code
stringlengths 87
55.2k
| code_codestyle
int64 0
349
| style_context
stringlengths 135
49.1k
| style_context_codestyle
int64 0
349
| label
int64 0
1
|
|---|---|---|---|---|
import importlib.metadata
import warnings
from copy import deepcopy
from packaging import version
from ..utils import logging
from .import_utils import is_accelerate_available, is_bitsandbytes_available
if is_bitsandbytes_available():
import bitsandbytes as bnb
import torch
import torch.nn as nn
from ..pytorch_utils import ConvaD
if is_accelerate_available():
from accelerate import init_empty_weights
from accelerate.utils import find_tied_parameters
a_ = logging.get_logger(__name__)
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : Tuple ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : int=None ,_UpperCamelCase : Any=None ):
# Recurse if needed
if "." in tensor_name:
__lowerCamelCase = tensor_name.split('''.''' )
for split in splits[:-1]:
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if new_module is None:
raise ValueError(F"""{module} has no attribute {split}.""" )
__lowerCamelCase = new_module
__lowerCamelCase = splits[-1]
if tensor_name not in module._parameters and tensor_name not in module._buffers:
raise ValueError(F"""{module} does not have a parameter or a buffer named {tensor_name}.""" )
__lowerCamelCase = tensor_name in module._buffers
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if old_value.device == torch.device('''meta''' ) and device not in ["meta", torch.device('''meta''' )] and value is None:
raise ValueError(F"""{tensor_name} is on the meta device, we need a `value` to put in on {device}.""" )
__lowerCamelCase = False
__lowerCamelCase = False
if is_buffer or not is_bitsandbytes_available():
__lowerCamelCase = False
__lowerCamelCase = False
else:
__lowerCamelCase = hasattr(bnb.nn ,'''Params4bit''' ) and isinstance(module._parameters[tensor_name] ,bnb.nn.Paramsabit )
__lowerCamelCase = isinstance(module._parameters[tensor_name] ,bnb.nn.IntaParams )
if is_abit or is_abit:
__lowerCamelCase = module._parameters[tensor_name]
if param.device.type != "cuda":
if value is None:
__lowerCamelCase = old_value.to(_UpperCamelCase )
elif isinstance(_UpperCamelCase ,torch.Tensor ):
__lowerCamelCase = value.to('''cpu''' )
if value.dtype == torch.inta:
__lowerCamelCase = version.parse(importlib.metadata.version('''bitsandbytes''' ) ) > version.parse(
'''0.37.2''' )
if not is_abit_serializable:
raise ValueError(
'''Detected int8 weights but the version of bitsandbytes is not compatible with int8 serialization. '''
'''Make sure to download the latest `bitsandbytes` version. `pip install --upgrade bitsandbytes`.''' )
else:
__lowerCamelCase = torch.tensor(_UpperCamelCase ,device='''cpu''' )
# Support models using `Conv1D` in place of `nn.Linear` (e.g. gpt2) by transposing the weight matrix prior to quantization.
# Since weights are saved in the correct "orientation", we skip transposing when loading.
if issubclass(module.source_cls ,_UpperCamelCase ) and fpaa_statistics is None:
__lowerCamelCase = new_value.T
__lowerCamelCase = old_value.__dict__
if is_abit:
__lowerCamelCase = bnb.nn.IntaParams(_UpperCamelCase ,requires_grad=_UpperCamelCase ,**_UpperCamelCase ).to(_UpperCamelCase )
elif is_abit:
__lowerCamelCase = bnb.nn.Paramsabit(_UpperCamelCase ,requires_grad=_UpperCamelCase ,**_UpperCamelCase ).to(_UpperCamelCase )
__lowerCamelCase = new_value
if fpaa_statistics is not None:
setattr(module.weight ,'''SCB''' ,fpaa_statistics.to(_UpperCamelCase ) )
else:
if value is None:
__lowerCamelCase = old_value.to(_UpperCamelCase )
elif isinstance(_UpperCamelCase ,torch.Tensor ):
__lowerCamelCase = value.to(_UpperCamelCase )
else:
__lowerCamelCase = torch.tensor(_UpperCamelCase ,device=_UpperCamelCase )
if is_buffer:
__lowerCamelCase = new_value
else:
__lowerCamelCase = nn.Parameter(_UpperCamelCase ,requires_grad=old_value.requires_grad )
__lowerCamelCase = new_value
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : Any=None ,_UpperCamelCase : Optional[Any]=None ,_UpperCamelCase : str=None ,_UpperCamelCase : str=False ):
for name, module in model.named_children():
if current_key_name is None:
__lowerCamelCase = []
current_key_name.append(_UpperCamelCase )
if (isinstance(_UpperCamelCase ,nn.Linear ) or isinstance(_UpperCamelCase ,_UpperCamelCase )) and name not in modules_to_not_convert:
# Check if the current key is not in the `modules_to_not_convert`
if not any(key in '''.'''.join(_UpperCamelCase ) for key in modules_to_not_convert ):
with init_empty_weights():
if isinstance(_UpperCamelCase ,_UpperCamelCase ):
__lowerCamelCase ,__lowerCamelCase = module.weight.shape
else:
__lowerCamelCase = module.in_features
__lowerCamelCase = module.out_features
if quantization_config.quantization_method() == "llm_int8":
__lowerCamelCase = bnb.nn.LinearabitLt(
_UpperCamelCase ,_UpperCamelCase ,module.bias is not None ,has_fpaa_weights=quantization_config.llm_inta_has_fpaa_weight ,threshold=quantization_config.llm_inta_threshold ,)
__lowerCamelCase = True
else:
if (
quantization_config.llm_inta_skip_modules is not None
and name in quantization_config.llm_inta_skip_modules
):
pass
else:
__lowerCamelCase = bnb.nn.Linearabit(
_UpperCamelCase ,_UpperCamelCase ,module.bias is not None ,quantization_config.bnb_abit_compute_dtype ,compress_statistics=quantization_config.bnb_abit_use_double_quant ,quant_type=quantization_config.bnb_abit_quant_type ,)
__lowerCamelCase = True
# Store the module class in case we need to transpose the weight later
__lowerCamelCase = type(_UpperCamelCase )
# Force requires grad to False to avoid unexpected errors
model._modules[name].requires_grad_(_UpperCamelCase )
if len(list(module.children() ) ) > 0:
__lowerCamelCase ,__lowerCamelCase = _replace_with_bnb_linear(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,has_been_replaced=_UpperCamelCase ,)
# Remove the last key for recursion
current_key_name.pop(-1 )
return model, has_been_replaced
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : Optional[int]=None ,_UpperCamelCase : str=None ,_UpperCamelCase : List[str]=None ):
__lowerCamelCase = ['''lm_head'''] if modules_to_not_convert is None else modules_to_not_convert
__lowerCamelCase ,__lowerCamelCase = _replace_with_bnb_linear(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
if not has_been_replaced:
logger.warning(
'''You are loading your model in 8bit or 4bit but no linear modules were found in your model.'''
''' Please double check your model architecture, or submit an issue on github if you think this is'''
''' a bug.''' )
return model
def a__ ( *_UpperCamelCase : Dict ,**_UpperCamelCase : Optional[int] ):
warnings.warn(
'''`replace_8bit_linear` will be deprecated in a future version, please use `replace_with_bnb_linear` instead''' ,_UpperCamelCase ,)
return replace_with_bnb_linear(*_UpperCamelCase ,**_UpperCamelCase )
def a__ ( *_UpperCamelCase : Any ,**_UpperCamelCase : int ):
warnings.warn(
'''`set_module_8bit_tensor_to_device` will be deprecated in a future version, please use `set_module_quantized_tensor_to_device` instead''' ,_UpperCamelCase ,)
return set_module_quantized_tensor_to_device(*_UpperCamelCase ,**_UpperCamelCase )
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = deepcopy(_UpperCamelCase ) # this has 0 cost since it is done inside `init_empty_weights` context manager`
tied_model.tie_weights()
__lowerCamelCase = find_tied_parameters(_UpperCamelCase )
# For compatibility with Accelerate < 0.18
if isinstance(_UpperCamelCase ,_UpperCamelCase ):
__lowerCamelCase = sum(list(tied_params.values() ) ,[] ) + list(tied_params.keys() )
else:
__lowerCamelCase = sum(_UpperCamelCase ,[] )
__lowerCamelCase = len(_UpperCamelCase ) > 0
# Check if it is a base model
__lowerCamelCase = not hasattr(_UpperCamelCase ,model.base_model_prefix )
# Ignore this for base models (BertModel, GPT2Model, etc.)
if (not has_tied_params) and is_base_model:
return []
# otherwise they have an attached head
__lowerCamelCase = list(model.named_children() )
__lowerCamelCase = [list_modules[-1][0]]
# add last module together with tied weights
__lowerCamelCase = set(_UpperCamelCase ) - set(_UpperCamelCase )
__lowerCamelCase = list(set(_UpperCamelCase ) ) + list(_UpperCamelCase )
# remove ".weight" from the keys
__lowerCamelCase = ['''.weight''', '''.bias''']
__lowerCamelCase = []
for name in list_untouched:
for name_to_remove in names_to_remove:
if name_to_remove in name:
__lowerCamelCase = name.replace(_UpperCamelCase ,'''''' )
filtered_module_names.append(_UpperCamelCase )
return filtered_module_names
| 330 |
import argparse
import fairseq
import torch
from transformers import UniSpeechSatConfig, UniSpeechSatForCTC, UniSpeechSatForPreTraining, logging
logging.set_verbosity_info()
a_ = logging.get_logger(__name__)
a_ = {
"""post_extract_proj""": """feature_projection.projection""",
"""encoder.pos_conv.0""": """encoder.pos_conv_embed.conv""",
"""self_attn.k_proj""": """encoder.layers.*.attention.k_proj""",
"""self_attn.v_proj""": """encoder.layers.*.attention.v_proj""",
"""self_attn.q_proj""": """encoder.layers.*.attention.q_proj""",
"""self_attn.out_proj""": """encoder.layers.*.attention.out_proj""",
"""self_attn_layer_norm""": """encoder.layers.*.layer_norm""",
"""fc1""": """encoder.layers.*.feed_forward.intermediate_dense""",
"""fc2""": """encoder.layers.*.feed_forward.output_dense""",
"""final_layer_norm""": """encoder.layers.*.final_layer_norm""",
"""encoder.layer_norm""": """encoder.layer_norm""",
"""encoder.layer_norm_for_extract""": """layer_norm_for_extract""",
"""w2v_model.layer_norm""": """feature_projection.layer_norm""",
"""quantizer.weight_proj""": """quantizer.weight_proj""",
"""quantizer.vars""": """quantizer.codevectors""",
"""project_q""": """project_q""",
"""final_proj""": """project_hid""",
"""w2v_encoder.proj""": """lm_head""",
"""label_embs_concat""": """label_embeddings_concat""",
"""mask_emb""": """masked_spec_embed""",
"""spk_proj""": """speaker_proj""",
}
a_ = [
"""lm_head""",
"""quantizer.weight_proj""",
"""quantizer.codevectors""",
"""project_q""",
"""project_hid""",
"""label_embeddings_concat""",
"""speaker_proj""",
"""layer_norm_for_extract""",
]
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : Tuple ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Dict ):
for attribute in key.split('''.''' ):
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if weight_type is not None:
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase ).shape
else:
__lowerCamelCase = hf_pointer.shape
if hf_shape != value.shape:
raise ValueError(
F"""Shape of hf {key + "." + weight_type if weight_type is not None else ""} is {hf_shape}, but should be"""
F""" {value.shape} for {full_name}""" )
if weight_type == "weight":
__lowerCamelCase = value
elif weight_type == "weight_g":
__lowerCamelCase = value
elif weight_type == "weight_v":
__lowerCamelCase = value
elif weight_type == "bias":
__lowerCamelCase = value
else:
__lowerCamelCase = value
logger.info(F"""{key + "." + weight_type if weight_type is not None else ""} was initialized from {full_name}.""" )
def a__ ( _UpperCamelCase : Any ,_UpperCamelCase : Any ):
__lowerCamelCase = []
__lowerCamelCase = fairseq_model.state_dict()
__lowerCamelCase = hf_model.unispeech_sat.feature_extractor
for name, value in fairseq_dict.items():
__lowerCamelCase = False
if "conv_layers" in name:
load_conv_layer(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,hf_model.config.feat_extract_norm == '''group''' ,)
__lowerCamelCase = True
else:
for key, mapped_key in MAPPING.items():
__lowerCamelCase = '''unispeech_sat.''' + mapped_key if mapped_key not in TOP_LEVEL_KEYS else mapped_key
if key in name or key.split('''w2v_model.''' )[-1] == name.split('''.''' )[0]:
if "layer_norm_for_extract" in name and (".".join(name.split('''.''' )[:-1] ) != key):
# special case since naming is very similar
continue
__lowerCamelCase = True
if "*" in mapped_key:
__lowerCamelCase = name.split(_UpperCamelCase )[0].split('''.''' )[-2]
__lowerCamelCase = mapped_key.replace('''*''' ,_UpperCamelCase )
if "weight_g" in name:
__lowerCamelCase = '''weight_g'''
elif "weight_v" in name:
__lowerCamelCase = '''weight_v'''
elif "bias" in name:
__lowerCamelCase = '''bias'''
elif "weight" in name:
# TODO: don't match quantizer.weight_proj
__lowerCamelCase = '''weight'''
else:
__lowerCamelCase = None
set_recursively(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
continue
if not is_used:
unused_weights.append(_UpperCamelCase )
logger.warning(F"""Unused weights: {unused_weights}""" )
def a__ ( _UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Dict ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = full_name.split('''conv_layers.''' )[-1]
__lowerCamelCase = name.split('''.''' )
__lowerCamelCase = int(items[0] )
__lowerCamelCase = int(items[1] )
if type_id == 0:
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif (type_id == 2 and not use_group_norm) or (type_id == 2 and layer_id == 0 and use_group_norm):
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(_UpperCamelCase )
@torch.no_grad()
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : List[str] ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : List[Any]=True ):
if config_path is not None:
__lowerCamelCase = UniSpeechSatConfig.from_pretrained(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatConfig()
__lowerCamelCase = ''''''
if is_finetuned:
__lowerCamelCase = UniSpeechSatForCTC(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatForPreTraining(_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
__lowerCamelCase = model[0].eval()
recursively_load_weights(_UpperCamelCase ,_UpperCamelCase )
hf_wavavec.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument("""--checkpoint_path""", default=None, type=str, help="""Path to fairseq checkpoint""")
parser.add_argument("""--dict_path""", default=None, type=str, help="""Path to dict of fine-tuned model""")
parser.add_argument("""--config_path""", default=None, type=str, help="""Path to hf config.json of model to convert""")
parser.add_argument(
"""--not_finetuned""", action="""store_true""", help="""Whether the model to convert is a fine-tuned model or not"""
)
a_ = parser.parse_args()
convert_unispeech_sat_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 330 | 1 |
import json
import os
import re
import sys
import urllib.request
import requests
from bsa import BeautifulSoup
a_ = {
"""User-Agent""": """Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"""
""" (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582"""
}
def a__ ( _UpperCamelCase : str = "dhaka" ,_UpperCamelCase : int = 5 ):
__lowerCamelCase = min(_UpperCamelCase ,50 ) # Prevent abuse!
__lowerCamelCase = {
'''q''': query,
'''tbm''': '''isch''',
'''hl''': '''en''',
'''ijn''': '''0''',
}
__lowerCamelCase = requests.get('''https://www.google.com/search''' ,params=_UpperCamelCase ,headers=_UpperCamelCase )
__lowerCamelCase = BeautifulSoup(html.text ,'''html.parser''' )
__lowerCamelCase = ''''''.join(
re.findall(R'''AF_initDataCallback\(([^<]+)\);''' ,str(soup.select('''script''' ) ) ) )
__lowerCamelCase = json.dumps(_UpperCamelCase )
__lowerCamelCase = json.loads(_UpperCamelCase )
__lowerCamelCase = re.findall(
R'''\[\"GRID_STATE0\",null,\[\[1,\[0,\".*?\",(.*),\"All\",''' ,_UpperCamelCase ,)
if not matched_google_image_data:
return 0
__lowerCamelCase = re.sub(
R'''\[\"(https\:\/\/encrypted-tbn0\.gstatic\.com\/images\?.*?)\",\d+,\d+\]''' ,'''''' ,str(_UpperCamelCase ) ,)
__lowerCamelCase = re.findall(
R'''(?:\'|,),\[\"(https:|http.*?)\",\d+,\d+\]''' ,_UpperCamelCase ,)
for index, fixed_full_res_image in enumerate(_UpperCamelCase ):
if index >= max_images:
return index
__lowerCamelCase = bytes(_UpperCamelCase ,'''ascii''' ).decode(
'''unicode-escape''' )
__lowerCamelCase = bytes(_UpperCamelCase ,'''ascii''' ).decode(
'''unicode-escape''' )
__lowerCamelCase = urllib.request.build_opener()
__lowerCamelCase = [
(
'''User-Agent''',
'''Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36'''
''' (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582''',
)
]
urllib.request.install_opener(_UpperCamelCase )
__lowerCamelCase = F"""query_{query.replace(" " ,"_" )}"""
if not os.path.exists(_UpperCamelCase ):
os.makedirs(_UpperCamelCase )
urllib.request.urlretrieve( # noqa: S310
_UpperCamelCase ,F"""{path_name}/original_size_img_{index}.jpg""" )
return index
if __name__ == "__main__":
try:
a_ = download_images_from_google_query(sys.argv[1])
print(f"{image_count} images were downloaded to disk.")
