAmelCase = MraForMultipleChoice(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowercase ( self : Tuple ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'input_ids': input_ids, 'token_type_ids': token_type_ids, 'attention_mask': input_mask}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
a_ = False
a_ = False
a_ = False
a_ = False
a_ = ()
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = MraModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : Tuple ) -> List[str]:
self.config_tester.run_common_tests()
def lowercase ( self : Optional[int] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
__lowerCAmelCase = type
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Any ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*lowerCAmelCase_ )
def lowercase ( self : List[str] ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*lowerCAmelCase_ )
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = MraModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
@unittest.skip(reason='MRA does not output attentions' )
def lowercase ( self : Optional[int] ) -> Tuple:
return
@require_torch
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@slow
def lowercase ( self : Optional[Any] ) -> List[str]:
__lowerCAmelCase = MraModel.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = torch.Size((1, 2_5_6, 7_6_8) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[-0.01_40, 0.08_30, -0.03_81], [0.15_46, 0.14_02, 0.02_20], [0.11_62, 0.08_51, 0.01_65]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : int ) -> Optional[int]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 2_5_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[9.25_95, -3.60_38, 11.88_19], [9.38_69, -3.26_93, 11.09_56], [11.85_24, -3.49_38, 13.12_10]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : Any ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-4096-8-d3' )
__lowerCAmelCase = torch.arange(4_0_9_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 4_0_9_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[5.47_89, -2.35_64, 7.50_64], [7.90_67, -1.33_69, 9.96_68], [9.07_12, -1.81_06, 7.03_80]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import re
def a_ ( lowerCAmelCase_ : str ):
return [char.split() for char in re.split(R'[^ a-z A-Z 0-9 \s]', str_ )]
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = split_input(str_ )
return "".join(
[''.join([char.capitalize() for char in sub_str] ) for sub_str in string_split] )
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : bool, lowerCAmelCase_ : str ):
try:
__lowerCAmelCase = split_input(lowerCAmelCase_ )
if upper:
__lowerCAmelCase = ''.join(
[
separator.join([char.upper() for char in sub_str] )
for sub_str in string_split
] )
else:
__lowerCAmelCase = ''.join(
[
separator.join([char.lower() for char in sub_str] )
for sub_str in string_split
] )
return res_str
except IndexError:
return "not valid string"
def a_ ( lowerCAmelCase_ : str ):
return to_simple_case(lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : str ):
try:
__lowerCAmelCase = to_simple_case(lowerCAmelCase_ )
return res_str[0].lower() + res_str[1:]
except IndexError:
return "not valid string"
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : bool ):
return to_complex_case(lowerCAmelCase_, lowerCAmelCase_, '_' )
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : bool ):
return to_complex_case(lowerCAmelCase_, lowerCAmelCase_, '-' )
if __name__ == "__main__":
__import__('doctest').testmod()
| 53 |
import argparse
import json
import os
import re
import shutil
import torch
from transformers import BioGptConfig, BioGptForCausalLM
from transformers.models.biogpt.tokenization_biogpt import VOCAB_FILES_NAMES
from transformers.tokenization_utils_base import TOKENIZER_CONFIG_FILE
from transformers.utils import WEIGHTS_NAME, logging
logging.set_verbosity_warning()
_snake_case : Union[str, Any] = 2
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , *, # begin keyword-only arguments
lowerCAmelCase_ : str="<s>" , lowerCAmelCase_ : Dict="<pad>" , lowerCAmelCase_ : Any="</s>" , lowerCAmelCase_ : List[str]="<unk>" , lowerCAmelCase_ : Optional[Any]=None , ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = bos, unk, pad, eos
__lowerCAmelCase = []
__lowerCAmelCase = []
__lowerCAmelCase = {}
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
if extra_special_symbols:
for s in extra_special_symbols:
self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = len(self.symbols )
def __eq__( self : Dict , lowerCAmelCase_ : Dict ) -> str:
return self.indices == other.indices
def __getitem__( self : List[Any] , lowerCAmelCase_ : int ) -> Union[str, Any]:
if idx < len(self.symbols ):
return self.symbols[idx]
return self.unk_word
def __len__( self : Tuple ) -> List[Any]:
return len(self.symbols )
def __contains__( self : Optional[Any] , lowerCAmelCase_ : Dict ) -> Optional[int]:
return sym in self.indices
@classmethod
def lowercase ( cls : Dict , lowerCAmelCase_ : str ) -> str:
__lowerCAmelCase = cls()
d.add_from_file(lowerCAmelCase_ )
return d
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int=1 , lowerCAmelCase_ : Any=False ) -> Optional[Any]:
if word in self.indices and not overwrite:
__lowerCAmelCase = self.indices[word]
__lowerCAmelCase = self.count[idx] + n
return idx
else:
__lowerCAmelCase = len(self.symbols )
__lowerCAmelCase = idx
self.symbols.append(lowerCAmelCase_ )
self.count.append(lowerCAmelCase_ )
return idx
def lowercase ( self : str , lowerCAmelCase_ : Union[str, Any] ) -> Dict:
return 0
def lowercase ( self : Tuple , lowerCAmelCase_ : Union[str, Any] ) -> int:
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
try:
with open(lowerCAmelCase_ , 'r' , encoding='utf-8' ) as fd:
self.add_from_file(lowerCAmelCase_ )
except FileNotFoundError as fnfe:
raise fnfe
except UnicodeError:
raise Exception('Incorrect encoding detected in {}, please rebuild the dataset'.format(lowerCAmelCase_ ) )
return
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = self._load_meta(lowerCAmelCase_ )
for line in lines[indices_start_line:]:
try:
__lowerCAmelCase , __lowerCAmelCase = line.rstrip().rsplit(' ' , 1 )
if field == "#fairseq:overwrite":
__lowerCAmelCase = True
__lowerCAmelCase , __lowerCAmelCase = line.rsplit(' ' , 1 )
else:
__lowerCAmelCase = False
__lowerCAmelCase = int(lowerCAmelCase_ )
__lowerCAmelCase = line
if word in self and not overwrite:
raise RuntimeError(
'Duplicate word found when loading Dictionary: \'{}\'. '
'Duplicate words can overwrite earlier ones by adding the '
'#fairseq:overwrite flag at the end of the corresponding row '
'in the dictionary file. If using the Camembert model, please '
'download an updated copy of the model file.'.format(lowerCAmelCase_ ) )
self.add_symbol(lowerCAmelCase_ , n=lowerCAmelCase_ , overwrite=lowerCAmelCase_ )
except ValueError:
raise ValueError('Incorrect dictionary format, expected \'<token> <cnt> [flags]\'' )
def a_ ( lowerCAmelCase_ : List[str] ):
# (1) remove word breaking symbol, (2) add word ending symbol where the word is not broken up,
# e.g.: d = {'le@@': 5, 'tt@@': 6, 'er': 7} => {'le': 5, 'tt': 6, 'er</w>': 7}
__lowerCAmelCase = dict((re.sub(R'@@$', '', lowerCAmelCase_ ), v) if k.endswith('@@' ) else (re.sub(R'$', '</w>', lowerCAmelCase_ ), v) for k, v in d.items() )
__lowerCAmelCase = '<s> <pad> </s> <unk>'.split()
# restore the special tokens
for k in keep_keys:
del da[F"""{k}</w>"""]
__lowerCAmelCase = d[k] # restore
return da
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str] ):
# prep
if not os.path.exists(lowerCAmelCase_ ):
raise ValueError(F"""path {biogpt_checkpoint_path} does not exist!""" )
os.makedirs(lowerCAmelCase_, exist_ok=lowerCAmelCase_ )
print(F"""Writing results to {pytorch_dump_folder_path}""" )
# handle various types of models
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'checkpoint.pt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {checkpoint_file} does not exist!""" )
__lowerCAmelCase = torch.load(lowerCAmelCase_, map_location='cpu' )
__lowerCAmelCase = chkpt['cfg']['model']
# dicts
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'dict.txt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {dict_file} does not exist!""" )
__lowerCAmelCase = Dictionary.load(lowerCAmelCase_ )
__lowerCAmelCase = rewrite_dict_keys(src_dict.indices )
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['vocab_file'] )
print(F"""Generating {src_vocab_file} of {src_vocab_size} records""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# merges_file (bpecodes)
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'bpecodes' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {bpecodes_file} does not exist!""" )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['merges_file'] )
shutil.copyfile(lowerCAmelCase_, lowerCAmelCase_ )
# model config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'config.json' )
__lowerCAmelCase = {
'activation_dropout': args['activation_dropout'],
'architectures': ['BioGptForCausalLM'],
'attention_probs_dropout_prob': args['attention_dropout'],
'bos_token_id': 0,
'eos_token_id': 2,
'hidden_act': args['activation_fn'],
'hidden_dropout_prob': args['dropout'],
'hidden_size': args['decoder_embed_dim'],
'initializer_range': 0.02,
'intermediate_size': args['decoder_ffn_embed_dim'],
'layer_norm_eps': 1E-12,
'layerdrop': args['decoder_layerdrop'],
'max_position_embeddings': args['max_target_positions'],
'model_type': 'biogpt',
'num_attention_heads': args['decoder_attention_heads'],
'num_hidden_layers': args['decoder_layers'],
'pad_token_id': 1,
'scale_embedding': not args['no_scale_embedding'],
'tie_word_embeddings': args['share_decoder_input_output_embed'],
'vocab_size': src_vocab_size,
}
# good hparam defaults to start with
print(F"""Generating {biogpt_model_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# tokenizer config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = {
'bos_token': '<s>',
'eos_token': '</s>',
'model_max_length': 1024,
'pad_token': '<pad>',
'special_tokens_map_file': None,
'tokenizer_class': 'BioGptTokenizer',
'unk_token': '<unk>',
}
print(F"""Generating {biogpt_tokenizer_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# model
__lowerCAmelCase = chkpt['model']
# remove unneeded keys
__lowerCAmelCase = [
'decoder.version',
]
for k in ignore_keys:
model_state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = list(model_state_dict.keys() )
for layer_name in layer_names:
if layer_name.endswith('output_projection.weight' ):
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
else:
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
__lowerCAmelCase = BioGptConfig.from_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = BioGptForCausalLM(lowerCAmelCase_ )
# check that it loads ok
model_new.load_state_dict(lowerCAmelCase_ )
# save
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
print(F"""Generating {pytorch_weights_dump_path}""" )
torch.save(lowerCAmelCase_, lowerCAmelCase_ )
print('Conversion is done!' )
if __name__ == "__main__":
_snake_case : str = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--biogpt_checkpoint_path',
default=None,
type=str,
required=True,
help=(
'Path to the official PyTorch checkpoint file which is expected to reside in the dump dir with dicts,'
' bpecodes, etc.'
),
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_biogpt_checkpoint_to_pytorch(args.biogpt_checkpoint_path, args.pytorch_dump_folder_path)
| 53 | 1 |
import gc
import tempfile
import unittest
import numpy as np
import torch
from diffusers import VersatileDiffusionTextToImagePipeline
from diffusers.utils.testing_utils import nightly, require_torch_gpu, torch_device
_snake_case : Union[str, Any] = False
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
pass
@nightly
@require_torch_gpu
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Optional[int] ) -> Optional[int]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def lowercase ( self : Optional[int] ) -> List[str]:
__lowerCAmelCase = VersatileDiffusionTextToImagePipeline.from_pretrained('shi-labs/versatile-diffusion' )
# remove text_unet
pipe.remove_unused_weights()
pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = 'A painting of a squirrel eating a burger '
__lowerCAmelCase = torch.manual_seed(0 )
__lowerCAmelCase = pipe(
prompt=lowerCAmelCase_ , generator=lowerCAmelCase_ , guidance_scale=7.5 , num_inference_steps=2 , output_type='numpy' ).images
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = VersatileDiffusionTextToImagePipeline.from_pretrained(lowerCAmelCase_ )
pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = generator.manual_seed(0 )
__lowerCAmelCase = pipe(
prompt=lowerCAmelCase_ , generator=lowerCAmelCase_ , guidance_scale=7.5 , num_inference_steps=2 , output_type='numpy' ).images
assert np.abs(image - new_image ).sum() < 1e-5, "Models don't have the same forward pass"
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = VersatileDiffusionTextToImagePipeline.from_pretrained(
'shi-labs/versatile-diffusion' , torch_dtype=torch.floataa )
pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = 'A painting of a squirrel eating a burger '
__lowerCAmelCase = torch.manual_seed(0 )
__lowerCAmelCase = pipe(
prompt=lowerCAmelCase_ , generator=lowerCAmelCase_ , guidance_scale=7.5 , num_inference_steps=5_0 , output_type='numpy' ).images
__lowerCAmelCase = image[0, 2_5_3:2_5_6, 2_5_3:2_5_6, -1]
assert image.shape == (1, 5_1_2, 5_1_2, 3)
__lowerCAmelCase = np.array([0.33_67, 0.31_69, 0.26_56, 0.38_70, 0.47_90, 0.37_96, 0.40_09, 0.48_78, 0.47_78] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
| 53 |
from typing import Tuple, Union
from ...modeling_outputs import BackboneOutput
from ...modeling_utils import PreTrainedModel
from ...utils import is_timm_available, is_torch_available, requires_backends
from ...utils.backbone_utils import BackboneMixin
from .configuration_timm_backbone import TimmBackboneConfig
if is_timm_available():
import timm
if is_torch_available():
from torch import Tensor
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = """pixel_values"""
a_ = False
a_ = TimmBackboneConfig
def __init__( self : Tuple , lowerCAmelCase_ : Any , **lowerCAmelCase_ : Optional[int] ) -> Optional[Any]:
requires_backends(self , 'timm' )
super().__init__(lowerCAmelCase_ )
__lowerCAmelCase = config
if config.backbone is None:
raise ValueError('backbone is not set in the config. Please set it to a timm model name.' )
if config.backbone not in timm.list_models():
raise ValueError(f"""backbone {config.backbone} is not supported by timm.""" )
if hasattr(lowerCAmelCase_ , 'out_features' ) and config.out_features is not None:
raise ValueError('out_features is not supported by TimmBackbone. Please use out_indices instead.' )
__lowerCAmelCase = getattr(lowerCAmelCase_ , 'use_pretrained_backbone' , lowerCAmelCase_ )
if pretrained is None:
raise ValueError('use_pretrained_backbone is not set in the config. Please set it to True or False.' )
# We just take the final layer by default. This matches the default for the transformers models.
__lowerCAmelCase = config.out_indices if getattr(lowerCAmelCase_ , 'out_indices' , lowerCAmelCase_ ) is not None else (-1,)
__lowerCAmelCase = timm.create_model(
config.backbone , pretrained=lowerCAmelCase_ , features_only=config.features_only , in_chans=config.num_channels , out_indices=lowerCAmelCase_ , **lowerCAmelCase_ , )
# These are used to control the output of the model when called. If output_hidden_states is True, then
# return_layers is modified to include all layers.
__lowerCAmelCase = self._backbone.return_layers
__lowerCAmelCase = {layer['module']: str(lowerCAmelCase_ ) for i, layer in enumerate(self._backbone.feature_info.info )}
super()._init_backbone(lowerCAmelCase_ )
@classmethod
def lowercase ( cls : int , lowerCAmelCase_ : Dict , *lowerCAmelCase_ : Dict , **lowerCAmelCase_ : Union[str, Any] ) -> Optional[int]:
requires_backends(cls , ['vision', 'timm'] )
from ...models.timm_backbone import TimmBackboneConfig
__lowerCAmelCase = kwargs.pop('config' , TimmBackboneConfig() )
__lowerCAmelCase = kwargs.pop('use_timm_backbone' , lowerCAmelCase_ )
if not use_timm:
raise ValueError('use_timm_backbone must be True for timm backbones' )
__lowerCAmelCase = kwargs.pop('num_channels' , config.num_channels )
__lowerCAmelCase = kwargs.pop('features_only' , config.features_only )
__lowerCAmelCase = kwargs.pop('use_pretrained_backbone' , config.use_pretrained_backbone )
__lowerCAmelCase = kwargs.pop('out_indices' , config.out_indices )
__lowerCAmelCase = TimmBackboneConfig(
backbone=lowerCAmelCase_ , num_channels=lowerCAmelCase_ , features_only=lowerCAmelCase_ , use_pretrained_backbone=lowerCAmelCase_ , out_indices=lowerCAmelCase_ , )
return super()._from_config(lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Tuple , lowerCAmelCase_ : int ) -> Dict:
pass
def lowercase ( self : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : Any=None , **lowerCAmelCase_ : Dict ) -> Union[BackboneOutput, Tuple[Tensor, ...]]:
__lowerCAmelCase = return_dict if return_dict is not None else self.config.use_return_dict
__lowerCAmelCase = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__lowerCAmelCase = output_attentions if output_attentions is not None else self.config.output_attentions
if output_attentions:
raise ValueError('Cannot output attentions for timm backbones at the moment' )
if output_hidden_states:
# We modify the return layers to include all the stages of the backbone
__lowerCAmelCase = self._all_layers
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = self._return_layers
__lowerCAmelCase = tuple(hidden_states[i] for i in self.out_indices )
else:
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = tuple(lowerCAmelCase_ )
__lowerCAmelCase = tuple(lowerCAmelCase_ ) if hidden_states is not None else None
if not return_dict:
__lowerCAmelCase = (feature_maps,)
if output_hidden_states:
__lowerCAmelCase = output + (hidden_states,)
return output
return BackboneOutput(feature_maps=lowerCAmelCase_ , hidden_states=lowerCAmelCase_ , attentions=lowerCAmelCase_ )
| 53 | 1 |
def a_ ( lowerCAmelCase_ : list, lowerCAmelCase_ : int = 0 ):
__lowerCAmelCase = length or len(lowerCAmelCase_ )
__lowerCAmelCase = False
for i in range(length - 1 ):
if list_data[i] > list_data[i + 1]:
__lowerCAmelCase , __lowerCAmelCase = list_data[i + 1], list_data[i]
__lowerCAmelCase = True
return list_data if not swapped else bubble_sort(lowerCAmelCase_, length - 1 )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 |
from __future__ import annotations
def a_ ( lowerCAmelCase_ : list[float] ):
if len(lowerCAmelCase_ ) < 2:
raise ValueError('Monogons and Digons are not polygons in the Euclidean space' )
if any(i <= 0 for i in nums ):
raise ValueError('All values must be greater than 0' )
__lowerCAmelCase = nums.copy()
copy_nums.sort()
return copy_nums[-1] < sum(copy_nums[:-1] )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 | 1 |
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_sentencepiece_available
_snake_case : List[str] = {}
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : int = ['BartphoTokenizer']
if TYPE_CHECKING:
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_bartpho import BartphoTokenizer
else:
import sys
_snake_case : Tuple = _LazyModule(__name__, globals()['__file__'], _import_structure, module_spec=__spec__)
| 53 |
import inspect
import unittest
from transformers import RegNetConfig, is_flax_available
from transformers.testing_utils import require_flax, slow
from transformers.utils import cached_property, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor
if is_flax_available():
import jax
import jax.numpy as jnp
from transformers.models.regnet.modeling_flax_regnet import FlaxRegNetForImageClassification, FlaxRegNetModel
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int]=3 , lowerCAmelCase_ : Dict=3_2 , lowerCAmelCase_ : Tuple=3 , lowerCAmelCase_ : Union[str, Any]=1_0 , lowerCAmelCase_ : List[str]=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Optional[int]=[1, 1, 2, 1] , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : Tuple="relu" , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : Optional[int]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = embeddings_size
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = hidden_act
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = len(lowerCAmelCase_ )
def lowercase ( self : Optional[int] ) -> List[Any]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : Tuple ) -> List[Any]:
return RegNetConfig(
num_channels=self.num_channels , embeddings_size=self.embeddings_size , hidden_sizes=self.hidden_sizes , depths=self.depths , hidden_act=self.hidden_act , num_labels=self.num_labels , image_size=self.image_size , )
def lowercase ( self : List[str] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : List[str] ) -> str:
__lowerCAmelCase = FlaxRegNetModel(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
# Output shape (b, c, h, w)
self.parent.assertEqual(
result.last_hidden_state.shape , (self.batch_size, self.hidden_sizes[-1], self.image_size // 3_2, self.image_size // 3_2) , )
def lowercase ( self : str , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : int ) -> Tuple:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = FlaxRegNetForImageClassification(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_flax
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (FlaxRegNetModel, FlaxRegNetForImageClassification) if is_flax_available() else ()
a_ = False
a_ = False
a_ = False
def lowercase ( self : Dict ) -> None:
__lowerCAmelCase = FlaxRegNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : int ) -> Optional[int]:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : str ) -> Union[str, Any]:
return
def lowercase ( self : Dict ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCAmelCase_ )
@unittest.skip(reason='RegNet does not use inputs_embeds' )
def lowercase ( self : Union[str, Any] ) -> Any:
pass
@unittest.skip(reason='RegNet does not support input and output embeddings' )
def lowercase ( self : Tuple ) -> Tuple:
pass
def lowercase ( self : Optional[Any] ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.__call__ )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
def check_hidden_states_output(lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : str ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
with self.subTest(model_class.__name__ ):
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model_class(lowerCAmelCase_ )
@jax.jit
def model_jitted(lowerCAmelCase_ : Optional[int] , **lowerCAmelCase_ : Dict ):
return model(pixel_values=lowerCAmelCase_ , **lowerCAmelCase_ )
with self.subTest('JIT Enabled' ):
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
with self.subTest('JIT Disabled' ):
with jax.disable_jit():
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
self.assertEqual(len(lowerCAmelCase_ ) , len(lowerCAmelCase_ ) )
for jitted_output, output in zip(lowerCAmelCase_ , lowerCAmelCase_ ):
self.assertEqual(jitted_output.shape , output.shape )
def a_ ( ):
__lowerCAmelCase = Image.open('./tests/fixtures/tests_samples/COCO/000000039769.png' )
return image
@require_flax
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
return AutoImageProcessor.from_pretrained('facebook/regnet-y-040' ) if is_vision_available() else None
@slow
def lowercase ( self : Optional[Any] ) -> Union[str, Any]:
__lowerCAmelCase = FlaxRegNetForImageClassification.from_pretrained('facebook/regnet-y-040' )
__lowerCAmelCase = self.default_image_processor
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = image_processor(images=lowerCAmelCase_ , return_tensors='np' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
# verify the logits
__lowerCAmelCase = (1, 1_0_0_0)
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = jnp.array([-0.41_80, -1.50_51, -3.48_36] )
self.assertTrue(jnp.allclose(outputs.logits[0, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import logging
from transformers import PretrainedConfig
_snake_case : List[str] = logging.getLogger(__name__)
_snake_case : Union[str, Any] = {
'bertabs-finetuned-cnndm': 'https://huggingface.co/remi/bertabs-finetuned-cnndm-extractive-abstractive-summarization/resolve/main/config.json',
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = """bertabs"""
def __init__( self : List[str] , lowerCAmelCase_ : List[Any]=3_0_5_2_2 , lowerCAmelCase_ : List[Any]=5_1_2 , lowerCAmelCase_ : str=6 , lowerCAmelCase_ : Union[str, Any]=5_1_2 , lowerCAmelCase_ : int=8 , lowerCAmelCase_ : List[str]=5_1_2 , lowerCAmelCase_ : Tuple=0.2 , lowerCAmelCase_ : Dict=6 , lowerCAmelCase_ : Dict=7_6_8 , lowerCAmelCase_ : Tuple=8 , lowerCAmelCase_ : Optional[Any]=2_0_4_8 , lowerCAmelCase_ : int=0.2 , **lowerCAmelCase_ : Optional[Any] , ) -> Optional[Any]:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = vocab_size
__lowerCAmelCase = max_pos
__lowerCAmelCase = enc_layers
__lowerCAmelCase = enc_hidden_size
__lowerCAmelCase = enc_heads
__lowerCAmelCase = enc_ff_size
__lowerCAmelCase = enc_dropout
__lowerCAmelCase = dec_layers
__lowerCAmelCase = dec_hidden_size
__lowerCAmelCase = dec_heads
__lowerCAmelCase = dec_ff_size
__lowerCAmelCase = dec_dropout
| 53 |
import json
import logging
import math
import os
import sys
from dataclasses import dataclass, field
from typing import Optional
from datasets import Dataset, load_dataset
import transformers
from transformers import (
CONFIG_MAPPING,
MODEL_FOR_MASKED_LM_MAPPING,
AutoConfig,
AutoModelForMaskedLM,
AutoTokenizer,
DataCollatorForWholeWordMask,
HfArgumentParser,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
_snake_case : Optional[int] = logging.getLogger(__name__)
_snake_case : Dict = list(MODEL_FOR_MASKED_LM_MAPPING.keys())
_snake_case : List[Any] = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """If training from scratch, pass a model type from the list: """ + """, """.join(_UpperCamelCase )} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Override some existing default config settings when a model is trained from scratch. Example: """
"""n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained config name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained tokenizer name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."""} , )
a_ = field(
default="""main""" , metadata={"""help""": """The specific model version to use (can be a branch name, tag name or commit id)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Will use the token generated when running `huggingface-cli login` (necessary to use this script """
"""with private models)."""
)
} , )
def lowercase ( self : List[Any] ) -> List[Any]:
if self.config_overrides is not None and (self.config_name is not None or self.model_name_or_path is not None):
raise ValueError(
'--config_overrides can\'t be used in combination with --config_name or --model_name_or_path' )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The name of the dataset to use (via the datasets library)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(default=_UpperCamelCase , metadata={"""help""": """The input training data file (a text file)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input evaluation data file to evaluate the perplexity on (a text file)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input train ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input validation ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached training and evaluation sets"""} )
a_ = field(
default=5 , metadata={
"""help""": """The percentage of the train set used as validation set in case there's no validation split"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The maximum total input sequence length after tokenization. Sequences longer """
"""than this will be truncated. Default to the max input length of the model."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=0.15 , metadata={"""help""": """Ratio of tokens to mask for masked language modeling loss"""} )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Whether to pad all samples to `max_seq_length`. """
"""If False, will pad the samples dynamically when batching to the maximum length in the batch."""
)
} , )
def lowercase ( self : int ) -> int:
if self.train_file is not None:
__lowerCAmelCase = self.train_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`train_file` should be a csv, a json or a txt file."
if self.validation_file is not None:
__lowerCAmelCase = self.validation_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`validation_file` should be a csv, a json or a txt file."
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : Union[str, Any] ):
with open(lowerCAmelCase_, 'r', encoding='utf-8' ) as f:
__lowerCAmelCase = [json.loads(lowerCAmelCase_ ) for line in f.read().splitlines() if (len(lowerCAmelCase_ ) > 0 and not line.isspace())]
assert len(lowerCAmelCase_ ) == len(lowerCAmelCase_ )
__lowerCAmelCase = {c: dataset[c] for c in dataset.column_names}
__lowerCAmelCase = refs
return Dataset.from_dict(lowerCAmelCase_ )
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
__lowerCAmelCase = load_dataset(data_args.dataset_name, data_args.dataset_config_name )
if "validation" not in datasets.keys():
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[:{data_args.validation_split_percentage}%]""", )
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[{data_args.validation_split_percentage}%:]""", )
else:
__lowerCAmelCase = {}
if data_args.train_file is not None:
__lowerCAmelCase = data_args.train_file
if data_args.validation_file is not None:
__lowerCAmelCase = data_args.validation_file
__lowerCAmelCase = data_args.train_file.split('.' )[-1]
if extension == "txt":
__lowerCAmelCase = 'text'
__lowerCAmelCase = load_dataset(lowerCAmelCase_, data_files=lowerCAmelCase_ )
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.config_name:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.config_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
__lowerCAmelCase = CONFIG_MAPPING[model_args.model_type]()
logger.warning('You are instantiating a new config instance from scratch.' )
if model_args.config_overrides is not None:
logger.info(F"""Overriding config: {model_args.config_overrides}""" )
config.update_from_string(model_args.config_overrides )
logger.info(F"""New config: {config}""" )
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'use_fast': model_args.use_fast_tokenizer,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.tokenizer_name:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.tokenizer_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
raise ValueError(
'You are instantiating a new tokenizer from scratch. This is not supported by this script.'