except IndexError:
print("""Please provide a search term.""")
raise
| 330 |
from typing import List, Union
import numpy as np
from ..utils import add_end_docstrings, is_torch_available, is_vision_available, logging, requires_backends
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_torch_available():
import torch
from ..models.auto.modeling_auto import MODEL_FOR_DEPTH_ESTIMATION_MAPPING
a_ = logging.get_logger(__name__)
@add_end_docstrings(lowerCAmelCase__ )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
requires_backends(self , '''vision''' )
self.check_model_type(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return super().__call__(__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
return {}, {}, {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = load_image(__UpperCAmelCase )
__lowerCamelCase = image.size
__lowerCamelCase = self.image_processor(images=__UpperCAmelCase , return_tensors=self.framework )
return model_inputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.model(**__UpperCAmelCase )
return model_outputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = model_outputs.predicted_depth
__lowerCamelCase = torch.nn.functional.interpolate(
predicted_depth.unsqueeze(1 ) , size=self.image_size[::-1] , mode='''bicubic''' , align_corners=__UpperCAmelCase )
__lowerCamelCase = prediction.squeeze().cpu().numpy()
__lowerCamelCase = (output * 255 / np.max(__UpperCAmelCase )).astype('''uint8''' )
__lowerCamelCase = Image.fromarray(__UpperCAmelCase )
__lowerCamelCase = {}
__lowerCamelCase = predicted_depth
__lowerCamelCase = depth
return output_dict
| 330 | 1 |
import os
import pytest
from attr import dataclass
a_ = """us-east-1""" # defaults region
@dataclass
class __lowerCAmelCase :
lowerCAmelCase__ = 42
lowerCAmelCase__ = """arn:aws:iam::558105141721:role/sagemaker_execution_role"""
lowerCAmelCase__ = {
"""task_name""": """mnli""",
"""per_device_train_batch_size""": 1_6,
"""per_device_eval_batch_size""": 1_6,
"""do_train""": True,
"""do_eval""": True,
"""do_predict""": True,
"""output_dir""": """/opt/ml/model""",
"""overwrite_output_dir""": True,
"""max_steps""": 5_0_0,
"""save_steps""": 5_5_0_0,
}
lowerCAmelCase__ = {**hyperparameters, """max_steps""": 1_0_0_0}
@property
def lowerCamelCase ( self ):
'''simple docstring'''
if self.framework == "pytorch":
return [
{"Name": "train_runtime", "Regex": r"train_runtime.*=\D*(.*?)$"},
{"Name": "eval_accuracy", "Regex": r"eval_accuracy.*=\D*(.*?)$"},
{"Name": "eval_loss", "Regex": r"eval_loss.*=\D*(.*?)$"},
]
else:
return [
{"Name": "train_runtime", "Regex": r"train_runtime.*=\D*(.*?)$"},
{"Name": "eval_accuracy", "Regex": r"loss.*=\D*(.*?)]?$"},
{"Name": "eval_loss", "Regex": r"sparse_categorical_accuracy.*=\D*(.*?)]?$"},
]
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return F"""{self.framework}-transfromers-test"""
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return F"""./tests/sagemaker/scripts/{self.framework}"""
@property
def lowerCamelCase ( self ):
'''simple docstring'''
if self.framework == "pytorch":
return "763104351884.dkr.ecr.us-east-1.amazonaws.com/huggingface-pytorch-training:1.7.1-transformers4.6.1-gpu-py36-cu110-ubuntu18.04"
else:
return "763104351884.dkr.ecr.us-east-1.amazonaws.com/huggingface-tensorflow-training:2.4.1-transformers4.6.1-gpu-py37-cu110-ubuntu18.04"
@pytest.fixture(scope='''class''' )
def a__ ( _UpperCamelCase : Dict ):
__lowerCamelCase = SageMakerTestEnvironment(framework=request.cls.framework )
| 330 |
from typing import Dict, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
convert_to_rgb,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
OPENAI_CLIP_MEAN,
OPENAI_CLIP_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_vision_available, logging
a_ = logging.get_logger(__name__)
if is_vision_available():
import PIL
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""pixel_values"""]
def __init__( self , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = True , __UpperCAmelCase = 1 / 255 , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = True , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = size if size is not None else {'''shortest_edge''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
__lowerCamelCase = crop_size if crop_size is not None else {'''height''': 224, '''width''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase , 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 OPENAI_CLIP_MEAN
__lowerCamelCase = image_std if image_std is not None else OPENAI_CLIP_STD
__lowerCamelCase = do_convert_rgb
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
if "shortest_edge" not in size:
raise ValueError(F"""The `size` parameter must contain the key `shortest_edge`. Got {size.keys()}""" )
__lowerCamelCase = get_resize_output_image_size(__UpperCAmelCase , size=size['''shortest_edge'''] , default_to_square=__UpperCAmelCase )
return resize(__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase )
if "height" not in size or "width" not in size:
raise ValueError(F"""The `size` parameter must contain the keys (height, width). Got {size.keys()}""" )
return center_crop(__UpperCAmelCase , size=(size['''height'''], size['''width''']) , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return rescale(__UpperCAmelCase , scale=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return normalize(__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = ChannelDimension.FIRST , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = do_resize if do_resize is not None else self.do_resize
__lowerCamelCase = size if size is not None else self.size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''size''' , default_to_square=__UpperCAmelCase )
__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 = crop_size if crop_size is not None else self.crop_size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''crop_size''' , default_to_square=__UpperCAmelCase )
__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 = do_convert_rgb if do_convert_rgb is not None else self.do_convert_rgb
__lowerCamelCase = make_list_of_images(__UpperCAmelCase )
if not valid_images(__UpperCAmelCase ):
raise ValueError(
'''Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '''
'''torch.Tensor, tf.Tensor or jax.ndarray.''' )
if do_resize and size is None:
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.''' )
# PIL RGBA images are converted to RGB
if do_convert_rgb:
__lowerCamelCase = [convert_to_rgb(__UpperCAmelCase ) for image in images]
# All transformations expect numpy arrays.
__lowerCamelCase = [to_numpy_array(__UpperCAmelCase ) for image in images]
if do_resize:
__lowerCamelCase = [self.resize(image=__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase ) for image in images]
if do_center_crop:
__lowerCamelCase = [self.center_crop(image=__UpperCAmelCase , size=__UpperCAmelCase ) for image in images]
if do_rescale:
__lowerCamelCase = [self.rescale(image=__UpperCAmelCase , scale=__UpperCAmelCase ) for image in images]
if do_normalize:
__lowerCamelCase = [self.normalize(image=__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase ) for image in images]
__lowerCamelCase = [to_channel_dimension_format(__UpperCAmelCase , __UpperCAmelCase ) for image in images]
__lowerCamelCase = {'''pixel_values''': images}
return BatchFeature(data=__UpperCAmelCase , tensor_type=__UpperCAmelCase )
| 330 | 1 |
import unittest
from transformers import SPIECE_UNDERLINE
from transformers.models.speechta import SpeechTaTokenizer
from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_tokenizers, slow
from transformers.tokenization_utils import AddedToken
from ...test_tokenization_common import TokenizerTesterMixin
a_ = get_tests_dir("""fixtures/test_sentencepiece_bpe_char.model""")
@require_sentencepiece
@require_tokenizers
class __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = SpeechTaTokenizer
lowerCAmelCase__ = False
lowerCAmelCase__ = True
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# We have a SentencePiece fixture for testing
__lowerCamelCase = SpeechTaTokenizer(__UpperCAmelCase )
__lowerCamelCase = AddedToken('''<mask>''' , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase )
__lowerCamelCase = mask_token
tokenizer.add_special_tokens({'''mask_token''': mask_token} )
tokenizer.add_tokens(['''<ctc_blank>'''] )
tokenizer.save_pretrained(self.tmpdirname )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = '''this is a test'''
__lowerCamelCase = '''this is a test'''
return input_text, output_text
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=20 , __UpperCAmelCase=5 ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_input_output_texts(__UpperCAmelCase )
__lowerCamelCase = tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.decode(__UpperCAmelCase , clean_up_tokenization_spaces=__UpperCAmelCase )
return text, ids
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''<pad>'''
__lowerCamelCase = 1
self.assertEqual(self.get_tokenizer()._convert_token_to_id(__UpperCAmelCase ) , __UpperCAmelCase )
self.assertEqual(self.get_tokenizer()._convert_id_to_token(__UpperCAmelCase ) , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = list(self.get_tokenizer().get_vocab().keys() )
self.assertEqual(vocab_keys[0] , '''<s>''' )
self.assertEqual(vocab_keys[1] , '''<pad>''' )
self.assertEqual(vocab_keys[-4] , '''œ''' )
self.assertEqual(vocab_keys[-2] , '''<mask>''' )
self.assertEqual(vocab_keys[-1] , '''<ctc_blank>''' )
self.assertEqual(len(__UpperCAmelCase ) , 81 )
def lowerCamelCase ( self ):
'''simple docstring'''
self.assertEqual(self.get_tokenizer().vocab_size , 79 )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_tokenizers(do_lower_case=__UpperCAmelCase )
for tokenizer in tokenizers:
with self.subTest(F"""{tokenizer.__class__.__name__}""" ):
__lowerCamelCase = tokenizer.vocab_size
__lowerCamelCase = len(__UpperCAmelCase )
self.assertNotEqual(__UpperCAmelCase , 0 )
# We usually have added tokens from the start in tests because our vocab fixtures are
# smaller than the original vocabs - let's not assert this
# self.assertEqual(vocab_size, all_size)
__lowerCamelCase = ['''aaaaa bbbbbb''', '''cccccccccdddddddd''']
__lowerCamelCase = tokenizer.add_tokens(__UpperCAmelCase )
__lowerCamelCase = tokenizer.vocab_size
__lowerCamelCase = len(__UpperCAmelCase )
self.assertNotEqual(__UpperCAmelCase , 0 )
self.assertEqual(__UpperCAmelCase , __UpperCAmelCase )
self.assertEqual(__UpperCAmelCase , len(__UpperCAmelCase ) )
self.assertEqual(__UpperCAmelCase , all_size + len(__UpperCAmelCase ) )
__lowerCamelCase = tokenizer.encode('''aaaaa bbbbbb low cccccccccdddddddd l''' , add_special_tokens=__UpperCAmelCase )
self.assertGreaterEqual(len(__UpperCAmelCase ) , 4 )
self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 )
self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 )
__lowerCamelCase = {'''eos_token''': '''>>>>|||<||<<|<<''', '''pad_token''': '''<<<<<|||>|>>>>|>'''}
__lowerCamelCase = tokenizer.add_special_tokens(__UpperCAmelCase )
__lowerCamelCase = tokenizer.vocab_size
__lowerCamelCase = len(__UpperCAmelCase )
self.assertNotEqual(__UpperCAmelCase , 0 )
self.assertEqual(__UpperCAmelCase , __UpperCAmelCase )
self.assertEqual(__UpperCAmelCase , len(__UpperCAmelCase ) )
self.assertEqual(__UpperCAmelCase , all_size_a + len(__UpperCAmelCase ) )
__lowerCamelCase = tokenizer.encode(
'''>>>>|||<||<<|<< aaaaabbbbbb low cccccccccdddddddd <<<<<|||>|>>>>|> l''' , add_special_tokens=__UpperCAmelCase )
self.assertGreaterEqual(len(__UpperCAmelCase ) , 6 )
self.assertGreater(tokens[0] , tokenizer.vocab_size - 1 )
self.assertGreater(tokens[0] , tokens[1] )
self.assertGreater(tokens[-3] , tokenizer.vocab_size - 1 )
self.assertGreater(tokens[-3] , tokens[-4] )
self.assertEqual(tokens[0] , tokenizer.eos_token_id )
self.assertEqual(tokens[-3] , tokenizer.pad_token_id )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_tokenizer()
__lowerCamelCase = tokenizer.tokenize('''This is a test''' )
# fmt: off
self.assertListEqual(__UpperCAmelCase , [SPIECE_UNDERLINE, '''T''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''a''', SPIECE_UNDERLINE, '''t''', '''e''', '''s''', '''t'''] )
# fmt: on
self.assertListEqual(
tokenizer.convert_tokens_to_ids(__UpperCAmelCase ) , [4, 32, 11, 10, 12, 4, 10, 12, 4, 7, 4, 6, 5, 12, 6] , )
__lowerCamelCase = tokenizer.tokenize('''I was born in 92000, and this is falsé.''' )
self.assertListEqual(
__UpperCAmelCase , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''92000''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] )
__lowerCamelCase = tokenizer.convert_tokens_to_ids(__UpperCAmelCase )
# fmt: off
self.assertListEqual(__UpperCAmelCase , [4, 30, 4, 20, 7, 12, 4, 25, 8, 13, 9, 4, 10, 9, 4, 3, 23, 4, 7, 9, 14, 4, 6, 11, 10, 12, 4, 10, 12, 4, 19, 7, 15, 12, 73, 26] )
# fmt: on
__lowerCamelCase = tokenizer.convert_ids_to_tokens(__UpperCAmelCase )
self.assertListEqual(
__UpperCAmelCase , [SPIECE_UNDERLINE, '''I''', SPIECE_UNDERLINE, '''w''', '''a''', '''s''', SPIECE_UNDERLINE, '''b''', '''o''', '''r''', '''n''', SPIECE_UNDERLINE, '''i''', '''n''', SPIECE_UNDERLINE, '''<unk>''', ''',''', SPIECE_UNDERLINE, '''a''', '''n''', '''d''', SPIECE_UNDERLINE, '''t''', '''h''', '''i''', '''s''', SPIECE_UNDERLINE, '''i''', '''s''', SPIECE_UNDERLINE, '''f''', '''a''', '''l''', '''s''', '''é''', '''.'''] )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