'You can do it from another script, save it, and load it from here, using --tokenizer_name.' )
if model_args.model_name_or_path:
__lowerCAmelCase = AutoModelForMaskedLM.from_pretrained(
model_args.model_name_or_path, from_tf=bool('.ckpt' in model_args.model_name_or_path ), config=lowerCAmelCase_, cache_dir=model_args.cache_dir, revision=model_args.model_revision, use_auth_token=True if model_args.use_auth_token else None, )
else:
logger.info('Training new model from scratch' )
__lowerCAmelCase = AutoModelForMaskedLM.from_config(lowerCAmelCase_ )
model.resize_token_embeddings(len(lowerCAmelCase_ ) )
# Preprocessing the datasets.
# First we tokenize all the texts.
if training_args.do_train:
__lowerCAmelCase = datasets['train'].column_names
else:
__lowerCAmelCase = datasets['validation'].column_names
__lowerCAmelCase = 'text' if 'text' in column_names else column_names[0]
__lowerCAmelCase = 'max_length' if data_args.pad_to_max_length else False
def tokenize_function(lowerCAmelCase_ : str ):
# Remove empty lines
__lowerCAmelCase = [line for line in examples['text'] if len(lowerCAmelCase_ ) > 0 and not line.isspace()]
return tokenizer(examples['text'], padding=lowerCAmelCase_, truncation=lowerCAmelCase_, max_length=data_args.max_seq_length )
__lowerCAmelCase = datasets.map(
lowerCAmelCase_, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, remove_columns=[text_column_name], load_from_cache_file=not data_args.overwrite_cache, )
# Add the chinese references if provided
if data_args.train_ref_file is not None:
__lowerCAmelCase = add_chinese_references(tokenized_datasets['train'], data_args.train_ref_file )
if data_args.validation_ref_file is not None:
__lowerCAmelCase = add_chinese_references(
tokenized_datasets['validation'], data_args.validation_ref_file )
# If we have ref files, need to avoid it removed by trainer
__lowerCAmelCase = data_args.train_ref_file or data_args.validation_ref_file
if has_ref:
__lowerCAmelCase = False
# Data collator
# This one will take care of randomly masking the tokens.
__lowerCAmelCase = DataCollatorForWholeWordMask(tokenizer=lowerCAmelCase_, mlm_probability=data_args.mlm_probability )
# Initialize our Trainer
__lowerCAmelCase = Trainer(
model=lowerCAmelCase_, args=lowerCAmelCase_, train_dataset=tokenized_datasets['train'] if training_args.do_train else None, eval_dataset=tokenized_datasets['validation'] if training_args.do_eval else None, tokenizer=lowerCAmelCase_, data_collator=lowerCAmelCase_, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif model_args.model_name_or_path is not None and os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model() # Saves the tokenizer too for easy upload
__lowerCAmelCase = os.path.join(training_args.output_dir, 'train_results.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Train results *****' )
for key, value in sorted(train_result.metrics.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
# Need to save the state, since Trainer.save_model saves only the tokenizer with the model
trainer.state.save_to_json(os.path.join(training_args.output_dir, 'trainer_state.json' ) )
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = math.exp(eval_output['eval_loss'] )
__lowerCAmelCase = perplexity
__lowerCAmelCase = os.path.join(training_args.output_dir, 'eval_results_mlm_wwm.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Eval results *****' )
for key, value in sorted(results.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
return results
def a_ ( lowerCAmelCase_ : Tuple ):
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()
| 53 | 1 |
def a_ ( lowerCAmelCase_ : list[list] ):
__lowerCAmelCase = current_set.copy()
for row_index, row in enumerate(lowerCAmelCase_ ):
__lowerCAmelCase = row[0]
for column_index, column in enumerate(lowerCAmelCase_ ):
if magnitude == 0:
__lowerCAmelCase = column
continue
__lowerCAmelCase = column / magnitude
# Subtract to cancel term
__lowerCAmelCase = current_set[0]
__lowerCAmelCase = [first_row]
__lowerCAmelCase = current_set[1::]
for row in current_set:
__lowerCAmelCase = []
# If first term is 0, it is already in form we want, so we preserve it
if row[0] == 0:
final_set.append(lowerCAmelCase_ )
continue
for column_index in range(len(lowerCAmelCase_ ) ):
temp_row.append(first_row[column_index] - row[column_index] )
final_set.append(lowerCAmelCase_ )
# Create next recursion iteration set
if len(final_set[0] ) != 3:
__lowerCAmelCase = final_set[0]
__lowerCAmelCase = []
__lowerCAmelCase = []
for row in final_set[1::]:
current_first_column.append(row[0] )
next_iteration.append(row[1::] )
__lowerCAmelCase = simplify(lowerCAmelCase_ )
for i in range(len(lowerCAmelCase_ ) ):
resultant[i].insert(0, current_first_column[i] )
resultant.insert(0, lowerCAmelCase_ )
__lowerCAmelCase = resultant
return final_set
def a_ ( lowerCAmelCase_ : list[list] ):
if len(lowerCAmelCase_ ) == 0:
raise IndexError('solve_simultaneous() requires n lists of length n+1' )
__lowerCAmelCase = len(lowerCAmelCase_ ) + 1
if any(len(lowerCAmelCase_ ) != _length for item in equations ):
raise IndexError('solve_simultaneous() requires n lists of length n+1' )
for row in equations:
if any(not isinstance(lowerCAmelCase_, (int, float) ) for column in row ):
raise ValueError('solve_simultaneous() requires lists of integers' )
if len(lowerCAmelCase_ ) == 1:
return [equations[0][-1] / equations[0][0]]
__lowerCAmelCase = equations.copy()
if any(0 in row for row in data_set ):
__lowerCAmelCase = data_set.copy()
__lowerCAmelCase = []
for row_index, row in enumerate(lowerCAmelCase_ ):
if 0 not in row:
__lowerCAmelCase = data_set.pop(lowerCAmelCase_ )
break
if not full_row:
raise ValueError('solve_simultaneous() requires at least 1 full equation' )
data_set.insert(0, lowerCAmelCase_ )
__lowerCAmelCase = data_set.copy()
__lowerCAmelCase = simplify(lowerCAmelCase_ )
__lowerCAmelCase = simplified[::-1]
__lowerCAmelCase = []
for row in simplified:
__lowerCAmelCase = row[-1]
if not solutions:
if row[-2] == 0:
solutions.append(0 )
continue
solutions.append(current_solution / row[-2] )
continue
__lowerCAmelCase = row.copy()[: len(lowerCAmelCase_ ) - 1 :]
while temp_row[0] == 0:
temp_row.pop(0 )
if len(lowerCAmelCase_ ) == 0:
solutions.append(0 )
continue
__lowerCAmelCase = temp_row[1::]
__lowerCAmelCase = temp_row[::-1]
for column_index, column in enumerate(lowerCAmelCase_ ):
current_solution -= column * solutions[column_index]
solutions.append(lowerCAmelCase_ )
__lowerCAmelCase = []
for item in solutions:
final.append(float(round(lowerCAmelCase_, 5 ) ) )
return final[::-1]
if __name__ == "__main__":
import doctest
doctest.testmod()
_snake_case : List[Any] = [
[2, 1, 1, 1, 1, 4],
[1, 2, 1, 1, 1, 5],
[1, 1, 2, 1, 1, 6],
[1, 1, 1, 2, 1, 7],
[1, 1, 1, 1, 2, 8],
]
print(solve_simultaneous(eq))
print(solve_simultaneous([[4, 2]]))
| 53 |
def a_ ( lowerCAmelCase_ : int = 200_0000 ):
__lowerCAmelCase = [0 for i in range(n + 1 )]
__lowerCAmelCase = 1
__lowerCAmelCase = 1
for i in range(2, int(n**0.5 ) + 1 ):
if primality_list[i] == 0:
for j in range(i * i, n + 1, lowerCAmelCase_ ):
__lowerCAmelCase = 1
__lowerCAmelCase = 0
for i in range(lowerCAmelCase_ ):
if primality_list[i] == 0:
sum_of_primes += i
return sum_of_primes
if __name__ == "__main__":
print(F"""{solution() = }""")
| 53 | 1 |
from dataclasses import dataclass, field
from typing import TYPE_CHECKING, Any, ClassVar, Dict, List, Optional, Union
import pyarrow as pa
if TYPE_CHECKING:
from .features import FeatureType
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = 42
a_ = None
# Automatically constructed
a_ = "dict"
a_ = None
a_ = field(default="""Translation""" , init=_UpperCamelCase , repr=_UpperCamelCase )
def __call__( self : int ) -> Dict:
return pa.struct({lang: pa.string() for lang in sorted(self.languages )} )
def lowercase ( self : List[str] ) -> Union["FeatureType", Dict[str, "FeatureType"]]:
from .features import Value
return {k: Value('string' ) for k in sorted(self.languages )}
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = None
a_ = None
a_ = None
# Automatically constructed
a_ = "dict"
a_ = None
a_ = field(default="""TranslationVariableLanguages""" , init=_UpperCamelCase , repr=_UpperCamelCase )
def lowercase ( self : str ) -> Union[str, Any]:
__lowerCAmelCase = sorted(set(self.languages ) ) if self.languages else None
__lowerCAmelCase = len(self.languages ) if self.languages else None
def __call__( self : Union[str, Any] ) -> str:
return pa.struct({'language': pa.list_(pa.string() ), 'translation': pa.list_(pa.string() )} )
def lowercase ( self : Optional[int] , lowerCAmelCase_ : List[Any] ) -> List[str]:
__lowerCAmelCase = set(self.languages )
if self.languages and set(lowerCAmelCase_ ) - lang_set:
raise ValueError(
f"""Some languages in example ({", ".join(sorted(set(lowerCAmelCase_ ) - lang_set ) )}) are not in valid set ({", ".join(lowerCAmelCase_ )}).""" )
# Convert dictionary into tuples, splitting out cases where there are
# multiple translations for a single language.
__lowerCAmelCase = []
for lang, text in translation_dict.items():
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
translation_tuples.append((lang, text) )
else:
translation_tuples.extend([(lang, el) for el in text] )
# Ensure translations are in ascending order by language code.
__lowerCAmelCase , __lowerCAmelCase = zip(*sorted(lowerCAmelCase_ ) )
return {"language": languages, "translation": translations}
def lowercase ( self : int ) -> Union["FeatureType", Dict[str, "FeatureType"]]:
from .features import Sequence, Value
return {
"language": Sequence(Value('string' ) ),
"translation": Sequence(Value('string' ) ),
}
| 53 |
import json
import logging
import os
import sys
from pathlib import Path
import finetune_rag
from transformers.file_utils import is_apex_available
from transformers.testing_utils import (
TestCasePlus,
execute_subprocess_async,
require_ray,
require_torch_gpu,
require_torch_multi_gpu,
)
logging.basicConfig(level=logging.DEBUG)
_snake_case : Tuple = logging.getLogger()
_snake_case : Any = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Any , lowerCAmelCase_ : Dict ) -> Optional[int]:
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = {'source': 'What is love ?', 'target': 'life'}
__lowerCAmelCase = {'train': 1_2, 'val': 2, 'test': 2}
for split in ["train", "test", "val"]:
for field in ["source", "target"]:
__lowerCAmelCase = '\n'.join([contents[field]] * n_lines[split] )
with open(os.path.join(lowerCAmelCase_ , f"""{split}.{field}""" ) , 'w' ) as f:
f.write(lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : int , lowerCAmelCase_ : str = "pytorch" ) -> List[str]:
__lowerCAmelCase = self.get_auto_remove_tmp_dir()
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'output' )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'data' )
self._create_dummy_data(data_dir=lowerCAmelCase_ )
__lowerCAmelCase = f"""
--data_dir {data_dir} \
--output_dir {output_dir} \
--model_name_or_path facebook/rag-sequence-base \
--model_type rag_sequence \
--do_train \
--do_predict \
--n_val -1 \
--val_check_interval 1.0 \
--train_batch_size 2 \
--eval_batch_size 1 \
--max_source_length 25 \
--max_target_length 25 \
--val_max_target_length 25 \
--test_max_target_length 25 \
--label_smoothing 0.1 \
--dropout 0.1 \
--attention_dropout 0.1 \
--weight_decay 0.001 \
--adam_epsilon 1e-08 \
--max_grad_norm 0.1 \
--lr_scheduler polynomial \
--learning_rate 3e-04 \
--num_train_epochs 1 \
--warmup_steps 4 \
--gradient_accumulation_steps 1 \
--distributed-port 8787 \
--use_dummy_dataset 1 \
--distributed_retriever {distributed_retriever} \
""".split()
if gpus > 0:
testargs.append(f"""--gpus={gpus}""" )
if is_apex_available():
testargs.append('--fp16' )
else:
testargs.append('--gpus=0' )
testargs.append('--distributed_backend=ddp_cpu' )
testargs.append('--num_processes=2' )
__lowerCAmelCase = [sys.executable, str(Path(finetune_rag.__file__ ).resolve() )] + testargs
execute_subprocess_async(lowerCAmelCase_ , env=self.get_env() )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'metrics.json' )
with open(lowerCAmelCase_ ) as f:
__lowerCAmelCase = json.load(lowerCAmelCase_ )
return result
@require_torch_gpu
def lowercase ( self : str ) -> int:
__lowerCAmelCase = self._run_finetune(gpus=1 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
def lowercase ( self : List[str] ) -> Dict:
__lowerCAmelCase = self._run_finetune(gpus=2 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_gpu
@require_ray
def lowercase ( self : int ) -> Tuple:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
@require_ray
def lowercase ( self : List[Any] ) -> str:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
| 53 | 1 |
import argparse
import json
import os
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
from accelerate.utils.deepspeed import DummyOptim, DummyScheduler
_snake_case : Optional[int] = 16
_snake_case : Tuple = 32
def a_ ( lowerCAmelCase_ : Accelerator, lowerCAmelCase_ : int = 16, lowerCAmelCase_ : str = "bert-base-cased" ):
__lowerCAmelCase = AutoTokenizer.from_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = load_dataset('glue', 'mrpc' )
def tokenize_function(lowerCAmelCase_ : List[Any] ):
# max_length=None => use the model max length (it's actually the default)
__lowerCAmelCase = tokenizer(examples['sentence1'], examples['sentence2'], truncation=lowerCAmelCase_, max_length=lowerCAmelCase_ )
return outputs
# Apply the method we just defined to all the examples in all the splits of the dataset
__lowerCAmelCase = datasets.map(
lowerCAmelCase_, batched=lowerCAmelCase_, remove_columns=['idx', 'sentence1', 'sentence2'], load_from_cache_file=lowerCAmelCase_ )
# We also rename the 'label' column to 'labels' which is the expected name for labels by the models of the
# transformers library
__lowerCAmelCase = tokenized_datasets.rename_column('label', 'labels' )
def collate_fn(lowerCAmelCase_ : str ):
# On TPU it's best to pad everything to the same length or training will be very slow.
if accelerator.distributed_type == DistributedType.TPU:
return tokenizer.pad(lowerCAmelCase_, padding='max_length', max_length=128, return_tensors='pt' )
return tokenizer.pad(lowerCAmelCase_, padding='longest', return_tensors='pt' )
# Instantiate dataloaders.
__lowerCAmelCase = DataLoader(
tokenized_datasets['train'], shuffle=lowerCAmelCase_, collate_fn=lowerCAmelCase_, batch_size=lowerCAmelCase_ )
__lowerCAmelCase = DataLoader(
tokenized_datasets['validation'], shuffle=lowerCAmelCase_, collate_fn=lowerCAmelCase_, batch_size=lowerCAmelCase_ )
return train_dataloader, eval_dataloader
def a_ ( lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : Any ):
model.eval()
__lowerCAmelCase = 0
for step, batch in enumerate(lowerCAmelCase_ ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs.logits.argmax(dim=-1 )
# It is slightly faster to call this once, than multiple times
__lowerCAmelCase , __lowerCAmelCase = accelerator.gather(
(predictions, batch['labels']) ) # If we are in a multiprocess environment, the last batch has duplicates
if accelerator.use_distributed:
if step == len(lowerCAmelCase_ ) - 1:
__lowerCAmelCase = predictions[: len(eval_dataloader.dataset ) - samples_seen]
__lowerCAmelCase = references[: len(eval_dataloader.dataset ) - samples_seen]
else:
samples_seen += references.shape[0]
metric.add_batch(
predictions=lowerCAmelCase_, references=lowerCAmelCase_, )
__lowerCAmelCase = metric.compute()
return eval_metric["accuracy"]
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : Optional[int] ):
# Initialize accelerator
__lowerCAmelCase = Accelerator()
# Sample hyper-parameters for learning rate, batch size, seed and a few other HPs
__lowerCAmelCase = config['lr']
__lowerCAmelCase = int(config['num_epochs'] )
__lowerCAmelCase = int(config['seed'] )
__lowerCAmelCase = int(config['batch_size'] )
__lowerCAmelCase = args.model_name_or_path
set_seed(lowerCAmelCase_ )
__lowerCAmelCase , __lowerCAmelCase = get_dataloaders(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# Instantiate the model (we build the model here so that the seed also control new weights initialization)
__lowerCAmelCase = AutoModelForSequenceClassification.from_pretrained(lowerCAmelCase_, return_dict=lowerCAmelCase_ )
# Instantiate optimizer
__lowerCAmelCase = (
AdamW
if accelerator.state.deepspeed_plugin is None
or 'optimizer' not in accelerator.state.deepspeed_plugin.deepspeed_config
else DummyOptim
)
__lowerCAmelCase = optimizer_cls(params=model.parameters(), lr=lowerCAmelCase_ )
if accelerator.state.deepspeed_plugin is not None:
__lowerCAmelCase = accelerator.state.deepspeed_plugin.deepspeed_config[
'gradient_accumulation_steps'
]
else:
__lowerCAmelCase = 1
__lowerCAmelCase = (len(lowerCAmelCase_ ) * num_epochs) // gradient_accumulation_steps
# Instantiate scheduler
if (
accelerator.state.deepspeed_plugin is None
or "scheduler" not in accelerator.state.deepspeed_plugin.deepspeed_config
):
__lowerCAmelCase = get_linear_schedule_with_warmup(
optimizer=lowerCAmelCase_, num_warmup_steps=0, num_training_steps=lowerCAmelCase_, )
else:
__lowerCAmelCase = DummyScheduler(lowerCAmelCase_, total_num_steps=lowerCAmelCase_, warmup_num_steps=0 )
# Prepare everything
# There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the
# prepare method.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = accelerator.prepare(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# We need to keep track of how many total steps we have iterated over
__lowerCAmelCase = 0
# We also need to keep track of the stating epoch so files are named properly
__lowerCAmelCase = 0
__lowerCAmelCase = evaluate.load('glue', 'mrpc' )
__lowerCAmelCase = num_epochs
if args.partial_train_epoch is not None:
__lowerCAmelCase = args.partial_train_epoch
if args.resume_from_checkpoint:
accelerator.load_state(args.resume_from_checkpoint )
__lowerCAmelCase = args.resume_from_checkpoint.split('epoch_' )[1]
__lowerCAmelCase = ''
for char in epoch_string:
if char.isdigit():
state_epoch_num += char
else:
break
__lowerCAmelCase = int(lowerCAmelCase_ ) + 1
__lowerCAmelCase = evaluation_loop(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
accelerator.print('resumed checkpoint performance:', lowerCAmelCase_ )
accelerator.print('resumed checkpoint\'s scheduler\'s lr:', lr_scheduler.get_lr()[0] )
accelerator.print('resumed optimizers\'s lr:', optimizer.param_groups[0]['lr'] )
with open(os.path.join(args.output_dir, F"""state_{starting_epoch-1}.json""" ), 'r' ) as f:
__lowerCAmelCase = json.load(lowerCAmelCase_ )
assert resumed_state["accuracy"] == accuracy, "Accuracy mismatch, loading from checkpoint failed"
assert (
resumed_state["lr"] == lr_scheduler.get_lr()[0]
), "Scheduler learning rate mismatch, loading from checkpoint failed"
assert (
resumed_state["optimizer_lr"] == optimizer.param_groups[0]["lr"]
), "Optimizer learning rate mismatch, loading from checkpoint failed"
assert resumed_state["epoch"] == starting_epoch - 1, "Epoch mismatch, loading from checkpoint failed"
return
# Now we train the model
__lowerCAmelCase = {}
for epoch in range(lowerCAmelCase_, lowerCAmelCase_ ):
model.train()
for step, batch in enumerate(lowerCAmelCase_ ):
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs.loss
__lowerCAmelCase = loss / gradient_accumulation_steps
accelerator.backward(lowerCAmelCase_ )
if step % gradient_accumulation_steps == 0:
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
overall_step += 1
__lowerCAmelCase = F"""epoch_{epoch}"""
__lowerCAmelCase = os.path.join(args.output_dir, lowerCAmelCase_ )
accelerator.save_state(lowerCAmelCase_ )
__lowerCAmelCase = evaluation_loop(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = accuracy
__lowerCAmelCase = lr_scheduler.get_lr()[0]
__lowerCAmelCase = optimizer.param_groups[0]['lr']
__lowerCAmelCase = epoch
__lowerCAmelCase = overall_step
accelerator.print(F"""epoch {epoch}:""", lowerCAmelCase_ )
accelerator.wait_for_everyone()
if accelerator.is_main_process:
with open(os.path.join(args.output_dir, F"""state_{epoch}.json""" ), 'w' ) as f:
json.dump(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( ):
__lowerCAmelCase = argparse.ArgumentParser(description='Simple example of training script tracking peak GPU memory usage.' )
parser.add_argument(
'--model_name_or_path', type=lowerCAmelCase_, default='bert-base-cased', help='Path to pretrained model or model identifier from huggingface.co/models.', required=lowerCAmelCase_, )
parser.add_argument(
'--output_dir', type=lowerCAmelCase_, default='.', help='Optional save directory where all checkpoint folders will be stored. Default is the current working directory.', )
parser.add_argument(
'--resume_from_checkpoint', type=lowerCAmelCase_, default=lowerCAmelCase_, help='If the training should continue from a checkpoint folder.', )
parser.add_argument(
'--partial_train_epoch', type=lowerCAmelCase_, default=lowerCAmelCase_, help='If passed, the training will stop after this number of epochs.', )
parser.add_argument(
'--num_epochs', type=lowerCAmelCase_, default=2, help='Number of train epochs.', )
__lowerCAmelCase = parser.parse_args()
__lowerCAmelCase = {'lr': 2E-5, 'num_epochs': args.num_epochs, 'seed': 42, 'batch_size': 16}
training_function(lowerCAmelCase_, lowerCAmelCase_ )
if __name__ == "__main__":
main()
| 53 |
import copy
import inspect
import unittest
from transformers import AutoBackbone
from transformers.configuration_utils import PretrainedConfig
from transformers.testing_utils import require_timm, require_torch, torch_device
from transformers.utils.import_utils import is_torch_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor
if is_torch_available():
import torch
from transformers import TimmBackbone, TimmBackboneConfig
from ...test_pipeline_mixin import PipelineTesterMixin
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : Optional[int]=None , lowerCAmelCase_ : Union[str, Any]="resnet50" , lowerCAmelCase_ : str=3 , lowerCAmelCase_ : List[str]=3_2 , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : Dict=True , lowerCAmelCase_ : Optional[Any]=True , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = out_indices if out_indices is not None else [4]
__lowerCAmelCase = stage_names
__lowerCAmelCase = out_features
__lowerCAmelCase = backbone
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = use_pretrained_backbone
__lowerCAmelCase = is_training
def lowercase ( self : List[str] ) -> Any:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : List[Any] ) -> Union[str, Any]:
return TimmBackboneConfig(
image_size=self.image_size , num_channels=self.num_channels , out_features=self.out_features , out_indices=self.out_indices , stage_names=self.stage_names , use_pretrained_backbone=self.use_pretrained_backbone , backbone=self.backbone , )
def lowercase ( self : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ) -> int:
__lowerCAmelCase = TimmBackbone(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.feature_map[-1].shape , (self.batch_size, model.channels[-1], 1_4, 1_4) , )
def lowercase ( self : List[str] ) -> str:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
@require_timm
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (TimmBackbone,) if is_torch_available() else ()
a_ = {"""feature-extraction""": TimmBackbone} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Tuple ) -> int:
__lowerCAmelCase = TimmBackboneModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : Dict ) -> List[str]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Union[str, Any] ) -> Optional[int]:
__lowerCAmelCase = 'resnet18'
__lowerCAmelCase = 'microsoft/resnet-18'
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(len(timm_model.stage_names ) , len(transformers_model.stage_names ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
# Out indices are set to the last layer by default. For timm models, we don't know
# the number of layers in advance, so we set it to (-1,), whereas for transformers
# models, we set it to [len(stage_names) - 1] (kept for backward compatibility).
self.assertEqual(timm_model.out_indices , (-1,) )
self.assertEqual(transformers_model.out_indices , [len(timm_model.stage_names ) - 1] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ , out_indices=[1, 2, 3] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , out_indices=[1, 2, 3] )
self.assertEqual(timm_model.out_indices , transformers_model.out_indices )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
@unittest.skip('TimmBackbone doesn\'t support feed forward chunking' )
def lowercase ( self : List[str] ) -> Tuple:
pass
@unittest.skip('TimmBackbone doesn\'t have num_hidden_layers attribute' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('TimmBackbone initialization is managed on the timm side' )
def lowercase ( self : str ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Any ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Optional[Any]:
pass
@unittest.skip('TimmBackbone model cannot be created without specifying a backbone checkpoint' )
def lowercase ( self : Dict ) -> Any:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Any ) -> Optional[int]:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : Union[str, Any] ) -> Tuple:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : List[str] ) -> Optional[int]:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Tuple ) -> List[str]:
pass
@unittest.skip('TimmBackbone doesn\'t have hidden size info in its configuration.' )
def lowercase ( self : int ) -> Optional[int]:
pass
@unittest.skip('TimmBackbone doesn\'t support output_attentions.' )
def lowercase ( self : Union[str, Any] ) -> str:
pass
@unittest.skip('Safetensors is not supported by timm.' )
def lowercase ( self : Dict ) -> str:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : List[str] ) -> Optional[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = True
__lowerCAmelCase = self.has_attentions
# no need to test all models as different heads yield the same functionality
__lowerCAmelCase = self.all_model_classes[0]
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs[0][-1]
# Encoder-/Decoder-only models
__lowerCAmelCase = outputs.hidden_states[0]
hidden_states.retain_grad()
if self.has_attentions:
__lowerCAmelCase = outputs.attentions[0]
attentions.retain_grad()
output.flatten()[0].backward(retain_graph=lowerCAmelCase_ )
self.assertIsNotNone(hidden_states.grad )
if self.has_attentions:
self.assertIsNotNone(attentions.grad )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , len(config.out_indices ) )
self.assertEqual(len(model.channels ) , len(config.out_indices ) )
# Check output of last stage is taken if out_features=None, out_indices=None
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , 1 )
self.assertEqual(len(model.channels ) , 1 )
# Check backbone can be initialized with fresh weights
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = False
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
| 53 | 1 |
import unittest
import numpy as np
from transformers.testing_utils import require_torch, require_vision
from transformers.utils import is_torch_available, is_vision_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin, prepare_image_inputs
if is_torch_available():
import torch
if is_vision_available():
from PIL import Image
from transformers import MobileNetVaImageProcessor
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Union[str, Any]=7 , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : List[str]=1_8 , lowerCAmelCase_ : Union[str, Any]=3_0 , lowerCAmelCase_ : int=4_0_0 , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : str=None , lowerCAmelCase_ : Optional[int]=True , lowerCAmelCase_ : List[Any]=None , ) -> Union[str, Any]:
__lowerCAmelCase = size if size is not None else {'shortest_edge': 2_0}
__lowerCAmelCase = crop_size if crop_size is not None else {'height': 1_8, 'width': 1_8}
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = image_size
__lowerCAmelCase = min_resolution
__lowerCAmelCase = max_resolution
__lowerCAmelCase = do_resize
__lowerCAmelCase = size
__lowerCAmelCase = do_center_crop
__lowerCAmelCase = crop_size
def lowercase ( self : Tuple ) -> Optional[Any]:
return {
"do_resize": self.do_resize,
"size": self.size,
"do_center_crop": self.do_center_crop,
"crop_size": self.crop_size,
}
@require_torch
@require_vision
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = MobileNetVaImageProcessor if is_vision_available() else None
def lowercase ( self : Dict ) -> int:
__lowerCAmelCase = MobileNetVaImageProcessingTester(self )
@property
def lowercase ( self : List[Any] ) -> Tuple:
return self.image_processor_tester.prepare_image_processor_dict()
def lowercase ( self : Any ) -> Optional[Any]:
__lowerCAmelCase = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(lowerCAmelCase_ , 'do_resize' ) )
self.assertTrue(hasattr(lowerCAmelCase_ , 'size' ) )
self.assertTrue(hasattr(lowerCAmelCase_ , 'do_center_crop' ) )
self.assertTrue(hasattr(lowerCAmelCase_ , 'crop_size' ) )
def lowercase ( self : int ) -> Optional[Any]:
__lowerCAmelCase = self.image_processing_class.from_dict(self.image_processor_dict )
self.assertEqual(image_processor.size , {'shortest_edge': 2_0} )
self.assertEqual(image_processor.crop_size , {'height': 1_8, 'width': 1_8} )
__lowerCAmelCase = self.image_processing_class.from_dict(self.image_processor_dict , size=4_2 , crop_size=8_4 )
self.assertEqual(image_processor.size , {'shortest_edge': 4_2} )
self.assertEqual(image_processor.crop_size , {'height': 8_4, 'width': 8_4} )
def lowercase ( self : List[Any] ) -> str:
pass
def lowercase ( self : Any ) -> str:
# Initialize image_processing
__lowerCAmelCase = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
__lowerCAmelCase = prepare_image_inputs(self.image_processor_tester , equal_resolution=lowerCAmelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCAmelCase_ , Image.Image )
# Test not batched input
__lowerCAmelCase = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
# Test batched
__lowerCAmelCase = image_processing(lowerCAmelCase_ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
def lowercase ( self : List[Any] ) -> List[Any]:
# Initialize image_processing
__lowerCAmelCase = self.image_processing_class(**self.image_processor_dict )
# create random numpy tensors
__lowerCAmelCase = prepare_image_inputs(self.image_processor_tester , equal_resolution=lowerCAmelCase_ , numpify=lowerCAmelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCAmelCase_ , np.ndarray )
# Test not batched input
__lowerCAmelCase = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
# Test batched
__lowerCAmelCase = image_processing(lowerCAmelCase_ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
def lowercase ( self : List[Any] ) -> List[Any]:
# Initialize image_processing
__lowerCAmelCase = self.image_processing_class(**self.image_processor_dict )
# create random PyTorch tensors
__lowerCAmelCase = prepare_image_inputs(self.image_processor_tester , equal_resolution=lowerCAmelCase_ , torchify=lowerCAmelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCAmelCase_ , torch.Tensor )
# Test not batched input
__lowerCAmelCase = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
# Test batched
__lowerCAmelCase = image_processing(lowerCAmelCase_ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
| 53 |
# Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import os
import platform
import numpy as np
import psutil
import torch
from accelerate import __version__ as version
from accelerate.commands.config import default_config_file, load_config_from_file
from ..utils import is_npu_available, is_xpu_available
def a_ ( lowerCAmelCase_ : str=None ):
if subparsers is not None:
__lowerCAmelCase = subparsers.add_parser('env' )
else:
__lowerCAmelCase = argparse.ArgumentParser('Accelerate env command' )
parser.add_argument(
'--config_file', default=lowerCAmelCase_, help='The config file to use for the default values in the launching script.' )
if subparsers is not None:
parser.set_defaults(func=lowerCAmelCase_ )
return parser
def a_ ( lowerCAmelCase_ : Optional[int] ):
__lowerCAmelCase = torch.__version__
__lowerCAmelCase = torch.cuda.is_available()
__lowerCAmelCase = is_xpu_available()
__lowerCAmelCase = is_npu_available()
__lowerCAmelCase = 'Not found'
# Get the default from the config file.
if args.config_file is not None or os.path.isfile(lowerCAmelCase_ ):
__lowerCAmelCase = load_config_from_file(args.config_file ).to_dict()
__lowerCAmelCase = {
'`Accelerate` version': version,
'Platform': platform.platform(),
'Python version': platform.python_version(),
'Numpy version': np.__version__,
'PyTorch version (GPU?)': F"""{pt_version} ({pt_cuda_available})""",
'PyTorch XPU available': str(lowerCAmelCase_ ),
'PyTorch NPU available': str(lowerCAmelCase_ ),
'System RAM': F"""{psutil.virtual_memory().total / 1024 ** 3:.2f} GB""",
}
if pt_cuda_available:
__lowerCAmelCase = torch.cuda.get_device_name()
print('\nCopy-and-paste the text below in your GitHub issue\n' )
print('\n'.join([F"""- {prop}: {val}""" for prop, val in info.items()] ) )
print('- `Accelerate` default config:' if args.config_file is None else '- `Accelerate` config passed:' )
__lowerCAmelCase = (
'\n'.join([F"""\t- {prop}: {val}""" for prop, val in accelerate_config.items()] )
if isinstance(lowerCAmelCase_, lowerCAmelCase_ )
else F"""\t{accelerate_config}"""
)
print(lowerCAmelCase_ )
__lowerCAmelCase = accelerate_config
return info
def a_ ( ):
__lowerCAmelCase = env_command_parser()
__lowerCAmelCase = parser.parse_args()
env_command(lowerCAmelCase_ )
return 0
if __name__ == "__main__":
raise SystemExit(main())
| 53 | 1 |
import requests
_snake_case : Tuple = '' # <-- Put your OpenWeatherMap appid here!