# Use custom sequence because this tokenizer does not handle numbers.
__lowerCamelCase = [
'''Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides '''
'''general-purpose architectures (BERT, GPT, RoBERTa, XLM, DistilBert, XLNet...) for Natural '''
'''Language Understanding (NLU) and Natural Language Generation (NLG) with over thirty-two pretrained '''
'''models in one hundred plus languages and deep interoperability between Jax, PyTorch and TensorFlow.''',
'''BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly '''
'''conditioning on both left and right context in all layers.''',
'''The quick brown fox jumps over the lazy dog.''',
]
# fmt: off
__lowerCamelCase = {
'''input_ids''': [
[4, 32, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 64, 19, 8, 13, 18, 5, 13, 15, 22, 4, 28, 9, 8, 20, 9, 4, 7, 12, 4, 24, 22, 6, 8, 13, 17, 11, 39, 6, 13, 7, 9, 12, 19, 8, 13, 18, 5, 13, 12, 4, 7, 9, 14, 4, 24, 22, 6, 8, 13, 17, 11, 39, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 39, 25, 5, 13, 6, 63, 4, 24, 13, 8, 27, 10, 14, 5, 12, 4, 21, 5, 9, 5, 13, 7, 15, 39, 24, 16, 13, 24, 8, 12, 5, 4, 7, 13, 17, 11, 10, 6, 5, 17, 6, 16, 13, 5, 12, 4, 64, 40, 47, 54, 32, 23, 4, 53, 49, 32, 23, 4, 54, 8, 40, 47, 54, 32, 7, 23, 4, 69, 52, 43, 23, 4, 51, 10, 12, 6, 10, 15, 40, 5, 13, 6, 23, 4, 69, 52, 48, 5, 6, 26, 26, 26, 63, 4, 19, 8, 13, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 61, 9, 14, 5, 13, 12, 6, 7, 9, 14, 10, 9, 21, 4, 64, 48, 52, 61, 63, 4, 7, 9, 14, 4, 48, 7, 6, 16, 13, 7, 15, 4, 52, 7, 9, 21, 16, 7, 21, 5, 4, 53, 5, 9, 5, 13, 7, 6, 10, 8, 9, 4, 64, 48, 52, 53, 63, 4, 20, 10, 6, 11, 4, 8, 27, 5, 13, 4, 6, 11, 10, 13, 6, 22, 39, 6, 20, 8, 4, 24, 13, 5, 6, 13, 7, 10, 9, 5, 14, 4, 18, 8, 14, 5, 15, 12, 4, 10, 9, 4, 8, 9, 5, 4, 11, 16, 9, 14, 13, 5, 14, 4, 24, 15, 16, 12, 4, 15, 7, 9, 21, 16, 7, 21, 5, 12, 4, 7, 9, 14, 4, 14, 5, 5, 24, 4, 10, 9, 6, 5, 13, 8, 24, 5, 13, 7, 25, 10, 15, 10, 6, 22, 4, 25, 5, 6, 20, 5, 5, 9, 4, 58, 7, 37, 23, 4, 49, 22, 32, 8, 13, 17, 11, 4, 7, 9, 14, 4, 32, 5, 9, 12, 8, 13, 55, 15, 8, 20, 26, 2],
[4, 40, 47, 54, 32, 4, 10, 12, 4, 14, 5, 12, 10, 21, 9, 5, 14, 4, 6, 8, 4, 24, 13, 5, 39, 6, 13, 7, 10, 9, 4, 14, 5, 5, 24, 4, 25, 10, 14, 10, 13, 5, 17, 6, 10, 8, 9, 7, 15, 4, 13, 5, 24, 13, 5, 12, 5, 9, 6, 7, 6, 10, 8, 9, 12, 4, 19, 13, 8, 18, 4, 16, 9, 15, 7, 25, 5, 15, 5, 14, 4, 6, 5, 37, 6, 4, 25, 22, 4, 46, 8, 10, 9, 6, 15, 22, 4, 17, 8, 9, 14, 10, 6, 10, 8, 9, 10, 9, 21, 4, 8, 9, 4, 25, 8, 6, 11, 4, 15, 5, 19, 6, 4, 7, 9, 14, 4, 13, 10, 21, 11, 6, 4, 17, 8, 9, 6, 5, 37, 6, 4, 10, 9, 4, 7, 15, 15, 4, 15, 7, 22, 5, 13, 12, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[4, 32, 11, 5, 4, 45, 16, 10, 17, 28, 4, 25, 13, 8, 20, 9, 4, 19, 8, 37, 4, 46, 16, 18, 24, 12, 4, 8, 27, 5, 13, 4, 6, 11, 5, 4, 15, 7, 57, 22, 4, 14, 8, 21, 26, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
],
'''attention_mask''': [
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
]
}
# fmt: on
self.tokenizer_integration_test_util(
expected_encoding=__UpperCAmelCase , model_name='''microsoft/speecht5_asr''' , revision='''c5ef64c71905caeccde0e4462ef3f9077224c524''' , sequences=__UpperCAmelCase , )
| 330 |
from __future__ import annotations
from typing import Generic, TypeVar
a_ = TypeVar("""T""")
class __lowerCAmelCase ( Generic[T] ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = data
__lowerCamelCase = self
__lowerCamelCase = 0
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# map from node name to the node object
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# create a new set with x as its member
__lowerCamelCase = DisjointSetTreeNode(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# find the set x belongs to (with path-compression)
__lowerCamelCase = self.map[data]
if elem_ref != elem_ref.parent:
__lowerCamelCase = self.find_set(elem_ref.parent.data )
return elem_ref.parent
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# helper function for union operation
if nodea.rank > nodea.rank:
__lowerCamelCase = nodea
else:
__lowerCamelCase = nodea
if nodea.rank == nodea.rank:
nodea.rank += 1
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# merge 2 disjoint sets
self.link(self.find_set(__UpperCAmelCase ) , self.find_set(__UpperCAmelCase ) )
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# connections: map from the node to the neighbouring nodes (with weights)
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# add a node ONLY if its not present in the graph
if node not in self.connections:
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# add an edge with the given weight
self.add_node(__UpperCAmelCase )
self.add_node(__UpperCAmelCase )
__lowerCamelCase = weight
__lowerCamelCase = weight
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = []
__lowerCamelCase = set()
for start in self.connections:
for end in self.connections[start]:
if (start, end) not in seen:
seen.add((end, start) )
edges.append((start, end, self.connections[start][end]) )
edges.sort(key=lambda __UpperCAmelCase : x[2] )
# creating the disjoint set
__lowerCamelCase = DisjointSetTree[T]()
for node in self.connections:
disjoint_set.make_set(__UpperCAmelCase )
# MST generation
__lowerCamelCase = 0
__lowerCamelCase = 0
__lowerCamelCase = GraphUndirectedWeighted[T]()
while num_edges < len(self.connections ) - 1:
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = edges[index]
index += 1
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
if parent_u != parent_v:
num_edges += 1
graph.add_edge(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
disjoint_set.union(__UpperCAmelCase , __UpperCAmelCase )
return graph
| 330 | 1 |
from typing import Optional
import pyspark
from .. import Features, NamedSplit
from ..download import DownloadMode
from ..packaged_modules.spark.spark import Spark
from .abc import AbstractDatasetReader
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = False , __UpperCAmelCase = None , __UpperCAmelCase = True , __UpperCAmelCase = "arrow" , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(
split=__UpperCAmelCase , features=__UpperCAmelCase , cache_dir=__UpperCAmelCase , keep_in_memory=__UpperCAmelCase , streaming=__UpperCAmelCase , **__UpperCAmelCase , )
__lowerCamelCase = load_from_cache_file
__lowerCamelCase = file_format
__lowerCamelCase = Spark(
df=__UpperCAmelCase , features=__UpperCAmelCase , cache_dir=__UpperCAmelCase , working_dir=__UpperCAmelCase , **__UpperCAmelCase , )
def lowerCamelCase ( self ):
'''simple docstring'''
if self.streaming:
return self.builder.as_streaming_dataset(split=self.split )
__lowerCamelCase = None if self._load_from_cache_file else DownloadMode.FORCE_REDOWNLOAD
self.builder.download_and_prepare(
download_mode=__UpperCAmelCase , file_format=self._file_format , )
return self.builder.as_dataset(split=self.split )
| 330 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=2 , __UpperCAmelCase=8 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=99 , __UpperCAmelCase=16 , __UpperCAmelCase=5 , __UpperCAmelCase=2 , __UpperCAmelCase=36 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.0 , __UpperCAmelCase=512 , __UpperCAmelCase=16 , __UpperCAmelCase=2 , __UpperCAmelCase=0.02 , __UpperCAmelCase=3 , __UpperCAmelCase=4 , __UpperCAmelCase=None , ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
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=__UpperCAmelCase , initializer_range=self.initializer_range , )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_config()
__lowerCamelCase = 300
return config
def lowerCamelCase ( self ):
'''simple docstring'''
(
(
__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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = True
__lowerCamelCase = MraModel(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , )
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , )
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForQuestionAnswering(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , start_positions=__UpperCAmelCase , end_positions=__UpperCAmelCase , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForSequenceClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForTokenClassification(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_choices
__lowerCamelCase = MraForMultipleChoice(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
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(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowerCamelCase ( self ):
'''simple docstring'''
__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 __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = ()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModelTester(self )
__lowerCamelCase = ConfigTester(self , config_class=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__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(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = MraModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
@unittest.skip(reason='''MRA does not output attentions''' )
def lowerCamelCase ( self ):
'''simple docstring'''
return
@require_torch
class __lowerCAmelCase ( unittest.TestCase ):
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = torch.Size((1, 256, 768) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[-0.0_140, 0.0_830, -0.0_381], [0.1_546, 0.1_402, 0.0_220], [0.1_162, 0.0_851, 0.0_165]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 256, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[9.2_595, -3.6_038, 11.8_819], [9.3_869, -3.2_693, 11.0_956], [11.8_524, -3.4_938, 13.1_210]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' )
__lowerCamelCase = torch.arange(4096 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 4096, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[5.4_789, -2.3_564, 7.5_064], [7.9_067, -1.3_369, 9.9_668], [9.0_712, -1.8_106, 7.0_380]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
import os
from shutil import copyfile
from typing import List, Optional, Tuple
from ...tokenization_utils import AddedToken
from ...tokenization_utils_fast import PreTrainedTokenizerFast
from ...utils import is_sentencepiece_available, logging
if is_sentencepiece_available():
from .tokenization_camembert import CamembertTokenizer
else:
a_ = None
a_ = logging.get_logger(__name__)
a_ = {"""vocab_file""": """sentencepiece.bpe.model""", """tokenizer_file""": """tokenizer.json"""}
a_ = {
"""vocab_file""": {
"""camembert-base""": """https://huggingface.co/camembert-base/resolve/main/sentencepiece.bpe.model""",
},
"""tokenizer_file""": {
"""camembert-base""": """https://huggingface.co/camembert-base/resolve/main/tokenizer.json""",
},
}
a_ = {
"""camembert-base""": 512,
}
a_ = """▁"""
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = VOCAB_FILES_NAMES
lowerCAmelCase__ = PRETRAINED_VOCAB_FILES_MAP
lowerCAmelCase__ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
lowerCAmelCase__ = ["""input_ids""", """attention_mask"""]
lowerCAmelCase__ = CamembertTokenizer
def __init__( self , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase="<s>" , __UpperCAmelCase="</s>" , __UpperCAmelCase="</s>" , __UpperCAmelCase="<s>" , __UpperCAmelCase="<unk>" , __UpperCAmelCase="<pad>" , __UpperCAmelCase="<mask>" , __UpperCAmelCase=["<s>NOTUSED", "</s>NOTUSED"] , **__UpperCAmelCase , ):
'''simple docstring'''
# Mask token behave like a normal word, i.e. include the space before it
__lowerCamelCase = AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else mask_token
super().__init__(
__UpperCAmelCase , tokenizer_file=__UpperCAmelCase , bos_token=__UpperCAmelCase , eos_token=__UpperCAmelCase , sep_token=__UpperCAmelCase , cls_token=__UpperCAmelCase , unk_token=__UpperCAmelCase , pad_token=__UpperCAmelCase , mask_token=__UpperCAmelCase , additional_special_tokens=__UpperCAmelCase , **__UpperCAmelCase , )
__lowerCamelCase = vocab_file
__lowerCamelCase = False if not self.vocab_file else True
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None ):
'''simple docstring'''
if token_ids_a is None:
return [self.cls_token_id] + token_ids_a + [self.sep_token_id]
__lowerCamelCase = [self.cls_token_id]
__lowerCamelCase = [self.sep_token_id]
return cls + token_ids_a + sep + sep + token_ids_a + sep
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None ):
'''simple docstring'''
__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 + sep + token_ids_a + sep ) * [0]
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None ):
'''simple docstring'''
if not self.can_save_slow_tokenizer:
raise ValueError(
'''Your fast tokenizer does not have the necessary information to save the vocabulary for a slow '''
'''tokenizer.''' )
if not os.path.isdir(__UpperCAmelCase ):
logger.error(F"""Vocabulary path ({save_directory}) should be a directory""" )
return
__lowerCamelCase = os.path.join(
__UpperCAmelCase , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] )
if os.path.abspath(self.vocab_file ) != os.path.abspath(__UpperCAmelCase ):
copyfile(self.vocab_file , __UpperCAmelCase )
return (out_vocab_file,)
| 330 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_torch_available,
)
a_ = {"""configuration_encoder_decoder""": ["""EncoderDecoderConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""EncoderDecoderModel"""]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TFEncoderDecoderModel"""]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""FlaxEncoderDecoderModel"""]
if TYPE_CHECKING:
from .configuration_encoder_decoder import EncoderDecoderConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_encoder_decoder import EncoderDecoderModel
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_encoder_decoder import TFEncoderDecoderModel
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_encoder_decoder import FlaxEncoderDecoderModel
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
import os
from datetime import datetime as dt
from github import Github
a_ = [
"""good first issue""",
"""good second issue""",
"""good difficult issue""",
"""enhancement""",
"""new pipeline/model""",
"""new scheduler""",
"""wip""",
]
def a__ ( ):
__lowerCamelCase = Github(os.environ['''GITHUB_TOKEN'''] )
__lowerCamelCase = g.get_repo('''huggingface/diffusers''' )
__lowerCamelCase = repo.get_issues(state='''open''' )
for issue in open_issues:
__lowerCamelCase = sorted(issue.get_comments() ,key=lambda _UpperCamelCase : i.created_at ,reverse=_UpperCamelCase )
__lowerCamelCase = comments[0] if len(_UpperCamelCase ) > 0 else None
if (
last_comment is not None
and last_comment.user.login == "github-actions[bot]"
and (dt.utcnow() - issue.updated_at).days > 7
and (dt.utcnow() - issue.created_at).days >= 30
and not any(label.name.lower() in LABELS_TO_EXEMPT for label in issue.get_labels() )
):
# Closes the issue after 7 days of inactivity since the Stalebot notification.
issue.edit(state='''closed''' )
elif (
"stale" in issue.get_labels()
and last_comment is not None
and last_comment.user.login != "github-actions[bot]"
):
# Opens the issue if someone other than Stalebot commented.
issue.edit(state='''open''' )
issue.remove_from_labels('''stale''' )
elif (
(dt.utcnow() - issue.updated_at).days > 23
and (dt.utcnow() - issue.created_at).days >= 30
and not any(label.name.lower() in LABELS_TO_EXEMPT for label in issue.get_labels() )
):
# Post a Stalebot notification after 23 days of inactivity.
issue.create_comment(
'''This issue has been automatically marked as stale because it has not had '''
'''recent activity. If you think this still needs to be addressed '''
'''please comment on this thread.\n\nPlease note that issues that do not follow the '''
'''[contributing guidelines](https://github.com/huggingface/diffusers/blob/main/CONTRIBUTING.md) '''
'''are likely to be ignored.''' )
issue.add_to_labels('''stale''' )
if __name__ == "__main__":
main()
| 330 |
from string import ascii_lowercase, ascii_uppercase
def a__ ( _UpperCamelCase : str ):
if not sentence:
return ""
__lowerCamelCase = dict(zip(_UpperCamelCase ,_UpperCamelCase ) )
return lower_to_upper.get(sentence[0] ,sentence[0] ) + sentence[1:]
if __name__ == "__main__":
from doctest import testmod
testmod()
| 330 | 1 |
import json
import os
from typing import Dict, List, Optional, Tuple
import regex as re
from ...tokenization_utils import PreTrainedTokenizer
from ...utils import logging
a_ = logging.get_logger(__name__)
a_ = {
"""vocab_file""": """vocab.json""",
"""merges_file""": """merges.txt""",
"""tokenizer_config_file""": """tokenizer_config.json""",
}
a_ = {
"""vocab_file""": {
"""facebook/blenderbot_small-90M""": """https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/vocab.json"""
},
"""merges_file""": {
"""facebook/blenderbot_small-90M""": """https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/merges.txt"""
},
"""tokenizer_config_file""": {
"""facebook/blenderbot_small-90M""": (
"""https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/tokenizer_config.json"""
)
},
}
a_ = {"""facebook/blenderbot_small-90M""": 512}
def a__ ( _UpperCamelCase : List[str] ):
__lowerCamelCase = set()
__lowerCamelCase = word[0]
for char in word[1:]:
pairs.add((prev_char, char) )
__lowerCamelCase = char
__lowerCamelCase = set(_UpperCamelCase )
return pairs
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = VOCAB_FILES_NAMES
lowerCAmelCase__ = PRETRAINED_VOCAB_FILES_MAP
lowerCAmelCase__ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
lowerCAmelCase__ = ["""input_ids""", """attention_mask"""]
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase="__start__" , __UpperCAmelCase="__end__" , __UpperCAmelCase="__unk__" , __UpperCAmelCase="__null__" , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(unk_token=__UpperCAmelCase , bos_token=__UpperCAmelCase , eos_token=__UpperCAmelCase , pad_token=__UpperCAmelCase , **__UpperCAmelCase )
with open(__UpperCAmelCase , encoding='''utf-8''' ) as vocab_handle:
__lowerCamelCase = json.load(__UpperCAmelCase )
__lowerCamelCase = {v: k for k, v in self.encoder.items()}
with open(__UpperCAmelCase , encoding='''utf-8''' ) as merges_handle:
__lowerCamelCase = merges_handle.read().split('''\n''' )[1:-1]
__lowerCamelCase = [tuple(merge.split() ) for merge in merges]
__lowerCamelCase = dict(zip(__UpperCAmelCase , range(len(__UpperCAmelCase ) ) ) )
__lowerCamelCase = {}
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return len(self.encoder )
def lowerCamelCase ( self ):
'''simple docstring'''
return dict(self.encoder , **self.added_tokens_encoder )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if token in self.cache:
return self.cache[token]
__lowerCamelCase = re.sub('''([.,!?()])''' , r''' \1''' , __UpperCAmelCase )
__lowerCamelCase = re.sub('''(\')''' , r''' \1 ''' , __UpperCAmelCase )
__lowerCamelCase = re.sub(r'''\s{2,}''' , ''' ''' , __UpperCAmelCase )
if "\n" in token:
__lowerCamelCase = token.replace('''\n''' , ''' __newln__''' )
__lowerCamelCase = token.split(''' ''' )
__lowerCamelCase = []
for token in tokens:
if not len(__UpperCAmelCase ):
continue
__lowerCamelCase = token.lower()
__lowerCamelCase = tuple(__UpperCAmelCase )
__lowerCamelCase = tuple(list(word[:-1] ) + [word[-1] + '''</w>'''] )
__lowerCamelCase = get_pairs(__UpperCAmelCase )
if not pairs:
words.append(__UpperCAmelCase )
continue
while True:
__lowerCamelCase = min(__UpperCAmelCase , key=lambda __UpperCAmelCase : self.bpe_ranks.get(__UpperCAmelCase , float('''inf''' ) ) )
if bigram not in self.bpe_ranks:
break
__lowerCamelCase ,__lowerCamelCase = bigram
__lowerCamelCase = []
__lowerCamelCase = 0
while i < len(__UpperCAmelCase ):
try:
__lowerCamelCase = word.index(__UpperCAmelCase , __UpperCAmelCase )
new_word.extend(word[i:j] )
__lowerCamelCase = j
except ValueError:
new_word.extend(word[i:] )
break
if word[i] == first and i < len(__UpperCAmelCase ) - 1 and word[i + 1] == second:
new_word.append(first + second )
i += 2
else:
new_word.append(word[i] )
i += 1
__lowerCamelCase = tuple(__UpperCAmelCase )
__lowerCamelCase = new_word
if len(__UpperCAmelCase ) == 1:
break
else:
__lowerCamelCase = get_pairs(__UpperCAmelCase )
__lowerCamelCase = '''@@ '''.join(__UpperCAmelCase )
__lowerCamelCase = word[:-4]
__lowerCamelCase = word
words.append(__UpperCAmelCase )
return " ".join(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = []
__lowerCamelCase = re.findall(r'''\S+\n?''' , __UpperCAmelCase )
for token in words:
split_tokens.extend(list(self.bpe(__UpperCAmelCase ).split(''' ''' ) ) )
return split_tokens
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = token.lower()
return self.encoder.get(__UpperCAmelCase , self.encoder.get(self.unk_token ) )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
return self.decoder.get(__UpperCAmelCase , self.unk_token )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = ''' '''.join(__UpperCAmelCase ).replace('''@@ ''' , '''''' ).strip()
return out_string
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None ):
'''simple docstring'''
if not os.path.isdir(__UpperCAmelCase ):
logger.error(F"""Vocabulary path ({save_directory}) should be a directory""" )
return
__lowerCamelCase = os.path.join(
__UpperCAmelCase , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] )
__lowerCamelCase = os.path.join(
__UpperCAmelCase , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''merges_file'''] )
with open(__UpperCAmelCase , '''w''' , encoding='''utf-8''' ) as f:
f.write(json.dumps(self.encoder , indent=2 , sort_keys=__UpperCAmelCase , ensure_ascii=__UpperCAmelCase ) + '''\n''' )
__lowerCamelCase = 0
with open(__UpperCAmelCase , '''w''' , encoding='''utf-8''' ) as writer:
writer.write('''#version: 0.2\n''' )
for bpe_tokens, token_index in sorted(self.bpe_ranks.items() , key=lambda __UpperCAmelCase : kv[1] ):
if index != token_index:
logger.warning(
F"""Saving vocabulary to {merge_file}: BPE merge indices are not consecutive."""
''' Please check that the tokenizer is not corrupted!''' )
__lowerCamelCase = token_index
writer.write(''' '''.join(__UpperCAmelCase ) + '''\n''' )
index += 1
return vocab_file, merge_file
| 330 |
from transformers import BertTokenizer, EncoderDecoderModel, SeqaSeqTrainer, SeqaSeqTrainingArguments
from transformers.testing_utils import TestCasePlus, require_torch, slow
from transformers.utils import is_datasets_available
if is_datasets_available():
import datasets
class __lowerCAmelCase ( lowerCAmelCase__ ):
@slow
@require_torch
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = EncoderDecoderModel.from_encoder_decoder_pretrained('''prajjwal1/bert-tiny''' , '''prajjwal1/bert-tiny''' )
__lowerCamelCase = BertTokenizer.from_pretrained('''bert-base-uncased''' )
__lowerCamelCase = bertabert.config.encoder.vocab_size
__lowerCamelCase = tokenizer.sep_token_id
__lowerCamelCase = tokenizer.cls_token_id
__lowerCamelCase = 128
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''train[:1%]''' )
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''validation[:1%]''' )
__lowerCamelCase = train_dataset.select(range(32 ) )
__lowerCamelCase = val_dataset.select(range(16 ) )
__lowerCamelCase = 4
def _map_to_encoder_decoder_inputs(__UpperCAmelCase ):
# Tokenizer will automatically set [BOS] <text> [EOS]
__lowerCamelCase = tokenizer(batch['''article'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=512 )
__lowerCamelCase = tokenizer(batch['''highlights'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=128 )
__lowerCamelCase = inputs.input_ids
__lowerCamelCase = inputs.attention_mask
__lowerCamelCase = outputs.input_ids
__lowerCamelCase = outputs.input_ids.copy()
__lowerCamelCase = [
[-100 if token == tokenizer.pad_token_id else token for token in labels] for labels in batch['''labels''']
]
__lowerCamelCase = outputs.attention_mask
assert all(len(__UpperCAmelCase ) == 512 for x in inputs.input_ids )
assert all(len(__UpperCAmelCase ) == 128 for x in outputs.input_ids )
return batch
def _compute_metrics(__UpperCAmelCase ):
__lowerCamelCase = pred.label_ids
__lowerCamelCase = pred.predictions
# all unnecessary tokens are removed
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = sum([int(pred_str[i] == label_str[i] ) for i in range(len(__UpperCAmelCase ) )] ) / len(__UpperCAmelCase )
return {"accuracy": accuracy}
# map train dataset
__lowerCamelCase = train_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
train_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
# same for validation dataset
__lowerCamelCase = val_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
val_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
__lowerCamelCase = self.get_auto_remove_tmp_dir()
__lowerCamelCase = SeqaSeqTrainingArguments(
output_dir=__UpperCAmelCase , per_device_train_batch_size=__UpperCAmelCase , per_device_eval_batch_size=__UpperCAmelCase , predict_with_generate=__UpperCAmelCase , evaluation_strategy='''steps''' , do_train=__UpperCAmelCase , do_eval=__UpperCAmelCase , warmup_steps=0 , eval_steps=2 , logging_steps=2 , )
# instantiate trainer
__lowerCamelCase = SeqaSeqTrainer(
model=__UpperCAmelCase , args=__UpperCAmelCase , compute_metrics=_compute_metrics , train_dataset=__UpperCAmelCase , eval_dataset=__UpperCAmelCase , tokenizer=__UpperCAmelCase , )
# start training
trainer.train()
| 330 | 1 |
import os
a_ = {"""I""": 1, """V""": 5, """X""": 10, """L""": 50, """C""": 100, """D""": 500, """M""": 1_000}
def a__ ( _UpperCamelCase : str ):
__lowerCamelCase = 0
__lowerCamelCase = 0
while index < len(_UpperCamelCase ) - 1:
__lowerCamelCase = SYMBOLS[numerals[index]]
__lowerCamelCase = SYMBOLS[numerals[index + 1]]
if current_value < next_value:
total_value -= current_value
else:
total_value += current_value
index += 1
total_value += SYMBOLS[numerals[index]]
return total_value
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = ''''''
__lowerCamelCase = num // 10_00
numerals += m_count * "M"
num %= 10_00
__lowerCamelCase = num // 1_00
if c_count == 9:
numerals += "CM"
c_count -= 9
elif c_count == 4:
numerals += "CD"
c_count -= 4
if c_count >= 5:
numerals += "D"
c_count -= 5
numerals += c_count * "C"
num %= 1_00
__lowerCamelCase = num // 10
if x_count == 9:
numerals += "XC"
x_count -= 9
elif x_count == 4:
numerals += "XL"
x_count -= 4
if x_count >= 5:
numerals += "L"
x_count -= 5
numerals += x_count * "X"
num %= 10
if num == 9:
numerals += "IX"
num -= 9
elif num == 4:
numerals += "IV"
num -= 4
if num >= 5:
numerals += "V"
num -= 5
numerals += num * "I"
return numerals
def a__ ( _UpperCamelCase : str = "/p089_roman.txt" ):
__lowerCamelCase = 0
with open(os.path.dirname(_UpperCamelCase ) + roman_numerals_filename ) as filea:
__lowerCamelCase = filea.readlines()
for line in lines:
__lowerCamelCase = line.strip()
__lowerCamelCase = parse_roman_numerals(_UpperCamelCase )
__lowerCamelCase = generate_roman_numerals(_UpperCamelCase )
savings += len(_UpperCamelCase ) - len(_UpperCamelCase )
return savings
if __name__ == "__main__":
print(f"{solution() = }")
| 330 |
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_timm_backbone""": ["""TimmBackboneConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TimmBackbone"""]
if TYPE_CHECKING:
from .configuration_timm_backbone import TimmBackboneConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_timm_backbone import TimmBackbone
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
def a__ ( _UpperCamelCase : list ):
for i in range(len(_UpperCamelCase ) - 1 ,0 ,-1 ):
__lowerCamelCase = False
for j in range(_UpperCamelCase ,0 ,-1 ):
if unsorted[j] < unsorted[j - 1]:
__lowerCamelCase ,__lowerCamelCase = unsorted[j - 1], unsorted[j]
__lowerCamelCase = True
for j in range(_UpperCamelCase ):
if unsorted[j] > unsorted[j + 1]:
__lowerCamelCase ,__lowerCamelCase = unsorted[j + 1], unsorted[j]
__lowerCamelCase = True
if not swapped:
break
return unsorted
if __name__ == "__main__":
import doctest
doctest.testmod()
a_ = input("""Enter numbers separated by a comma:\n""").strip()
a_ = [int(item) for item in user_input.split(""",""")]
print(f"{cocktail_shaker_sort(unsorted) = }")
| 330 |
import random
import unittest
from torch.utils.data import BatchSampler, DataLoader, IterableDataset
from accelerate import Accelerator
from accelerate.data_loader import (
BatchSamplerShard,
DataLoaderDispatcher,
DataLoaderShard,
IterableDatasetShard,
SkipBatchSampler,
SkipDataLoader,
skip_first_batches,
)
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase=0.01 , __UpperCAmelCase=1000 ):
'''simple docstring'''
__lowerCamelCase = p_stop
__lowerCamelCase = max_length
def __iter__( self ):
'''simple docstring'''
__lowerCamelCase = 0
__lowerCamelCase = False
while not stop and count < self.max_length:
yield count
count += 1
__lowerCamelCase = random.random() < self.p_stop
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = [
BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
for i in range(2 )
]
__lowerCamelCase = [list(__UpperCAmelCase ) for batch_sampler_shard in batch_sampler_shards]
if not split_batches:
self.assertListEqual([len(__UpperCAmelCase ) for shard in batch_sampler_shards] , [len(__UpperCAmelCase ) for e in expected] )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [0, 1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 0]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [1, 2, 3]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1, 0]], [[1, 0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 0]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], [[0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = [[0, 1, 2], [3, 4], [5, 6, 7, 8], [9, 10, 11], [12, 13]]
__lowerCamelCase = [BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , even_batches=__UpperCAmelCase ) for i in range(2 )]
self.assertEqual(len(batch_sampler_shards[0] ) , 3 )
self.assertEqual(len(batch_sampler_shards[1] ) , 2 )
self.assertListEqual(list(batch_sampler_shards[0] ) , [[0, 1, 2], [5, 6, 7, 8], [12, 13]] )
self.assertListEqual(list(batch_sampler_shards[1] ) , [[3, 4], [9, 10, 11]] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=2 , __UpperCAmelCase=False ):
'''simple docstring'''
random.seed(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = [
IterableDatasetShard(
__UpperCAmelCase , batch_size=__UpperCAmelCase , drop_last=__UpperCAmelCase , num_processes=__UpperCAmelCase , process_index=__UpperCAmelCase , split_batches=__UpperCAmelCase , )
for i in range(__UpperCAmelCase )
]
__lowerCamelCase = []
for iterable_dataset_shard in iterable_dataset_shards:
# Since our random iterable dataset will be... random... we need to use a seed to get reproducible results.
random.seed(__UpperCAmelCase )
iterable_dataset_lists.append(list(__UpperCAmelCase ) )
__lowerCamelCase = batch_size // num_processes if split_batches else batch_size
# All iterable dataset shard should have the same length, a round multiple of shard_batch_size
__lowerCamelCase = iterable_dataset_lists[0]
for l in iterable_dataset_lists[1:]:
self.assertEqual(len(__UpperCAmelCase ) , len(__UpperCAmelCase ) )
self.assertTrue(len(__UpperCAmelCase ) % shard_batch_size == 0 )
__lowerCamelCase = []
for idx in range(0 , len(__UpperCAmelCase ) , __UpperCAmelCase ):
for l in iterable_dataset_lists:
observed += l[idx : idx + shard_batch_size]
if not drop_last:
while len(__UpperCAmelCase ) < len(__UpperCAmelCase ):
reference += reference
self.assertListEqual(__UpperCAmelCase , reference[: len(__UpperCAmelCase )] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = 42
__lowerCamelCase = RandomIterableDataset()
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
# Edge case with a very small dataset
__lowerCamelCase = RandomIterableDataset(max_length=2 )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BatchSampler(range(16 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = SkipBatchSampler(__UpperCAmelCase , 2 )
self.assertListEqual(list(__UpperCAmelCase ) , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = SkipDataLoader(list(range(16 ) ) , batch_size=4 , skip_batches=2 )
self.assertListEqual([t.tolist() for t in dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoader(list(range(16 ) ) , batch_size=4 )
__lowerCamelCase = skip_first_batches(__UpperCAmelCase , num_batches=2 )
self.assertListEqual([t.tolist() for t in new_dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoaderShard(list(range(16 ) ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
def lowerCamelCase ( self ):
'''simple docstring'''
Accelerator()
__lowerCamelCase = DataLoaderDispatcher(range(16 ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
| 330 | 1 |
from numpy import exp, pi, sqrt
def a__ ( _UpperCamelCase : Union[str, Any] ,_UpperCamelCase : float = 0.0 ,_UpperCamelCase : float = 1.0 ):
return 1 / sqrt(2 * pi * sigma**2 ) * exp(-((x - mu) ** 2) / (2 * sigma**2) )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 |
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = n ** (1 / 3)
return (val * val * val) == n
if __name__ == "__main__":
print(perfect_cube(27))
print(perfect_cube(4))
| 330 | 1 |
def a__ ( _UpperCamelCase : Union[str, Any] ):
__lowerCamelCase ,__lowerCamelCase = [], []
while len(_UpperCamelCase ) > 1:
__lowerCamelCase ,__lowerCamelCase = min(_UpperCamelCase ), max(_UpperCamelCase )
start.append(_UpperCamelCase )
end.append(_UpperCamelCase )
collection.remove(_UpperCamelCase )
collection.remove(_UpperCamelCase )
end.reverse()
return start + collection + end
if __name__ == "__main__":
a_ = input("""Enter numbers separated by a comma:\n""").strip()
a_ = [int(item) for item in user_input.split(""",""")]
print(*merge_sort(unsorted), sep=""",""")
| 330 |
import argparse
import os
# New Code #
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
from accelerate.utils import find_executable_batch_size
########################################################################
# This is a fully working simple example to use Accelerate,
# specifically showcasing how to ensure out-of-memory errors never
# interrupt training, and builds off the `nlp_example.py` script.
#
# This example trains a Bert base model on GLUE MRPC
# in any of the following settings (with the same script):
# - single CPU or single GPU
# - multi GPUS (using PyTorch distributed mode)
# - (multi) TPUs
# - fp16 (mixed-precision) or fp32 (normal precision)
#
# New additions from the base script can be found quickly by
# looking for the # New Code # tags
#
# To run it in each of these various modes, follow the instructions
# in the readme for examples:
# https://github.com/huggingface/accelerate/tree/main/examples
#
########################################################################
a_ = 16
a_ = 32
def a__ ( _UpperCamelCase : Accelerator ,_UpperCamelCase : int = 16 ):
__lowerCamelCase = AutoTokenizer.from_pretrained('''bert-base-cased''' )
__lowerCamelCase = load_dataset('''glue''' ,'''mrpc''' )
def tokenize_function(_UpperCamelCase : Optional[Any] ):
# max_length=None => use the model max length (it's actually the default)
__lowerCamelCase = tokenizer(examples['''sentence1'''] ,examples['''sentence2'''] ,truncation=_UpperCamelCase ,max_length=_UpperCamelCase )
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(
_UpperCamelCase ,batched=_UpperCamelCase ,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(_UpperCamelCase : Any ):
# On TPU it's best to pad everything to the same length or training will be very slow.
__lowerCamelCase = 1_28 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(
_UpperCamelCase ,padding='''longest''' ,max_length=_UpperCamelCase ,pad_to_multiple_of=_UpperCamelCase ,return_tensors='''pt''' ,)
# Instantiate dataloaders.
__lowerCamelCase = DataLoader(
tokenized_datasets['''train'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
__lowerCamelCase = DataLoader(
tokenized_datasets['''validation'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
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
a_ = mocked_dataloaders # noqa: F811
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# For testing only
if os.environ.get('''TESTING_MOCKED_DATALOADERS''' ,_UpperCamelCase ) == "1":
__lowerCamelCase = 2
# Initialize accelerator
__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'''] )
__lowerCamelCase = evaluate.load('''glue''' ,'''mrpc''' )
# New Code #
# We now can define an inner training loop function. It should take a batch size as the only parameter,
# and build the dataloaders in there.
# It also gets our decorator
@find_executable_batch_size(starting_batch_size=_UpperCamelCase )
def inner_training_loop(_UpperCamelCase : Union[str, Any] ):
# And now just move everything below under this function
# We need to bring in the Accelerator object from earlier
nonlocal accelerator
# And reset all of its attributes that could hold onto any memory:
accelerator.free_memory()
# Then we can declare the model, optimizer, and everything else:
set_seed(_UpperCamelCase )
# 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=_UpperCamelCase )
# 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=_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = get_dataloaders(_UpperCamelCase ,_UpperCamelCase )
# Instantiate scheduler
__lowerCamelCase = get_linear_schedule_with_warmup(
optimizer=_UpperCamelCase ,num_warmup_steps=1_00 ,num_training_steps=(len(_UpperCamelCase ) * num_epochs) ,)
# Prepare everything
# There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the
# prepare method.
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Now we train the model
for epoch in range(_UpperCamelCase ):
model.train()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.loss
accelerator.backward(_UpperCamelCase )
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
model.eval()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
with torch.no_grad():
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.logits.argmax(dim=-1 )
__lowerCamelCase ,__lowerCamelCase = accelerator.gather_for_metrics((predictions, batch['''labels''']) )
metric.add_batch(
predictions=_UpperCamelCase ,references=_UpperCamelCase ,)
__lowerCamelCase = metric.compute()
# Use accelerator.print to print only on the main process.
accelerator.print(F"""epoch {epoch}:""" ,_UpperCamelCase )
# New Code #
# And call it at the end with no arguments
# Note: You could also refactor this outside of your training loop function
inner_training_loop()
def a__ ( ):
__lowerCamelCase = argparse.ArgumentParser(description='''Simple example of training script.''' )
parser.add_argument(
'''--mixed_precision''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,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.''' )
__lowerCamelCase = parser.parse_args()
__lowerCamelCase = {'''lr''': 2e-5, '''num_epochs''': 3, '''seed''': 42, '''batch_size''': 16}
training_function(_UpperCamelCase ,_UpperCamelCase )
if __name__ == "__main__":
main()
| 330 | 1 |
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_tokenizers_available, is_torch_available
a_ = {
"""configuration_altclip""": [
"""ALTCLIP_PRETRAINED_CONFIG_ARCHIVE_MAP""",
"""AltCLIPConfig""",
"""AltCLIPTextConfig""",
"""AltCLIPVisionConfig""",
],
"""processing_altclip""": ["""AltCLIPProcessor"""],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = [
"""ALTCLIP_PRETRAINED_MODEL_ARCHIVE_LIST""",
"""AltCLIPPreTrainedModel""",
"""AltCLIPModel""",
"""AltCLIPTextModel""",
"""AltCLIPVisionModel""",
]
if TYPE_CHECKING:
from .configuration_altclip import (
ALTCLIP_PRETRAINED_CONFIG_ARCHIVE_MAP,
AltCLIPConfig,
AltCLIPTextConfig,
AltCLIPVisionConfig,
)
from .processing_altclip import AltCLIPProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_altclip import (
ALTCLIP_PRETRAINED_MODEL_ARCHIVE_LIST,
AltCLIPModel,
AltCLIPPreTrainedModel,
AltCLIPTextModel,
AltCLIPVisionModel,
)
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 |
import logging
import os
import threading
import time
try:
import warnings
except ImportError:
a_ = None
try:
import msvcrt
except ImportError:
a_ = None
try:
import fcntl
except ImportError:
a_ = None
# Backward compatibility
# ------------------------------------------------
try:
TimeoutError
except NameError:
a_ = OSError
# Data
# ------------------------------------------------
a_ = [
"""Timeout""",
"""BaseFileLock""",
"""WindowsFileLock""",
"""UnixFileLock""",
"""SoftFileLock""",
"""FileLock""",
]
a_ = """3.0.12"""
a_ = None
def a__ ( ):
global _logger
__lowerCamelCase = _logger or logging.getLogger(__name__ )
return _logger
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = lock_file
return None
def __str__( self ):
'''simple docstring'''
__lowerCamelCase = F"""The file lock '{self.lock_file}' could not be acquired."""
return temp
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = lock
return None
def __enter__( self ):
'''simple docstring'''
return self.lock
def __exit__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
self.lock.release()
return None
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
__lowerCamelCase = max_filename_length if max_filename_length is not None else 255
# Hash the filename if it's too long
__lowerCamelCase = self.hash_filename_if_too_long(__UpperCAmelCase , __UpperCAmelCase )
# The path to the lock file.
__lowerCamelCase = lock_file
# The file descriptor for the *_lock_file* as it is returned by the
# os.open() function.
# This file lock is only NOT None, if the object currently holds the
# lock.
__lowerCamelCase = None
# The default timeout value.
__lowerCamelCase = timeout
# We use this lock primarily for the lock counter.
__lowerCamelCase = threading.Lock()
# The lock counter is used for implementing the nested locking
# mechanism. Whenever the lock is acquired, the counter is increased and
# the lock is only released, when this value is 0 again.
__lowerCamelCase = 0
return None
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._lock_file
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._timeout
@timeout.setter
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = float(__UpperCAmelCase )
return None
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError()
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError()
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._lock_file_fd is not None
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=0.05 ):
'''simple docstring'''
# Use the default timeout, if no timeout is provided.
if timeout is None:
__lowerCamelCase = self.timeout
# Increment the number right at the beginning.
# We can still undo it, if something fails.
with self._thread_lock:
self._lock_counter += 1
__lowerCamelCase = id(self )
__lowerCamelCase = self._lock_file
__lowerCamelCase = time.time()
try:
while True:
with self._thread_lock:
if not self.is_locked:
logger().debug(F"""Attempting to acquire lock {lock_id} on {lock_filename}""" )
self._acquire()
if self.is_locked:
logger().debug(F"""Lock {lock_id} acquired on {lock_filename}""" )
break
elif timeout >= 0 and time.time() - start_time > timeout:
logger().debug(F"""Timeout on acquiring lock {lock_id} on {lock_filename}""" )
raise Timeout(self._lock_file )
else:
logger().debug(
F"""Lock {lock_id} not acquired on {lock_filename}, waiting {poll_intervall} seconds ...""" )
time.sleep(__UpperCAmelCase )
except: # noqa
# Something did go wrong, so decrement the counter.
with self._thread_lock:
__lowerCamelCase = max(0 , self._lock_counter - 1 )
raise
return _Acquire_ReturnProxy(lock=self )
def lowerCamelCase ( self , __UpperCAmelCase=False ):
'''simple docstring'''
with self._thread_lock:
if self.is_locked:
self._lock_counter -= 1
if self._lock_counter == 0 or force:
__lowerCamelCase = id(self )
__lowerCamelCase = self._lock_file
logger().debug(F"""Attempting to release lock {lock_id} on {lock_filename}""" )
self._release()
__lowerCamelCase = 0
logger().debug(F"""Lock {lock_id} released on {lock_filename}""" )
return None
def __enter__( self ):
'''simple docstring'''
self.acquire()
return self
def __exit__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
self.release()
return None
def __del__( self ):
'''simple docstring'''
self.release(force=__UpperCAmelCase )
return None
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = os.path.basename(__UpperCAmelCase )
if len(__UpperCAmelCase ) > max_length and max_length > 0:
__lowerCamelCase = os.path.dirname(__UpperCAmelCase )
__lowerCamelCase = str(hash(__UpperCAmelCase ) )
__lowerCamelCase = filename[: max_length - len(__UpperCAmelCase ) - 8] + '''...''' + hashed_filename + '''.lock'''
return os.path.join(__UpperCAmelCase , __UpperCAmelCase )
else:
return path
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
from .file_utils import relative_to_absolute_path
super().__init__(__UpperCAmelCase , timeout=__UpperCAmelCase , max_filename_length=__UpperCAmelCase )
__lowerCamelCase = '''\\\\?\\''' + relative_to_absolute_path(self.lock_file )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_RDWR | os.O_CREAT | os.O_TRUNC
try:
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
except OSError:
pass
else:
try:
msvcrt.locking(__UpperCAmelCase , msvcrt.LK_NBLCK , 1 )
except OSError:
os.close(__UpperCAmelCase )
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self._lock_file_fd
__lowerCamelCase = None
msvcrt.locking(__UpperCAmelCase , msvcrt.LK_UNLCK , 1 )
os.close(__UpperCAmelCase )
try:
os.remove(self._lock_file )
# Probably another instance of the application
# that acquired the file lock.
except OSError:
pass
return None
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
__lowerCamelCase = os.statvfs(os.path.dirname(__UpperCAmelCase ) ).f_namemax
super().__init__(__UpperCAmelCase , timeout=__UpperCAmelCase , max_filename_length=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_RDWR | os.O_CREAT | os.O_TRUNC
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
try:
fcntl.flock(__UpperCAmelCase , fcntl.LOCK_EX | fcntl.LOCK_NB )
except OSError:
os.close(__UpperCAmelCase )
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
# Do not remove the lockfile:
#
# https://github.com/benediktschmitt/py-filelock/issues/31
# https://stackoverflow.com/questions/17708885/flock-removing-locked-file-without-race-condition
__lowerCamelCase = self._lock_file_fd
__lowerCamelCase = None
fcntl.flock(__UpperCAmelCase , fcntl.LOCK_UN )
os.close(__UpperCAmelCase )
return None
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_WRONLY | os.O_CREAT | os.O_EXCL | os.O_TRUNC
try:
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
except OSError:
pass
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
os.close(self._lock_file_fd )
__lowerCamelCase = None
try:
os.remove(self._lock_file )
# The file is already deleted and that's what we want.
except OSError:
pass
return None
a_ = None
if msvcrt:
a_ = WindowsFileLock
elif fcntl:
a_ = UnixFileLock
else:
a_ = SoftFileLock
if warnings is not None:
warnings.warn("""only soft file lock is available""")
| 330 | 1 |
from typing import Any
def a__ ( _UpperCamelCase : list ):
if not input_list:
return []