_snake_case : Dict = 'https://api.openweathermap.org/data/2.5/'
def a_ ( lowerCAmelCase_ : str = "Chicago", lowerCAmelCase_ : str = APPID ):
return requests.get(URL_BASE + 'weather', params=locals() ).json()
def a_ ( lowerCAmelCase_ : str = "Kolkata, India", lowerCAmelCase_ : str = APPID ):
return requests.get(URL_BASE + 'forecast', params=locals() ).json()
def a_ ( lowerCAmelCase_ : float = 55.68, lowerCAmelCase_ : float = 12.57, lowerCAmelCase_ : str = APPID ):
return requests.get(URL_BASE + 'onecall', params=locals() ).json()
if __name__ == "__main__":
from pprint import pprint
while True:
_snake_case : List[str] = input('Enter a location:').strip()
if location:
pprint(current_weather(location))
else:
break
| 53 |
import importlib
import sys
from argparse import REMAINDER, ArgumentParser
from pathlib import Path
import torch_xla.distributed.xla_multiprocessing as xmp
def a_ ( ):
__lowerCAmelCase = ArgumentParser(
description=(
'PyTorch TPU distributed training launch '
'helper utility that will spawn up '
'multiple distributed processes'
) )
# Optional arguments for the launch helper
parser.add_argument('--num_cores', type=lowerCAmelCase_, default=1, help='Number of TPU cores to use (1 or 8).' )
# positional
parser.add_argument(
'training_script', type=lowerCAmelCase_, help=(
'The full path to the single TPU training '
'program/script to be launched in parallel, '
'followed by all the arguments for the '
'training script'
), )
# rest from the training program
parser.add_argument('training_script_args', nargs=lowerCAmelCase_ )
return parser.parse_args()
def a_ ( ):
__lowerCAmelCase = parse_args()
# Import training_script as a module.
__lowerCAmelCase = Path(args.training_script )
sys.path.append(str(script_fpath.parent.resolve() ) )
__lowerCAmelCase = script_fpath.stem
__lowerCAmelCase = importlib.import_module(lowerCAmelCase_ )
# Patch sys.argv
__lowerCAmelCase = [args.training_script] + args.training_script_args + ['--tpu_num_cores', str(args.num_cores )]
xmp.spawn(mod._mp_fn, args=(), nprocs=args.num_cores )
if __name__ == "__main__":
main()
| 53 | 1 |
import unittest
from accelerate import debug_launcher
from accelerate.test_utils import require_cpu, test_ops, test_script
@require_cpu
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : List[Any] ) -> Tuple:
debug_launcher(test_script.main )
def lowercase ( self : Any ) -> List[Any]:
debug_launcher(test_ops.main )
| 53 |
import inspect
import unittest
from huggingface_hub import hf_hub_download
from transformers import ConvNextConfig, UperNetConfig
from transformers.testing_utils import require_torch, require_torch_multi_gpu, require_vision, slow, torch_device
from transformers.utils import is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import UperNetForSemanticSegmentation
from transformers.models.upernet.modeling_upernet import UPERNET_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Dict=1_3 , lowerCAmelCase_ : str=3_2 , lowerCAmelCase_ : Optional[Any]=3 , lowerCAmelCase_ : Any=4 , lowerCAmelCase_ : str=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Tuple=[2, 2, 3, 2] , lowerCAmelCase_ : str=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[int]=3_7 , lowerCAmelCase_ : Dict="gelu" , lowerCAmelCase_ : List[Any]=1_0 , lowerCAmelCase_ : str=0.02 , lowerCAmelCase_ : Dict=["stage2", "stage3", "stage4"] , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : List[Any]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = num_stages
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = out_features
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = num_stages
def lowercase ( self : Dict ) -> List[str]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = self.get_config()
return config, pixel_values, labels
def lowercase ( self : List[str] ) -> Union[str, Any]:
return ConvNextConfig(
num_channels=self.num_channels , num_stages=self.num_stages , hidden_sizes=self.hidden_sizes , depths=self.depths , is_training=self.is_training , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , out_features=self.out_features , )
def lowercase ( self : Dict ) -> List[str]:
return UperNetConfig(
backbone_config=self.get_backbone_config() , hidden_size=5_1_2 , pool_scales=[1, 2, 3, 6] , use_auxiliary_head=lowerCAmelCase_ , auxiliary_loss_weight=0.4 , auxiliary_in_channels=4_0 , auxiliary_channels=2_5_6 , auxiliary_num_convs=1 , auxiliary_concat_input=lowerCAmelCase_ , loss_ignore_index=2_5_5 , num_labels=self.num_labels , )
def lowercase ( self : List[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : int ) -> Optional[Any]:
__lowerCAmelCase = UperNetForSemanticSegmentation(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.logits.shape , (self.batch_size, self.num_labels, self.image_size, self.image_size) )
def lowercase ( self : Union[str, Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (UperNetForSemanticSegmentation,) if is_torch_available() else ()
a_ = {"""image-segmentation""": UperNetForSemanticSegmentation} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = UperNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : List[str] ) -> int:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Tuple ) -> Union[str, Any]:
return
def lowercase ( self : Optional[int] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_semantic_segmentation(*lowerCAmelCase_ )
@unittest.skip(reason='UperNet does not use inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Dict:
pass
@unittest.skip(reason='UperNet does not support input and output embeddings' )
def lowercase ( self : Optional[Any] ) -> Dict:
pass
@unittest.skip(reason='UperNet does not have a base model' )
def lowercase ( self : Optional[int] ) -> List[Any]:
pass
@unittest.skip(reason='UperNet does not have a base model' )
def lowercase ( self : str ) -> Dict:
pass
@require_torch_multi_gpu
@unittest.skip(reason='UperNet has some layers using `add_module` which doesn\'t work well with `nn.DataParallel`' )
def lowercase ( self : Optional[Any] ) -> Optional[int]:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : Tuple ) -> List[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Tuple:
def check_hidden_states_output(lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
# ConvNext's feature maps are of shape (batch_size, num_channels, height, width)
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [self.model_tester.image_size // 4, self.model_tester.image_size // 4] , )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Any ) -> Tuple:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = _config_zero_init(lowerCAmelCase_ )
__lowerCAmelCase = _config_zero_init(configs_no_init.backbone_config )
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(config=lowerCAmelCase_ )
for name, param in model.named_parameters():
if param.requires_grad:
self.assertIn(
((param.data.mean() * 1e9).round() / 1e9).item() , [0.0, 1.0] , msg=f"""Parameter {name} of model {model_class} seems not properly initialized""" , )
@unittest.skip(reason='UperNet does not have tied weights' )
def lowercase ( self : Any ) -> int:
pass
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in UPERNET_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
def a_ ( ):
__lowerCAmelCase = hf_hub_download(
repo_id='hf-internal-testing/fixtures_ade20k', repo_type='dataset', filename='ADE_val_00000001.jpg' )
__lowerCAmelCase = Image.open(lowerCAmelCase_ ).convert('RGB' )
return image
@require_torch
@require_vision
@slow
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Dict ) -> Union[str, Any]:
__lowerCAmelCase = AutoImageProcessor.from_pretrained('openmmlab/upernet-swin-tiny' )
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained('openmmlab/upernet-swin-tiny' ).to(lowerCAmelCase_ )
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = torch.Size((1, model.config.num_labels, 5_1_2, 5_1_2) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-7.59_58, -7.59_58, -7.43_02], [-7.59_58, -7.59_58, -7.43_02], [-7.47_97, -7.47_97, -7.30_68]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, 0, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = AutoImageProcessor.from_pretrained('openmmlab/upernet-convnext-tiny' )
__lowerCAmelCase = UperNetForSemanticSegmentation.from_pretrained('openmmlab/upernet-convnext-tiny' ).to(lowerCAmelCase_ )
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = torch.Size((1, model.config.num_labels, 5_1_2, 5_1_2) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-8.81_10, -8.81_10, -8.65_21], [-8.81_10, -8.81_10, -8.65_21], [-8.77_46, -8.77_46, -8.61_30]] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, 0, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import inspect
import logging
import os
import random
import shutil
import tempfile
import unittest
import pytest
import torch
from torch import nn
from torch.utils.data import DataLoader, TensorDataset
from accelerate import Accelerator
from accelerate.test_utils import execute_subprocess_async, require_cuda
from accelerate.utils import ProjectConfiguration, set_seed
_snake_case : Optional[int] = logging.getLogger(__name__)
def a_ ( lowerCAmelCase_ : List[Any]=2, lowerCAmelCase_ : List[str]=3, lowerCAmelCase_ : List[Any]=16, lowerCAmelCase_ : int = 10, lowerCAmelCase_ : int = 2 ):
def get_dataset(lowerCAmelCase_ : int ):
__lowerCAmelCase = torch.randn(batch_size * n_batches, 1 )
return TensorDataset(lowerCAmelCase_, a * x + b + 0.1 * torch.randn(batch_size * n_batches, 1 ) )
__lowerCAmelCase = get_dataset(lowerCAmelCase_ )
__lowerCAmelCase = get_dataset(lowerCAmelCase_ )
__lowerCAmelCase = DataLoader(lowerCAmelCase_, shuffle=lowerCAmelCase_, batch_size=lowerCAmelCase_, num_workers=4 )
__lowerCAmelCase = DataLoader(lowerCAmelCase_, shuffle=lowerCAmelCase_, batch_size=lowerCAmelCase_, num_workers=4 )
return (train_dataloader, valid_dataloader)
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : Any, lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int, lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str]=None ):
__lowerCAmelCase = []
for epoch in range(lowerCAmelCase_ ):
# Train quickly
model.train()
for batch in dataloader:
__lowerCAmelCase , __lowerCAmelCase = batch
__lowerCAmelCase = model(lowerCAmelCase_ )
__lowerCAmelCase = torch.nn.functional.mse_loss(lowerCAmelCase_, lowerCAmelCase_ )
accelerator.backward(lowerCAmelCase_ )
optimizer.step()
optimizer.zero_grad()
rands.append(random.random() ) # Introduce some randomness
if scheduler is not None:
scheduler.step()
return rands
class _UpperCAmelCase ( nn.Module ):
"""simple docstring"""
def __init__( self : List[str] ) -> Any:
super().__init__()
__lowerCAmelCase = nn.Parameter(torch.randn(1 ) )
__lowerCAmelCase = nn.Parameter(torch.randn(1 ) )
def lowercase ( self : int , lowerCAmelCase_ : Tuple ) -> List[str]:
return x * self.a + self.b
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : str ) -> Any:
with tempfile.TemporaryDirectory() as tmpdir:
set_seed(4_2 )
__lowerCAmelCase = DummyModel()
__lowerCAmelCase = torch.optim.Adam(params=model.parameters() , lr=1e-3 )
__lowerCAmelCase , __lowerCAmelCase = dummy_dataloaders()
__lowerCAmelCase = ProjectConfiguration(total_limit=1 , project_dir=lowerCAmelCase_ , automatic_checkpoint_naming=lowerCAmelCase_ )
# Train baseline
__lowerCAmelCase = Accelerator(project_config=lowerCAmelCase_ )
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = accelerator.prepare(
lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# Save initial
accelerator.save_state()
# Save second state
accelerator.save_state()
self.assertEqual(len(os.listdir(accelerator.project_dir ) ) , 1 )
def lowercase ( self : Dict ) -> Dict:
with tempfile.TemporaryDirectory() as tmpdir:
set_seed(4_2 )
__lowerCAmelCase = DummyModel()
__lowerCAmelCase = torch.optim.Adam(params=model.parameters() , lr=1e-3 )
__lowerCAmelCase , __lowerCAmelCase = dummy_dataloaders()
# Train baseline
__lowerCAmelCase = Accelerator()
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = accelerator.prepare(
lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# Save initial
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'initial' )
accelerator.save_state(lowerCAmelCase_ )
((__lowerCAmelCase) , (__lowerCAmelCase)) = model.a.item(), model.b.item()
__lowerCAmelCase = optimizer.state_dict()
__lowerCAmelCase = train(3 , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
((__lowerCAmelCase) , (__lowerCAmelCase)) = model.a.item(), model.b.item()
__lowerCAmelCase = optimizer.state_dict()
# Train partially
set_seed(4_2 )
__lowerCAmelCase = DummyModel()
__lowerCAmelCase = torch.optim.Adam(params=model.parameters() , lr=1e-3 )
__lowerCAmelCase , __lowerCAmelCase = dummy_dataloaders()
__lowerCAmelCase = Accelerator()
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = accelerator.prepare(
lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
accelerator.load_state(lowerCAmelCase_ )
((__lowerCAmelCase) , (__lowerCAmelCase)) = model.a.item(), model.b.item()
__lowerCAmelCase = optimizer.state_dict()
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = train(2 , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# Save everything
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'checkpoint' )
accelerator.save_state(lowerCAmelCase_ )
# Load everything back in and make sure all states work
accelerator.load_state(lowerCAmelCase_ )
test_rands += train(1 , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
((__lowerCAmelCase) , (__lowerCAmelCase)) = model.a.item(), model.b.item()
__lowerCAmelCase = optimizer.state_dict()
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> Tuple:
with tempfile.TemporaryDirectory() as tmpdir:
set_seed(4_2 )
__lowerCAmelCase = DummyModel()
__lowerCAmelCase = torch.optim.Adam(params=model.parameters() , lr=1e-3 )
__lowerCAmelCase , __lowerCAmelCase = dummy_dataloaders()
__lowerCAmelCase = ProjectConfiguration(automatic_checkpoint_naming=lowerCAmelCase_ )
# Train baseline
__lowerCAmelCase = Accelerator(project_dir=lowerCAmelCase_ , project_config=lowerCAmelCase_ )
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = accelerator.prepare(
lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# Save initial
accelerator.save_state()
((__lowerCAmelCase) , (__lowerCAmelCase)) = model.a.item(), model.b.item()
__lowerCAmelCase = optimizer.state_dict()
__lowerCAmelCase = train(3 , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
((__lowerCAmelCase) , (__lowerCAmelCase)) = model.a.item(), model.b.item()
__lowerCAmelCase = optimizer.state_dict()
# Train partially
set_seed(4_2 )
__lowerCAmelCase = DummyModel()
__lowerCAmelCase = torch.optim.Adam(params=model.parameters() , lr=1e-3 )
__lowerCAmelCase , __lowerCAmelCase = dummy_dataloaders()
__lowerCAmelCase = ProjectConfiguration(iteration=1 , automatic_checkpoint_naming=lowerCAmelCase_ )
__lowerCAmelCase = Accelerator(project_dir=lowerCAmelCase_ , project_config=lowerCAmelCase_ )
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = accelerator.prepare(
lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
accelerator.load_state(os.path.join(lowerCAmelCase_ , 'checkpoints' , 'checkpoint_0' ) )
((__lowerCAmelCase) , (__lowerCAmelCase)) = model.a.item(), model.b.item()
__lowerCAmelCase = optimizer.state_dict()
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = train(2 , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# Save everything
accelerator.save_state()
# Load everything back in and make sure all states work
accelerator.load_state(os.path.join(lowerCAmelCase_ , 'checkpoints' , 'checkpoint_1' ) )
test_rands += train(1 , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
((__lowerCAmelCase) , (__lowerCAmelCase)) = model.a.item(), model.b.item()
__lowerCAmelCase = optimizer.state_dict()
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
self.assertEqual(lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Any ) -> Any:
__lowerCAmelCase = torch.tensor([1, 2, 3] )
__lowerCAmelCase = torch.tensor([2, 3, 4] )
__lowerCAmelCase = DummyModel()
__lowerCAmelCase = torch.optim.Adam(net.parameters() )
__lowerCAmelCase = Accelerator()
with self.assertRaises(lowerCAmelCase_ ) as ve:
accelerator.register_for_checkpointing(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = str(ve.exception )
self.assertTrue('Item at index 0' in message )
self.assertTrue('Item at index 1' in message )
self.assertFalse('Item at index 2' in message )
self.assertFalse('Item at index 3' in message )
def lowercase ( self : Any ) -> List[Any]:
with tempfile.TemporaryDirectory() as tmpdir:
set_seed(4_2 )
__lowerCAmelCase = DummyModel()
__lowerCAmelCase = torch.optim.Adam(params=model.parameters() , lr=1e-3 )
__lowerCAmelCase = torch.optim.lr_scheduler.StepLR(lowerCAmelCase_ , step_size=1 , gamma=0.99 )
__lowerCAmelCase , __lowerCAmelCase = dummy_dataloaders()
__lowerCAmelCase = ProjectConfiguration(automatic_checkpoint_naming=lowerCAmelCase_ )
# Train baseline
__lowerCAmelCase = Accelerator(project_dir=lowerCAmelCase_ , project_config=lowerCAmelCase_ )
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = accelerator.prepare(
lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# Save initial
accelerator.save_state()
__lowerCAmelCase = scheduler.state_dict()
train(3 , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
self.assertNotEqual(lowerCAmelCase_ , scheduler.state_dict() )
# Load everything back in and make sure all states work
accelerator.load_state(os.path.join(lowerCAmelCase_ , 'checkpoints' , 'checkpoint_0' ) )
self.assertEqual(lowerCAmelCase_ , scheduler.state_dict() )
def lowercase ( self : int ) -> Union[str, Any]:
with tempfile.TemporaryDirectory() as tmpdir:
set_seed(4_2 )
__lowerCAmelCase = DummyModel()
__lowerCAmelCase = ProjectConfiguration(automatic_checkpoint_naming=lowerCAmelCase_ , total_limit=2 )
# Train baseline
__lowerCAmelCase = Accelerator(project_dir=lowerCAmelCase_ , project_config=lowerCAmelCase_ )
__lowerCAmelCase = accelerator.prepare(lowerCAmelCase_ )
# Save 3 states:
for _ in range(1_1 ):
accelerator.save_state()
self.assertTrue(not os.path.exists(os.path.join(lowerCAmelCase_ , 'checkpoints' , 'checkpoint_0' ) ) )
self.assertTrue(os.path.exists(os.path.join(lowerCAmelCase_ , 'checkpoints' , 'checkpoint_9' ) ) )
self.assertTrue(os.path.exists(os.path.join(lowerCAmelCase_ , 'checkpoints' , 'checkpoint_10' ) ) )
@require_cuda
def lowercase ( self : List[str] ) -> Any:
__lowerCAmelCase = ['torchrun', f"""--nproc_per_node={torch.cuda.device_count()}""", inspect.getfile(self.__class__ )]
execute_subprocess_async(lowerCAmelCase_ , env=os.environ.copy() )
if __name__ == "__main__":
_snake_case : List[str] = '/tmp/accelerate/state_checkpointing'
_snake_case : int = DummyModel()
_snake_case : Union[str, Any] = torch.optim.Adam(params=model.parameters(), lr=1e-3)
_snake_case : str = torch.optim.lr_scheduler.StepLR(optimizer, step_size=1, gamma=0.99)
_snake_case , _snake_case : Optional[int] = dummy_dataloaders()
_snake_case : Optional[int] = ProjectConfiguration(automatic_checkpoint_naming=True)
# Train baseline
_snake_case : int = Accelerator(project_dir=savedir, project_config=project_config, mixed_precision='no')
if accelerator.process_index == 0:
if os.path.exists(savedir):
shutil.rmtree(savedir)
os.makedirs(savedir)
_snake_case , _snake_case , _snake_case , _snake_case , _snake_case : Dict = accelerator.prepare(
model, optimizer, train_dataloader, valid_dataloader, scheduler
)
_snake_case , _snake_case : Tuple = accelerator.prepare(model, optimizer)
train(3, model, train_dataloader, optimizer, accelerator, scheduler)
# Check that the intial optimizer is loaded on the GPU
for group in optimizer.param_groups:
_snake_case : List[Any] = group['params'][0].device
break
assert param_device.type == accelerator.device.type
_snake_case : Optional[Any] = model.cpu()
accelerator.wait_for_everyone()
accelerator.save_state()
accelerator.wait_for_everyone()
# Check CPU state
accelerator.load_state(os.path.join(savedir, 'checkpoints', 'checkpoint_0'), map_location='cpu')
for group in optimizer.param_groups:
_snake_case : List[str] = group['params'][0].device
break
assert (
param_device.type == torch.device('cpu').type
), F"Loaded optimizer states did not match, expected to be loaded on the CPU but got {param_device}"
# Check device state
model.to(accelerator.device)
accelerator.load_state(os.path.join(savedir, 'checkpoints', 'checkpoint_0'), map_location='on_device')
for group in optimizer.param_groups:
_snake_case : List[str] = group['params'][0].device
break
assert (
param_device.type == accelerator.device.type
), F"Loaded optimizer states did not match, expected to be loaded on {accelerator.device} but got {param_device}"
# Check error
with pytest.raises(TypeError, match='Unsupported optimizer map location passed'):
accelerator.load_state(os.path.join(savedir, 'checkpoints', 'checkpoint_0'), map_location='invalid')
accelerator.wait_for_everyone()
if accelerator.process_index == 0:
shutil.rmtree(savedir)
accelerator.wait_for_everyone()
| 53 |
import pytest
from datasets import Dataset, DatasetDict, Features, NamedSplit, Value
from datasets.io.text import TextDatasetReader
from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : Optional[Any] ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ )
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('keep_in_memory', [False, True] )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : int ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_, keep_in_memory=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize(
'features', [
None,
{'text': 'string'},
{'text': 'int32'},
{'text': 'float32'},
], )
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Dict, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = features.copy() if features else default_expected_features
__lowerCAmelCase = (
Features({feature: Value(lowerCAmelCase_ ) for feature, dtype in features.items()} ) if features is not None else None
)
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, features=lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize('split', [None, NamedSplit('train' ), 'train', 'test'] )
def a_ ( lowerCAmelCase_ : List[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Any ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_, split=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
assert dataset.split == split if split else "train"
@pytest.mark.parametrize('path_type', [str, list] )
def a_ ( lowerCAmelCase_ : Dict, lowerCAmelCase_ : Any, lowerCAmelCase_ : Dict ):
if issubclass(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = text_path
elif issubclass(lowerCAmelCase_, lowerCAmelCase_ ):
__lowerCAmelCase = [text_path]
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_dataset(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : int, lowerCAmelCase_ : Tuple=("train",) ):
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ )
for split in splits:
__lowerCAmelCase = dataset_dict[split]
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('keep_in_memory', [False, True] )
def a_ ( lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : Dict ):
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__lowerCAmelCase = TextDatasetReader({'train': text_path}, cache_dir=lowerCAmelCase_, keep_in_memory=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize(
'features', [
None,
{'text': 'string'},
{'text': 'int32'},
{'text': 'float32'},
], )
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[Any] ):
__lowerCAmelCase = tmp_path / 'cache'
# CSV file loses col_1 string dtype information: default now is "int64" instead of "string"
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = features.copy() if features else default_expected_features
__lowerCAmelCase = (
Features({feature: Value(lowerCAmelCase_ ) for feature, dtype in features.items()} ) if features is not None else None
)
__lowerCAmelCase = TextDatasetReader({'train': text_path}, features=lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_ )
@pytest.mark.parametrize('split', [None, NamedSplit('train' ), 'train', 'test'] )
def a_ ( lowerCAmelCase_ : int, lowerCAmelCase_ : str, lowerCAmelCase_ : Optional[int] ):
if split:
__lowerCAmelCase = {split: text_path}
else:
__lowerCAmelCase = 'train'
__lowerCAmelCase = {'train': text_path, 'test': text_path}
__lowerCAmelCase = tmp_path / 'cache'
__lowerCAmelCase = {'text': 'string'}
__lowerCAmelCase = TextDatasetReader(lowerCAmelCase_, cache_dir=lowerCAmelCase_ ).read()
_check_text_datasetdict(lowerCAmelCase_, lowerCAmelCase_, splits=list(path.keys() ) )
assert all(dataset[split].split == split for split in path.keys() )
| 53 | 1 |
import unittest
import numpy as np
from transformers.testing_utils import require_torch, require_vision
from transformers.utils import is_torch_available, is_vision_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin, prepare_image_inputs
if is_torch_available():
import torch
if is_vision_available():
from PIL import Image
from transformers import PoolFormerImageProcessor
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : int , lowerCAmelCase_ : str , lowerCAmelCase_ : List[Any]=7 , lowerCAmelCase_ : Optional[int]=3 , lowerCAmelCase_ : List[Any]=3_0 , lowerCAmelCase_ : List[str]=4_0_0 , lowerCAmelCase_ : int=True , lowerCAmelCase_ : str=None , lowerCAmelCase_ : List[Any]=0.9 , lowerCAmelCase_ : Any=None , lowerCAmelCase_ : Dict=True , lowerCAmelCase_ : Optional[Any]=[0.5, 0.5, 0.5] , lowerCAmelCase_ : List[str]=[0.5, 0.5, 0.5] , ) -> List[str]:
__lowerCAmelCase = size if size is not None else {'shortest_edge': 3_0}
__lowerCAmelCase = crop_size if crop_size is not None else {'height': 3_0, 'width': 3_0}
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = min_resolution
__lowerCAmelCase = max_resolution
__lowerCAmelCase = do_resize_and_center_crop
__lowerCAmelCase = size
__lowerCAmelCase = crop_pct
__lowerCAmelCase = crop_size
__lowerCAmelCase = do_normalize
__lowerCAmelCase = image_mean
__lowerCAmelCase = image_std
def lowercase ( self : str ) -> Tuple:
return {
"size": self.size,
"do_resize_and_center_crop": self.do_resize_and_center_crop,
"crop_pct": self.crop_pct,
"crop_size": self.crop_size,
"do_normalize": self.do_normalize,
"image_mean": self.image_mean,
"image_std": self.image_std,
}
@require_torch
@require_vision
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = PoolFormerImageProcessor if is_vision_available() else None
def lowercase ( self : int ) -> str:
__lowerCAmelCase = PoolFormerImageProcessingTester(self )
@property
def lowercase ( self : List[Any] ) -> Any:
return self.image_processor_tester.prepare_image_processor_dict()
def lowercase ( self : List[str] ) -> Optional[int]:
__lowerCAmelCase = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(lowerCAmelCase_ , 'do_resize_and_center_crop' ) )
self.assertTrue(hasattr(lowerCAmelCase_ , 'size' ) )
self.assertTrue(hasattr(lowerCAmelCase_ , 'crop_pct' ) )
self.assertTrue(hasattr(lowerCAmelCase_ , 'do_normalize' ) )
self.assertTrue(hasattr(lowerCAmelCase_ , 'image_mean' ) )
self.assertTrue(hasattr(lowerCAmelCase_ , 'image_std' ) )
def lowercase ( self : Optional[Any] ) -> List[str]:
__lowerCAmelCase = self.image_processing_class.from_dict(self.image_processor_dict )
self.assertEqual(image_processor.size , {'shortest_edge': 3_0} )
self.assertEqual(image_processor.crop_size , {'height': 3_0, 'width': 3_0} )
__lowerCAmelCase = self.image_processing_class.from_dict(self.image_processor_dict , size=4_2 , crop_size=8_4 )
self.assertEqual(image_processor.size , {'shortest_edge': 4_2} )
self.assertEqual(image_processor.crop_size , {'height': 8_4, 'width': 8_4} )
def lowercase ( self : List[Any] ) -> str:
pass
def lowercase ( self : Tuple ) -> Tuple:
# Initialize image_processing
__lowerCAmelCase = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
__lowerCAmelCase = prepare_image_inputs(self.image_processor_tester , equal_resolution=lowerCAmelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCAmelCase_ , Image.Image )
# Test not batched input
__lowerCAmelCase = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
# Test batched
__lowerCAmelCase = image_processing(lowerCAmelCase_ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
def lowercase ( self : Tuple ) -> Optional[int]:
# Initialize image_processing
__lowerCAmelCase = self.image_processing_class(**self.image_processor_dict )
# create random numpy tensors
__lowerCAmelCase = prepare_image_inputs(self.image_processor_tester , equal_resolution=lowerCAmelCase_ , numpify=lowerCAmelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCAmelCase_ , np.ndarray )
# Test not batched input
__lowerCAmelCase = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
# Test batched
__lowerCAmelCase = image_processing(lowerCAmelCase_ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
def lowercase ( self : Dict ) -> Optional[Any]:
# Initialize image_processing
__lowerCAmelCase = self.image_processing_class(**self.image_processor_dict )
# create random PyTorch tensors
__lowerCAmelCase = prepare_image_inputs(self.image_processor_tester , equal_resolution=lowerCAmelCase_ , torchify=lowerCAmelCase_ )
for image in image_inputs:
self.assertIsInstance(lowerCAmelCase_ , torch.Tensor )
# Test not batched input
__lowerCAmelCase = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
# Test batched
__lowerCAmelCase = image_processing(lowerCAmelCase_ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size['height'],
self.image_processor_tester.crop_size['width'],
) , )
| 53 |
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 ViTConfig, ViTForImageClassification, ViTImageProcessor, ViTModel
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Union[str, Any] = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : int=False ):
__lowerCAmelCase = []
for i in range(config.num_hidden_layers ):
# encoder layers: output projection, 2 feedforward neural networks and 2 layernorms
rename_keys.append((F"""blocks.{i}.norm1.weight""", F"""vit.encoder.layer.{i}.layernorm_before.weight""") )
rename_keys.append((F"""blocks.{i}.norm1.bias""", F"""vit.encoder.layer.{i}.layernorm_before.bias""") )
rename_keys.append((F"""blocks.{i}.attn.proj.weight""", F"""vit.encoder.layer.{i}.attention.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.attn.proj.bias""", F"""vit.encoder.layer.{i}.attention.output.dense.bias""") )
rename_keys.append((F"""blocks.{i}.norm2.weight""", F"""vit.encoder.layer.{i}.layernorm_after.weight""") )
rename_keys.append((F"""blocks.{i}.norm2.bias""", F"""vit.encoder.layer.{i}.layernorm_after.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.weight""", F"""vit.encoder.layer.{i}.intermediate.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc1.bias""", F"""vit.encoder.layer.{i}.intermediate.dense.bias""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.weight""", F"""vit.encoder.layer.{i}.output.dense.weight""") )
rename_keys.append((F"""blocks.{i}.mlp.fc2.bias""", F"""vit.encoder.layer.{i}.output.dense.bias""") )
# projection layer + position embeddings
rename_keys.extend(
[
('cls_token', 'vit.embeddings.cls_token'),
('patch_embed.proj.weight', 'vit.embeddings.patch_embeddings.projection.weight'),
('patch_embed.proj.bias', 'vit.embeddings.patch_embeddings.projection.bias'),
('pos_embed', 'vit.embeddings.position_embeddings'),
] )
if base_model:
# layernorm + pooler
rename_keys.extend(
[
('norm.weight', 'layernorm.weight'),
('norm.bias', 'layernorm.bias'),
] )
# if just the base model, we should remove "vit" from all keys that start with "vit"
__lowerCAmelCase = [(pair[0], pair[1][4:]) if pair[1].startswith('vit' ) else pair for pair in rename_keys]
else:
# layernorm + classification head
rename_keys.extend(
[
('norm.weight', 'vit.layernorm.weight'),
('norm.bias', 'vit.layernorm.bias'),
('head.weight', 'classifier.weight'),
('head.bias', 'classifier.bias'),
] )
return rename_keys
def a_ ( lowerCAmelCase_ : Any, lowerCAmelCase_ : Tuple, lowerCAmelCase_ : Optional[int]=False ):
for i in range(config.num_hidden_layers ):
if base_model:
__lowerCAmelCase = ''
else:
__lowerCAmelCase = 'vit.'
# read in weights + bias of input projection layer (in timm, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""blocks.{i}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[
: config.hidden_size, :
]
__lowerCAmelCase = in_proj_bias[: config.hidden_size]
__lowerCAmelCase = in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
__lowerCAmelCase = in_proj_bias[
config.hidden_size : config.hidden_size * 2
]
__lowerCAmelCase = in_proj_weight[
-config.hidden_size :, :
]
__lowerCAmelCase = in_proj_bias[-config.hidden_size :]
def a_ ( lowerCAmelCase_ : List[str] ):
__lowerCAmelCase = ['head.weight', 'head.bias']
for k in ignore_keys:
state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int, lowerCAmelCase_ : Union[str, Any] ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Optional[Any]=True ):
__lowerCAmelCase = ViTConfig()
# patch_size
if model_name[-1] == "8":
__lowerCAmelCase = 8
# set labels if required
if not base_model:
__lowerCAmelCase = 1000
__lowerCAmelCase = 'huggingface/label-files'
__lowerCAmelCase = 'imagenet-1k-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
__lowerCAmelCase = idalabel
__lowerCAmelCase = {v: k for k, v in idalabel.items()}
# size of the architecture
if model_name in ["dino_vits8", "dino_vits16"]:
__lowerCAmelCase = 384
__lowerCAmelCase = 1536
__lowerCAmelCase = 12
__lowerCAmelCase = 6
# load original model from torch hub
__lowerCAmelCase = torch.hub.load('facebookresearch/dino:main', lowerCAmelCase_ )
original_model.eval()
# load state_dict of original model, remove and rename some keys
__lowerCAmelCase = original_model.state_dict()
if base_model:
remove_classification_head_(lowerCAmelCase_ )
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_, base_model=lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_q_k_v(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# load HuggingFace model
if base_model:
__lowerCAmelCase = ViTModel(lowerCAmelCase_, add_pooling_layer=lowerCAmelCase_ ).eval()
else:
__lowerCAmelCase = ViTForImageClassification(lowerCAmelCase_ ).eval()
model.load_state_dict(lowerCAmelCase_ )
# Check outputs on an image, prepared by ViTImageProcessor
__lowerCAmelCase = ViTImageProcessor()
__lowerCAmelCase = image_processor(images=prepare_img(), return_tensors='pt' )
__lowerCAmelCase = encoding['pixel_values']
__lowerCAmelCase = model(lowerCAmelCase_ )
if base_model:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert torch.allclose(lowerCAmelCase_, outputs.last_hidden_state[:, 0, :], atol=1E-1 )
else:
__lowerCAmelCase = original_model(lowerCAmelCase_ )
assert logits.shape == outputs.logits.shape
assert torch.allclose(lowerCAmelCase_, outputs.logits, atol=1E-3 )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
print(F"""Saving model {model_name} to {pytorch_dump_folder_path}""" )
model.save_pretrained(lowerCAmelCase_ )
print(F"""Saving image processor to {pytorch_dump_folder_path}""" )
image_processor.save_pretrained(lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : Dict = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='dino_vitb16',
type=str,
help='Name of the model trained with DINO you\'d like to convert.',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--base_model',
action='store_true',
help='Whether to only convert the base model (no projection head weights).',
)
parser.set_defaults(base_model=True)
_snake_case : List[Any] = parser.parse_args()
convert_vit_checkpoint(args.model_name, args.pytorch_dump_folder_path, args.base_model)
| 53 | 1 |
import inspect
import unittest
from transformers import DecisionTransformerConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import DecisionTransformerModel
from transformers.models.decision_transformer.modeling_decision_transformer import (
DECISION_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
)
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : str , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any]=1_3 , lowerCAmelCase_ : Optional[int]=7 , lowerCAmelCase_ : List[str]=6 , lowerCAmelCase_ : str=1_7 , lowerCAmelCase_ : Dict=2_3 , lowerCAmelCase_ : Dict=1_1 , lowerCAmelCase_ : List[str]=True , ) -> str:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = act_dim
__lowerCAmelCase = state_dim
__lowerCAmelCase = hidden_size
__lowerCAmelCase = max_length
__lowerCAmelCase = is_training
def lowercase ( self : Optional[int] ) -> int:
__lowerCAmelCase = floats_tensor((self.batch_size, self.seq_length, self.state_dim) )
__lowerCAmelCase = floats_tensor((self.batch_size, self.seq_length, self.act_dim) )
__lowerCAmelCase = floats_tensor((self.batch_size, self.seq_length, 1) )
__lowerCAmelCase = floats_tensor((self.batch_size, self.seq_length, 1) )
__lowerCAmelCase = ids_tensor((self.batch_size, self.seq_length) , vocab_size=1_0_0_0 )
__lowerCAmelCase = random_attention_mask((self.batch_size, self.seq_length) )
__lowerCAmelCase = self.get_config()
return (
config,
states,
actions,
rewards,
returns_to_go,
timesteps,
attention_mask,
)
def lowercase ( self : str ) -> Union[str, Any]:
return DecisionTransformerConfig(
batch_size=self.batch_size , seq_length=self.seq_length , act_dim=self.act_dim , state_dim=self.state_dim , hidden_size=self.hidden_size , max_length=self.max_length , )
def lowercase ( self : Optional[int] , lowerCAmelCase_ : str , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[Any] , ) -> Dict:
__lowerCAmelCase = DecisionTransformerModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
self.parent.assertEqual(result.state_preds.shape , states.shape )
self.parent.assertEqual(result.action_preds.shape , actions.shape )
self.parent.assertEqual(result.return_preds.shape , returns_to_go.shape )
self.parent.assertEqual(
result.last_hidden_state.shape , (self.batch_size, self.seq_length * 3, self.hidden_size) ) # seq length *3 as there are 3 modelities: states, returns and actions
def lowercase ( self : int ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {
'states': states,
'actions': actions,
'rewards': rewards,
'returns_to_go': returns_to_go,
'timesteps': timesteps,
'attention_mask': attention_mask,
}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (DecisionTransformerModel,) if is_torch_available() else ()
a_ = ()
a_ = {"""feature-extraction""": DecisionTransformerModel} if is_torch_available() else {}
# Ignoring of a failing test from GenerationTesterMixin, as the model does not use inputs_ids
a_ = False
# Ignoring of a failing tests from ModelTesterMixin, as the model does not implement these features
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : int ) -> Optional[Any]:
__lowerCAmelCase = DecisionTransformerModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : Optional[int] ) -> Optional[int]:
self.config_tester.run_common_tests()
def lowercase ( self : Optional[int] ) -> int:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
@slow
def lowercase ( self : Union[str, Any] ) -> Union[str, Any]:
for model_name in DECISION_TRANSFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = DecisionTransformerModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
def lowercase ( self : Optional[Any] ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = [
'states',
'actions',
'rewards',
'returns_to_go',
'timesteps',
'attention_mask',
]
self.assertListEqual(arg_names[: len(lowerCAmelCase_ )] , lowerCAmelCase_ )
@require_torch
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@slow
def lowercase ( self : List[str] ) -> Any:
__lowerCAmelCase = 2 # number of steps of autoregressive prediction we will perform
__lowerCAmelCase = 1_0 # defined by the RL environment, may be normalized
__lowerCAmelCase = DecisionTransformerModel.from_pretrained('edbeeching/decision-transformer-gym-hopper-expert' )
__lowerCAmelCase = model.to(lowerCAmelCase_ )
__lowerCAmelCase = model.config
torch.manual_seed(0 )
__lowerCAmelCase = torch.randn(1 , 1 , config.state_dim ).to(device=lowerCAmelCase_ , dtype=torch.floataa ) # env.reset()
__lowerCAmelCase = torch.tensor(
[[0.24_27_93, -0.28_69_30_74, 0.8_74_26_13], [0.67_81_52_74, -0.08_10_10_85, -0.12_95_21_47]] , device=lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(lowerCAmelCase_ , device=lowerCAmelCase_ , dtype=torch.floataa ).reshape(1 , 1 , 1 )
__lowerCAmelCase = state
__lowerCAmelCase = torch.zeros(1 , 0 , config.act_dim , device=lowerCAmelCase_ , dtype=torch.floataa )
__lowerCAmelCase = torch.zeros(1 , 0 , device=lowerCAmelCase_ , dtype=torch.floataa )
__lowerCAmelCase = torch.tensor(0 , device=lowerCAmelCase_ , dtype=torch.long ).reshape(1 , 1 )
for step in range(lowerCAmelCase_ ):
__lowerCAmelCase = torch.cat([actions, torch.zeros(1 , 1 , config.act_dim , device=lowerCAmelCase_ )] , dim=1 )
__lowerCAmelCase = torch.cat([rewards, torch.zeros(1 , 1 , device=lowerCAmelCase_ )] , dim=1 )
__lowerCAmelCase = torch.ones(1 , states.shape[1] ).to(dtype=torch.long , device=states.device )
with torch.no_grad():
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = model(
states=lowerCAmelCase_ , actions=lowerCAmelCase_ , rewards=lowerCAmelCase_ , returns_to_go=lowerCAmelCase_ , timesteps=lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , return_dict=lowerCAmelCase_ , )
self.assertEqual(action_pred.shape , actions.shape )
self.assertTrue(torch.allclose(action_pred[0, -1] , expected_outputs[step] , atol=1e-4 ) )
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = ( # env.step(action)
torch.randn(1 , 1 , config.state_dim ).to(device=lowerCAmelCase_ , dtype=torch.floataa ),
1.0,
False,
{},
)
__lowerCAmelCase = action_pred[0, -1]
__lowerCAmelCase = torch.cat([states, state] , dim=1 )
__lowerCAmelCase = returns_to_go[0, -1] - reward
__lowerCAmelCase = torch.cat([returns_to_go, pred_return.reshape(1 , 1 , 1 )] , dim=1 )
__lowerCAmelCase = torch.cat(
[timesteps, torch.ones((1, 1) , device=lowerCAmelCase_ , dtype=torch.long ) * (step + 1)] , dim=1 )
| 53 |
import unittest
from diffusers.pipelines.pipeline_utils import is_safetensors_compatible
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Union[str, Any] ) -> List[str]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Any:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
# Removed: 'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> List[str]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : List[Any] ) -> int:
__lowerCAmelCase = [
'safety_checker/pytorch_model.bin',
'safety_checker/model.safetensors',
'vae/diffusion_pytorch_model.bin',
'vae/diffusion_pytorch_model.safetensors',
'text_encoder/pytorch_model.bin',
# Removed: 'text_encoder/model.safetensors',
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ ) )
def lowercase ( self : str ) -> str:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[str]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'unet/diffusion_pytorch_model.bin',
'unet/diffusion_pytorch_model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> Union[str, Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
# Removed: 'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : str ) -> List[Any]:
__lowerCAmelCase = [
'text_encoder/pytorch_model.fp16.bin',
'text_encoder/model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : List[str] ) -> List[Any]:
# pass variant but use the non-variant filenames
__lowerCAmelCase = [
'text_encoder/pytorch_model.bin',
'text_encoder/model.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertTrue(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
def lowercase ( self : Optional[Any] ) -> Optional[Any]:
__lowerCAmelCase = [
'safety_checker/pytorch_model.fp16.bin',
'safety_checker/model.fp16.safetensors',
'vae/diffusion_pytorch_model.fp16.bin',
'vae/diffusion_pytorch_model.fp16.safetensors',
'text_encoder/pytorch_model.fp16.bin',
# 'text_encoder/model.fp16.safetensors',
'unet/diffusion_pytorch_model.fp16.bin',
'unet/diffusion_pytorch_model.fp16.safetensors',
]
__lowerCAmelCase = 'fp16'
self.assertFalse(is_safetensors_compatible(lowerCAmelCase_ , variant=lowerCAmelCase_ ) )
| 53 | 1 |
from math import factorial
def a_ ( lowerCAmelCase_ : int = 20 ):
__lowerCAmelCase = 2 * n # middle entry of odd rows starting at row 3 is the solution for n = 1,
# 2, 3,...
__lowerCAmelCase = n // 2
return int(factorial(lowerCAmelCase_ ) / (factorial(lowerCAmelCase_ ) * factorial(n - k )) )
if __name__ == "__main__":
import sys
if len(sys.argv) == 1:
print(solution(20))
else:
try:
_snake_case : Union[str, Any] = int(sys.argv[1])
print(solution(n))
except ValueError:
print('Invalid entry - please enter a number.')
| 53 |
import math
def a_ ( lowerCAmelCase_ : list, lowerCAmelCase_ : int ):
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = int(math.floor(math.sqrt(lowerCAmelCase_ ) ) )
__lowerCAmelCase = 0
while arr[min(lowerCAmelCase_, lowerCAmelCase_ ) - 1] < x:
__lowerCAmelCase = step
step += int(math.floor(math.sqrt(lowerCAmelCase_ ) ) )
if prev >= n:
return -1
while arr[prev] < x:
__lowerCAmelCase = prev + 1
if prev == min(lowerCAmelCase_, lowerCAmelCase_ ):
return -1
if arr[prev] == x:
return prev
return -1
if __name__ == "__main__":
_snake_case : List[str] = input('Enter numbers separated by a comma:\n').strip()
_snake_case : Optional[Any] = [int(item) for item in user_input.split(',')]
_snake_case : List[str] = int(input('Enter the number to be searched:\n'))
_snake_case : Optional[int] = jump_search(arr, x)
if res == -1:
print('Number not found!')
else:
print(F"""Number {x} is at index {res}""")
| 53 | 1 |
import argparse
import torch
from torch import nn
from transformers import MBartConfig, MBartForConditionalGeneration
def a_ ( lowerCAmelCase_ : Dict ):
__lowerCAmelCase = [
'encoder.version',
'decoder.version',
'model.encoder.version',
'model.decoder.version',
'_float_tensor',
'decoder.output_projection.weight',
]
for k in ignore_keys:
state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Dict ):
__lowerCAmelCase , __lowerCAmelCase = emb.weight.shape
__lowerCAmelCase = nn.Linear(lowerCAmelCase_, lowerCAmelCase_, bias=lowerCAmelCase_ )
__lowerCAmelCase = emb.weight.data
return lin_layer
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : Any="facebook/mbart-large-en-ro", lowerCAmelCase_ : List[str]=False, lowerCAmelCase_ : int=False ):
__lowerCAmelCase = torch.load(lowerCAmelCase_, map_location='cpu' )['model']
remove_ignore_keys_(lowerCAmelCase_ )
__lowerCAmelCase = state_dict['encoder.embed_tokens.weight'].shape[0]
__lowerCAmelCase = MBartConfig.from_pretrained(lowerCAmelCase_, vocab_size=lowerCAmelCase_ )
if mbart_aa and finetuned:
__lowerCAmelCase = 'relu'
__lowerCAmelCase = state_dict['decoder.embed_tokens.weight']
__lowerCAmelCase = MBartForConditionalGeneration(lowerCAmelCase_ )
model.model.load_state_dict(lowerCAmelCase_ )
if finetuned:
__lowerCAmelCase = make_linear_from_emb(model.model.shared )
return model
if __name__ == "__main__":
_snake_case : int = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'fairseq_path', type=str, help='bart.large, bart.large.cnn or a path to a model.pt on local filesystem.'
)
parser.add_argument('pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model.')
parser.add_argument(
'--hf_config',
default='facebook/mbart-large-cc25',
type=str,
help='Which huggingface architecture to use: mbart-large',
)
parser.add_argument('--mbart_50', action='store_true', help='whether the model is mMART-50 checkpoint')
parser.add_argument('--finetuned', action='store_true', help='whether the model is a fine-tuned checkpoint')
_snake_case : List[str] = parser.parse_args()
_snake_case : Dict = convert_fairseq_mbart_checkpoint_from_disk(
args.fairseq_path, hf_config_path=args.hf_config, finetuned=args.finetuned, mbart_aa=args.mbart_aa
)
model.save_pretrained(args.pytorch_dump_folder_path)
| 53 |
import argparse
import torch
from transformers import RemBertConfig, RemBertModel, load_tf_weights_in_rembert
from transformers.utils import logging
logging.set_verbosity_info()
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : List[Any], lowerCAmelCase_ : str ):
# Initialise PyTorch model
__lowerCAmelCase = RemBertConfig.from_json_file(lowerCAmelCase_ )
print('Building PyTorch model from configuration: {}'.format(str(lowerCAmelCase_ ) ) )
__lowerCAmelCase = RemBertModel(lowerCAmelCase_ )
# Load weights from tf checkpoint
load_tf_weights_in_rembert(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
# Save pytorch-model
print('Save PyTorch model to {}'.format(lowerCAmelCase_ ) )
torch.save(model.state_dict(), lowerCAmelCase_ )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--tf_checkpoint_path', default=None, type=str, required=True, help='Path to the TensorFlow checkpoint path.'
)
parser.add_argument(
'--rembert_config_file',
default=None,
type=str,
required=True,
help=(
'The config json file corresponding to the pre-trained RemBERT model. \n'
'This specifies the model architecture.'
),
)
parser.add_argument(
'--pytorch_dump_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_rembert_tf_checkpoint_to_pytorch(args.tf_checkpoint_path, args.rembert_config_file, args.pytorch_dump_path)
| 53 | 1 |
import unittest
import numpy as np
from transformers import RobertaPreLayerNormConfig, is_flax_available
from transformers.testing_utils import require_flax, slow
from ...test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_flax_available():
import jax.numpy as jnp
from transformers.models.roberta_prelayernorm.modeling_flax_roberta_prelayernorm import (
FlaxRobertaPreLayerNormForCausalLM,
FlaxRobertaPreLayerNormForMaskedLM,
FlaxRobertaPreLayerNormForMultipleChoice,
FlaxRobertaPreLayerNormForQuestionAnswering,
FlaxRobertaPreLayerNormForSequenceClassification,
FlaxRobertaPreLayerNormForTokenClassification,
FlaxRobertaPreLayerNormModel,
)
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : int , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : List[Any]=1_3 , lowerCAmelCase_ : List[str]=7 , lowerCAmelCase_ : Optional[int]=True , lowerCAmelCase_ : Dict=True , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : str=9_9 , lowerCAmelCase_ : Union[str, Any]=3_2 , lowerCAmelCase_ : List[Any]=5 , lowerCAmelCase_ : Any=4 , lowerCAmelCase_ : List[str]=3_7 , lowerCAmelCase_ : Dict="gelu" , lowerCAmelCase_ : Any=0.1 , lowerCAmelCase_ : List[Any]=0.1 , lowerCAmelCase_ : str=5_1_2 , lowerCAmelCase_ : Optional[int]=1_6 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : Union[str, Any]=0.02 , lowerCAmelCase_ : List[str]=4 , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_attention_mask
__lowerCAmelCase = use_token_type_ids
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = type_vocab_size
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = num_choices
def lowercase ( self : int ) -> List[Any]:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = None
if self.use_attention_mask:
__lowerCAmelCase = random_attention_mask([self.batch_size, self.seq_length] )
__lowerCAmelCase = None
if self.use_token_type_ids:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
__lowerCAmelCase = RobertaPreLayerNormConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , )
return config, input_ids, token_type_ids, attention_mask
def lowercase ( self : Tuple ) -> Union[str, Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'input_ids': input_ids, 'token_type_ids': token_type_ids, 'attention_mask': attention_mask}
return config, inputs_dict
def lowercase ( self : Any ) -> List[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = True
__lowerCAmelCase = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
encoder_hidden_states,
encoder_attention_mask,
)
@require_flax
# Copied from tests.models.roberta.test_modelling_flax_roberta.FlaxRobertaPreLayerNormModelTest with ROBERTA->ROBERTA_PRELAYERNORM,Roberta->RobertaPreLayerNorm,roberta-base->andreasmadsen/efficient_mlm_m0.40
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = True
a_ = (
(
FlaxRobertaPreLayerNormModel,
FlaxRobertaPreLayerNormForCausalLM,
FlaxRobertaPreLayerNormForMaskedLM,
FlaxRobertaPreLayerNormForSequenceClassification,
FlaxRobertaPreLayerNormForTokenClassification,
FlaxRobertaPreLayerNormForMultipleChoice,
FlaxRobertaPreLayerNormForQuestionAnswering,
)
if is_flax_available()
else ()
)
def lowercase ( self : Tuple ) -> Optional[Any]:
__lowerCAmelCase = FlaxRobertaPreLayerNormModelTester(self )
@slow
def lowercase ( self : Tuple ) -> List[Any]:
for model_class_name in self.all_model_classes:
__lowerCAmelCase = model_class_name.from_pretrained('andreasmadsen/efficient_mlm_m0.40' , from_pt=lowerCAmelCase_ )
__lowerCAmelCase = model(np.ones((1, 1) ) )
self.assertIsNotNone(lowerCAmelCase_ )
@require_flax
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@slow
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
__lowerCAmelCase = FlaxRobertaPreLayerNormForMaskedLM.from_pretrained('andreasmadsen/efficient_mlm_m0.40' , from_pt=lowerCAmelCase_ )
__lowerCAmelCase = np.array([[0, 3_1_4_1_4, 2_3_2, 3_2_8, 7_4_0, 1_1_4_0, 1_2_6_9_5, 6_9, 4_6_0_7_8, 1_5_8_8, 2]] , dtype=jnp.intaa )
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = [1, 1_1, 5_0_2_6_5]
self.assertEqual(list(output.shape ) , lowerCAmelCase_ )
# compare the actual values for a slice.
__lowerCAmelCase = np.array(
[[[40.48_80, 18.01_99, -5.23_67], [-1.88_77, -4.08_85, 10.70_85], [-2.26_13, -5.61_10, 7.26_65]]] , dtype=np.floataa )
self.assertTrue(np.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : Tuple ) -> Dict:
__lowerCAmelCase = FlaxRobertaPreLayerNormModel.from_pretrained('andreasmadsen/efficient_mlm_m0.40' , from_pt=lowerCAmelCase_ )
__lowerCAmelCase = np.array([[0, 3_1_4_1_4, 2_3_2, 3_2_8, 7_4_0, 1_1_4_0, 1_2_6_9_5, 6_9, 4_6_0_7_8, 1_5_8_8, 2]] , dtype=jnp.intaa )
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
# compare the actual values for a slice.
__lowerCAmelCase = np.array(
[[[0.02_08, -0.03_56, 0.02_37], [-0.15_69, -0.04_11, -0.26_26], [0.18_79, 0.01_25, -0.00_89]]] , dtype=np.floataa )
self.assertTrue(np.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 |
import argparse
import json
import pickle
from pathlib import Path
import requests
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import MaskFormerConfig, MaskFormerForInstanceSegmentation, MaskFormerImageProcessor, SwinConfig
from transformers.utils import logging
logging.set_verbosity_info()
_snake_case : Any = logging.get_logger(__name__)
def a_ ( lowerCAmelCase_ : str ):
__lowerCAmelCase = SwinConfig.from_pretrained(
'microsoft/swin-tiny-patch4-window7-224', out_features=['stage1', 'stage2', 'stage3', 'stage4'] )
__lowerCAmelCase = MaskFormerConfig(backbone_config=lowerCAmelCase_ )
__lowerCAmelCase = 'huggingface/label-files'
if "ade20k-full" in model_name:
# this should be ok
__lowerCAmelCase = 847
__lowerCAmelCase = 'maskformer-ade20k-full-id2label.json'
elif "ade" in model_name:
# this should be ok
__lowerCAmelCase = 150
__lowerCAmelCase = 'ade20k-id2label.json'
elif "coco-stuff" in model_name:
# this should be ok
__lowerCAmelCase = 171
__lowerCAmelCase = 'maskformer-coco-stuff-id2label.json'
elif "coco" in model_name:
# TODO
__lowerCAmelCase = 133
__lowerCAmelCase = 'coco-panoptic-id2label.json'
elif "cityscapes" in model_name:
# this should be ok
__lowerCAmelCase = 19
__lowerCAmelCase = 'cityscapes-id2label.json'
elif "vistas" in model_name:
# this should be ok
__lowerCAmelCase = 65
__lowerCAmelCase = 'mapillary-vistas-id2label.json'
__lowerCAmelCase = json.load(open(hf_hub_download(lowerCAmelCase_, lowerCAmelCase_, repo_type='dataset' ), 'r' ) )
__lowerCAmelCase = {int(lowerCAmelCase_ ): v for k, v in idalabel.items()}
return config
def a_ ( lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = []
# stem
# fmt: off
rename_keys.append(('backbone.patch_embed.proj.weight', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.weight') )
rename_keys.append(('backbone.patch_embed.proj.bias', 'model.pixel_level_module.encoder.model.embeddings.patch_embeddings.projection.bias') )
rename_keys.append(('backbone.patch_embed.norm.weight', 'model.pixel_level_module.encoder.model.embeddings.norm.weight') )
rename_keys.append(('backbone.patch_embed.norm.bias', 'model.pixel_level_module.encoder.model.embeddings.norm.bias') )
# stages
for i in range(len(config.backbone_config.depths ) ):
for j in range(config.backbone_config.depths[i] ):
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_before.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_bias_table""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_bias_table""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.relative_position_index""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.self.relative_position_index""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.attn.proj.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.attention.output.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.norm2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.layernorm_after.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc1.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.intermediate.dense.bias""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.weight""") )
rename_keys.append((F"""backbone.layers.{i}.blocks.{j}.mlp.fc2.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.blocks.{j}.output.dense.bias""") )
if i < 3:
rename_keys.append((F"""backbone.layers.{i}.downsample.reduction.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.reduction.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.weight""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.weight""") )
rename_keys.append((F"""backbone.layers.{i}.downsample.norm.bias""", F"""model.pixel_level_module.encoder.model.encoder.layers.{i}.downsample.norm.bias""") )
rename_keys.append((F"""backbone.norm{i}.weight""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.weight""") )
rename_keys.append((F"""backbone.norm{i}.bias""", F"""model.pixel_level_module.encoder.hidden_states_norms.{i}.bias""") )
# FPN
rename_keys.append(('sem_seg_head.layer_4.weight', 'model.pixel_level_module.decoder.fpn.stem.0.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.weight', 'model.pixel_level_module.decoder.fpn.stem.1.weight') )
rename_keys.append(('sem_seg_head.layer_4.norm.bias', 'model.pixel_level_module.decoder.fpn.stem.1.bias') )
for source_index, target_index in zip(range(3, 0, -1 ), range(0, 3 ) ):
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.0.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.weight""") )
rename_keys.append((F"""sem_seg_head.adapter_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.proj.1.bias""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.0.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.weight""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.weight""") )
rename_keys.append((F"""sem_seg_head.layer_{source_index}.norm.bias""", F"""model.pixel_level_module.decoder.fpn.layers.{target_index}.block.1.bias""") )
rename_keys.append(('sem_seg_head.mask_features.weight', 'model.pixel_level_module.decoder.mask_projection.weight') )
rename_keys.append(('sem_seg_head.mask_features.bias', 'model.pixel_level_module.decoder.mask_projection.bias') )
# Transformer decoder
for idx in range(config.decoder_config.decoder_layers ):
# self-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn.out_proj.bias""") )
# cross-attention out projection
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.out_proj.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn.out_proj.bias""") )
# MLP 1
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc1.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear1.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc1.bias""") )
# MLP 2
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.weight""", F"""model.transformer_module.decoder.layers.{idx}.fc2.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.linear2.bias""", F"""model.transformer_module.decoder.layers.{idx}.fc2.bias""") )
# layernorm 1 (self-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.weight""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm1.bias""", F"""model.transformer_module.decoder.layers.{idx}.self_attn_layer_norm.bias""") )
# layernorm 2 (cross-attention layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.weight""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm2.bias""", F"""model.transformer_module.decoder.layers.{idx}.encoder_attn_layer_norm.bias""") )
# layernorm 3 (final layernorm)
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.weight""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.norm3.bias""", F"""model.transformer_module.decoder.layers.{idx}.final_layer_norm.bias""") )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.weight', 'model.transformer_module.decoder.layernorm.weight') )
rename_keys.append(('sem_seg_head.predictor.transformer.decoder.norm.bias', 'model.transformer_module.decoder.layernorm.bias') )
# heads on top
rename_keys.append(('sem_seg_head.predictor.query_embed.weight', 'model.transformer_module.queries_embedder.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.weight', 'model.transformer_module.input_projection.weight') )
rename_keys.append(('sem_seg_head.predictor.input_proj.bias', 'model.transformer_module.input_projection.bias') )
rename_keys.append(('sem_seg_head.predictor.class_embed.weight', 'class_predictor.weight') )
rename_keys.append(('sem_seg_head.predictor.class_embed.bias', 'class_predictor.bias') )
for i in range(3 ):
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.weight""", F"""mask_embedder.{i}.0.weight""") )
rename_keys.append((F"""sem_seg_head.predictor.mask_embed.layers.{i}.bias""", F"""mask_embedder.{i}.0.bias""") )
# fmt: on
return rename_keys
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = dct.pop(lowerCAmelCase_ )
__lowerCAmelCase = val
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : int ):
__lowerCAmelCase = [int(backbone_config.embed_dim * 2**i ) for i in range(len(backbone_config.depths ) )]
for i in range(len(backbone_config.depths ) ):
__lowerCAmelCase = num_features[i]
for j in range(backbone_config.depths[i] ):
# fmt: off
# read in weights + bias of input projection layer (in original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.weight""" )
__lowerCAmelCase = state_dict.pop(F"""backbone.layers.{i}.blocks.{j}.attn.qkv.bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[:dim, :]
__lowerCAmelCase = in_proj_bias[: dim]
__lowerCAmelCase = in_proj_weight[
dim : dim * 2, :
]
__lowerCAmelCase = in_proj_bias[
dim : dim * 2
]
__lowerCAmelCase = in_proj_weight[
-dim :, :
]
__lowerCAmelCase = in_proj_bias[-dim :]
# fmt: on
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : Dict ):
# fmt: off
__lowerCAmelCase = config.decoder_config.hidden_size
for idx in range(config.decoder_config.decoder_layers ):
# read in weights + bias of self-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.self_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# read in weights + bias of cross-attention input projection layer (in the original implementation, this is a single matrix + bias)
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_weight""" )
__lowerCAmelCase = state_dict.pop(F"""sem_seg_head.predictor.transformer.decoder.layers.{idx}.multihead_attn.in_proj_bias""" )
# next, add query, keys and values (in that order) to the state dict
__lowerCAmelCase = in_proj_weight[: hidden_size, :]
__lowerCAmelCase = in_proj_bias[:config.hidden_size]
__lowerCAmelCase = in_proj_weight[hidden_size : hidden_size * 2, :]
__lowerCAmelCase = in_proj_bias[hidden_size : hidden_size * 2]
__lowerCAmelCase = in_proj_weight[-hidden_size :, :]
__lowerCAmelCase = in_proj_bias[-hidden_size :]
# fmt: on
def a_ ( ):
__lowerCAmelCase = 'http://images.cocodataset.org/val2017/000000039769.jpg'
__lowerCAmelCase = Image.open(requests.get(lowerCAmelCase_, stream=lowerCAmelCase_ ).raw )
return im
@torch.no_grad()
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : str, lowerCAmelCase_ : bool = False ):
__lowerCAmelCase = get_maskformer_config(lowerCAmelCase_ )
# load original state_dict
with open(lowerCAmelCase_, 'rb' ) as f:
__lowerCAmelCase = pickle.load(lowerCAmelCase_ )
__lowerCAmelCase = data['model']
# for name, param in state_dict.items():
# print(name, param.shape)
# rename keys
__lowerCAmelCase = create_rename_keys(lowerCAmelCase_ )
for src, dest in rename_keys:
rename_key(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
read_in_swin_q_k_v(lowerCAmelCase_, config.backbone_config )
read_in_decoder_q_k_v(lowerCAmelCase_, lowerCAmelCase_ )
# update to torch tensors
for key, value in state_dict.items():
__lowerCAmelCase = torch.from_numpy(lowerCAmelCase_ )
# load 🤗 model
__lowerCAmelCase = MaskFormerForInstanceSegmentation(lowerCAmelCase_ )
model.eval()
for name, param in model.named_parameters():
print(lowerCAmelCase_, param.shape )
__lowerCAmelCase , __lowerCAmelCase = model.load_state_dict(lowerCAmelCase_, strict=lowerCAmelCase_ )
assert missing_keys == [
"model.pixel_level_module.encoder.model.layernorm.weight",
"model.pixel_level_module.encoder.model.layernorm.bias",
]
assert len(lowerCAmelCase_ ) == 0, F"""Unexpected keys: {unexpected_keys}"""
# verify results
__lowerCAmelCase = prepare_img()
if "vistas" in model_name:
__lowerCAmelCase = 65
elif "cityscapes" in model_name:
__lowerCAmelCase = 6_5535
else:
__lowerCAmelCase = 255
__lowerCAmelCase = True if 'ade' in model_name else False
__lowerCAmelCase = MaskFormerImageProcessor(ignore_index=lowerCAmelCase_, reduce_labels=lowerCAmelCase_ )
__lowerCAmelCase = image_processor(lowerCAmelCase_, return_tensors='pt' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
print('Logits:', outputs.class_queries_logits[0, :3, :3] )
if model_name == "maskformer-swin-tiny-ade":
__lowerCAmelCase = torch.tensor(
[[3.6353, -4.4770, -2.6065], [0.5081, -4.2394, -3.5343], [2.1909, -5.0353, -1.9323]] )
assert torch.allclose(outputs.class_queries_logits[0, :3, :3], lowerCAmelCase_, atol=1E-4 )
print('Looks ok!' )
if pytorch_dump_folder_path is not None:
print(F"""Saving model and image processor to {pytorch_dump_folder_path}""" )
Path(lowerCAmelCase_ ).mkdir(exist_ok=lowerCAmelCase_ )
model.save_pretrained(lowerCAmelCase_ )
image_processor.save_pretrained(lowerCAmelCase_ )
if push_to_hub:
print('Pushing model and image processor to the hub...' )
model.push_to_hub(F"""nielsr/{model_name}""" )
image_processor.push_to_hub(F"""nielsr/{model_name}""" )
if __name__ == "__main__":
_snake_case : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--model_name',
default='maskformer-swin-tiny-ade',
type=str,
help=('Name of the MaskFormer model you\'d like to convert',),
)
parser.add_argument(
'--checkpoint_path',
default='/Users/nielsrogge/Documents/MaskFormer_checkpoints/MaskFormer-Swin-tiny-ADE20k/model.pkl',
type=str,
help='Path to the original state dict (.pth file).',
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the output PyTorch model directory.'
)
parser.add_argument(
'--push_to_hub', action='store_true', help='Whether or not to push the converted model to the 🤗 hub.'
)
_snake_case : List[str] = parser.parse_args()
convert_maskformer_checkpoint(
args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub
)
| 53 | 1 |
from __future__ import annotations
def a_ ( lowerCAmelCase_ : list[float] ):
if len(lowerCAmelCase_ ) < 2:
raise ValueError('Monogons and Digons are not polygons in the Euclidean space' )
if any(i <= 0 for i in nums ):
raise ValueError('All values must be greater than 0' )
__lowerCAmelCase = nums.copy()
copy_nums.sort()
return copy_nums[-1] < sum(copy_nums[:-1] )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 |
import json
import logging
import os
import re
import sys
from dataclasses import dataclass, field
from typing import Any, Dict, List, Optional, Union
import datasets
import numpy as np
import torch
import torchaudio
from packaging import version
from torch import nn
import transformers
from transformers import (
HfArgumentParser,
Trainer,
TrainingArguments,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaForCTC,
WavaVecaProcessor,
is_apex_available,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
if is_apex_available():
from apex import amp
if version.parse(version.parse(torch.__version__).base_version) >= version.parse('1.6'):
_snake_case : List[Any] = True
from torch.cuda.amp import autocast
_snake_case : Dict = logging.getLogger(__name__)
def a_ ( lowerCAmelCase_ : str=None, lowerCAmelCase_ : str=None ):
return field(default_factory=lambda: default, metadata=lowerCAmelCase_ )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
metadata={"""help""": """Path to pretrained model or model identifier from huggingface.co/models"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to freeze the feature extractor layers of the model."""} )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout ratio for the attention probabilities."""} )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout ratio for activations inside the fully connected layer."""} )
a_ = field(
default=0.1 , metadata={
"""help""": """The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler."""
} , )
a_ = field(
default=0.1 , metadata={"""help""": """The dropout probabilitiy for all 1D convolutional layers in feature extractor."""} , )
a_ = field(
default=0.05 , metadata={
"""help""": (
"""Propability of each feature vector along the time axis to be chosen as the start of the vector"""
"""span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"""
"""vectors will be masked along the time axis. This is only relevant if ``apply_spec_augment is True``."""
)
} , )
a_ = field(default=0.0 , metadata={"""help""": """The LayerDrop probability."""} )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(
default="""train+validation""" , metadata={
"""help""": """The name of the training data set split to use (via the datasets library). Defaults to 'train'"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached preprocessed datasets or not."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""For debugging purposes or quicker training, truncate the number of training examples to this """
"""value if set."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""For debugging purposes or quicker training, truncate the number of validation examples to this """
"""value if set."""
)
} , )
a_ = list_field(
default=[""",""", """?""", """.""", """!""", """-""", """;""", """:""", """\"\"""", """%""", """'""", """\"""", """�"""] , metadata={"""help""": """A list of characters to remove from the transcripts."""} , )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = 42
a_ = True
a_ = None
a_ = None
a_ = None
a_ = None
def __call__( self : int , lowerCAmelCase_ : List[Dict[str, Union[List[int], torch.Tensor]]] ) -> Dict[str, torch.Tensor]:
# split inputs and labels since they have to be of different lenghts and need
# different padding methods
__lowerCAmelCase = [{'input_values': feature['input_values']} for feature in features]
__lowerCAmelCase = [{'input_ids': feature['labels']} for feature in features]
__lowerCAmelCase = self.processor.pad(
lowerCAmelCase_ , padding=self.padding , max_length=self.max_length , pad_to_multiple_of=self.pad_to_multiple_of , return_tensors='pt' , )
__lowerCAmelCase = self.processor.pad(
labels=lowerCAmelCase_ , padding=self.padding , max_length=self.max_length_labels , pad_to_multiple_of=self.pad_to_multiple_of_labels , return_tensors='pt' , )
# replace padding with -100 to ignore loss correctly
__lowerCAmelCase = labels_batch['input_ids'].masked_fill(labels_batch.attention_mask.ne(1 ) , -1_0_0 )
__lowerCAmelCase = labels
return batch
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Tuple , lowerCAmelCase_ : nn.Module , lowerCAmelCase_ : Dict[str, Union[torch.Tensor, Any]] ) -> torch.Tensor:
model.train()
__lowerCAmelCase = self._prepare_inputs(lowerCAmelCase_ )
if self.use_amp:
with autocast():
__lowerCAmelCase = self.compute_loss(lowerCAmelCase_ , lowerCAmelCase_ )
else:
__lowerCAmelCase = self.compute_loss(lowerCAmelCase_ , lowerCAmelCase_ )
if self.args.n_gpu > 1:
if model.module.config.ctc_loss_reduction == "mean":
__lowerCAmelCase = loss.mean()
elif model.module.config.ctc_loss_reduction == "sum":
__lowerCAmelCase = loss.sum() / (inputs['labels'] >= 0).sum()
else:
raise ValueError(f"""{model.config.ctc_loss_reduction} is not valid. Choose one of ['mean', 'sum']""" )
if self.args.gradient_accumulation_steps > 1:
__lowerCAmelCase = loss / self.args.gradient_accumulation_steps
if self.use_amp:
self.scaler.scale(lowerCAmelCase_ ).backward()
elif self.use_apex:
with amp.scale_loss(lowerCAmelCase_ , self.optimizer ) as scaled_loss:
scaled_loss.backward()
elif self.deepspeed:
self.deepspeed.backward(lowerCAmelCase_ )
else:
loss.backward()
return loss.detach()
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets:
__lowerCAmelCase = datasets.load_dataset(
'common_voice', data_args.dataset_config_name, split=data_args.train_split_name )
__lowerCAmelCase = datasets.load_dataset('common_voice', data_args.dataset_config_name, split='test' )
# Create and save tokenizer
__lowerCAmelCase = F"""[{"".join(data_args.chars_to_ignore )}]"""
def remove_special_characters(lowerCAmelCase_ : Any ):
__lowerCAmelCase = re.sub(lowerCAmelCase_, '', batch['sentence'] ).lower() + ' '
return batch
__lowerCAmelCase = train_dataset.map(lowerCAmelCase_, remove_columns=['sentence'] )
__lowerCAmelCase = eval_dataset.map(lowerCAmelCase_, remove_columns=['sentence'] )
def extract_all_chars(lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = ' '.join(batch['text'] )
__lowerCAmelCase = list(set(lowerCAmelCase_ ) )
return {"vocab": [vocab], "all_text": [all_text]}
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, batched=lowerCAmelCase_, batch_size=-1, keep_in_memory=lowerCAmelCase_, remove_columns=train_dataset.column_names, )
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, batched=lowerCAmelCase_, batch_size=-1, keep_in_memory=lowerCAmelCase_, remove_columns=eval_dataset.column_names, )
__lowerCAmelCase = list(set(vocab_train['vocab'][0] ) | set(vocab_test['vocab'][0] ) )
__lowerCAmelCase = {v: k for k, v in enumerate(lowerCAmelCase_ )}
__lowerCAmelCase = vocab_dict[' ']
del vocab_dict[" "]
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = len(lowerCAmelCase_ )
with open('vocab.json', 'w' ) as vocab_file:
json.dump(lowerCAmelCase_, lowerCAmelCase_ )
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = WavaVecaCTCTokenizer(
'vocab.json', unk_token='[UNK]', pad_token='[PAD]', word_delimiter_token='|', )
__lowerCAmelCase = WavaVecaFeatureExtractor(
feature_size=1, sampling_rate=1_6000, padding_value=0.0, do_normalize=lowerCAmelCase_, return_attention_mask=lowerCAmelCase_ )
__lowerCAmelCase = WavaVecaProcessor(feature_extractor=lowerCAmelCase_, tokenizer=lowerCAmelCase_ )
__lowerCAmelCase = WavaVecaForCTC.from_pretrained(
model_args.model_name_or_path, cache_dir=model_args.cache_dir, activation_dropout=model_args.activation_dropout, attention_dropout=model_args.attention_dropout, hidden_dropout=model_args.hidden_dropout, feat_proj_dropout=model_args.feat_proj_dropout, mask_time_prob=model_args.mask_time_prob, gradient_checkpointing=training_args.gradient_checkpointing, layerdrop=model_args.layerdrop, ctc_loss_reduction='mean', pad_token_id=processor.tokenizer.pad_token_id, vocab_size=len(processor.tokenizer ), )
if data_args.max_train_samples is not None:
__lowerCAmelCase = min(len(lowerCAmelCase_ ), data_args.max_train_samples )
__lowerCAmelCase = train_dataset.select(range(lowerCAmelCase_ ) )
if data_args.max_val_samples is not None:
__lowerCAmelCase = eval_dataset.select(range(data_args.max_val_samples ) )
__lowerCAmelCase = torchaudio.transforms.Resample(4_8000, 1_6000 )
# Preprocessing the datasets.
# We need to read the aduio files as arrays and tokenize the targets.
def speech_file_to_array_fn(lowerCAmelCase_ : int ):
__lowerCAmelCase , __lowerCAmelCase = torchaudio.load(batch['path'] )
__lowerCAmelCase = resampler(lowerCAmelCase_ ).squeeze().numpy()
__lowerCAmelCase = 1_6000
__lowerCAmelCase = batch['text']
return batch
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, remove_columns=train_dataset.column_names, num_proc=data_args.preprocessing_num_workers, )
__lowerCAmelCase = eval_dataset.map(
lowerCAmelCase_, remove_columns=eval_dataset.column_names, num_proc=data_args.preprocessing_num_workers, )
def prepare_dataset(lowerCAmelCase_ : Union[str, Any] ):
# check that all files have the correct sampling rate
assert (
len(set(batch['sampling_rate'] ) ) == 1
), F"""Make sure all inputs have the same sampling rate of {processor.feature_extractor.sampling_rate}."""
__lowerCAmelCase = processor(
audio=batch['speech'], text=batch['target_text'], sampling_rate=batch['sampling_rate'][0] )
batch.update(lowerCAmelCase_ )
return batch
__lowerCAmelCase = train_dataset.map(
lowerCAmelCase_, remove_columns=train_dataset.column_names, batch_size=training_args.per_device_train_batch_size, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, )
__lowerCAmelCase = eval_dataset.map(
lowerCAmelCase_, remove_columns=eval_dataset.column_names, batch_size=training_args.per_device_train_batch_size, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, )
# Metric
__lowerCAmelCase = datasets.load_metric('wer' )
def compute_metrics(lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = pred.predictions
__lowerCAmelCase = np.argmax(lowerCAmelCase_, axis=-1 )
__lowerCAmelCase = processor.tokenizer.pad_token_id
__lowerCAmelCase = processor.batch_decode(lowerCAmelCase_ )
# we do not want to group tokens when computing the metrics
__lowerCAmelCase = processor.batch_decode(pred.label_ids, group_tokens=lowerCAmelCase_ )
__lowerCAmelCase = wer_metric.compute(predictions=lowerCAmelCase_, references=lowerCAmelCase_ )
return {"wer": wer}
if model_args.freeze_feature_extractor:
model.freeze_feature_extractor()
# Data collator
__lowerCAmelCase = DataCollatorCTCWithPadding(processor=lowerCAmelCase_, padding=lowerCAmelCase_ )
# Initialize our Trainer
__lowerCAmelCase = CTCTrainer(
model=lowerCAmelCase_, data_collator=lowerCAmelCase_, args=lowerCAmelCase_, compute_metrics=lowerCAmelCase_, train_dataset=train_dataset if training_args.do_train else None, eval_dataset=eval_dataset if training_args.do_eval else None, tokenizer=processor.feature_extractor, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
# Save the feature_extractor and the tokenizer
if is_main_process(training_args.local_rank ):
processor.save_pretrained(training_args.output_dir )
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model()
__lowerCAmelCase = train_result.metrics
__lowerCAmelCase = (
data_args.max_train_samples if data_args.max_train_samples is not None else len(lowerCAmelCase_ )
)
__lowerCAmelCase = min(lowerCAmelCase_, len(lowerCAmelCase_ ) )
trainer.log_metrics('train', lowerCAmelCase_ )
trainer.save_metrics('train', lowerCAmelCase_ )
trainer.save_state()
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = data_args.max_val_samples if data_args.max_val_samples is not None else len(lowerCAmelCase_ )
__lowerCAmelCase = min(lowerCAmelCase_, len(lowerCAmelCase_ ) )
trainer.log_metrics('eval', lowerCAmelCase_ )
trainer.save_metrics('eval', lowerCAmelCase_ )
return results
if __name__ == "__main__":
main()
| 53 | 1 |
# Algorithm for the pigeonhole sorting
def a_ ( lowerCAmelCase_ : Optional[Any] ):
__lowerCAmelCase = min(lowerCAmelCase_ ) # min() finds the minimum value
__lowerCAmelCase = max(lowerCAmelCase_ ) # max() finds the maximum value
__lowerCAmelCase = max_val - min_val + 1 # size is difference of max and min values plus one
# list of pigeonholes of size equal to the variable size
__lowerCAmelCase = [0] * size
# Populate the pigeonholes.
for x in a:
assert isinstance(lowerCAmelCase_, lowerCAmelCase_ ), "integers only please"
holes[x - min_val] += 1
# Putting the elements back into the array in an order.
__lowerCAmelCase = 0
for count in range(lowerCAmelCase_ ):
while holes[count] > 0:
holes[count] -= 1
__lowerCAmelCase = count + min_val
i += 1
def a_ ( ):
__lowerCAmelCase = [8, 3, 2, 7, 4, 6, 8]
pigeonhole_sort(lowerCAmelCase_ )
print('Sorted order is:', ' '.join(lowerCAmelCase_ ) )
if __name__ == "__main__":
main()
| 53 |
import json
from typing import List, Optional, Tuple
from tokenizers import normalizers
from ....tokenization_utils_fast import PreTrainedTokenizerFast
from ....utils import logging
from .tokenization_retribert import RetriBertTokenizer
_snake_case : Any = logging.get_logger(__name__)
_snake_case : int = {'vocab_file': 'vocab.txt', 'tokenizer_file': 'tokenizer.json'}
_snake_case : Optional[Any] = {
'vocab_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/vocab.txt'
),
},
'tokenizer_file': {
'yjernite/retribert-base-uncased': (
'https://huggingface.co/yjernite/retribert-base-uncased/resolve/main/tokenizer.json'
),
},
}
_snake_case : str = {
'yjernite/retribert-base-uncased': 512,
}
_snake_case : Optional[int] = {
'yjernite/retribert-base-uncased': {'do_lower_case': True},
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = PRETRAINED_VOCAB_FILES_MAP
a_ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
a_ = PRETRAINED_INIT_CONFIGURATION
a_ = RetriBertTokenizer
a_ = ["""input_ids""", """attention_mask"""]
def __init__( self : Dict , lowerCAmelCase_ : Optional[Any]=None , lowerCAmelCase_ : Tuple=None , lowerCAmelCase_ : List[str]=True , lowerCAmelCase_ : str="[UNK]" , lowerCAmelCase_ : Optional[Any]="[SEP]" , lowerCAmelCase_ : List[str]="[PAD]" , lowerCAmelCase_ : Optional[int]="[CLS]" , lowerCAmelCase_ : List[Any]="[MASK]" , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : List[str]=None , **lowerCAmelCase_ : List[Any] , ) -> Dict:
super().__init__(
lowerCAmelCase_ , tokenizer_file=lowerCAmelCase_ , do_lower_case=lowerCAmelCase_ , unk_token=lowerCAmelCase_ , sep_token=lowerCAmelCase_ , pad_token=lowerCAmelCase_ , cls_token=lowerCAmelCase_ , mask_token=lowerCAmelCase_ , tokenize_chinese_chars=lowerCAmelCase_ , strip_accents=lowerCAmelCase_ , **lowerCAmelCase_ , )
__lowerCAmelCase = json.loads(self.backend_tokenizer.normalizer.__getstate__() )
if (
normalizer_state.get('lowercase' , lowerCAmelCase_ ) != do_lower_case
or normalizer_state.get('strip_accents' , lowerCAmelCase_ ) != strip_accents
or normalizer_state.get('handle_chinese_chars' , lowerCAmelCase_ ) != tokenize_chinese_chars
):
__lowerCAmelCase = getattr(lowerCAmelCase_ , normalizer_state.pop('type' ) )
__lowerCAmelCase = do_lower_case
__lowerCAmelCase = strip_accents
__lowerCAmelCase = tokenize_chinese_chars
__lowerCAmelCase = normalizer_class(**lowerCAmelCase_ )
__lowerCAmelCase = do_lower_case
def lowercase ( self : str , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Optional[int]=None ) -> Optional[int]:
__lowerCAmelCase = [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 lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[int] , lowerCAmelCase_ : Optional[List[int]] = None ) -> List[int]:
__lowerCAmelCase = [self.sep_token_id]
__lowerCAmelCase = [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 lowercase ( self : int , lowerCAmelCase_ : str , lowerCAmelCase_ : Optional[str] = None ) -> Tuple[str]:
__lowerCAmelCase = self._tokenizer.model.save(lowerCAmelCase_ , name=lowerCAmelCase_ )
return tuple(lowerCAmelCase_ )
| 53 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
_snake_case : Dict = logging.get_logger(__name__)
_snake_case : Dict = {
'caidas/swin2sr-classicalsr-x2-64': (
'https://huggingface.co/caidas/swin2sr-classicalsr-x2-64/resolve/main/config.json'
),
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = """swin2sr"""
a_ = {
"""hidden_size""": """embed_dim""",
"""num_attention_heads""": """num_heads""",
"""num_hidden_layers""": """num_layers""",
}
def __init__( self : Any , lowerCAmelCase_ : List[str]=6_4 , lowerCAmelCase_ : str=1 , lowerCAmelCase_ : int=3 , lowerCAmelCase_ : Optional[Any]=1_8_0 , lowerCAmelCase_ : Union[str, Any]=[6, 6, 6, 6, 6, 6] , lowerCAmelCase_ : int=[6, 6, 6, 6, 6, 6] , lowerCAmelCase_ : Optional[int]=8 , lowerCAmelCase_ : List[str]=2.0 , lowerCAmelCase_ : Tuple=True , lowerCAmelCase_ : Union[str, Any]=0.0 , lowerCAmelCase_ : Dict=0.0 , lowerCAmelCase_ : List[str]=0.1 , lowerCAmelCase_ : int="gelu" , lowerCAmelCase_ : List[Any]=False , lowerCAmelCase_ : List[Any]=0.02 , lowerCAmelCase_ : List[Any]=1e-5 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : Union[str, Any]=1.0 , lowerCAmelCase_ : List[Any]="1conv" , lowerCAmelCase_ : Optional[Any]="pixelshuffle" , **lowerCAmelCase_ : Dict , ) -> str:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = image_size
__lowerCAmelCase = patch_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = embed_dim
__lowerCAmelCase = depths
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = num_heads
__lowerCAmelCase = window_size
__lowerCAmelCase = mlp_ratio
__lowerCAmelCase = qkv_bias
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = drop_path_rate
__lowerCAmelCase = hidden_act
__lowerCAmelCase = use_absolute_embeddings
__lowerCAmelCase = layer_norm_eps
__lowerCAmelCase = initializer_range
__lowerCAmelCase = upscale
__lowerCAmelCase = img_range
__lowerCAmelCase = resi_connection
__lowerCAmelCase = upsampler
| 53 |
import numpy as np
from cva import COLOR_BGR2GRAY, cvtColor, imread
from numpy import array, uinta
from PIL import Image
from digital_image_processing import change_contrast as cc
from digital_image_processing import convert_to_negative as cn
from digital_image_processing import sepia as sp
from digital_image_processing.dithering import burkes as bs
from digital_image_processing.edge_detection import canny
from digital_image_processing.filters import convolve as conv
from digital_image_processing.filters import gaussian_filter as gg
from digital_image_processing.filters import local_binary_pattern as lbp
from digital_image_processing.filters import median_filter as med
from digital_image_processing.filters import sobel_filter as sob
from digital_image_processing.resize import resize as rs
_snake_case : Union[str, Any] = imread(R'digital_image_processing/image_data/lena_small.jpg')
_snake_case : Optional[int] = cvtColor(img, COLOR_BGR2GRAY)
def a_ ( ):
__lowerCAmelCase = cn.convert_to_negative(lowerCAmelCase_ )
# assert negative_img array for at least one True
assert negative_img.any()
def a_ ( ):
with Image.open('digital_image_processing/image_data/lena_small.jpg' ) as img:
# Work around assertion for response
assert str(cc.change_contrast(lowerCAmelCase_, 110 ) ).startswith(
'<PIL.Image.Image image mode=RGB size=100x100 at' )
def a_ ( ):
__lowerCAmelCase = canny.gen_gaussian_kernel(9, sigma=1.4 )
# Assert ambiguous array
assert resp.all()
def a_ ( ):
__lowerCAmelCase = imread('digital_image_processing/image_data/lena_small.jpg', 0 )
# assert ambiguous array for all == True
assert canny_img.all()
__lowerCAmelCase = canny.canny(lowerCAmelCase_ )
# assert canny array for at least one True
assert canny_array.any()
def a_ ( ):
assert gg.gaussian_filter(lowerCAmelCase_, 5, sigma=0.9 ).all()
def a_ ( ):
# laplace diagonals
__lowerCAmelCase = array([[0.25, 0.5, 0.25], [0.5, -3, 0.5], [0.25, 0.5, 0.25]] )
__lowerCAmelCase = conv.img_convolve(lowerCAmelCase_, lowerCAmelCase_ ).astype(lowerCAmelCase_ )
assert res.any()
def a_ ( ):
assert med.median_filter(lowerCAmelCase_, 3 ).any()
def a_ ( ):
__lowerCAmelCase , __lowerCAmelCase = sob.sobel_filter(lowerCAmelCase_ )
assert grad.any() and theta.any()
def a_ ( ):
__lowerCAmelCase = sp.make_sepia(lowerCAmelCase_, 20 )
assert sepia.all()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg" ):
__lowerCAmelCase = bs.Burkes(imread(lowerCAmelCase_, 1 ), 120 )
burkes.process()
assert burkes.output_img.any()
def a_ ( lowerCAmelCase_ : str = "digital_image_processing/image_data/lena_small.jpg", ):
__lowerCAmelCase = rs.NearestNeighbour(imread(lowerCAmelCase_, 1 ), 400, 200 )
nn.process()
assert nn.output.any()
def a_ ( ):
__lowerCAmelCase = 'digital_image_processing/image_data/lena.jpg'
# Reading the image and converting it to grayscale.
__lowerCAmelCase = imread(lowerCAmelCase_, 0 )
# Test for get_neighbors_pixel function() return not None
__lowerCAmelCase = 0
__lowerCAmelCase = 0
__lowerCAmelCase = image[x_coordinate][y_coordinate]
__lowerCAmelCase = lbp.get_neighbors_pixel(
lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert neighbors_pixels is not None
# Test for local_binary_pattern function()
# Create a numpy array as the same height and width of read image
__lowerCAmelCase = np.zeros((image.shape[0], image.shape[1]) )
# Iterating through the image and calculating the local binary pattern value
# for each pixel.
for i in range(0, image.shape[0] ):
for j in range(0, image.shape[1] ):
__lowerCAmelCase = lbp.local_binary_value(lowerCAmelCase_, lowerCAmelCase_, lowerCAmelCase_ )
assert lbp_image.any()
| 53 | 1 |
import argparse
import json
import os
import torch
from transformers.file_utils import has_file
from diffusers import UNetaDConditionModel, UNetaDModel
_snake_case : Optional[int] = False
_snake_case : Any = True
_snake_case : Optional[Any] = False
if __name__ == "__main__":
_snake_case : Dict = argparse.ArgumentParser()
parser.add_argument(
'--repo_path',
default=None,
type=str,
required=True,
help='The config json file corresponding to the architecture.',
)
parser.add_argument('--dump_path', default=None, type=str, required=True, help='Path to the output model.')
_snake_case : Tuple = parser.parse_args()
_snake_case : Any = {
'image_size': 'sample_size',
'num_res_blocks': 'layers_per_block',
'block_channels': 'block_out_channels',
'down_blocks': 'down_block_types',
'up_blocks': 'up_block_types',
'downscale_freq_shift': 'freq_shift',
'resnet_num_groups': 'norm_num_groups',
'resnet_act_fn': 'act_fn',
'resnet_eps': 'norm_eps',
'num_head_channels': 'attention_head_dim',
}
_snake_case : int = {
'time_steps': 'time_proj',
'mid': 'mid_block',
'downsample_blocks': 'down_blocks',
'upsample_blocks': 'up_blocks',
}
_snake_case : Optional[int] = '' if has_file(args.repo_path, 'config.json') else 'unet'
with open(os.path.join(args.repo_path, subfolder, 'config.json'), 'r', encoding='utf-8') as reader:
_snake_case : Any = reader.read()
_snake_case : Dict = json.loads(text)
if do_only_config:
for key in config_parameters_to_change.keys():
config.pop(key, None)
if has_file(args.repo_path, 'config.json'):
_snake_case : str = UNetaDModel(**config)
else:
_snake_case : Any = UNetaDConditionModel if 'ldm-text2im-large-256' in args.repo_path else UNetaDModel
_snake_case : Dict = class_name(**config)
if do_only_config:
model.save_config(os.path.join(args.repo_path, subfolder))
_snake_case : int = dict(model.config)
if do_only_renaming:
for key, value in config_parameters_to_change.items():
if key in config:
_snake_case : Optional[Any] = config[key]
del config[key]
_snake_case : Optional[Any] = [k.replace('UNetRes', '') for k in config['down_block_types']]
_snake_case : Tuple = [k.replace('UNetRes', '') for k in config['up_block_types']]
if do_only_weights:
_snake_case : int = torch.load(os.path.join(args.repo_path, subfolder, 'diffusion_pytorch_model.bin'))
_snake_case : Any = {}
for param_key, param_value in state_dict.items():
if param_key.endswith('.op.bias') or param_key.endswith('.op.weight'):
continue
_snake_case : Optional[int] = False
for key, new_key in key_parameters_to_change.items():
if not has_changed and param_key.split('.')[0] == key:
_snake_case : List[Any] = param_value
_snake_case : Optional[Any] = True
if not has_changed:
_snake_case : Union[str, Any] = param_value
model.load_state_dict(new_state_dict)
model.save_pretrained(os.path.join(args.repo_path, subfolder))
| 53 |
from typing import Dict, Iterable, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_DEFAULT_MEAN,
IMAGENET_DEFAULT_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, logging
_snake_case : List[Any] = logging.get_logger(__name__)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = ["""pixel_values"""]
def __init__( self : Optional[int] , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Dict[str, int] = None , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Union[int, float] = 1 / 2_5_5 , lowerCAmelCase_ : bool = True , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_MEAN , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = IMAGENET_DEFAULT_STD , **lowerCAmelCase_ : Any , ) -> None:
super().__init__(**lowerCAmelCase_ )
__lowerCAmelCase = size if size is not None else {'shortest_edge': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else {'height': 2_2_4, 'width': 2_2_4}
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = do_resize
__lowerCAmelCase = size
__lowerCAmelCase = resample
__lowerCAmelCase = do_center_crop
__lowerCAmelCase = crop_size
__lowerCAmelCase = do_rescale
__lowerCAmelCase = rescale_factor
__lowerCAmelCase = do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else IMAGENET_DEFAULT_MEAN
__lowerCAmelCase = image_std if image_std is not None else IMAGENET_DEFAULT_STD
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : PILImageResampling = PILImageResampling.BICUBIC , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : Optional[int] , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
# size_dict is a dict with either keys "height" and "width" or "shortest_edge"
if "shortest_edge" in size:
__lowerCAmelCase = int((2_5_6 / 2_2_4) * size['shortest_edge'] )
__lowerCAmelCase = get_resize_output_image_size(lowerCAmelCase_ , size=lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = {'height': output_size[0], 'width': output_size[1]}
if "height" not in size_dict or "width" not in size_dict:
raise ValueError(
f"""Size dict must have keys 'height' and 'width' or 'shortest_edge'. Got {size_dict.keys()}""" )
return resize(
lowerCAmelCase_ , size=(size_dict['height'], size_dict['width']) , resample=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : str , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Dict[str, int] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : str , ) -> np.ndarray:
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ )
if "height" not in size or "width" not in size:
raise ValueError(f"""Size dict must have keys 'height' and 'width'. Got {size.keys()}""" )
return center_crop(lowerCAmelCase_ , size=(size['height'], size['width']) , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[int, float] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : int , ) -> np.ndarray:
return rescale(lowerCAmelCase_ , scale=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : int , lowerCAmelCase_ : np.ndarray , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Union[float, List[float]] , lowerCAmelCase_ : Optional[Union[str, ChannelDimension]] = None , **lowerCAmelCase_ : List[str] , ) -> np.ndarray:
return normalize(lowerCAmelCase_ , mean=lowerCAmelCase_ , std=lowerCAmelCase_ , data_format=lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : ImageInput , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : PILImageResampling = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Dict[str, int]] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[float] = None , lowerCAmelCase_ : Optional[bool] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[Union[float, Iterable[float]]] = None , lowerCAmelCase_ : Optional[TensorType] = None , lowerCAmelCase_ : ChannelDimension = ChannelDimension.FIRST , **lowerCAmelCase_ : str , ) -> BatchFeature:
__lowerCAmelCase = do_resize if do_resize is not None else self.do_resize
__lowerCAmelCase = resample if resample is not None else self.resample
__lowerCAmelCase = do_center_crop if do_center_crop is not None else self.do_center_crop
__lowerCAmelCase = do_rescale if do_rescale is not None else self.do_rescale
__lowerCAmelCase = rescale_factor if rescale_factor is not None else self.rescale_factor
__lowerCAmelCase = do_normalize if do_normalize is not None else self.do_normalize
__lowerCAmelCase = image_mean if image_mean is not None else self.image_mean
__lowerCAmelCase = image_std if image_std is not None else self.image_std
__lowerCAmelCase = size if size is not None else self.size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , default_to_square=lowerCAmelCase_ )
__lowerCAmelCase = crop_size if crop_size is not None else self.crop_size
__lowerCAmelCase = get_size_dict(lowerCAmelCase_ , param_name='crop_size' )
__lowerCAmelCase = make_list_of_images(lowerCAmelCase_ )
if not valid_images(lowerCAmelCase_ ):
raise ValueError(
'Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '
'torch.Tensor, tf.Tensor or jax.ndarray.' )
if do_resize and size is None:
raise ValueError('Size must be specified if do_resize is True.' )
if do_center_crop and crop_size is None:
raise ValueError('Crop size must be specified if do_center_crop is True.' )
if do_rescale and rescale_factor is None:
raise ValueError('Rescale factor must be specified if do_rescale is True.' )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError('Image mean and std must be specified if do_normalize is True.' )