__lowerCamelCase = [input_list.count(_UpperCamelCase ) for value in input_list]
__lowerCamelCase = max(_UpperCamelCase ) # Gets the maximum count in the input list.
# Gets values of modes
return sorted({input_list[i] for i, value in enumerate(_UpperCamelCase ) if value == y} )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 |
import copy
import inspect
import unittest
import numpy as np
from huggingface_hub import hf_hub_download
from transformers import TimesformerConfig
from transformers.models.auto import get_values
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 (
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING,
TimesformerForVideoClassification,
TimesformerModel,
)
from transformers.models.timesformer.modeling_timesformer import TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from transformers import VideoMAEImageProcessor
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=13 , __UpperCAmelCase=10 , __UpperCAmelCase=3 , __UpperCAmelCase=2 , __UpperCAmelCase=2 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=32 , __UpperCAmelCase=5 , __UpperCAmelCase=4 , __UpperCAmelCase=37 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=10 , __UpperCAmelCase=0.02 , __UpperCAmelCase="divided_space_time" , __UpperCAmelCase=None , ):
'''simple docstring'''
__lowerCamelCase = parent
__lowerCamelCase = batch_size
__lowerCamelCase = image_size
__lowerCamelCase = num_channels
__lowerCamelCase = patch_size
__lowerCamelCase = num_frames
__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 = attention_type
__lowerCamelCase = initializer_range
__lowerCamelCase = scope
__lowerCamelCase = num_labels
# in TimeSformer, the number of spatial tokens equals num_frames * num_patches per frame + 1 CLS token
__lowerCamelCase = (image_size // patch_size) ** 2
__lowerCamelCase = (num_frames) * self.num_patches_per_frame + 1
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = floats_tensor(
[self.batch_size, self.num_frames, 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 lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , num_frames=self.num_frames , 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 , initializer_range=self.initializer_range , attention_type=self.attention_type , )
__lowerCamelCase = self.num_labels
return config
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
# verify the logits shape
__lowerCamelCase = torch.Size((self.batch_size, self.num_labels) )
self.parent.assertEqual(result.logits.shape , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = config_and_inputs
__lowerCamelCase = {'''pixel_values''': pixel_values}
return config, inputs_dict
@require_torch
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (TimesformerModel, TimesformerForVideoClassification) if is_torch_available() else ()
lowerCAmelCase__ = (
{"""feature-extraction""": TimesformerModel, """video-classification""": TimesformerForVideoClassification}
if is_torch_available()
else {}
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerModelTester(self )
__lowerCamelCase = ConfigTester(
self , config_class=__UpperCAmelCase , has_text_modality=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = copy.deepcopy(__UpperCAmelCase )
if return_labels:
if model_class in get_values(__UpperCAmelCase ):
__lowerCamelCase = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=__UpperCAmelCase )
return inputs_dict
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
@unittest.skip(reason='''TimeSformer does not use inputs_embeds''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
__lowerCamelCase = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(__UpperCAmelCase , nn.Linear ) )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
__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] , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_video_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = TimesformerModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.has_attentions:
pass
else:
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCamelCase = True
for model_class in self.all_model_classes:
__lowerCamelCase = self.model_tester.seq_length
__lowerCamelCase = self.model_tester.num_frames
__lowerCamelCase = True
__lowerCamelCase = False
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
__lowerCamelCase = len(__UpperCAmelCase )
# Check attention is always last and order is fine
__lowerCamelCase = True
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
self.assertEqual(out_len + 1 , len(__UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
def lowerCamelCase ( self ):
'''simple docstring'''
def check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.hidden_states
__lowerCamelCase = self.model_tester.num_hidden_layers + 1
self.assertEqual(len(__UpperCAmelCase ) , __UpperCAmelCase )
__lowerCamelCase = self.model_tester.seq_length
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [seq_length, self.model_tester.hidden_size] , )
__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(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCamelCase = True
check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def a__ ( ):
__lowerCamelCase = hf_hub_download(
repo_id='''hf-internal-testing/spaghetti-video''' ,filename='''eating_spaghetti.npy''' ,repo_type='''dataset''' )
__lowerCamelCase = np.load(_UpperCamelCase )
return list(_UpperCamelCase )
@require_torch
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
@cached_property
def lowerCamelCase ( self ):
'''simple docstring'''
# logits were tested with a different mean and std, so we use the same here
return (
VideoMAEImageProcessor(image_mean=[0.5, 0.5, 0.5] , image_std=[0.5, 0.5, 0.5] )
if is_vision_available()
else None
)
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification.from_pretrained('''facebook/timesformer-base-finetuned-k400''' ).to(
__UpperCAmelCase )
__lowerCamelCase = self.default_image_processor
__lowerCamelCase = prepare_video()
__lowerCamelCase = image_processor(video[:8] , return_tensors='''pt''' ).to(__UpperCAmelCase )
# forward pass
with torch.no_grad():
__lowerCamelCase = model(**__UpperCAmelCase )
# verify the logits
__lowerCamelCase = torch.Size((1, 400) )
self.assertEqual(outputs.logits.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor([-0.3_016, -0.7_713, -0.4_205] ).to(__UpperCAmelCase )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a_ = logging.get_logger(__name__)
a_ = {
"""google/realm-cc-news-pretrained-embedder""": (
"""https://huggingface.co/google/realm-cc-news-pretrained-embedder/resolve/main/config.json"""
),
"""google/realm-cc-news-pretrained-encoder""": (
"""https://huggingface.co/google/realm-cc-news-pretrained-encoder/resolve/main/config.json"""
),
"""google/realm-cc-news-pretrained-scorer""": (
"""https://huggingface.co/google/realm-cc-news-pretrained-scorer/resolve/main/config.json"""
),
"""google/realm-cc-news-pretrained-openqa""": (
"""https://huggingface.co/google/realm-cc-news-pretrained-openqa/aresolve/main/config.json"""
),
"""google/realm-orqa-nq-openqa""": """https://huggingface.co/google/realm-orqa-nq-openqa/resolve/main/config.json""",
"""google/realm-orqa-nq-reader""": """https://huggingface.co/google/realm-orqa-nq-reader/resolve/main/config.json""",
"""google/realm-orqa-wq-openqa""": """https://huggingface.co/google/realm-orqa-wq-openqa/resolve/main/config.json""",
"""google/realm-orqa-wq-reader""": """https://huggingface.co/google/realm-orqa-wq-reader/resolve/main/config.json""",
# See all REALM models at https://huggingface.co/models?filter=realm
}
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """realm"""
def __init__( self , __UpperCAmelCase=30522 , __UpperCAmelCase=768 , __UpperCAmelCase=128 , __UpperCAmelCase=12 , __UpperCAmelCase=12 , __UpperCAmelCase=8 , __UpperCAmelCase=3072 , __UpperCAmelCase="gelu_new" , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=512 , __UpperCAmelCase=2 , __UpperCAmelCase=0.02 , __UpperCAmelCase=1E-1_2 , __UpperCAmelCase=256 , __UpperCAmelCase=10 , __UpperCAmelCase=1E-3 , __UpperCAmelCase=5 , __UpperCAmelCase=320 , __UpperCAmelCase=13353718 , __UpperCAmelCase=5000 , __UpperCAmelCase=1 , __UpperCAmelCase=0 , __UpperCAmelCase=2 , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(pad_token_id=__UpperCAmelCase , bos_token_id=__UpperCAmelCase , eos_token_id=__UpperCAmelCase , **__UpperCAmelCase )
# Common config
__lowerCamelCase = vocab_size
__lowerCamelCase = max_position_embeddings
__lowerCamelCase = hidden_size
__lowerCamelCase = retriever_proj_size
__lowerCamelCase = num_hidden_layers
__lowerCamelCase = num_attention_heads
__lowerCamelCase = num_candidates
__lowerCamelCase = intermediate_size
__lowerCamelCase = hidden_act
__lowerCamelCase = hidden_dropout_prob
__lowerCamelCase = attention_probs_dropout_prob
__lowerCamelCase = initializer_range
__lowerCamelCase = type_vocab_size
__lowerCamelCase = layer_norm_eps
# Reader config
__lowerCamelCase = span_hidden_size
__lowerCamelCase = max_span_width
__lowerCamelCase = reader_layer_norm_eps
__lowerCamelCase = reader_beam_size
__lowerCamelCase = reader_seq_len
# Retrieval config
__lowerCamelCase = num_block_records
__lowerCamelCase = searcher_beam_size
| 330 |
def a__ ( _UpperCamelCase : int ):
if not isinstance(_UpperCamelCase ,_UpperCamelCase ):
__lowerCamelCase = F"""Input value of [number={number}] must be an integer"""
raise TypeError(_UpperCamelCase )
if number < 0:
return False
__lowerCamelCase = number * number
while number > 0:
if number % 10 != number_square % 10:
return False
number //= 10
number_square //= 10
return True
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 | 1 |
import warnings
warnings.warn(
"""memory_utils has been reorganized to utils.memory. Import `find_executable_batchsize` from the main `__init__`: """
"""`from accelerate import find_executable_batch_size` to avoid this warning.""",
FutureWarning,
)
| 330 |
import gc
import unittest
from parameterized import parameterized
from diffusers import FlaxUNetaDConditionModel
from diffusers.utils import is_flax_available
from diffusers.utils.testing_utils import load_hf_numpy, require_flax, slow
if is_flax_available():
import jax
import jax.numpy as jnp
@slow
@require_flax
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
return F"""gaussian_noise_s={seed}_shape={"_".join([str(__UpperCAmelCase ) for s in shape] )}.npy"""
def lowerCamelCase ( self ):
'''simple docstring'''
# clean up the VRAM after each test
super().tearDown()
gc.collect()
def lowerCamelCase ( self , __UpperCAmelCase=0 , __UpperCAmelCase=(4, 4, 64, 64) , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = jnp.array(load_hf_numpy(self.get_file_format(__UpperCAmelCase , __UpperCAmelCase ) ) , dtype=__UpperCAmelCase )
return image
def lowerCamelCase ( self , __UpperCAmelCase=False , __UpperCAmelCase="CompVis/stable-diffusion-v1-4" ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = '''bf16''' if fpaa else None
__lowerCamelCase ,__lowerCamelCase = FlaxUNetaDConditionModel.from_pretrained(
__UpperCAmelCase , subfolder='''unet''' , dtype=__UpperCAmelCase , revision=__UpperCAmelCase )
return model, params
def lowerCamelCase ( self , __UpperCAmelCase=0 , __UpperCAmelCase=(4, 77, 768) , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = jnp.array(load_hf_numpy(self.get_file_format(__UpperCAmelCase , __UpperCAmelCase ) ) , dtype=__UpperCAmelCase )
return hidden_states
@parameterized.expand(
[
# fmt: off
[83, 4, [-0.2_323, -0.1_304, 0.0_813, -0.3_093, -0.0_919, -0.1_571, -0.1_125, -0.5_806]],
[17, 0.55, [-0.0_831, -0.2_443, 0.0_901, -0.0_919, 0.3_396, 0.0_103, -0.3_743, 0.0_701]],
[8, 0.89, [-0.4_863, 0.0_859, 0.0_875, -0.1_658, 0.9_199, -0.0_114, 0.4_839, 0.4_639]],
[3, 1000, [-0.5_649, 0.2_402, -0.5_518, 0.1_248, 1.1_328, -0.2_443, -0.0_325, -1.0_078]],
# fmt: on
] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_unet_model(model_id='''CompVis/stable-diffusion-v1-4''' , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_latents(__UpperCAmelCase , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_encoder_hidden_states(__UpperCAmelCase , fpaa=__UpperCAmelCase )
__lowerCamelCase = model.apply(
{'''params''': params} , __UpperCAmelCase , jnp.array(__UpperCAmelCase , dtype=jnp.intaa ) , encoder_hidden_states=__UpperCAmelCase , ).sample
assert sample.shape == latents.shape
__lowerCamelCase = jnp.asarray(jax.device_get((sample[-1, -2:, -2:, :2].flatten()) ) , dtype=jnp.floataa )
__lowerCamelCase = jnp.array(__UpperCAmelCase , dtype=jnp.floataa )
# Found torch (float16) and flax (bfloat16) outputs to be within this tolerance, in the same hardware
assert jnp.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-2 )
@parameterized.expand(
[
# fmt: off
[83, 4, [0.1_514, 0.0_807, 0.1_624, 0.1_016, -0.1_896, 0.0_263, 0.0_677, 0.2_310]],
[17, 0.55, [0.1_164, -0.0_216, 0.0_170, 0.1_589, -0.3_120, 0.1_005, -0.0_581, -0.1_458]],
[8, 0.89, [-0.1_758, -0.0_169, 0.1_004, -0.1_411, 0.1_312, 0.1_103, -0.1_996, 0.2_139]],
[3, 1000, [0.1_214, 0.0_352, -0.0_731, -0.1_562, -0.0_994, -0.0_906, -0.2_340, -0.0_539]],
# fmt: on
] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_unet_model(model_id='''stabilityai/stable-diffusion-2''' , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_latents(__UpperCAmelCase , shape=(4, 4, 96, 96) , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_encoder_hidden_states(__UpperCAmelCase , shape=(4, 77, 1024) , fpaa=__UpperCAmelCase )
__lowerCamelCase = model.apply(
{'''params''': params} , __UpperCAmelCase , jnp.array(__UpperCAmelCase , dtype=jnp.intaa ) , encoder_hidden_states=__UpperCAmelCase , ).sample
assert sample.shape == latents.shape
__lowerCamelCase = jnp.asarray(jax.device_get((sample[-1, -2:, -2:, :2].flatten()) ) , dtype=jnp.floataa )
__lowerCamelCase = jnp.array(__UpperCAmelCase , dtype=jnp.floataa )
# Found torch (float16) and flax (bfloat16) outputs to be within this tolerance, on the same hardware
assert jnp.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-2 )
| 330 | 1 |
import unittest
from parameterized import parameterized
from transformers import OpenLlamaConfig, is_torch_available, set_seed
from transformers.testing_utils import require_torch, 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 OpenLlamaForCausalLM, OpenLlamaForSequenceClassification, OpenLlamaModel
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=13 , __UpperCAmelCase=7 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=False , __UpperCAmelCase=True , __UpperCAmelCase=99 , __UpperCAmelCase=32 , __UpperCAmelCase=5 , __UpperCAmelCase=4 , __UpperCAmelCase=37 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=512 , __UpperCAmelCase=16 , __UpperCAmelCase=2 , __UpperCAmelCase=0.02 , __UpperCAmelCase=3 , __UpperCAmelCase=4 , __UpperCAmelCase=None , ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
return OpenLlamaConfig(
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=__UpperCAmelCase , initializer_range=self.initializer_range , use_stable_embedding=__UpperCAmelCase , )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = OpenLlamaModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = True
__lowerCamelCase = OpenLlamaModel(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , )
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , )
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = OpenLlamaForCausalLM(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = True
__lowerCamelCase = True
__lowerCamelCase = OpenLlamaForCausalLM(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
# first forward pass
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , use_cache=__UpperCAmelCase , )
__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(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , output_hidden_states=__UpperCAmelCase , )['''hidden_states'''][0]
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , past_key_values=__UpperCAmelCase , output_hidden_states=__UpperCAmelCase , )['''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(__UpperCAmelCase , __UpperCAmelCase , atol=1E-3 ) )
def lowerCamelCase ( self ):
'''simple docstring'''
__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 __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (
(OpenLlamaModel, OpenLlamaForCausalLM, OpenLlamaForSequenceClassification) if is_torch_available() else ()
)
lowerCAmelCase__ = (OpenLlamaForCausalLM,) if is_torch_available() else ()
lowerCAmelCase__ = (
{
"""feature-extraction""": OpenLlamaModel,
"""text-classification""": OpenLlamaForSequenceClassification,
"""text-generation""": OpenLlamaForCausalLM,
"""zero-shot""": OpenLlamaForSequenceClassification,
}
if is_torch_available()
else {}
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = OpenLlamaModelTester(self )