# All transformations expect numpy arrays.
__lowerCAmelCase = [to_numpy_array(lowerCAmelCase_ ) for image in images]
if do_resize:
__lowerCAmelCase = [self.resize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_center_crop:
__lowerCAmelCase = [self.center_crop(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_rescale:
__lowerCAmelCase = [self.rescale(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
if do_normalize:
__lowerCAmelCase = [self.normalize(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = [to_channel_dimension_format(lowerCAmelCase_ , lowerCAmelCase_ ) for image in images]
__lowerCAmelCase = {'pixel_values': images}
return BatchFeature(data=lowerCAmelCase_ , tensor_type=lowerCAmelCase_ )
| 53 | 1 |
import gc
import unittest
import numpy as np
import torch
from diffusers import DanceDiffusionPipeline, IPNDMScheduler, UNetaDModel
from diffusers.utils import slow, torch_device
from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu, skip_mps
from ..pipeline_params import UNCONDITIONAL_AUDIO_GENERATION_BATCH_PARAMS, UNCONDITIONAL_AUDIO_GENERATION_PARAMS
from ..test_pipelines_common import PipelineTesterMixin
enable_full_determinism()
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = DanceDiffusionPipeline
a_ = UNCONDITIONAL_AUDIO_GENERATION_PARAMS
a_ = PipelineTesterMixin.required_optional_params - {
"""callback""",
"""latents""",
"""callback_steps""",
"""output_type""",
"""num_images_per_prompt""",
}
a_ = UNCONDITIONAL_AUDIO_GENERATION_BATCH_PARAMS
a_ = False
a_ = False
def lowercase ( self : int ) -> Union[str, Any]:
torch.manual_seed(0 )
__lowerCAmelCase = UNetaDModel(
block_out_channels=(3_2, 3_2, 6_4) , extra_in_channels=1_6 , sample_size=5_1_2 , sample_rate=1_6_0_0_0 , in_channels=2 , out_channels=2 , flip_sin_to_cos=lowerCAmelCase_ , use_timestep_embedding=lowerCAmelCase_ , time_embedding_type='fourier' , mid_block_type='UNetMidBlock1D' , down_block_types=('DownBlock1DNoSkip', 'DownBlock1D', 'AttnDownBlock1D') , up_block_types=('AttnUpBlock1D', 'UpBlock1D', 'UpBlock1DNoSkip') , )
__lowerCAmelCase = IPNDMScheduler()
__lowerCAmelCase = {
'unet': unet,
'scheduler': scheduler,
}
return components
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Union[str, Any]=0 ) -> int:
if str(lowerCAmelCase_ ).startswith('mps' ):
__lowerCAmelCase = torch.manual_seed(lowerCAmelCase_ )
else:
__lowerCAmelCase = torch.Generator(device=lowerCAmelCase_ ).manual_seed(lowerCAmelCase_ )
__lowerCAmelCase = {
'batch_size': 1,
'generator': generator,
'num_inference_steps': 4,
}
return inputs
def lowercase ( self : List[Any] ) -> Dict:
__lowerCAmelCase = 'cpu' # ensure determinism for the device-dependent torch.Generator
__lowerCAmelCase = self.get_dummy_components()
__lowerCAmelCase = DanceDiffusionPipeline(**lowerCAmelCase_ )
__lowerCAmelCase = pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = self.get_dummy_inputs(lowerCAmelCase_ )
__lowerCAmelCase = pipe(**lowerCAmelCase_ )
__lowerCAmelCase = output.audios
__lowerCAmelCase = audio[0, -3:, -3:]
assert audio.shape == (1, 2, components["unet"].sample_size)
__lowerCAmelCase = np.array([-0.72_65, 1.00_00, -0.83_88, 0.11_75, 0.94_98, -1.00_00] )
assert np.abs(audio_slice.flatten() - expected_slice ).max() < 1e-2
@skip_mps
def lowercase ( self : Any ) -> Any:
return super().test_save_load_local()
@skip_mps
def lowercase ( self : Union[str, Any] ) -> Union[str, Any]:
return super().test_dict_tuple_outputs_equivalent(expected_max_difference=3e-3 )
@skip_mps
def lowercase ( self : Optional[int] ) -> Dict:
return super().test_save_load_optional_components()
@skip_mps
def lowercase ( self : Tuple ) -> Any:
return super().test_attention_slicing_forward_pass()
def lowercase ( self : str ) -> Union[str, Any]:
super().test_inference_batch_single_identical(expected_max_diff=3e-3 )
@slow
@require_torch_gpu
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def lowercase ( self : Optional[int] ) -> Optional[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def lowercase ( self : int ) -> Dict:
__lowerCAmelCase = torch_device
__lowerCAmelCase = DanceDiffusionPipeline.from_pretrained('harmonai/maestro-150k' )
__lowerCAmelCase = pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = torch.manual_seed(0 )
__lowerCAmelCase = pipe(generator=lowerCAmelCase_ , num_inference_steps=1_0_0 , audio_length_in_s=4.0_96 )
__lowerCAmelCase = output.audios
__lowerCAmelCase = audio[0, -3:, -3:]
assert audio.shape == (1, 2, pipe.unet.sample_size)
__lowerCAmelCase = np.array([-0.01_92, -0.02_31, -0.03_18, -0.00_59, 0.00_02, -0.00_20] )
assert np.abs(audio_slice.flatten() - expected_slice ).max() < 1e-2
def lowercase ( self : Tuple ) -> Optional[int]:
__lowerCAmelCase = torch_device
__lowerCAmelCase = DanceDiffusionPipeline.from_pretrained('harmonai/maestro-150k' , torch_dtype=torch.floataa )
__lowerCAmelCase = pipe.to(lowerCAmelCase_ )
pipe.set_progress_bar_config(disable=lowerCAmelCase_ )
__lowerCAmelCase = torch.manual_seed(0 )
__lowerCAmelCase = pipe(generator=lowerCAmelCase_ , num_inference_steps=1_0_0 , audio_length_in_s=4.0_96 )
__lowerCAmelCase = output.audios
__lowerCAmelCase = audio[0, -3:, -3:]
assert audio.shape == (1, 2, pipe.unet.sample_size)
__lowerCAmelCase = np.array([-0.03_67, -0.04_88, -0.07_71, -0.05_25, -0.04_44, -0.03_41] )
assert np.abs(audio_slice.flatten() - expected_slice ).max() < 1e-2
| 53 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Dict=2 , lowerCAmelCase_ : Optional[int]=8 , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : str=True , lowerCAmelCase_ : Optional[Any]=9_9 , lowerCAmelCase_ : List[Any]=1_6 , lowerCAmelCase_ : int=5 , lowerCAmelCase_ : Optional[Any]=2 , lowerCAmelCase_ : str=3_6 , lowerCAmelCase_ : Optional[int]="gelu" , lowerCAmelCase_ : Any=0.0 , lowerCAmelCase_ : Optional[int]=0.0 , lowerCAmelCase_ : str=5_1_2 , lowerCAmelCase_ : List[str]=1_6 , lowerCAmelCase_ : str=2 , lowerCAmelCase_ : Tuple=0.02 , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : List[Any]=4 , lowerCAmelCase_ : List[str]=None , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = seq_length
__lowerCAmelCase = is_training
__lowerCAmelCase = use_input_mask
__lowerCAmelCase = use_token_type_ids
__lowerCAmelCase = use_labels
__lowerCAmelCase = vocab_size
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = max_position_embeddings
__lowerCAmelCase = type_vocab_size
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = num_labels
__lowerCAmelCase = num_choices
__lowerCAmelCase = scope
def lowercase ( self : Optional[int] ) -> Dict:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
__lowerCAmelCase = None
if self.use_input_mask:
__lowerCAmelCase = random_attention_mask([self.batch_size, self.seq_length] )
__lowerCAmelCase = None
if self.use_token_type_ids:
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
__lowerCAmelCase = None
__lowerCAmelCase = None
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
__lowerCAmelCase = ids_tensor([self.batch_size] , self.num_choices )
__lowerCAmelCase = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowercase ( self : Any ) -> Union[str, Any]:
return MraConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , )
def lowercase ( self : Dict ) -> List[Any]:
__lowerCAmelCase = self.get_config()
__lowerCAmelCase = 3_0_0
return config
def lowercase ( self : Optional[int] ) -> Union[str, Any]:
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = self.prepare_config_and_inputs()
__lowerCAmelCase = True
__lowerCAmelCase = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
__lowerCAmelCase = ids_tensor([self.batch_size, self.seq_length] , vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
encoder_hidden_states,
encoder_attention_mask,
)
def lowercase ( self : Optional[int] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , ) -> Tuple:
__lowerCAmelCase = True
__lowerCAmelCase = MraModel(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , encoder_attention_mask=lowerCAmelCase_ , )
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , encoder_hidden_states=lowerCAmelCase_ , )
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : Tuple ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Dict ) -> str:
__lowerCAmelCase = MraForQuestionAnswering(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , start_positions=lowerCAmelCase_ , end_positions=lowerCAmelCase_ , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowercase ( self : int , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Dict , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : List[Any] , lowerCAmelCase_ : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForSequenceClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Any , lowerCAmelCase_ : int , lowerCAmelCase_ : int , lowerCAmelCase_ : Dict ) -> Any:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = MraForTokenClassification(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowercase ( self : int , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : Any , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : Optional[Any] , lowerCAmelCase_ : Optional[Any] ) -> List[Any]:
__lowerCAmelCase = self.num_choices
__lowerCAmelCase = MraForMultipleChoice(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
__lowerCAmelCase = model(
lowerCAmelCase_ , attention_mask=lowerCAmelCase_ , token_type_ids=lowerCAmelCase_ , labels=lowerCAmelCase_ , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowercase ( self : Tuple ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
(
(
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) , (
__lowerCAmelCase
) ,
) = config_and_inputs
__lowerCAmelCase = {'input_ids': input_ids, 'token_type_ids': token_type_ids, 'attention_mask': input_mask}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
a_ = False
a_ = False
a_ = False
a_ = False
a_ = ()
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = MraModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : Tuple ) -> List[str]:
self.config_tester.run_common_tests()
def lowercase ( self : Optional[int] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
__lowerCAmelCase = type
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Any ) -> Union[str, Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*lowerCAmelCase_ )
def lowercase ( self : List[str] ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*lowerCAmelCase_ )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*lowerCAmelCase_ )
@slow
def lowercase ( self : Optional[int] ) -> Optional[int]:
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = MraModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
@unittest.skip(reason='MRA does not output attentions' )
def lowercase ( self : Optional[int] ) -> Tuple:
return
@require_torch
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@slow
def lowercase ( self : Optional[Any] ) -> List[str]:
__lowerCAmelCase = MraModel.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = torch.Size((1, 2_5_6, 7_6_8) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[-0.01_40, 0.08_30, -0.03_81], [0.15_46, 0.14_02, 0.02_20], [0.11_62, 0.08_51, 0.01_65]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : int ) -> Optional[int]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-512-4' )
__lowerCAmelCase = torch.arange(2_5_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 2_5_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[9.25_95, -3.60_38, 11.88_19], [9.38_69, -3.26_93, 11.09_56], [11.85_24, -3.49_38, 13.12_10]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
@slow
def lowercase ( self : Any ) -> List[str]:
__lowerCAmelCase = MraForMaskedLM.from_pretrained('uw-madison/mra-base-4096-8-d3' )
__lowerCAmelCase = torch.arange(4_0_9_6 ).unsqueeze(0 )
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )[0]
__lowerCAmelCase = 5_0_2_6_5
__lowerCAmelCase = torch.Size((1, 4_0_9_6, vocab_size) )
self.assertEqual(output.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[[5.47_89, -2.35_64, 7.50_64], [7.90_67, -1.33_69, 9.96_68], [9.07_12, -1.81_06, 7.03_80]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
def a_ ( lowerCAmelCase_ : int = 200_0000 ):
__lowerCAmelCase = [0 for i in range(n + 1 )]
__lowerCAmelCase = 1
__lowerCAmelCase = 1
for i in range(2, int(n**0.5 ) + 1 ):
if primality_list[i] == 0:
for j in range(i * i, n + 1, lowerCAmelCase_ ):
__lowerCAmelCase = 1
__lowerCAmelCase = 0
for i in range(lowerCAmelCase_ ):
if primality_list[i] == 0:
sum_of_primes += i
return sum_of_primes
if __name__ == "__main__":
print(F"""{solution() = }""")
| 53 |
import argparse
import json
import os
import re
import shutil
import torch
from transformers import BioGptConfig, BioGptForCausalLM
from transformers.models.biogpt.tokenization_biogpt import VOCAB_FILES_NAMES
from transformers.tokenization_utils_base import TOKENIZER_CONFIG_FILE
from transformers.utils import WEIGHTS_NAME, logging
logging.set_verbosity_warning()
_snake_case : Union[str, Any] = 2
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , *, # begin keyword-only arguments
lowerCAmelCase_ : str="<s>" , lowerCAmelCase_ : Dict="<pad>" , lowerCAmelCase_ : Any="</s>" , lowerCAmelCase_ : List[str]="<unk>" , lowerCAmelCase_ : Optional[Any]=None , ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = bos, unk, pad, eos
__lowerCAmelCase = []
__lowerCAmelCase = []
__lowerCAmelCase = {}
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
if extra_special_symbols:
for s in extra_special_symbols:
self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = len(self.symbols )
def __eq__( self : Dict , lowerCAmelCase_ : Dict ) -> str:
return self.indices == other.indices
def __getitem__( self : List[Any] , lowerCAmelCase_ : int ) -> Union[str, Any]:
if idx < len(self.symbols ):
return self.symbols[idx]
return self.unk_word
def __len__( self : Tuple ) -> List[Any]:
return len(self.symbols )
def __contains__( self : Optional[Any] , lowerCAmelCase_ : Dict ) -> Optional[int]:
return sym in self.indices
@classmethod
def lowercase ( cls : Dict , lowerCAmelCase_ : str ) -> str:
__lowerCAmelCase = cls()
d.add_from_file(lowerCAmelCase_ )
return d
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int=1 , lowerCAmelCase_ : Any=False ) -> Optional[Any]:
if word in self.indices and not overwrite:
__lowerCAmelCase = self.indices[word]
__lowerCAmelCase = self.count[idx] + n
return idx
else:
__lowerCAmelCase = len(self.symbols )
__lowerCAmelCase = idx
self.symbols.append(lowerCAmelCase_ )
self.count.append(lowerCAmelCase_ )
return idx
def lowercase ( self : str , lowerCAmelCase_ : Union[str, Any] ) -> Dict:
return 0
def lowercase ( self : Tuple , lowerCAmelCase_ : Union[str, Any] ) -> int:
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
try:
with open(lowerCAmelCase_ , 'r' , encoding='utf-8' ) as fd:
self.add_from_file(lowerCAmelCase_ )
except FileNotFoundError as fnfe:
raise fnfe
except UnicodeError:
raise Exception('Incorrect encoding detected in {}, please rebuild the dataset'.format(lowerCAmelCase_ ) )
return
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = self._load_meta(lowerCAmelCase_ )
for line in lines[indices_start_line:]:
try:
__lowerCAmelCase , __lowerCAmelCase = line.rstrip().rsplit(' ' , 1 )
if field == "#fairseq:overwrite":
__lowerCAmelCase = True
__lowerCAmelCase , __lowerCAmelCase = line.rsplit(' ' , 1 )
else:
__lowerCAmelCase = False
__lowerCAmelCase = int(lowerCAmelCase_ )
__lowerCAmelCase = line
if word in self and not overwrite:
raise RuntimeError(
'Duplicate word found when loading Dictionary: \'{}\'. '
'Duplicate words can overwrite earlier ones by adding the '
'#fairseq:overwrite flag at the end of the corresponding row '
'in the dictionary file. If using the Camembert model, please '
'download an updated copy of the model file.'.format(lowerCAmelCase_ ) )
self.add_symbol(lowerCAmelCase_ , n=lowerCAmelCase_ , overwrite=lowerCAmelCase_ )
except ValueError:
raise ValueError('Incorrect dictionary format, expected \'<token> <cnt> [flags]\'' )
def a_ ( lowerCAmelCase_ : List[str] ):
# (1) remove word breaking symbol, (2) add word ending symbol where the word is not broken up,
# e.g.: d = {'le@@': 5, 'tt@@': 6, 'er': 7} => {'le': 5, 'tt': 6, 'er</w>': 7}
__lowerCAmelCase = dict((re.sub(R'@@$', '', lowerCAmelCase_ ), v) if k.endswith('@@' ) else (re.sub(R'$', '</w>', lowerCAmelCase_ ), v) for k, v in d.items() )
__lowerCAmelCase = '<s> <pad> </s> <unk>'.split()
# restore the special tokens
for k in keep_keys:
del da[F"""{k}</w>"""]
__lowerCAmelCase = d[k] # restore
return da
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str] ):
# prep
if not os.path.exists(lowerCAmelCase_ ):
raise ValueError(F"""path {biogpt_checkpoint_path} does not exist!""" )
os.makedirs(lowerCAmelCase_, exist_ok=lowerCAmelCase_ )
print(F"""Writing results to {pytorch_dump_folder_path}""" )
# handle various types of models
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'checkpoint.pt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {checkpoint_file} does not exist!""" )
__lowerCAmelCase = torch.load(lowerCAmelCase_, map_location='cpu' )
__lowerCAmelCase = chkpt['cfg']['model']
# dicts
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'dict.txt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {dict_file} does not exist!""" )
__lowerCAmelCase = Dictionary.load(lowerCAmelCase_ )
__lowerCAmelCase = rewrite_dict_keys(src_dict.indices )
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['vocab_file'] )
print(F"""Generating {src_vocab_file} of {src_vocab_size} records""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# merges_file (bpecodes)
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'bpecodes' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {bpecodes_file} does not exist!""" )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['merges_file'] )
shutil.copyfile(lowerCAmelCase_, lowerCAmelCase_ )
# model config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'config.json' )
__lowerCAmelCase = {
'activation_dropout': args['activation_dropout'],
'architectures': ['BioGptForCausalLM'],
'attention_probs_dropout_prob': args['attention_dropout'],
'bos_token_id': 0,
'eos_token_id': 2,
'hidden_act': args['activation_fn'],
'hidden_dropout_prob': args['dropout'],
'hidden_size': args['decoder_embed_dim'],
'initializer_range': 0.02,
'intermediate_size': args['decoder_ffn_embed_dim'],
'layer_norm_eps': 1E-12,
'layerdrop': args['decoder_layerdrop'],
'max_position_embeddings': args['max_target_positions'],
'model_type': 'biogpt',
'num_attention_heads': args['decoder_attention_heads'],
'num_hidden_layers': args['decoder_layers'],
'pad_token_id': 1,
'scale_embedding': not args['no_scale_embedding'],
'tie_word_embeddings': args['share_decoder_input_output_embed'],
'vocab_size': src_vocab_size,
}
# good hparam defaults to start with
print(F"""Generating {biogpt_model_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# tokenizer config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = {
'bos_token': '<s>',
'eos_token': '</s>',
'model_max_length': 1024,
'pad_token': '<pad>',
'special_tokens_map_file': None,
'tokenizer_class': 'BioGptTokenizer',
'unk_token': '<unk>',
}
print(F"""Generating {biogpt_tokenizer_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# model
__lowerCAmelCase = chkpt['model']
# remove unneeded keys
__lowerCAmelCase = [
'decoder.version',
]
for k in ignore_keys:
model_state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = list(model_state_dict.keys() )
for layer_name in layer_names:
if layer_name.endswith('output_projection.weight' ):
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
else:
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
__lowerCAmelCase = BioGptConfig.from_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = BioGptForCausalLM(lowerCAmelCase_ )
# check that it loads ok
model_new.load_state_dict(lowerCAmelCase_ )
# save
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
print(F"""Generating {pytorch_weights_dump_path}""" )
torch.save(lowerCAmelCase_, lowerCAmelCase_ )
print('Conversion is done!' )
if __name__ == "__main__":
_snake_case : str = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--biogpt_checkpoint_path',
default=None,
type=str,
required=True,
help=(
'Path to the official PyTorch checkpoint file which is expected to reside in the dump dir with dicts,'
' bpecodes, etc.'
),
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_biogpt_checkpoint_to_pytorch(args.biogpt_checkpoint_path, args.pytorch_dump_folder_path)
| 53 | 1 |
import warnings
from ...utils import is_sklearn_available, requires_backends
if is_sklearn_available():
from scipy.stats import pearsonr, spearmanr
from sklearn.metrics import fa_score, matthews_corrcoef
_snake_case : Union[str, Any] = (
'This metric will be removed from the library soon, metrics should be handled with the 🤗 Evaluate '
'library. You can have a look at this example script for pointers: '
'https://github.com/huggingface/transformers/blob/main/examples/pytorch/text-classification/run_glue.py'
)
def a_ ( lowerCAmelCase_ : Optional[int], lowerCAmelCase_ : Union[str, Any] ):
warnings.warn(lowerCAmelCase_, lowerCAmelCase_ )
requires_backends(lowerCAmelCase_, 'sklearn' )
return (preds == labels).mean()
def a_ ( lowerCAmelCase_ : str, lowerCAmelCase_ : List[str] ):
warnings.warn(lowerCAmelCase_, lowerCAmelCase_ )
requires_backends(lowerCAmelCase_, 'sklearn' )
__lowerCAmelCase = simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = fa_score(y_true=lowerCAmelCase_, y_pred=lowerCAmelCase_ )
return {
"acc": acc,
"f1": fa,
"acc_and_f1": (acc + fa) / 2,
}
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : Optional[int] ):
warnings.warn(lowerCAmelCase_, lowerCAmelCase_ )
requires_backends(lowerCAmelCase_, 'sklearn' )
__lowerCAmelCase = pearsonr(lowerCAmelCase_, lowerCAmelCase_ )[0]
__lowerCAmelCase = spearmanr(lowerCAmelCase_, lowerCAmelCase_ )[0]
return {
"pearson": pearson_corr,
"spearmanr": spearman_corr,
"corr": (pearson_corr + spearman_corr) / 2,
}
def a_ ( lowerCAmelCase_ : List[Any], lowerCAmelCase_ : List[str], lowerCAmelCase_ : Union[str, Any] ):
warnings.warn(lowerCAmelCase_, lowerCAmelCase_ )
requires_backends(lowerCAmelCase_, 'sklearn' )
assert len(lowerCAmelCase_ ) == len(lowerCAmelCase_ ), F"""Predictions and labels have mismatched lengths {len(lowerCAmelCase_ )} and {len(lowerCAmelCase_ )}"""
if task_name == "cola":
return {"mcc": matthews_corrcoef(lowerCAmelCase_, lowerCAmelCase_ )}
elif task_name == "sst-2":
return {"acc": simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )}
elif task_name == "mrpc":
return acc_and_fa(lowerCAmelCase_, lowerCAmelCase_ )
elif task_name == "sts-b":
return pearson_and_spearman(lowerCAmelCase_, lowerCAmelCase_ )
elif task_name == "qqp":
return acc_and_fa(lowerCAmelCase_, lowerCAmelCase_ )
elif task_name == "mnli":
return {"mnli/acc": simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )}
elif task_name == "mnli-mm":
return {"mnli-mm/acc": simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )}
elif task_name == "qnli":
return {"acc": simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )}
elif task_name == "rte":
return {"acc": simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )}
elif task_name == "wnli":
return {"acc": simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )}
elif task_name == "hans":
return {"acc": simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )}
else:
raise KeyError(lowerCAmelCase_ )
def a_ ( lowerCAmelCase_ : Optional[Any], lowerCAmelCase_ : int, lowerCAmelCase_ : Union[str, Any] ):
warnings.warn(lowerCAmelCase_, lowerCAmelCase_ )
requires_backends(lowerCAmelCase_, 'sklearn' )
if len(lowerCAmelCase_ ) != len(lowerCAmelCase_ ):
raise ValueError(F"""Predictions and labels have mismatched lengths {len(lowerCAmelCase_ )} and {len(lowerCAmelCase_ )}""" )
if task_name == "xnli":
return {"acc": simple_accuracy(lowerCAmelCase_, lowerCAmelCase_ )}
else:
raise KeyError(lowerCAmelCase_ )
| 53 |
from typing import Tuple, Union
from ...modeling_outputs import BackboneOutput
from ...modeling_utils import PreTrainedModel
from ...utils import is_timm_available, is_torch_available, requires_backends
from ...utils.backbone_utils import BackboneMixin
from .configuration_timm_backbone import TimmBackboneConfig
if is_timm_available():
import timm
if is_torch_available():
from torch import Tensor
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = """pixel_values"""
a_ = False
a_ = TimmBackboneConfig
def __init__( self : Tuple , lowerCAmelCase_ : Any , **lowerCAmelCase_ : Optional[int] ) -> Optional[Any]:
requires_backends(self , 'timm' )
super().__init__(lowerCAmelCase_ )
__lowerCAmelCase = config
if config.backbone is None:
raise ValueError('backbone is not set in the config. Please set it to a timm model name.' )
if config.backbone not in timm.list_models():
raise ValueError(f"""backbone {config.backbone} is not supported by timm.""" )
if hasattr(lowerCAmelCase_ , 'out_features' ) and config.out_features is not None:
raise ValueError('out_features is not supported by TimmBackbone. Please use out_indices instead.' )
__lowerCAmelCase = getattr(lowerCAmelCase_ , 'use_pretrained_backbone' , lowerCAmelCase_ )
if pretrained is None:
raise ValueError('use_pretrained_backbone is not set in the config. Please set it to True or False.' )
# We just take the final layer by default. This matches the default for the transformers models.
__lowerCAmelCase = config.out_indices if getattr(lowerCAmelCase_ , 'out_indices' , lowerCAmelCase_ ) is not None else (-1,)
__lowerCAmelCase = timm.create_model(
config.backbone , pretrained=lowerCAmelCase_ , features_only=config.features_only , in_chans=config.num_channels , out_indices=lowerCAmelCase_ , **lowerCAmelCase_ , )
# These are used to control the output of the model when called. If output_hidden_states is True, then
# return_layers is modified to include all layers.
__lowerCAmelCase = self._backbone.return_layers
__lowerCAmelCase = {layer['module']: str(lowerCAmelCase_ ) for i, layer in enumerate(self._backbone.feature_info.info )}
super()._init_backbone(lowerCAmelCase_ )
@classmethod
def lowercase ( cls : int , lowerCAmelCase_ : Dict , *lowerCAmelCase_ : Dict , **lowerCAmelCase_ : Union[str, Any] ) -> Optional[int]:
requires_backends(cls , ['vision', 'timm'] )
from ...models.timm_backbone import TimmBackboneConfig
__lowerCAmelCase = kwargs.pop('config' , TimmBackboneConfig() )
__lowerCAmelCase = kwargs.pop('use_timm_backbone' , lowerCAmelCase_ )
if not use_timm:
raise ValueError('use_timm_backbone must be True for timm backbones' )
__lowerCAmelCase = kwargs.pop('num_channels' , config.num_channels )
__lowerCAmelCase = kwargs.pop('features_only' , config.features_only )
__lowerCAmelCase = kwargs.pop('use_pretrained_backbone' , config.use_pretrained_backbone )
__lowerCAmelCase = kwargs.pop('out_indices' , config.out_indices )
__lowerCAmelCase = TimmBackboneConfig(
backbone=lowerCAmelCase_ , num_channels=lowerCAmelCase_ , features_only=lowerCAmelCase_ , use_pretrained_backbone=lowerCAmelCase_ , out_indices=lowerCAmelCase_ , )
return super()._from_config(lowerCAmelCase_ , **lowerCAmelCase_ )
def lowercase ( self : Tuple , lowerCAmelCase_ : int ) -> Dict:
pass
def lowercase ( self : Optional[int] , lowerCAmelCase_ : Union[str, Any] , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : Any=None , **lowerCAmelCase_ : Dict ) -> Union[BackboneOutput, Tuple[Tensor, ...]]:
__lowerCAmelCase = return_dict if return_dict is not None else self.config.use_return_dict
__lowerCAmelCase = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__lowerCAmelCase = output_attentions if output_attentions is not None else self.config.output_attentions
if output_attentions:
raise ValueError('Cannot output attentions for timm backbones at the moment' )
if output_hidden_states:
# We modify the return layers to include all the stages of the backbone
__lowerCAmelCase = self._all_layers
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = self._return_layers
__lowerCAmelCase = tuple(hidden_states[i] for i in self.out_indices )
else:
__lowerCAmelCase = self._backbone(lowerCAmelCase_ , **lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = tuple(lowerCAmelCase_ )
__lowerCAmelCase = tuple(lowerCAmelCase_ ) if hidden_states is not None else None
if not return_dict:
__lowerCAmelCase = (feature_maps,)
if output_hidden_states:
__lowerCAmelCase = output + (hidden_states,)
return output
return BackboneOutput(feature_maps=lowerCAmelCase_ , hidden_states=lowerCAmelCase_ , attentions=lowerCAmelCase_ )
| 53 | 1 |
from dataclasses import dataclass
from typing import Optional, Tuple, Union
import torch
import torch.nn as nn
from ..configuration_utils import ConfigMixin, register_to_config
from ..utils import BaseOutput, apply_forward_hook
from .modeling_utils import ModelMixin
from .vae import Decoder, DecoderOutput, Encoder, VectorQuantizer
@dataclass
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = 42
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
@register_to_config
def __init__( self : Union[str, Any] , lowerCAmelCase_ : int = 3 , lowerCAmelCase_ : int = 3 , lowerCAmelCase_ : Tuple[str] = ("DownEncoderBlock2D",) , lowerCAmelCase_ : Tuple[str] = ("UpDecoderBlock2D",) , lowerCAmelCase_ : Tuple[int] = (6_4,) , lowerCAmelCase_ : int = 1 , lowerCAmelCase_ : str = "silu" , lowerCAmelCase_ : int = 3 , lowerCAmelCase_ : int = 3_2 , lowerCAmelCase_ : int = 2_5_6 , lowerCAmelCase_ : int = 3_2 , lowerCAmelCase_ : Optional[int] = None , lowerCAmelCase_ : float = 0.1_82_15 , lowerCAmelCase_ : str = "group" , ) -> List[Any]:
super().__init__()
# pass init params to Encoder
__lowerCAmelCase = Encoder(
in_channels=lowerCAmelCase_ , out_channels=lowerCAmelCase_ , down_block_types=lowerCAmelCase_ , block_out_channels=lowerCAmelCase_ , layers_per_block=lowerCAmelCase_ , act_fn=lowerCAmelCase_ , norm_num_groups=lowerCAmelCase_ , double_z=lowerCAmelCase_ , )
__lowerCAmelCase = vq_embed_dim if vq_embed_dim is not None else latent_channels
__lowerCAmelCase = nn.Convad(lowerCAmelCase_ , lowerCAmelCase_ , 1 )
__lowerCAmelCase = VectorQuantizer(lowerCAmelCase_ , lowerCAmelCase_ , beta=0.25 , remap=lowerCAmelCase_ , sane_index_shape=lowerCAmelCase_ )
__lowerCAmelCase = nn.Convad(lowerCAmelCase_ , lowerCAmelCase_ , 1 )
# pass init params to Decoder
__lowerCAmelCase = Decoder(
in_channels=lowerCAmelCase_ , out_channels=lowerCAmelCase_ , up_block_types=lowerCAmelCase_ , block_out_channels=lowerCAmelCase_ , layers_per_block=lowerCAmelCase_ , act_fn=lowerCAmelCase_ , norm_num_groups=lowerCAmelCase_ , norm_type=lowerCAmelCase_ , )
@apply_forward_hook
def lowercase ( self : Optional[int] , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = True ) -> VQEncoderOutput:
__lowerCAmelCase = self.encoder(lowerCAmelCase_ )
__lowerCAmelCase = self.quant_conv(lowerCAmelCase_ )
if not return_dict:
return (h,)
return VQEncoderOutput(latents=lowerCAmelCase_ )
@apply_forward_hook
def lowercase ( self : Optional[Any] , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = False , lowerCAmelCase_ : bool = True ) -> Union[DecoderOutput, torch.FloatTensor]:
# also go through quantization layer
if not force_not_quantize:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = self.quantize(lowerCAmelCase_ )
else:
__lowerCAmelCase = h
__lowerCAmelCase = self.post_quant_conv(lowerCAmelCase_ )
__lowerCAmelCase = self.decoder(lowerCAmelCase_ , quant if self.config.norm_type == 'spatial' else None )
if not return_dict:
return (dec,)
return DecoderOutput(sample=lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : torch.FloatTensor , lowerCAmelCase_ : bool = True ) -> Union[DecoderOutput, torch.FloatTensor]:
__lowerCAmelCase = sample
__lowerCAmelCase = self.encode(lowerCAmelCase_ ).latents
__lowerCAmelCase = self.decode(lowerCAmelCase_ ).sample
if not return_dict:
return (dec,)
return DecoderOutput(sample=lowerCAmelCase_ )
| 53 |
from __future__ import annotations
def a_ ( lowerCAmelCase_ : list[float] ):
if len(lowerCAmelCase_ ) < 2:
raise ValueError('Monogons and Digons are not polygons in the Euclidean space' )
if any(i <= 0 for i in nums ):
raise ValueError('All values must be greater than 0' )
__lowerCAmelCase = nums.copy()
copy_nums.sort()
return copy_nums[-1] < sum(copy_nums[:-1] )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 53 | 1 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_sentencepiece_available,
is_tokenizers_available,
is_torch_available,
)
_snake_case : str = {}
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : List[str] = ['NllbTokenizer']
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
_snake_case : List[Any] = ['NllbTokenizerFast']
if TYPE_CHECKING:
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_nllb import NllbTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_nllb_fast import NllbTokenizerFast
else:
import sys
_snake_case : Dict = _LazyModule(__name__, globals()['__file__'], _import_structure, module_spec=__spec__)
| 53 |
import inspect
import unittest
from transformers import RegNetConfig, is_flax_available
from transformers.testing_utils import require_flax, slow
from transformers.utils import cached_property, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor
if is_flax_available():
import jax
import jax.numpy as jnp
from transformers.models.regnet.modeling_flax_regnet import FlaxRegNetForImageClassification, FlaxRegNetModel
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : Optional[int]=3 , lowerCAmelCase_ : Dict=3_2 , lowerCAmelCase_ : Tuple=3 , lowerCAmelCase_ : Union[str, Any]=1_0 , lowerCAmelCase_ : List[str]=[1_0, 2_0, 3_0, 4_0] , lowerCAmelCase_ : Optional[int]=[1, 1, 2, 1] , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Any=True , lowerCAmelCase_ : Tuple="relu" , lowerCAmelCase_ : Union[str, Any]=3 , lowerCAmelCase_ : Optional[int]=None , ) -> int:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = embeddings_size
__lowerCAmelCase = hidden_sizes
__lowerCAmelCase = depths
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = hidden_act
__lowerCAmelCase = num_labels
__lowerCAmelCase = scope
__lowerCAmelCase = len(lowerCAmelCase_ )
def lowercase ( self : Optional[int] ) -> List[Any]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : Tuple ) -> List[Any]:
return RegNetConfig(
num_channels=self.num_channels , embeddings_size=self.embeddings_size , hidden_sizes=self.hidden_sizes , depths=self.depths , hidden_act=self.hidden_act , num_labels=self.num_labels , image_size=self.image_size , )
def lowercase ( self : List[str] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : List[str] ) -> str:
__lowerCAmelCase = FlaxRegNetModel(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
# Output shape (b, c, h, w)
self.parent.assertEqual(
result.last_hidden_state.shape , (self.batch_size, self.hidden_sizes[-1], self.image_size // 3_2, self.image_size // 3_2) , )
def lowercase ( self : str , lowerCAmelCase_ : Optional[int] , lowerCAmelCase_ : int ) -> Tuple:
__lowerCAmelCase = self.num_labels
__lowerCAmelCase = FlaxRegNetForImageClassification(config=lowerCAmelCase_ )
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowercase ( self : List[Any] ) -> Optional[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_flax
class _UpperCAmelCase ( _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (FlaxRegNetModel, FlaxRegNetForImageClassification) if is_flax_available() else ()
a_ = False
a_ = False
a_ = False
def lowercase ( self : Dict ) -> None:
__lowerCAmelCase = FlaxRegNetModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : int ) -> Optional[int]:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : str ) -> Union[str, Any]:
return
def lowercase ( self : Dict ) -> str:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> Tuple:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCAmelCase_ )
@unittest.skip(reason='RegNet does not use inputs_embeds' )
def lowercase ( self : Union[str, Any] ) -> Any:
pass
@unittest.skip(reason='RegNet does not support input and output embeddings' )
def lowercase ( self : Tuple ) -> Tuple:
pass
def lowercase ( self : Optional[Any] ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.__call__ )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> Union[str, Any]:
def check_hidden_states_output(lowerCAmelCase_ : Any , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ):
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCAmelCase = self.model_tester.num_stages
self.assertEqual(len(lowerCAmelCase_ ) , expected_num_stages + 1 )
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCAmelCase = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : str ) -> str:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
with self.subTest(model_class.__name__ ):
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model_class(lowerCAmelCase_ )
@jax.jit
def model_jitted(lowerCAmelCase_ : Optional[int] , **lowerCAmelCase_ : Dict ):
return model(pixel_values=lowerCAmelCase_ , **lowerCAmelCase_ )
with self.subTest('JIT Enabled' ):
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
with self.subTest('JIT Disabled' ):
with jax.disable_jit():
__lowerCAmelCase = model_jitted(**lowerCAmelCase_ ).to_tuple()
self.assertEqual(len(lowerCAmelCase_ ) , len(lowerCAmelCase_ ) )
for jitted_output, output in zip(lowerCAmelCase_ , lowerCAmelCase_ ):
self.assertEqual(jitted_output.shape , output.shape )
def a_ ( ):
__lowerCAmelCase = Image.open('./tests/fixtures/tests_samples/COCO/000000039769.png' )
return image
@require_flax
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
return AutoImageProcessor.from_pretrained('facebook/regnet-y-040' ) if is_vision_available() else None
@slow
def lowercase ( self : Optional[Any] ) -> Union[str, Any]:
__lowerCAmelCase = FlaxRegNetForImageClassification.from_pretrained('facebook/regnet-y-040' )
__lowerCAmelCase = self.default_image_processor
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = image_processor(images=lowerCAmelCase_ , return_tensors='np' )
__lowerCAmelCase = model(**lowerCAmelCase_ )
# verify the logits
__lowerCAmelCase = (1, 1_0_0_0)
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = jnp.array([-0.41_80, -1.50_51, -3.48_36] )
self.assertTrue(jnp.allclose(outputs.logits[0, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 53 | 1 |
import collections
from typing import List, Optional, Union
from ...tokenization_utils_base import BatchEncoding
from ...utils import TensorType, add_end_docstrings, add_start_docstrings, logging
from ..bert.tokenization_bert_fast import BertTokenizerFast
from .tokenization_dpr import DPRContextEncoderTokenizer, DPRQuestionEncoderTokenizer, DPRReaderTokenizer
_snake_case : List[str] = logging.get_logger(__name__)
_snake_case : Dict = {'vocab_file': 'vocab.txt', 'tokenizer_file': 'tokenizer.json'}
_snake_case : Optional[Any] = {
'vocab_file': {
'facebook/dpr-ctx_encoder-single-nq-base': (
'https://huggingface.co/facebook/dpr-ctx_encoder-single-nq-base/resolve/main/vocab.txt'
),
'facebook/dpr-ctx_encoder-multiset-base': (
'https://huggingface.co/facebook/dpr-ctx_encoder-multiset-base/resolve/main/vocab.txt'
),
},
'tokenizer_file': {
'facebook/dpr-ctx_encoder-single-nq-base': (
'https://huggingface.co/facebook/dpr-ctx_encoder-single-nq-base/resolve/main/tokenizer.json'
),
'facebook/dpr-ctx_encoder-multiset-base': (
'https://huggingface.co/facebook/dpr-ctx_encoder-multiset-base/resolve/main/tokenizer.json'
),
},
}
_snake_case : Union[str, Any] = {
'vocab_file': {
'facebook/dpr-question_encoder-single-nq-base': (
'https://huggingface.co/facebook/dpr-question_encoder-single-nq-base/resolve/main/vocab.txt'
),
'facebook/dpr-question_encoder-multiset-base': (
'https://huggingface.co/facebook/dpr-question_encoder-multiset-base/resolve/main/vocab.txt'
),
},
'tokenizer_file': {
'facebook/dpr-question_encoder-single-nq-base': (
'https://huggingface.co/facebook/dpr-question_encoder-single-nq-base/resolve/main/tokenizer.json'
),
'facebook/dpr-question_encoder-multiset-base': (
'https://huggingface.co/facebook/dpr-question_encoder-multiset-base/resolve/main/tokenizer.json'
),
},
}
_snake_case : Dict = {
'vocab_file': {
'facebook/dpr-reader-single-nq-base': (
'https://huggingface.co/facebook/dpr-reader-single-nq-base/resolve/main/vocab.txt'
),
'facebook/dpr-reader-multiset-base': (
'https://huggingface.co/facebook/dpr-reader-multiset-base/resolve/main/vocab.txt'
),
},
'tokenizer_file': {
'facebook/dpr-reader-single-nq-base': (
'https://huggingface.co/facebook/dpr-reader-single-nq-base/resolve/main/tokenizer.json'
),
'facebook/dpr-reader-multiset-base': (
'https://huggingface.co/facebook/dpr-reader-multiset-base/resolve/main/tokenizer.json'
),
},
}
_snake_case : Tuple = {
'facebook/dpr-ctx_encoder-single-nq-base': 512,
'facebook/dpr-ctx_encoder-multiset-base': 512,
}
_snake_case : Tuple = {
'facebook/dpr-question_encoder-single-nq-base': 512,
'facebook/dpr-question_encoder-multiset-base': 512,
}
_snake_case : Dict = {
'facebook/dpr-reader-single-nq-base': 512,
'facebook/dpr-reader-multiset-base': 512,
}
_snake_case : Optional[int] = {
'facebook/dpr-ctx_encoder-single-nq-base': {'do_lower_case': True},
'facebook/dpr-ctx_encoder-multiset-base': {'do_lower_case': True},
}
_snake_case : int = {
'facebook/dpr-question_encoder-single-nq-base': {'do_lower_case': True},
'facebook/dpr-question_encoder-multiset-base': {'do_lower_case': True},
}
_snake_case : Tuple = {
'facebook/dpr-reader-single-nq-base': {'do_lower_case': True},
'facebook/dpr-reader-multiset-base': {'do_lower_case': True},
}
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = CONTEXT_ENCODER_PRETRAINED_VOCAB_FILES_MAP
a_ = CONTEXT_ENCODER_PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
a_ = CONTEXT_ENCODER_PRETRAINED_INIT_CONFIGURATION
a_ = DPRContextEncoderTokenizer
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = QUESTION_ENCODER_PRETRAINED_VOCAB_FILES_MAP
a_ = QUESTION_ENCODER_PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
a_ = QUESTION_ENCODER_PRETRAINED_INIT_CONFIGURATION
a_ = DPRQuestionEncoderTokenizer
_snake_case : Tuple = collections.namedtuple(
'DPRSpanPrediction', ['span_score', 'relevance_score', 'doc_id', 'start_index', 'end_index', 'text']
)
_snake_case : Optional[Any] = collections.namedtuple('DPRReaderOutput', ['start_logits', 'end_logits', 'relevance_logits'])
_snake_case : Any = R'\n Return a dictionary with the token ids of the input strings and other information to give to `.decode_best_spans`.\n It converts the strings of a question and different passages (title and text) in a sequence of IDs (integers),\n using the tokenizer and vocabulary. The resulting `input_ids` is a matrix of size `(n_passages, sequence_length)`\n with the format:\n\n [CLS] <question token ids> [SEP] <titles ids> [SEP] <texts ids>\n\n Args:\n questions (`str` or `List[str]`):\n The questions to be encoded. You can specify one question for many passages. In this case, the question\n will be duplicated like `[questions] * n_passages`. Otherwise you have to specify as many questions as in\n `titles` or `texts`.\n titles (`str` or `List[str]`):\n The passages titles to be encoded. This can be a string or a list of strings if there are several passages.\n texts (`str` or `List[str]`):\n The passages texts to be encoded. This can be a string or a list of strings if there are several passages.\n padding (`bool`, `str` or [`~utils.PaddingStrategy`], *optional*, defaults to `False`):\n Activates and controls padding. Accepts the following values:\n\n - `True` or `\'longest\'`: Pad to the longest sequence in the batch (or no padding if only a single sequence\n if provided).\n - `\'max_length\'`: Pad to a maximum length specified with the argument `max_length` or to the maximum\n acceptable input length for the model if that argument is not provided.\n - `False` or `\'do_not_pad\'` (default): No padding (i.e., can output a batch with sequences of different\n lengths).\n truncation (`bool`, `str` or [`~tokenization_utils_base.TruncationStrategy`], *optional*, defaults to `False`):\n Activates and controls truncation. Accepts the following values:\n\n - `True` or `\'longest_first\'`: Truncate to a maximum length specified with the argument `max_length` or to\n the maximum acceptable input length for the model if that argument is not provided. This will truncate\n token by token, removing a token from the longest sequence in the pair if a pair of sequences (or a batch\n of pairs) is provided.\n - `\'only_first\'`: Truncate to a maximum length specified with the argument `max_length` or to the maximum\n acceptable input length for the model if that argument is not provided. This will only truncate the first\n sequence of a pair if a pair of sequences (or a batch of pairs) is provided.\n - `\'only_second\'`: Truncate to a maximum length specified with the argument `max_length` or to the maximum\n acceptable input length for the model if that argument is not provided. This will only truncate the\n second sequence of a pair if a pair of sequences (or a batch of pairs) is provided.\n - `False` or `\'do_not_truncate\'` (default): No truncation (i.e., can output batch with sequence lengths\n greater than the model maximum admissible input size).\n max_length (`int`, *optional*):\n Controls the maximum length to use by one of the truncation/padding parameters.\n\n If left unset or set to `None`, this will use the predefined model maximum length if a maximum length\n is required by one of the truncation/padding parameters. If the model has no specific maximum input\n length (like XLNet) truncation/padding to a maximum length will be deactivated.\n return_tensors (`str` or [`~utils.TensorType`], *optional*):\n If set, will return tensors instead of list of python integers. Acceptable values are:\n\n - `\'tf\'`: Return TensorFlow `tf.constant` objects.\n - `\'pt\'`: Return PyTorch `torch.Tensor` objects.\n - `\'np\'`: Return Numpy `np.ndarray` objects.\n return_attention_mask (`bool`, *optional*):\n Whether or not to return the attention mask. If not set, will return the attention mask according to the\n specific tokenizer\'s default, defined by the `return_outputs` attribute.\n\n [What are attention masks?](../glossary#attention-mask)\n\n Return:\n `Dict[str, List[List[int]]]`: A dictionary with the following keys:\n\n - `input_ids`: List of token ids to be fed to a model.\n - `attention_mask`: List of indices specifying which tokens should be attended to by the model.\n '
@add_start_docstrings(_UpperCamelCase )
class _UpperCAmelCase :
"""simple docstring"""
def __call__( self : Any , lowerCAmelCase_ : Any , lowerCAmelCase_ : Optional[str] = None , lowerCAmelCase_ : Optional[str] = None , lowerCAmelCase_ : Union[bool, str] = False , lowerCAmelCase_ : Union[bool, str] = False , lowerCAmelCase_ : Optional[int] = None , lowerCAmelCase_ : Optional[Union[str, TensorType]] = None , lowerCAmelCase_ : Optional[bool] = None , **lowerCAmelCase_ : Optional[Any] , ) -> BatchEncoding:
if titles is None and texts is None:
return super().__call__(
lowerCAmelCase_ , padding=lowerCAmelCase_ , truncation=lowerCAmelCase_ , max_length=lowerCAmelCase_ , return_tensors=lowerCAmelCase_ , return_attention_mask=lowerCAmelCase_ , **lowerCAmelCase_ , )
elif titles is None or texts is None:
__lowerCAmelCase = titles if texts is None else texts
return super().__call__(
lowerCAmelCase_ , lowerCAmelCase_ , padding=lowerCAmelCase_ , truncation=lowerCAmelCase_ , max_length=lowerCAmelCase_ , return_tensors=lowerCAmelCase_ , return_attention_mask=lowerCAmelCase_ , **lowerCAmelCase_ , )
__lowerCAmelCase = titles if not isinstance(lowerCAmelCase_ , lowerCAmelCase_ ) else [titles]
__lowerCAmelCase = texts if not isinstance(lowerCAmelCase_ , lowerCAmelCase_ ) else [texts]
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = questions if not isinstance(lowerCAmelCase_ , lowerCAmelCase_ ) else [questions] * n_passages
assert len(lowerCAmelCase_ ) == len(
lowerCAmelCase_ ), f"""There should be as many titles than texts but got {len(lowerCAmelCase_ )} titles and {len(lowerCAmelCase_ )} texts."""
__lowerCAmelCase = super().__call__(lowerCAmelCase_ , lowerCAmelCase_ , padding=lowerCAmelCase_ , truncation=lowerCAmelCase_ )['input_ids']
__lowerCAmelCase = super().__call__(lowerCAmelCase_ , add_special_tokens=lowerCAmelCase_ , padding=lowerCAmelCase_ , truncation=lowerCAmelCase_ )['input_ids']
__lowerCAmelCase = {
'input_ids': [
(encoded_question_and_title + encoded_text)[:max_length]
if max_length is not None and truncation
else encoded_question_and_title + encoded_text
for encoded_question_and_title, encoded_text in zip(lowerCAmelCase_ , lowerCAmelCase_ )
]
}
if return_attention_mask is not False:
__lowerCAmelCase = []
for input_ids in encoded_inputs["input_ids"]:
attention_mask.append([int(input_id != self.pad_token_id ) for input_id in input_ids] )
__lowerCAmelCase = attention_mask
return self.pad(lowerCAmelCase_ , padding=lowerCAmelCase_ , max_length=lowerCAmelCase_ , return_tensors=lowerCAmelCase_ )
def lowercase ( self : int , lowerCAmelCase_ : BatchEncoding , lowerCAmelCase_ : DPRReaderOutput , lowerCAmelCase_ : int = 1_6 , lowerCAmelCase_ : int = 6_4 , lowerCAmelCase_ : int = 4 , ) -> List[DPRSpanPrediction]:
__lowerCAmelCase = reader_input['input_ids']
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = reader_output[:3]
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = sorted(range(lowerCAmelCase_ ) , reverse=lowerCAmelCase_ , key=relevance_logits.__getitem__ )
__lowerCAmelCase = []
for doc_id in sorted_docs:
__lowerCAmelCase = list(input_ids[doc_id] )
# assuming question & title information is at the beginning of the sequence
__lowerCAmelCase = sequence_ids.index(self.sep_token_id , 2 ) + 1 # second sep id
if sequence_ids[-1] == self.pad_token_id:
__lowerCAmelCase = sequence_ids.index(self.pad_token_id )
else:
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = self._get_best_spans(
start_logits=start_logits[doc_id][passage_offset:sequence_len] , end_logits=end_logits[doc_id][passage_offset:sequence_len] , max_answer_length=lowerCAmelCase_ , top_spans=lowerCAmelCase_ , )
for start_index, end_index in best_spans:
start_index += passage_offset
end_index += passage_offset
nbest_spans_predictions.append(
DPRSpanPrediction(
span_score=start_logits[doc_id][start_index] + end_logits[doc_id][end_index] , relevance_score=relevance_logits[doc_id] , doc_id=lowerCAmelCase_ , start_index=lowerCAmelCase_ , end_index=lowerCAmelCase_ , text=self.decode(sequence_ids[start_index : end_index + 1] ) , ) )
if len(lowerCAmelCase_ ) >= num_spans:
break
return nbest_spans_predictions[:num_spans]
def lowercase ( self : Any , lowerCAmelCase_ : List[int] , lowerCAmelCase_ : List[int] , lowerCAmelCase_ : int , lowerCAmelCase_ : int , ) -> List[DPRSpanPrediction]:
__lowerCAmelCase = []
for start_index, start_score in enumerate(lowerCAmelCase_ ):
for answer_length, end_score in enumerate(end_logits[start_index : start_index + max_answer_length] ):
scores.append(((start_index, start_index + answer_length), start_score + end_score) )
__lowerCAmelCase = sorted(lowerCAmelCase_ , key=lambda lowerCAmelCase_ : x[1] , reverse=lowerCAmelCase_ )
__lowerCAmelCase = []
for (start_index, end_index), score in scores:
assert start_index <= end_index, f"""Wrong span indices: [{start_index}:{end_index}]"""
__lowerCAmelCase = end_index - start_index + 1
assert length <= max_answer_length, f"""Span is too long: {length} > {max_answer_length}"""
if any(
start_index <= prev_start_index <= prev_end_index <= end_index
or prev_start_index <= start_index <= end_index <= prev_end_index
for (prev_start_index, prev_end_index) in chosen_span_intervals ):
continue
chosen_span_intervals.append((start_index, end_index) )
if len(lowerCAmelCase_ ) == top_spans:
break
return chosen_span_intervals
@add_end_docstrings(_UpperCamelCase )
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase ):
"""simple docstring"""
a_ = VOCAB_FILES_NAMES
a_ = READER_PRETRAINED_VOCAB_FILES_MAP
a_ = READER_PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
a_ = READER_PRETRAINED_INIT_CONFIGURATION
a_ = ["""input_ids""", """attention_mask"""]
a_ = DPRReaderTokenizer
| 53 |
import json
import logging
import math
import os
import sys
from dataclasses import dataclass, field
from typing import Optional
from datasets import Dataset, load_dataset
import transformers
from transformers import (
CONFIG_MAPPING,
MODEL_FOR_MASKED_LM_MAPPING,
AutoConfig,
AutoModelForMaskedLM,
AutoTokenizer,
DataCollatorForWholeWordMask,
HfArgumentParser,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
_snake_case : Optional[int] = logging.getLogger(__name__)
_snake_case : Dict = list(MODEL_FOR_MASKED_LM_MAPPING.keys())
_snake_case : List[Any] = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """If training from scratch, pass a model type from the list: """ + """, """.join(_UpperCamelCase )} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Override some existing default config settings when a model is trained from scratch. Example: """
"""n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained config name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Pretrained tokenizer name or path if not the same as model_name"""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co"""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."""} , )
a_ = field(
default="""main""" , metadata={"""help""": """The specific model version to use (can be a branch name, tag name or commit id)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Will use the token generated when running `huggingface-cli login` (necessary to use this script """
"""with private models)."""
)
} , )
def lowercase ( self : List[Any] ) -> List[Any]:
if self.config_overrides is not None and (self.config_name is not None or self.model_name_or_path is not None):
raise ValueError(
'--config_overrides can\'t be used in combination with --config_name or --model_name_or_path' )
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The name of the dataset to use (via the datasets library)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The configuration name of the dataset to use (via the datasets library)."""} )
a_ = field(default=_UpperCamelCase , metadata={"""help""": """The input training data file (a text file)."""} )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input evaluation data file to evaluate the perplexity on (a text file)."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input train ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """An optional input validation ref data file for whole word masking in Chinese."""} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """Overwrite the cached training and evaluation sets"""} )
a_ = field(
default=5 , metadata={
"""help""": """The percentage of the train set used as validation set in case there's no validation split"""
} , )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""The maximum total input sequence length after tokenization. Sequences longer """
"""than this will be truncated. Default to the max input length of the model."""
)
} , )
a_ = field(
default=_UpperCamelCase , metadata={"""help""": """The number of processes to use for the preprocessing."""} , )
a_ = field(
default=0.15 , metadata={"""help""": """Ratio of tokens to mask for masked language modeling loss"""} )
a_ = field(
default=_UpperCamelCase , metadata={
"""help""": (
"""Whether to pad all samples to `max_seq_length`. """
"""If False, will pad the samples dynamically when batching to the maximum length in the batch."""
)
} , )
def lowercase ( self : int ) -> int:
if self.train_file is not None:
__lowerCAmelCase = self.train_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`train_file` should be a csv, a json or a txt file."
if self.validation_file is not None:
__lowerCAmelCase = self.validation_file.split('.' )[-1]
assert extension in ["csv", "json", "txt"], "`validation_file` should be a csv, a json or a txt file."
def a_ ( lowerCAmelCase_ : List[str], lowerCAmelCase_ : Union[str, Any] ):
with open(lowerCAmelCase_, 'r', encoding='utf-8' ) as f:
__lowerCAmelCase = [json.loads(lowerCAmelCase_ ) for line in f.read().splitlines() if (len(lowerCAmelCase_ ) > 0 and not line.isspace())]
assert len(lowerCAmelCase_ ) == len(lowerCAmelCase_ )
__lowerCAmelCase = {c: dataset[c] for c in dataset.column_names}
__lowerCAmelCase = refs
return Dataset.from_dict(lowerCAmelCase_ )
def a_ ( ):
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
__lowerCAmelCase = 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.
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
__lowerCAmelCase = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
__lowerCAmelCase = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F"""Output directory ({training_args.output_dir}) already exists and is not empty. """
'Use --overwrite_output_dir to overcome.' )
elif last_checkpoint is not None:
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.' )
# 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 )], )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# 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}""" )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info('Training/evaluation parameters %s', lowerCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets: you can either provide your own CSV/JSON/TXT training and evaluation files (see below)
# or just provide the name of one of the public datasets available on the hub at https://huggingface.co/datasets/
# (the dataset will be downloaded automatically from the datasets Hub).
#
# For CSV/JSON files, this script will use the column called 'text' or the first column if no column called
# 'text' is found. You can easily tweak this behavior (see below).
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
__lowerCAmelCase = load_dataset(data_args.dataset_name, data_args.dataset_config_name )
if "validation" not in datasets.keys():
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[:{data_args.validation_split_percentage}%]""", )
__lowerCAmelCase = load_dataset(
data_args.dataset_name, data_args.dataset_config_name, split=F"""train[{data_args.validation_split_percentage}%:]""", )
else:
__lowerCAmelCase = {}
if data_args.train_file is not None:
__lowerCAmelCase = data_args.train_file
if data_args.validation_file is not None:
__lowerCAmelCase = data_args.validation_file
__lowerCAmelCase = data_args.train_file.split('.' )[-1]
if extension == "txt":
__lowerCAmelCase = 'text'
__lowerCAmelCase = load_dataset(lowerCAmelCase_, data_files=lowerCAmelCase_ )
# See more about loading any type of standard or custom dataset (from files, python dict, pandas DataFrame, etc) at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.config_name:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.config_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoConfig.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
__lowerCAmelCase = CONFIG_MAPPING[model_args.model_type]()
logger.warning('You are instantiating a new config instance from scratch.' )
if model_args.config_overrides is not None:
logger.info(F"""Overriding config: {model_args.config_overrides}""" )
config.update_from_string(model_args.config_overrides )
logger.info(F"""New config: {config}""" )
__lowerCAmelCase = {
'cache_dir': model_args.cache_dir,
'use_fast': model_args.use_fast_tokenizer,
'revision': model_args.model_revision,
'use_auth_token': True if model_args.use_auth_token else None,
}
if model_args.tokenizer_name:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.tokenizer_name, **lowerCAmelCase_ )
elif model_args.model_name_or_path:
__lowerCAmelCase = AutoTokenizer.from_pretrained(model_args.model_name_or_path, **lowerCAmelCase_ )
else:
raise ValueError(
'You are instantiating a new tokenizer from scratch. This is not supported by this script.'