__lowerCamelCase = ConfigTester(self , config_class=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__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(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCamelCase = 3
__lowerCamelCase = input_dict['''input_ids''']
__lowerCamelCase = input_ids.ne(1 ).to(__UpperCAmelCase )
__lowerCamelCase = ids_tensor([self.model_tester.batch_size] , self.model_tester.type_sequence_label_size )
__lowerCamelCase = OpenLlamaForSequenceClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , labels=__UpperCAmelCase )
self.assertEqual(result.logits.shape , (self.model_tester.batch_size, self.model_tester.num_labels) )
def lowerCamelCase ( self ):
'''simple docstring'''
__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(__UpperCAmelCase )
__lowerCamelCase = ids_tensor([self.model_tester.batch_size] , self.model_tester.type_sequence_label_size )
__lowerCamelCase = OpenLlamaForSequenceClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , labels=__UpperCAmelCase )
self.assertEqual(result.logits.shape , (self.model_tester.batch_size, self.model_tester.num_labels) )
def lowerCamelCase ( self ):
'''simple docstring'''
__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(__UpperCAmelCase )
__lowerCamelCase = ids_tensor(
[self.model_tester.batch_size, config.num_labels] , self.model_tester.type_sequence_label_size ).to(torch.float )
__lowerCamelCase = OpenLlamaForSequenceClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , labels=__UpperCAmelCase )
self.assertEqual(result.logits.shape , (self.model_tester.batch_size, self.model_tester.num_labels) )
@unittest.skip('''Open-Llama buffers include complex numbers, which breaks this test''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
@parameterized.expand([('''linear''',), ('''dynamic''',)] )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCamelCase = ids_tensor([1, 10] , config.vocab_size )
__lowerCamelCase = ids_tensor([1, int(config.max_position_embeddings * 1.5 )] , config.vocab_size )
set_seed(42 ) # Fixed seed at init time so the two models get the same random weights
__lowerCamelCase = OpenLlamaModel(__UpperCAmelCase )
original_model.to(__UpperCAmelCase )
original_model.eval()
__lowerCamelCase = original_model(__UpperCAmelCase ).last_hidden_state
__lowerCamelCase = original_model(__UpperCAmelCase ).last_hidden_state
set_seed(42 ) # Fixed seed at init time so the two models get the same random weights
__lowerCamelCase = {'''type''': scaling_type, '''factor''': 10.0}
__lowerCamelCase = OpenLlamaModel(__UpperCAmelCase )
scaled_model.to(__UpperCAmelCase )
scaled_model.eval()
__lowerCamelCase = scaled_model(__UpperCAmelCase ).last_hidden_state
__lowerCamelCase = scaled_model(__UpperCAmelCase ).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(__UpperCAmelCase , __UpperCAmelCase , atol=1E-5 ) )
else:
self.assertFalse(torch.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-5 ) )
# The output should be different for long inputs
self.assertFalse(torch.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-5 ) )
| 330 |
from typing import TYPE_CHECKING
from ....utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_mmbt""": ["""MMBTConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""MMBTForClassification""", """MMBTModel""", """ModalEmbeddings"""]
if TYPE_CHECKING:
from .configuration_mmbt import MMBTConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_mmbt import MMBTForClassification, MMBTModel, ModalEmbeddings
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
import collections
import tempfile
import unittest
import numpy as np
from transformers.testing_utils import (
is_pt_flax_cross_test,
require_flax,
require_torch,
require_vision,
slow,
torch_device,
)
from transformers.utils import is_flax_available, is_torch_available, is_vision_available
from ...test_modeling_flax_common import floats_tensor, ids_tensor, random_attention_mask
from ..bert.test_modeling_flax_bert import FlaxBertModelTester
from ..clip.test_modeling_flax_clip import FlaxCLIPVisionModelTester
from ..vit.test_modeling_flax_vit import FlaxViTModelTester
if is_flax_available():
from transformers import (
FlaxBertModel,
FlaxCLIPVisionModel,
FlaxVisionTextDualEncoderModel,
FlaxViTModel,
VisionTextDualEncoderConfig,
VisionTextDualEncoderProcessor,
)
from transformers.modeling_flax_pytorch_utils import (
convert_pytorch_state_dict_to_flax,
load_flax_weights_in_pytorch_model,
)
if is_torch_available():
import torch
from transformers import VisionTextDualEncoderModel
if is_vision_available():
from PIL import Image
def a__ ( _UpperCamelCase : List[Any] ):
if isinstance(_UpperCamelCase ,collections.abc.Iterable ):
return x
return (x, x)
@require_flax
class __lowerCAmelCase :
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = np.abs((a - b) ).max()
self.assertLessEqual(__UpperCAmelCase , __UpperCAmelCase , F"""Difference between torch and flax is {diff} (>= {tol}).""" )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = VisionTextDualEncoderConfig.from_vision_text_configs(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = FlaxVisionTextDualEncoderModel(__UpperCAmelCase )
__lowerCamelCase = model(input_ids=__UpperCAmelCase , pixel_values=__UpperCAmelCase , attention_mask=__UpperCAmelCase )
self.assertEqual(output['''text_embeds'''].shape , (input_ids.shape[0], config.projection_dim) )
self.assertEqual(output['''image_embeds'''].shape , (pixel_values.shape[0], config.projection_dim) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_vision_text_model(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = {'''vision_model''': vision_model, '''text_model''': text_model}
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_vision_text_pretrained(**__UpperCAmelCase )
__lowerCamelCase = model(input_ids=__UpperCAmelCase , pixel_values=__UpperCAmelCase , attention_mask=__UpperCAmelCase )
self.assertEqual(output['''text_embeds'''].shape , (input_ids.shape[0], model.config.projection_dim) )
self.assertEqual(output['''image_embeds'''].shape , (pixel_values.shape[0], model.config.projection_dim) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_vision_text_model(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = {'''vision_model''': vision_model, '''text_model''': text_model}
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_vision_text_pretrained(**__UpperCAmelCase )
__lowerCamelCase = model(input_ids=__UpperCAmelCase , pixel_values=__UpperCAmelCase , attention_mask=__UpperCAmelCase )
__lowerCamelCase = output[0]
with tempfile.TemporaryDirectory() as tmpdirname:
model.save_pretrained(__UpperCAmelCase )
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_pretrained(__UpperCAmelCase )
__lowerCamelCase = model(input_ids=__UpperCAmelCase , pixel_values=__UpperCAmelCase , attention_mask=__UpperCAmelCase )
__lowerCamelCase = after_output[0]
__lowerCamelCase = np.amax(np.abs(out_a - out_a ) )
self.assertLessEqual(__UpperCAmelCase , 1E-3 )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_vision_text_model(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = {'''vision_model''': vision_model, '''text_model''': text_model}
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_vision_text_pretrained(**__UpperCAmelCase )
__lowerCamelCase = model(
input_ids=__UpperCAmelCase , pixel_values=__UpperCAmelCase , attention_mask=__UpperCAmelCase , output_attentions=__UpperCAmelCase )
__lowerCamelCase = output.vision_model_output.attentions
self.assertEqual(len(__UpperCAmelCase ) , vision_config.num_hidden_layers )
# in ViT, the seq_len equals the number of patches + 1 (we add 1 for the [CLS] token)
__lowerCamelCase = to_atuple(vision_model.config.image_size )
__lowerCamelCase = to_atuple(vision_model.config.patch_size )
__lowerCamelCase = (image_size[1] // patch_size[1]) * (image_size[0] // patch_size[0])
__lowerCamelCase = num_patches + 1
self.assertEqual(vision_attentions[0].shape[-3:] , (vision_config.num_attention_heads, seq_len, seq_len) )
__lowerCamelCase = output.text_model_output.attentions
self.assertEqual(len(__UpperCAmelCase ) , text_config.num_hidden_layers )
self.assertEqual(
text_attentions[0].shape[-3:] , (text_config.num_attention_heads, input_ids.shape[-1], input_ids.shape[-1]) , )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
pt_model.to(__UpperCAmelCase )
pt_model.eval()
# prepare inputs
__lowerCamelCase = inputs_dict
__lowerCamelCase = {k: torch.tensor(v.tolist() ) for k, v in flax_inputs.items()}
with torch.no_grad():
__lowerCamelCase = pt_model(**__UpperCAmelCase ).to_tuple()
__lowerCamelCase = fx_model(**__UpperCAmelCase ).to_tuple()
self.assertEqual(len(__UpperCAmelCase ) , len(__UpperCAmelCase ) , '''Output lengths differ between Flax and PyTorch''' )
for fx_output, pt_output in zip(fx_outputs[:4] , pt_outputs[:4] ):
self.assert_almost_equals(__UpperCAmelCase , pt_output.numpy() , 4E-2 )
# PT -> Flax
with tempfile.TemporaryDirectory() as tmpdirname:
pt_model.save_pretrained(__UpperCAmelCase )
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_pretrained(__UpperCAmelCase , from_pt=__UpperCAmelCase )
__lowerCamelCase = fx_model_loaded(**__UpperCAmelCase ).to_tuple()
self.assertEqual(len(__UpperCAmelCase ) , len(__UpperCAmelCase ) , '''Output lengths differ between Flax and PyTorch''' )
for fx_output_loaded, pt_output in zip(fx_outputs_loaded[:4] , pt_outputs[:4] ):
self.assert_almost_equals(__UpperCAmelCase , pt_output.numpy() , 4E-2 )
# Flax -> PT
with tempfile.TemporaryDirectory() as tmpdirname:
fx_model.save_pretrained(__UpperCAmelCase )
__lowerCamelCase = VisionTextDualEncoderModel.from_pretrained(__UpperCAmelCase , from_flax=__UpperCAmelCase )
pt_model_loaded.to(__UpperCAmelCase )
pt_model_loaded.eval()
with torch.no_grad():
__lowerCamelCase = pt_model_loaded(**__UpperCAmelCase ).to_tuple()
self.assertEqual(len(__UpperCAmelCase ) , len(__UpperCAmelCase ) , '''Output lengths differ between Flax and PyTorch''' )
for fx_output, pt_output_loaded in zip(fx_outputs[:4] , pt_outputs_loaded[:4] ):
self.assert_almost_equals(__UpperCAmelCase , pt_output_loaded.numpy() , 4E-2 )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = VisionTextDualEncoderConfig.from_vision_text_configs(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = VisionTextDualEncoderModel(__UpperCAmelCase )
__lowerCamelCase = FlaxVisionTextDualEncoderModel(__UpperCAmelCase )
__lowerCamelCase = convert_pytorch_state_dict_to_flax(pt_model.state_dict() , __UpperCAmelCase )
__lowerCamelCase = fx_state
self.check_pt_flax_equivalence(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = VisionTextDualEncoderConfig.from_vision_text_configs(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = VisionTextDualEncoderModel(__UpperCAmelCase )
__lowerCamelCase = FlaxVisionTextDualEncoderModel(__UpperCAmelCase )
__lowerCamelCase = load_flax_weights_in_pytorch_model(__UpperCAmelCase , fx_model.params )
self.check_pt_flax_equivalence(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
self.check_model_from_pretrained_configs(**__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
self.check_vision_text_dual_encoder_from_pretrained(**__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
self.check_save_load(**__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
self.check_vision_text_output_attention(**__UpperCAmelCase )
@is_pt_flax_cross_test
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
__lowerCamelCase = config_inputs_dict.pop('''vision_config''' )
__lowerCamelCase = config_inputs_dict.pop('''text_config''' )
__lowerCamelCase = config_inputs_dict
self.check_equivalence_pt_to_flax(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
self.check_equivalence_flax_to_pt(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_pretrained_model_and_inputs()
__lowerCamelCase = model_a(**__UpperCAmelCase )
__lowerCamelCase = outputs[0]
with tempfile.TemporaryDirectory() as tmp_dirname:
model_a.save_pretrained(__UpperCAmelCase )
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_pretrained(__UpperCAmelCase )
__lowerCamelCase = model_a(**__UpperCAmelCase )
__lowerCamelCase = after_outputs[0]
__lowerCamelCase = np.amax(np.abs(out_a - out_a ) )
self.assertLessEqual(__UpperCAmelCase , 1E-5 )
@require_flax
class __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_vision_text_pretrained(
'''hf-internal-testing/tiny-random-vit''' , '''hf-internal-testing/tiny-bert''' , vision_from_pt=__UpperCAmelCase , text_from_pt=__UpperCAmelCase , )
__lowerCamelCase = 13
__lowerCamelCase = floats_tensor(
[
batch_size,
model.config.vision_config.num_channels,
model.config.vision_config.image_size,
model.config.vision_config.image_size,
] )
__lowerCamelCase = ids_tensor([batch_size, 4] , model.config.text_config.vocab_size )
__lowerCamelCase = random_attention_mask([batch_size, 4] )
__lowerCamelCase = {'''pixel_values''': pixel_values, '''input_ids''': input_ids, '''attention_mask''': attention_mask}
return model, inputs
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = FlaxViTModel(__UpperCAmelCase )
__lowerCamelCase = FlaxBertModel(__UpperCAmelCase )
return vision_model, text_model
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = FlaxViTModelTester(self )
__lowerCamelCase = FlaxBertModelTester(self )
__lowerCamelCase = vit_model_tester.prepare_config_and_inputs()
__lowerCamelCase = bert_model_tester.prepare_config_and_inputs()
__lowerCamelCase ,__lowerCamelCase = vision_config_and_inputs
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = text_config_and_inputs
# make sure that cross attention layers are added
return {
"text_config": text_config,
"vision_config": vision_config,
"pixel_values": pixel_values,
"attention_mask": attention_mask,
"input_ids": input_ids,
"token_type_ids": token_type_ids,
}
@require_torch
class __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_vision_text_pretrained(
'''hf-internal-testing/tiny-random-clip''' , '''hf-internal-testing/tiny-bert''' , vision_from_pt=__UpperCAmelCase , text_from_pt=__UpperCAmelCase , )
__lowerCamelCase = 13
__lowerCamelCase = floats_tensor(
[
batch_size,
model.config.vision_config.num_channels,
model.config.vision_config.image_size,
model.config.vision_config.image_size,
] )
__lowerCamelCase = ids_tensor([batch_size, 4] , model.config.text_config.vocab_size )
__lowerCamelCase = random_attention_mask([batch_size, 4] )
__lowerCamelCase = {'''pixel_values''': pixel_values, '''input_ids''': input_ids, '''attention_mask''': attention_mask}
return model, inputs
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = FlaxCLIPVisionModel(__UpperCAmelCase )
__lowerCamelCase = FlaxBertModel(__UpperCAmelCase )
return vision_model, text_model
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = FlaxCLIPVisionModelTester(self )
__lowerCamelCase = FlaxBertModelTester(self )
__lowerCamelCase = clip_model_tester.prepare_config_and_inputs()
__lowerCamelCase = bert_model_tester.prepare_config_and_inputs()
__lowerCamelCase ,__lowerCamelCase = vision_config_and_inputs
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = text_config_and_inputs
# make sure that cross attention layers are added
return {
"text_config": text_config,
"vision_config": vision_config,
"pixel_values": pixel_values,
"attention_mask": attention_mask,
"input_ids": input_ids,
"token_type_ids": token_type_ids,
}
@require_flax
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = FlaxVisionTextDualEncoderModel.from_pretrained('''clip-italian/clip-italian''' , logit_scale_init_value=1.0 )
__lowerCamelCase = VisionTextDualEncoderProcessor.from_pretrained('''clip-italian/clip-italian''' )
__lowerCamelCase = Image.open('''./tests/fixtures/tests_samples/COCO/000000039769.png''' )
__lowerCamelCase = processor(
text=['''una foto di un gatto''', '''una foto di un cane'''] , images=__UpperCAmelCase , padding=__UpperCAmelCase , return_tensors='''np''' )
__lowerCamelCase = model(**__UpperCAmelCase )
# verify the logits
self.assertEqual(outputs.logits_per_image.shape , (inputs.pixel_values.shape[0], inputs.input_ids.shape[0]) )
self.assertEqual(
outputs.logits_per_text.shape , (inputs.input_ids.shape[0], inputs.pixel_values.shape[0]) , )
__lowerCamelCase = np.array([[1.2_284_727, 0.3_104_122]] )
self.assertTrue(np.allclose(outputs.logits_per_image , __UpperCAmelCase , atol=1E-3 ) )
| 330 |
import gc
import importlib.metadata
import tempfile
import unittest
from packaging import version
from transformers import (
AutoModel,
AutoModelForCausalLM,
AutoModelForSeqaSeqLM,
AutoModelForSequenceClassification,
AutoTokenizer,
BitsAndBytesConfig,
pipeline,
)
from transformers.testing_utils import (
is_torch_available,
require_accelerate,
require_bitsandbytes,
require_torch,