'You can do it from another script, save it, and load it from here, using --tokenizer_name.' )
if model_args.model_name_or_path:
__lowerCAmelCase = AutoModelForMaskedLM.from_pretrained(
model_args.model_name_or_path, from_tf=bool('.ckpt' in model_args.model_name_or_path ), config=lowerCAmelCase_, cache_dir=model_args.cache_dir, revision=model_args.model_revision, use_auth_token=True if model_args.use_auth_token else None, )
else:
logger.info('Training new model from scratch' )
__lowerCAmelCase = AutoModelForMaskedLM.from_config(lowerCAmelCase_ )
model.resize_token_embeddings(len(lowerCAmelCase_ ) )
# Preprocessing the datasets.
# First we tokenize all the texts.
if training_args.do_train:
__lowerCAmelCase = datasets['train'].column_names
else:
__lowerCAmelCase = datasets['validation'].column_names
__lowerCAmelCase = 'text' if 'text' in column_names else column_names[0]
__lowerCAmelCase = 'max_length' if data_args.pad_to_max_length else False
def tokenize_function(lowerCAmelCase_ : str ):
# Remove empty lines
__lowerCAmelCase = [line for line in examples['text'] if len(lowerCAmelCase_ ) > 0 and not line.isspace()]
return tokenizer(examples['text'], padding=lowerCAmelCase_, truncation=lowerCAmelCase_, max_length=data_args.max_seq_length )
__lowerCAmelCase = datasets.map(
lowerCAmelCase_, batched=lowerCAmelCase_, num_proc=data_args.preprocessing_num_workers, remove_columns=[text_column_name], load_from_cache_file=not data_args.overwrite_cache, )
# Add the chinese references if provided
if data_args.train_ref_file is not None:
__lowerCAmelCase = add_chinese_references(tokenized_datasets['train'], data_args.train_ref_file )
if data_args.validation_ref_file is not None:
__lowerCAmelCase = add_chinese_references(
tokenized_datasets['validation'], data_args.validation_ref_file )
# If we have ref files, need to avoid it removed by trainer
__lowerCAmelCase = data_args.train_ref_file or data_args.validation_ref_file
if has_ref:
__lowerCAmelCase = False
# Data collator
# This one will take care of randomly masking the tokens.
__lowerCAmelCase = DataCollatorForWholeWordMask(tokenizer=lowerCAmelCase_, mlm_probability=data_args.mlm_probability )
# Initialize our Trainer
__lowerCAmelCase = Trainer(
model=lowerCAmelCase_, args=lowerCAmelCase_, train_dataset=tokenized_datasets['train'] if training_args.do_train else None, eval_dataset=tokenized_datasets['validation'] if training_args.do_eval else None, tokenizer=lowerCAmelCase_, data_collator=lowerCAmelCase_, )
# Training
if training_args.do_train:
if last_checkpoint is not None:
__lowerCAmelCase = last_checkpoint
elif model_args.model_name_or_path is not None and os.path.isdir(model_args.model_name_or_path ):
__lowerCAmelCase = model_args.model_name_or_path
else:
__lowerCAmelCase = None
__lowerCAmelCase = trainer.train(resume_from_checkpoint=lowerCAmelCase_ )
trainer.save_model() # Saves the tokenizer too for easy upload
__lowerCAmelCase = os.path.join(training_args.output_dir, 'train_results.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Train results *****' )
for key, value in sorted(train_result.metrics.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
# Need to save the state, since Trainer.save_model saves only the tokenizer with the model
trainer.state.save_to_json(os.path.join(training_args.output_dir, 'trainer_state.json' ) )
# Evaluation
__lowerCAmelCase = {}
if training_args.do_eval:
logger.info('*** Evaluate ***' )
__lowerCAmelCase = trainer.evaluate()
__lowerCAmelCase = math.exp(eval_output['eval_loss'] )
__lowerCAmelCase = perplexity
__lowerCAmelCase = os.path.join(training_args.output_dir, 'eval_results_mlm_wwm.txt' )
if trainer.is_world_process_zero():
with open(lowerCAmelCase_, 'w' ) as writer:
logger.info('***** Eval results *****' )
for key, value in sorted(results.items() ):
logger.info(F""" {key} = {value}""" )
writer.write(F"""{key} = {value}\n""" )
return results
def a_ ( lowerCAmelCase_ : Tuple ):
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()
| 53 | 1 |
import inspect
import math
import tempfile
import unittest
import numpy as np
from transformers import ViTMAEConfig
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 torch import nn
from transformers import ViTMAEForPreTraining, ViTMAEModel
from transformers.models.vit.modeling_vit import VIT_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import ViTImageProcessor
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[Any] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Optional[int]=1_3 , lowerCAmelCase_ : Any=3_0 , lowerCAmelCase_ : Dict=2 , lowerCAmelCase_ : List[Any]=3 , lowerCAmelCase_ : Optional[Any]=True , lowerCAmelCase_ : Union[str, Any]=True , lowerCAmelCase_ : Tuple=3_2 , lowerCAmelCase_ : Union[str, Any]=5 , lowerCAmelCase_ : str=4 , lowerCAmelCase_ : Tuple=3_7 , lowerCAmelCase_ : str="gelu" , lowerCAmelCase_ : Optional[int]=0.1 , lowerCAmelCase_ : Dict=0.1 , lowerCAmelCase_ : Any=1_0 , lowerCAmelCase_ : List[str]=0.02 , lowerCAmelCase_ : Tuple=3 , lowerCAmelCase_ : int=0.6 , lowerCAmelCase_ : Tuple=None , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = patch_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = is_training
__lowerCAmelCase = use_labels
__lowerCAmelCase = hidden_size
__lowerCAmelCase = num_hidden_layers
__lowerCAmelCase = num_attention_heads
__lowerCAmelCase = intermediate_size
__lowerCAmelCase = hidden_act
__lowerCAmelCase = hidden_dropout_prob
__lowerCAmelCase = attention_probs_dropout_prob
__lowerCAmelCase = type_sequence_label_size
__lowerCAmelCase = initializer_range
__lowerCAmelCase = mask_ratio
__lowerCAmelCase = scope
# in ViTMAE, the expected sequence length = (num_patches + 1) * (1 - config.mask_ratio), rounded above
# (we add 1 for the [CLS] token)
__lowerCAmelCase = (image_size // patch_size) ** 2
__lowerCAmelCase = int(math.ceil((1 - mask_ratio) * (num_patches + 1) ) )
def lowercase ( self : Any ) -> Optional[Any]:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = None
if self.use_labels:
__lowerCAmelCase = ids_tensor([self.batch_size] , self.type_sequence_label_size )
__lowerCAmelCase = self.get_config()
return config, pixel_values, labels
def lowercase ( self : Any ) -> int:
return ViTMAEConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , is_decoder=lowerCAmelCase_ , initializer_range=self.initializer_range , mask_ratio=self.mask_ratio , )
def lowercase ( self : List[str] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : Optional[int] ) -> Union[str, Any]:
__lowerCAmelCase = ViTMAEModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowercase ( self : Dict , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : str , lowerCAmelCase_ : int ) -> List[str]:
__lowerCAmelCase = ViTMAEForPreTraining(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(lowerCAmelCase_ )
__lowerCAmelCase = (self.image_size // self.patch_size) ** 2
__lowerCAmelCase = self.patch_size**2 * self.num_channels
self.parent.assertEqual(result.logits.shape , (self.batch_size, num_patches, expected_num_channels) )
# test greyscale images
__lowerCAmelCase = 1
__lowerCAmelCase = ViTMAEForPreTraining(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
__lowerCAmelCase = model(lowerCAmelCase_ )
__lowerCAmelCase = self.patch_size**2
self.parent.assertEqual(result.logits.shape , (self.batch_size, num_patches, expected_num_channels) )
def lowercase ( self : Tuple ) -> List[Any]:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (ViTMAEModel, ViTMAEForPreTraining) if is_torch_available() else ()
a_ = {"""feature-extraction""": ViTMAEModel} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : str ) -> Optional[Any]:
__lowerCAmelCase = ViTMAEModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ , hidden_size=3_7 )
def lowercase ( self : List[Any] ) -> List[str]:
self.config_tester.run_common_tests()
@unittest.skip(reason='ViTMAE does not use inputs_embeds' )
def lowercase ( self : List[str] ) -> Dict:
pass
def lowercase ( self : List[str] ) -> List[str]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
__lowerCAmelCase = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCAmelCase_ , nn.Linear ) )
def lowercase ( self : int ) -> int:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] ) -> Any:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def lowercase ( self : List[Any] ) -> int:
__lowerCAmelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_pretraining(*lowerCAmelCase_ )
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : int , lowerCAmelCase_ : str , lowerCAmelCase_ : List[Any] ) -> List[Any]:
# make masks reproducible
np.random.seed(2 )
__lowerCAmelCase = int((pt_model.config.image_size // pt_model.config.patch_size) ** 2 )
__lowerCAmelCase = np.random.uniform(size=(self.model_tester.batch_size, num_patches) )
__lowerCAmelCase = torch.from_numpy(lowerCAmelCase_ )
# Add `noise` argument.
# PT inputs will be prepared in `super().check_pt_tf_models()` with this added `noise` argument
__lowerCAmelCase = pt_noise
super().check_pt_tf_models(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def lowercase ( self : Tuple ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
# make random mask reproducible
torch.manual_seed(2 )
with torch.no_grad():
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
__lowerCAmelCase = outputs[0].cpu().numpy()
__lowerCAmelCase = 0
with tempfile.TemporaryDirectory() as tmpdirname:
model.save_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = model_class.from_pretrained(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
# make random mask reproducible
torch.manual_seed(2 )
with torch.no_grad():
__lowerCAmelCase = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
# Make sure we don't have nans
__lowerCAmelCase = after_outputs[0].cpu().numpy()
__lowerCAmelCase = 0
__lowerCAmelCase = np.amax(np.abs(out_a - out_a ) )
self.assertLessEqual(lowerCAmelCase_ , 1e-5 )
@unittest.skip(
reason='ViTMAE returns a random mask + ids_restore in each forward pass. See test_save_load\n to get deterministic results.' )
def lowercase ( self : Dict ) -> Optional[Any]:
pass
@unittest.skip(
reason='ViTMAE returns a random mask + ids_restore in each forward pass. See test_save_load\n to get deterministic results.' )
def lowercase ( self : str ) -> List[Any]:
pass
@unittest.skip(
reason='ViTMAE returns a random mask + ids_restore in each forward pass. See test_save_load\n to get deterministic results.' )
def lowercase ( self : int ) -> List[str]:
pass
@unittest.skip(reason='ViTMAE returns a random mask + ids_restore in each forward pass. See test_save_load' )
def lowercase ( self : List[str] ) -> Optional[Any]:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : List[Any] ) -> Any:
pass
@slow
def lowercase ( self : Dict ) -> List[Any]:
for model_name in VIT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCAmelCase = ViTMAEModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
def a_ ( ):
__lowerCAmelCase = Image.open('./tests/fixtures/tests_samples/COCO/000000039769.png' )
return image
@require_torch
@require_vision
class _UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def lowercase ( self : Tuple ) -> List[Any]:
return ViTImageProcessor.from_pretrained('facebook/vit-mae-base' ) if is_vision_available() else None
@slow
def lowercase ( self : Optional[int] ) -> Union[str, Any]:
# make random mask reproducible across the PT and TF model
np.random.seed(2 )
__lowerCAmelCase = ViTMAEForPreTraining.from_pretrained('facebook/vit-mae-base' ).to(lowerCAmelCase_ )
__lowerCAmelCase = self.default_image_processor
__lowerCAmelCase = prepare_img()
__lowerCAmelCase = image_processor(images=lowerCAmelCase_ , return_tensors='pt' ).to(lowerCAmelCase_ )
# prepare a noise vector that will be also used for testing the TF model
# (this way we can ensure that the PT and TF models operate on the same inputs)
__lowerCAmelCase = ViTMAEConfig()
__lowerCAmelCase = int((vit_mae_config.image_size // vit_mae_config.patch_size) ** 2 )
__lowerCAmelCase = np.random.uniform(size=(1, num_patches) )
# forward pass
with torch.no_grad():
__lowerCAmelCase = model(**lowerCAmelCase_ , noise=torch.from_numpy(lowerCAmelCase_ ).to(device=lowerCAmelCase_ ) )
# verify the logits
__lowerCAmelCase = torch.Size((1, 1_9_6, 7_6_8) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
__lowerCAmelCase = torch.tensor(
[[-0.05_48, -1.70_23, -0.93_25], [0.37_21, -0.56_70, -0.22_33], [0.82_35, -1.38_78, -0.35_24]] )
self.assertTrue(torch.allclose(outputs.logits[0, :3, :3] , expected_slice.to(lowerCAmelCase_ ) , atol=1e-4 ) )
| 53 |
def a_ ( lowerCAmelCase_ : int = 200_0000 ):
__lowerCAmelCase = [0 for i in range(n + 1 )]
__lowerCAmelCase = 1
__lowerCAmelCase = 1
for i in range(2, int(n**0.5 ) + 1 ):
if primality_list[i] == 0:
for j in range(i * i, n + 1, lowerCAmelCase_ ):
__lowerCAmelCase = 1
__lowerCAmelCase = 0
for i in range(lowerCAmelCase_ ):
if primality_list[i] == 0:
sum_of_primes += i
return sum_of_primes
if __name__ == "__main__":
print(F"""{solution() = }""")
| 53 | 1 |
from __future__ import annotations
from collections.abc import Iterable, Iterator
from dataclasses import dataclass
_snake_case : Dict = (3, 9, -11, 0, 7, 5, 1, -1)
_snake_case : List[str] = (4, 6, 2, 0, 8, 10, 3, -2)
@dataclass
class _UpperCAmelCase :
"""simple docstring"""
a_ = 42
a_ = 42
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : int , lowerCAmelCase_ : Iterable[int] ) -> None:
__lowerCAmelCase = None
for i in sorted(lowerCAmelCase_ , reverse=lowerCAmelCase_ ):
__lowerCAmelCase = Node(lowerCAmelCase_ , self.head )
def __iter__( self : List[Any] ) -> Iterator[int]:
__lowerCAmelCase = self.head
while node:
yield node.data
__lowerCAmelCase = node.next_node
def __len__( self : str ) -> int:
return sum(1 for _ in self )
def __str__( self : str ) -> str:
return " -> ".join([str(lowerCAmelCase_ ) for node in self] )
def a_ ( lowerCAmelCase_ : SortedLinkedList, lowerCAmelCase_ : SortedLinkedList ):
return SortedLinkedList(list(lowerCAmelCase_ ) + list(lowerCAmelCase_ ) )
if __name__ == "__main__":
import doctest
doctest.testmod()
_snake_case : Optional[Any] = SortedLinkedList
print(merge_lists(SSL(test_data_odd), SSL(test_data_even)))
| 53 |
import json
import logging
import os
import sys
from pathlib import Path
import finetune_rag
from transformers.file_utils import is_apex_available
from transformers.testing_utils import (
TestCasePlus,
execute_subprocess_async,
require_ray,
require_torch_gpu,
require_torch_multi_gpu,
)
logging.basicConfig(level=logging.DEBUG)
_snake_case : Tuple = logging.getLogger()
_snake_case : Any = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class _UpperCAmelCase ( _UpperCamelCase ):
"""simple docstring"""
def lowercase ( self : Any , lowerCAmelCase_ : Dict ) -> Optional[int]:
os.makedirs(lowerCAmelCase_ , exist_ok=lowerCAmelCase_ )
__lowerCAmelCase = {'source': 'What is love ?', 'target': 'life'}
__lowerCAmelCase = {'train': 1_2, 'val': 2, 'test': 2}
for split in ["train", "test", "val"]:
for field in ["source", "target"]:
__lowerCAmelCase = '\n'.join([contents[field]] * n_lines[split] )
with open(os.path.join(lowerCAmelCase_ , f"""{split}.{field}""" ) , 'w' ) as f:
f.write(lowerCAmelCase_ )
def lowercase ( self : Dict , lowerCAmelCase_ : int , lowerCAmelCase_ : str = "pytorch" ) -> List[str]:
__lowerCAmelCase = self.get_auto_remove_tmp_dir()
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'output' )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'data' )
self._create_dummy_data(data_dir=lowerCAmelCase_ )
__lowerCAmelCase = f"""
--data_dir {data_dir} \
--output_dir {output_dir} \
--model_name_or_path facebook/rag-sequence-base \
--model_type rag_sequence \
--do_train \
--do_predict \
--n_val -1 \
--val_check_interval 1.0 \
--train_batch_size 2 \
--eval_batch_size 1 \
--max_source_length 25 \
--max_target_length 25 \
--val_max_target_length 25 \
--test_max_target_length 25 \
--label_smoothing 0.1 \
--dropout 0.1 \
--attention_dropout 0.1 \
--weight_decay 0.001 \
--adam_epsilon 1e-08 \
--max_grad_norm 0.1 \
--lr_scheduler polynomial \
--learning_rate 3e-04 \
--num_train_epochs 1 \
--warmup_steps 4 \
--gradient_accumulation_steps 1 \
--distributed-port 8787 \
--use_dummy_dataset 1 \
--distributed_retriever {distributed_retriever} \
""".split()
if gpus > 0:
testargs.append(f"""--gpus={gpus}""" )
if is_apex_available():
testargs.append('--fp16' )
else:
testargs.append('--gpus=0' )
testargs.append('--distributed_backend=ddp_cpu' )
testargs.append('--num_processes=2' )
__lowerCAmelCase = [sys.executable, str(Path(finetune_rag.__file__ ).resolve() )] + testargs
execute_subprocess_async(lowerCAmelCase_ , env=self.get_env() )
__lowerCAmelCase = os.path.join(lowerCAmelCase_ , 'metrics.json' )
with open(lowerCAmelCase_ ) as f:
__lowerCAmelCase = json.load(lowerCAmelCase_ )
return result
@require_torch_gpu
def lowercase ( self : str ) -> int:
__lowerCAmelCase = self._run_finetune(gpus=1 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
def lowercase ( self : List[str] ) -> Dict:
__lowerCAmelCase = self._run_finetune(gpus=2 )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_gpu
@require_ray
def lowercase ( self : int ) -> Tuple:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
@require_torch_multi_gpu
@require_ray
def lowercase ( self : List[Any] ) -> str:
__lowerCAmelCase = self._run_finetune(gpus=1 , distributed_retriever='ray' )
self.assertGreaterEqual(result['test'][0]['test_avg_em'] , 0.2 )
| 53 | 1 |
import argparse
import json
import os
import re
import shutil
import torch
from transformers import BioGptConfig, BioGptForCausalLM
from transformers.models.biogpt.tokenization_biogpt import VOCAB_FILES_NAMES
from transformers.tokenization_utils_base import TOKENIZER_CONFIG_FILE
from transformers.utils import WEIGHTS_NAME, logging
logging.set_verbosity_warning()
_snake_case : Union[str, Any] = 2
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : Tuple , *, # begin keyword-only arguments
lowerCAmelCase_ : str="<s>" , lowerCAmelCase_ : Dict="<pad>" , lowerCAmelCase_ : Any="</s>" , lowerCAmelCase_ : List[str]="<unk>" , lowerCAmelCase_ : Optional[Any]=None , ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase = bos, unk, pad, eos
__lowerCAmelCase = []
__lowerCAmelCase = []
__lowerCAmelCase = {}
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = self.add_symbol(lowerCAmelCase_ )
if extra_special_symbols:
for s in extra_special_symbols:
self.add_symbol(lowerCAmelCase_ )
__lowerCAmelCase = len(self.symbols )
def __eq__( self : Dict , lowerCAmelCase_ : Dict ) -> str:
return self.indices == other.indices
def __getitem__( self : List[Any] , lowerCAmelCase_ : int ) -> Union[str, Any]:
if idx < len(self.symbols ):
return self.symbols[idx]
return self.unk_word
def __len__( self : Tuple ) -> List[Any]:
return len(self.symbols )
def __contains__( self : Optional[Any] , lowerCAmelCase_ : Dict ) -> Optional[int]:
return sym in self.indices
@classmethod
def lowercase ( cls : Dict , lowerCAmelCase_ : str ) -> str:
__lowerCAmelCase = cls()
d.add_from_file(lowerCAmelCase_ )
return d
def lowercase ( self : Union[str, Any] , lowerCAmelCase_ : List[str] , lowerCAmelCase_ : int=1 , lowerCAmelCase_ : Any=False ) -> Optional[Any]:
if word in self.indices and not overwrite:
__lowerCAmelCase = self.indices[word]
__lowerCAmelCase = self.count[idx] + n
return idx
else:
__lowerCAmelCase = len(self.symbols )
__lowerCAmelCase = idx
self.symbols.append(lowerCAmelCase_ )
self.count.append(lowerCAmelCase_ )
return idx
def lowercase ( self : str , lowerCAmelCase_ : Union[str, Any] ) -> Dict:
return 0
def lowercase ( self : Tuple , lowerCAmelCase_ : Union[str, Any] ) -> int:
if isinstance(lowerCAmelCase_ , lowerCAmelCase_ ):
try:
with open(lowerCAmelCase_ , 'r' , encoding='utf-8' ) as fd:
self.add_from_file(lowerCAmelCase_ )
except FileNotFoundError as fnfe:
raise fnfe
except UnicodeError:
raise Exception('Incorrect encoding detected in {}, please rebuild the dataset'.format(lowerCAmelCase_ ) )
return
__lowerCAmelCase = f.readlines()
__lowerCAmelCase = self._load_meta(lowerCAmelCase_ )
for line in lines[indices_start_line:]:
try:
__lowerCAmelCase , __lowerCAmelCase = line.rstrip().rsplit(' ' , 1 )
if field == "#fairseq:overwrite":
__lowerCAmelCase = True
__lowerCAmelCase , __lowerCAmelCase = line.rsplit(' ' , 1 )
else:
__lowerCAmelCase = False
__lowerCAmelCase = int(lowerCAmelCase_ )
__lowerCAmelCase = line
if word in self and not overwrite:
raise RuntimeError(
'Duplicate word found when loading Dictionary: \'{}\'. '
'Duplicate words can overwrite earlier ones by adding the '
'#fairseq:overwrite flag at the end of the corresponding row '
'in the dictionary file. If using the Camembert model, please '
'download an updated copy of the model file.'.format(lowerCAmelCase_ ) )
self.add_symbol(lowerCAmelCase_ , n=lowerCAmelCase_ , overwrite=lowerCAmelCase_ )
except ValueError:
raise ValueError('Incorrect dictionary format, expected \'<token> <cnt> [flags]\'' )
def a_ ( lowerCAmelCase_ : List[str] ):
# (1) remove word breaking symbol, (2) add word ending symbol where the word is not broken up,
# e.g.: d = {'le@@': 5, 'tt@@': 6, 'er': 7} => {'le': 5, 'tt': 6, 'er</w>': 7}
__lowerCAmelCase = dict((re.sub(R'@@$', '', lowerCAmelCase_ ), v) if k.endswith('@@' ) else (re.sub(R'$', '</w>', lowerCAmelCase_ ), v) for k, v in d.items() )
__lowerCAmelCase = '<s> <pad> </s> <unk>'.split()
# restore the special tokens
for k in keep_keys:
del da[F"""{k}</w>"""]
__lowerCAmelCase = d[k] # restore
return da
def a_ ( lowerCAmelCase_ : Union[str, Any], lowerCAmelCase_ : List[str] ):
# prep
if not os.path.exists(lowerCAmelCase_ ):
raise ValueError(F"""path {biogpt_checkpoint_path} does not exist!""" )
os.makedirs(lowerCAmelCase_, exist_ok=lowerCAmelCase_ )
print(F"""Writing results to {pytorch_dump_folder_path}""" )
# handle various types of models
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'checkpoint.pt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {checkpoint_file} does not exist!""" )
__lowerCAmelCase = torch.load(lowerCAmelCase_, map_location='cpu' )
__lowerCAmelCase = chkpt['cfg']['model']
# dicts
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'dict.txt' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {dict_file} does not exist!""" )
__lowerCAmelCase = Dictionary.load(lowerCAmelCase_ )
__lowerCAmelCase = rewrite_dict_keys(src_dict.indices )
__lowerCAmelCase = len(lowerCAmelCase_ )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['vocab_file'] )
print(F"""Generating {src_vocab_file} of {src_vocab_size} records""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# merges_file (bpecodes)
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'bpecodes' )
if not os.path.isfile(lowerCAmelCase_ ):
raise ValueError(F"""path to the file {bpecodes_file} does not exist!""" )
__lowerCAmelCase = os.path.join(lowerCAmelCase_, VOCAB_FILES_NAMES['merges_file'] )
shutil.copyfile(lowerCAmelCase_, lowerCAmelCase_ )
# model config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, 'config.json' )
__lowerCAmelCase = {
'activation_dropout': args['activation_dropout'],
'architectures': ['BioGptForCausalLM'],
'attention_probs_dropout_prob': args['attention_dropout'],
'bos_token_id': 0,
'eos_token_id': 2,
'hidden_act': args['activation_fn'],
'hidden_dropout_prob': args['dropout'],
'hidden_size': args['decoder_embed_dim'],
'initializer_range': 0.02,
'intermediate_size': args['decoder_ffn_embed_dim'],
'layer_norm_eps': 1E-12,
'layerdrop': args['decoder_layerdrop'],
'max_position_embeddings': args['max_target_positions'],
'model_type': 'biogpt',
'num_attention_heads': args['decoder_attention_heads'],
'num_hidden_layers': args['decoder_layers'],
'pad_token_id': 1,
'scale_embedding': not args['no_scale_embedding'],
'tie_word_embeddings': args['share_decoder_input_output_embed'],
'vocab_size': src_vocab_size,
}
# good hparam defaults to start with
print(F"""Generating {biogpt_model_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# tokenizer config
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = {
'bos_token': '<s>',
'eos_token': '</s>',
'model_max_length': 1024,
'pad_token': '<pad>',
'special_tokens_map_file': None,
'tokenizer_class': 'BioGptTokenizer',
'unk_token': '<unk>',
}
print(F"""Generating {biogpt_tokenizer_config_file}""" )
with open(lowerCAmelCase_, 'w', encoding='utf-8' ) as f:
f.write(json.dumps(lowerCAmelCase_, ensure_ascii=lowerCAmelCase_, indent=lowerCAmelCase_ ) )
# model
__lowerCAmelCase = chkpt['model']
# remove unneeded keys
__lowerCAmelCase = [
'decoder.version',
]
for k in ignore_keys:
model_state_dict.pop(lowerCAmelCase_, lowerCAmelCase_ )
__lowerCAmelCase = list(model_state_dict.keys() )
for layer_name in layer_names:
if layer_name.endswith('output_projection.weight' ):
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
else:
__lowerCAmelCase = model_state_dict.pop(lowerCAmelCase_ )
__lowerCAmelCase = BioGptConfig.from_pretrained(lowerCAmelCase_ )
__lowerCAmelCase = BioGptForCausalLM(lowerCAmelCase_ )
# check that it loads ok
model_new.load_state_dict(lowerCAmelCase_ )
# save
__lowerCAmelCase = os.path.join(lowerCAmelCase_, lowerCAmelCase_ )
print(F"""Generating {pytorch_weights_dump_path}""" )
torch.save(lowerCAmelCase_, lowerCAmelCase_ )
print('Conversion is done!' )
if __name__ == "__main__":
_snake_case : str = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'--biogpt_checkpoint_path',
default=None,
type=str,
required=True,
help=(
'Path to the official PyTorch checkpoint file which is expected to reside in the dump dir with dicts,'
' bpecodes, etc.'
),
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, required=True, help='Path to the output PyTorch model.'
)
_snake_case : int = parser.parse_args()
convert_biogpt_checkpoint_to_pytorch(args.biogpt_checkpoint_path, args.pytorch_dump_folder_path)
| 53 |
import copy
import inspect
import unittest
from transformers import AutoBackbone
from transformers.configuration_utils import PretrainedConfig
from transformers.testing_utils import require_timm, require_torch, torch_device
from transformers.utils.import_utils import is_torch_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor
if is_torch_available():
import torch
from transformers import TimmBackbone, TimmBackboneConfig
from ...test_pipeline_mixin import PipelineTesterMixin
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self : List[str] , lowerCAmelCase_ : int , lowerCAmelCase_ : List[str]=None , lowerCAmelCase_ : List[Any]=None , lowerCAmelCase_ : Optional[int]=None , lowerCAmelCase_ : Union[str, Any]="resnet50" , lowerCAmelCase_ : str=3 , lowerCAmelCase_ : List[str]=3_2 , lowerCAmelCase_ : Dict=3 , lowerCAmelCase_ : Dict=True , lowerCAmelCase_ : Optional[Any]=True , ) -> List[Any]:
__lowerCAmelCase = parent
__lowerCAmelCase = out_indices if out_indices is not None else [4]
__lowerCAmelCase = stage_names
__lowerCAmelCase = out_features
__lowerCAmelCase = backbone
__lowerCAmelCase = batch_size
__lowerCAmelCase = image_size
__lowerCAmelCase = num_channels
__lowerCAmelCase = use_pretrained_backbone
__lowerCAmelCase = is_training
def lowercase ( self : List[str] ) -> Any:
__lowerCAmelCase = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__lowerCAmelCase = self.get_config()
return config, pixel_values
def lowercase ( self : List[Any] ) -> Union[str, Any]:
return TimmBackboneConfig(
image_size=self.image_size , num_channels=self.num_channels , out_features=self.out_features , out_indices=self.out_indices , stage_names=self.stage_names , use_pretrained_backbone=self.use_pretrained_backbone , backbone=self.backbone , )
def lowercase ( self : List[str] , lowerCAmelCase_ : Tuple , lowerCAmelCase_ : Tuple ) -> int:
__lowerCAmelCase = TimmBackbone(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
__lowerCAmelCase = model(lowerCAmelCase_ )
self.parent.assertEqual(
result.feature_map[-1].shape , (self.batch_size, model.channels[-1], 1_4, 1_4) , )
def lowercase ( self : List[str] ) -> str:
__lowerCAmelCase = self.prepare_config_and_inputs()
__lowerCAmelCase , __lowerCAmelCase = config_and_inputs
__lowerCAmelCase = {'pixel_values': pixel_values}
return config, inputs_dict
@require_torch
@require_timm
class _UpperCAmelCase ( _UpperCamelCase , _UpperCamelCase , _UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
a_ = (TimmBackbone,) if is_torch_available() else ()
a_ = {"""feature-extraction""": TimmBackbone} if is_torch_available() else {}
a_ = False
a_ = False
a_ = False
a_ = False
def lowercase ( self : Tuple ) -> int:
__lowerCAmelCase = TimmBackboneModelTester(self )
__lowerCAmelCase = ConfigTester(self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ )
def lowercase ( self : Dict ) -> List[str]:
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def lowercase ( self : Union[str, Any] ) -> Optional[int]:
__lowerCAmelCase = 'resnet18'
__lowerCAmelCase = 'microsoft/resnet-18'
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(len(timm_model.stage_names ) , len(transformers_model.stage_names ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
# Out indices are set to the last layer by default. For timm models, we don't know
# the number of layers in advance, so we set it to (-1,), whereas for transformers
# models, we set it to [len(stage_names) - 1] (kept for backward compatibility).
self.assertEqual(timm_model.out_indices , (-1,) )
self.assertEqual(transformers_model.out_indices , [len(timm_model.stage_names ) - 1] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , use_timm_backbone=lowerCAmelCase_ , out_indices=[1, 2, 3] )
__lowerCAmelCase = AutoBackbone.from_pretrained(lowerCAmelCase_ , out_indices=[1, 2, 3] )
self.assertEqual(timm_model.out_indices , transformers_model.out_indices )
self.assertEqual(len(timm_model.out_features ) , len(transformers_model.out_features ) )
self.assertEqual(timm_model.channels , transformers_model.channels )
@unittest.skip('TimmBackbone doesn\'t support feed forward chunking' )
def lowercase ( self : List[str] ) -> Tuple:
pass
@unittest.skip('TimmBackbone doesn\'t have num_hidden_layers attribute' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('TimmBackbone initialization is managed on the timm side' )
def lowercase ( self : str ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Any ) -> str:
pass
@unittest.skip('TimmBackbone models doesn\'t have inputs_embeds' )
def lowercase ( self : Optional[int] ) -> Optional[Any]:
pass
@unittest.skip('TimmBackbone model cannot be created without specifying a backbone checkpoint' )
def lowercase ( self : Dict ) -> Any:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Any ) -> Optional[int]:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : Union[str, Any] ) -> Tuple:
pass
@unittest.skip('model weights aren\'t tied in TimmBackbone.' )
def lowercase ( self : List[str] ) -> Optional[int]:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Dict ) -> int:
pass
@unittest.skip('Only checkpoints on timm can be loaded into TimmBackbone' )
def lowercase ( self : Tuple ) -> List[str]:
pass
@unittest.skip('TimmBackbone doesn\'t have hidden size info in its configuration.' )
def lowercase ( self : int ) -> Optional[int]:
pass
@unittest.skip('TimmBackbone doesn\'t support output_attentions.' )
def lowercase ( self : Union[str, Any] ) -> str:
pass
@unittest.skip('Safetensors is not supported by timm.' )
def lowercase ( self : Dict ) -> str:
pass
@unittest.skip('Will be fixed soon by reducing the size of the model used for common tests.' )
def lowercase ( self : List[str] ) -> Optional[Any]:
pass
def lowercase ( self : Union[str, Any] ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
__lowerCAmelCase = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__lowerCAmelCase = [*signature.parameters.keys()]
__lowerCAmelCase = ['pixel_values']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def lowercase ( self : int ) -> Union[str, Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCAmelCase = True
__lowerCAmelCase = self.has_attentions
# no need to test all models as different heads yield the same functionality
__lowerCAmelCase = self.all_model_classes[0]
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
__lowerCAmelCase = self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ )
__lowerCAmelCase = model(**lowerCAmelCase_ )
__lowerCAmelCase = outputs[0][-1]
# Encoder-/Decoder-only models
__lowerCAmelCase = outputs.hidden_states[0]
hidden_states.retain_grad()
if self.has_attentions:
__lowerCAmelCase = outputs.attentions[0]
attentions.retain_grad()
output.flatten()[0].backward(retain_graph=lowerCAmelCase_ )
self.assertIsNotNone(hidden_states.grad )
if self.has_attentions:
self.assertIsNotNone(attentions.grad )
def lowercase ( self : Dict ) -> Optional[Any]:
__lowerCAmelCase , __lowerCAmelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , len(config.out_indices ) )
self.assertEqual(len(model.channels ) , len(config.out_indices ) )
# Check output of last stage is taken if out_features=None, out_indices=None
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = None
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
self.assertEqual(len(result.feature_maps ) , 1 )
self.assertEqual(len(model.channels ) , 1 )
# Check backbone can be initialized with fresh weights
__lowerCAmelCase = copy.deepcopy(lowerCAmelCase_ )
__lowerCAmelCase = False
__lowerCAmelCase = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
__lowerCAmelCase = model(**lowerCAmelCase_ )
| 53 | 1 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.