require_torch_gpu,
require_torch_multi_gpu,
slow,
)
def a__ ( _UpperCamelCase : Optional[int] ):
if model.config.model_type == "gpt2":
return model.transformer.h[0].mlp.c_fc
return model.transformer.h[0].mlp.dense_ah_to_h
if is_torch_available():
import torch
import torch.nn as nn
class __lowerCAmelCase ( nn.Module ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
super().__init__()
__lowerCamelCase = module
__lowerCamelCase = nn.Sequential(
nn.Linear(module.in_features , __UpperCAmelCase , bias=__UpperCAmelCase ) , nn.Linear(__UpperCAmelCase , module.out_features , bias=__UpperCAmelCase ) , )
__lowerCamelCase = (2.0 / (5 * min(module.in_features , module.out_features ))) ** 0.5
nn.init.normal_(self.adapter[0].weight , std=__UpperCAmelCase )
nn.init.zeros_(self.adapter[1].weight )
self.adapter.to(module.weight.device )
def lowerCamelCase ( self , __UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return self.module(__UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ) + self.adapter(__UpperCAmelCase )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class __lowerCAmelCase ( unittest.TestCase ):
# We keep the constants inside the init function and model loading inside setUp function
# We need to test on relatively large models (aka >1b parameters otherwise the quantiztion may not work as expected)
# Therefore here we use only bloom-1b3 to test our module
lowerCAmelCase__ = """bigscience/bloom-1b7"""
# Constant values
lowerCAmelCase__ = 2.1_09_65_95_52_69_25_74
lowerCAmelCase__ = """Hello my name is"""
lowerCAmelCase__ = set()
EXPECTED_OUTPUTS.add("""Hello my name is John and I am a professional photographer. I""" )
EXPECTED_OUTPUTS.add("""Hello my name is John.\nI am a friend of your father.\n""" )
EXPECTED_OUTPUTS.add("""Hello my name is John Doe, I am a student at the University""" )
lowerCAmelCase__ = 1_0
def lowerCamelCase ( self ):
'''simple docstring'''
# Models and tokenizer
__lowerCamelCase = AutoTokenizer.from_pretrained(self.model_name )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# Models and tokenizer
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , torch_dtype=torch.floataa , device_map='''auto''' )
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
def lowerCamelCase ( self ):
'''simple docstring'''
del self.model_fpaa
del self.model_abit
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_abit.config
self.assertTrue(hasattr(__UpperCAmelCase , '''quantization_config''' ) )
__lowerCamelCase = config.to_dict()
__lowerCamelCase = config.to_diff_dict()
__lowerCamelCase = config.to_json_string()
def lowerCamelCase ( self ):
'''simple docstring'''
from bitsandbytes.nn import Paramsabit
__lowerCamelCase = self.model_fpaa.get_memory_footprint()
__lowerCamelCase = self.model_abit.get_memory_footprint()
self.assertAlmostEqual(mem_fpaa / mem_abit , self.EXPECTED_RELATIVE_DIFFERENCE )
__lowerCamelCase = get_some_linear_layer(self.model_abit )
self.assertTrue(linear.weight.__class__ == Paramsabit )
def lowerCamelCase ( self ):
'''simple docstring'''
from transformers import TaPreTrainedModel
self.model_fpaa.get_memory_footprint()
self.model_abit.get_memory_footprint()
for name, module in self.model_abit.named_modules():
if isinstance(__UpperCAmelCase , torch.nn.Linear ):
if name not in ["lm_head"] + TaPreTrainedModel._keep_in_fpaa_modules:
# 4-bit parameters are packed in uint8 variables
self.assertTrue(module.weight.dtype == torch.uinta )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = self.model_abit.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_sequences[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitsAndBytesConfig()
__lowerCamelCase = True
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = model_abit_from_config.generate(
input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_sequences[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
def lowerCamelCase ( self ):
'''simple docstring'''
with self.assertRaises(__UpperCAmelCase ), tempfile.TemporaryDirectory() as tmpdirname:
self.model_abit.save_pretrained(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitsAndBytesConfig()
with self.assertRaises(__UpperCAmelCase ):
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=__UpperCAmelCase , load_in_abit=__UpperCAmelCase , device_map='''auto''' , bnb_abit_quant_type='''nf4''' , )
def lowerCamelCase ( self ):
'''simple docstring'''
with self.assertRaises(__UpperCAmelCase ):
# Tries with `str`
self.model_abit.to('''cpu''' )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `dtype``
self.model_abit.to(torch.floataa )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.to(torch.device('''cuda:0''' ) )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.float()
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.half()
# Test if we did not break anything
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = self.model_fpaa.to(torch.floataa )
__lowerCamelCase = self.model_fpaa.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.to('''cpu''' )
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.half()
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.float()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = AutoModelForSeqaSeqLM.from_pretrained('''t5-small''' , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
self.assertTrue(model.decoder.block[0].layer[2].DenseReluDense.wo.weight.dtype == torch.floataa )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class __lowerCAmelCase ( unittest.TestCase ):
@classmethod
def lowerCamelCase ( cls ):
'''simple docstring'''
__lowerCamelCase = '''t5-small'''
__lowerCamelCase = '''google/flan-t5-small''' # flan-t5 uses dense-act instead of dense-relu-dense
__lowerCamelCase = AutoTokenizer.from_pretrained(cls.model_name )
__lowerCamelCase = '''Translate in German: Hello, my dog is cute'''
def lowerCamelCase ( self ):
'''simple docstring'''
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
from transformers import TaForConditionalGeneration
__lowerCamelCase = TaForConditionalGeneration._keep_in_fpaa_modules
__lowerCamelCase = None
# test with `t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
# test with `flan-t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
__lowerCamelCase = modules
def lowerCamelCase ( self ):
'''simple docstring'''
import bitsandbytes as bnb
from transformers import TaForConditionalGeneration
# test with `t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# there was a bug with decoders - this test checks that it is fixed
self.assertTrue(isinstance(model.decoder.block[0].layer[0].SelfAttention.q , bnb.nn.Linearabit ) )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
# test with `flan-t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# model_name
__lowerCamelCase = '''bigscience/bloom-560m'''
__lowerCamelCase = '''t5-small'''
# Different types of model
__lowerCamelCase = AutoModel.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# Sequence classification model
__lowerCamelCase = AutoModelForSequenceClassification.from_pretrained(
self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# CausalLM model
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# Seq2seq model
__lowerCamelCase = AutoModelForSeqaSeqLM.from_pretrained(
self.seq_to_seq_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
def lowerCamelCase ( self ):
'''simple docstring'''
del self.base_model
del self.sequence_model
del self.model_abit
del self.seq_to_seq_model
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
from bitsandbytes.nn import Paramsabit
self.assertTrue(self.base_model.h[-1].mlp.dense_ah_to_h.weight.__class__ == Paramsabit )
# Other heads should be nn.Parameter
self.assertTrue(self.model_abit.lm_head.weight.__class__ == torch.nn.Parameter )
self.assertTrue(self.sequence_model.score.weight.__class__ == torch.nn.Parameter )
self.assertTrue(self.seq_to_seq_model.lm_head.weight.__class__ == torch.nn.Parameter )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
del self.pipe
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = pipeline(
'''text-generation''' , model=self.model_name , model_kwargs={'''device_map''': '''auto''', '''load_in_4bit''': True, '''torch_dtype''': torch.floataa} , max_new_tokens=self.MAX_NEW_TOKENS , )
# Real second forward pass
__lowerCamelCase = self.pipe(self.input_text )
self.assertIn(pipeline_output[0]['''generated_text'''] , self.EXPECTED_OUTPUTS )
@require_torch_multi_gpu
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , load_in_abit=__UpperCAmelCase , device_map='''balanced''' )
# Check correct device map
self.assertEqual(set(model_parallel.hf_device_map.values() ) , {0, 1} )
# Check that inference pass works on the model
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
# Second real batch
__lowerCamelCase = model_parallel.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_parallel[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''facebook/opt-350m'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
if version.parse(importlib.metadata.version('''bitsandbytes''' ) ) < version.parse('''0.37.0''' ):
return
# Step 1: freeze all parameters
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase )
self.assertEqual(set(model.hf_device_map.values() ) , {torch.cuda.current_device()} )
for param in model.parameters():
__lowerCamelCase = False # freeze the model - train adapters later
if param.ndim == 1:
# cast the small parameters (e.g. layernorm) to fp32 for stability
__lowerCamelCase = param.data.to(torch.floataa )
# Step 2: add adapters
for _, module in model.named_modules():
if "OPTAttention" in repr(type(__UpperCAmelCase ) ):
__lowerCamelCase = LoRALayer(module.q_proj , rank=16 )
__lowerCamelCase = LoRALayer(module.k_proj , rank=16 )
__lowerCamelCase = LoRALayer(module.v_proj , rank=16 )
# Step 3: dummy batch
__lowerCamelCase = self.tokenizer('''Test batch ''' , return_tensors='''pt''' ).to(0 )
# Step 4: Check if the gradient is not None
with torch.cuda.amp.autocast():
__lowerCamelCase = model.forward(**__UpperCAmelCase )
out.logits.norm().backward()
for module in model.modules():
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
self.assertTrue(module.adapter[1].weight.grad is not None )
self.assertTrue(module.adapter[1].weight.grad.norm().item() > 0 )
elif isinstance(__UpperCAmelCase , nn.Embedding ):
self.assertTrue(module.weight.grad is None )
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """gpt2-xl"""
lowerCAmelCase__ = 3.31_91_85_48_54_15_21_87
| 330 | 1 |
def a__ ( _UpperCamelCase : list[int] ):
__lowerCamelCase = []
if len(_UpperCamelCase ) == 1:
return [nums.copy()]
for _ in range(len(_UpperCamelCase ) ):
__lowerCamelCase = nums.pop(0 )
__lowerCamelCase = permute(_UpperCamelCase )
for perm in permutations:
perm.append(_UpperCamelCase )
result.extend(_UpperCamelCase )
nums.append(_UpperCamelCase )
return result
def a__ ( _UpperCamelCase : Any ):
def backtrack(_UpperCamelCase : List[Any] ):
if start == len(_UpperCamelCase ) - 1:
output.append(nums[:] )
else:
for i in range(_UpperCamelCase ,len(_UpperCamelCase ) ):
__lowerCamelCase ,__lowerCamelCase = nums[i], nums[start]
backtrack(start + 1 )
__lowerCamelCase ,__lowerCamelCase = nums[i], nums[start] # backtrack
__lowerCamelCase = []
backtrack(0 )
return output
if __name__ == "__main__":
import doctest
# use res to print the data in permute2 function
a_ = permutea([1, 2, 3])
print(res)
doctest.testmod()
| 330 |
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 __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 42
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ ):
lowerCAmelCase__ = True
@register_to_config
def __init__( self , __UpperCAmelCase = 3 , __UpperCAmelCase = 3 , __UpperCAmelCase = ("DownEncoderBlock2D",) , __UpperCAmelCase = ("UpDecoderBlock2D",) , __UpperCAmelCase = (64,) , __UpperCAmelCase = 1 , __UpperCAmelCase = "silu" , __UpperCAmelCase = 4 , __UpperCAmelCase = 32 , __UpperCAmelCase = 32 , __UpperCAmelCase = 0.18_215 , ):
'''simple docstring'''
super().__init__()
# pass init params to Encoder
__lowerCamelCase = Encoder(
in_channels=__UpperCAmelCase , out_channels=__UpperCAmelCase , down_block_types=__UpperCAmelCase , block_out_channels=__UpperCAmelCase , layers_per_block=__UpperCAmelCase , act_fn=__UpperCAmelCase , norm_num_groups=__UpperCAmelCase , double_z=__UpperCAmelCase , )
# pass init params to Decoder
__lowerCamelCase = Decoder(
in_channels=__UpperCAmelCase , out_channels=__UpperCAmelCase , up_block_types=__UpperCAmelCase , block_out_channels=__UpperCAmelCase , layers_per_block=__UpperCAmelCase , norm_num_groups=__UpperCAmelCase , act_fn=__UpperCAmelCase , )
__lowerCamelCase = nn.Convad(2 * latent_channels , 2 * latent_channels , 1 )
__lowerCamelCase = nn.Convad(__UpperCAmelCase , __UpperCAmelCase , 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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
if isinstance(__UpperCAmelCase , (Encoder, Decoder) ):
__lowerCamelCase = value
def lowerCamelCase ( self , __UpperCAmelCase = True ):
'''simple docstring'''
__lowerCamelCase = use_tiling
def lowerCamelCase ( self ):
'''simple docstring'''
self.enable_tiling(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = True
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = False
@property
# Copied from diffusers.models.unet_2d_condition.UNet2DConditionModel.attn_processors
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = {}
def fn_recursive_add_processors(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
if hasattr(__UpperCAmelCase , '''set_processor''' ):
__lowerCamelCase = module.processor
for sub_name, child in module.named_children():
fn_recursive_add_processors(F"""{name}.{sub_name}""" , __UpperCAmelCase , __UpperCAmelCase )
return processors
for name, module in self.named_children():
fn_recursive_add_processors(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
return processors
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = len(self.attn_processors.keys() )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) != count:
raise ValueError(
F"""A dict of processors was passed, but the number of processors {len(__UpperCAmelCase )} does not match the"""
F""" number of attention layers: {count}. Please make sure to pass {count} processor classes.""" )
def fn_recursive_attn_processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
if hasattr(__UpperCAmelCase , '''set_processor''' ):
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
module.set_processor(__UpperCAmelCase )
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}""" , __UpperCAmelCase , __UpperCAmelCase )
for name, module in self.named_children():
fn_recursive_attn_processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
self.set_attn_processor(AttnProcessor() )
@apply_forward_hook
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
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(__UpperCAmelCase , return_dict=__UpperCAmelCase )
if self.use_slicing and x.shape[0] > 1:
__lowerCamelCase = [self.encoder(__UpperCAmelCase ) for x_slice in x.split(1 )]
__lowerCamelCase = torch.cat(__UpperCAmelCase )
else:
__lowerCamelCase = self.encoder(__UpperCAmelCase )
__lowerCamelCase = self.quant_conv(__UpperCAmelCase )
__lowerCamelCase = DiagonalGaussianDistribution(__UpperCAmelCase )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
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(__UpperCAmelCase , return_dict=__UpperCAmelCase )
__lowerCamelCase = self.post_quant_conv(__UpperCAmelCase )
__lowerCamelCase = self.decoder(__UpperCAmelCase )
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
@apply_forward_hook
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
if self.use_slicing and z.shape[0] > 1:
__lowerCamelCase = [self._decode(__UpperCAmelCase ).sample for z_slice in z.split(1 )]
__lowerCamelCase = torch.cat(__UpperCAmelCase )
else:
__lowerCamelCase = self._decode(__UpperCAmelCase ).sample
if not return_dict:
return (decoded,)
return DecoderOutput(sample=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(a.shape[2] , b.shape[2] , __UpperCAmelCase )
for y in range(__UpperCAmelCase ):
__lowerCamelCase = a[:, :, -blend_extent + y, :] * (1 - y / blend_extent) + b[:, :, y, :] * (y / blend_extent)
return b
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(a.shape[3] , b.shape[3] , __UpperCAmelCase )
for x in range(__UpperCAmelCase ):
__lowerCamelCase = a[:, :, :, -blend_extent + x] * (1 - x / blend_extent) + b[:, :, :, x] * (x / blend_extent)
return b
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
__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] , __UpperCAmelCase ):
__lowerCamelCase = []
for j in range(0 , x.shape[3] , __UpperCAmelCase ):
__lowerCamelCase = x[:, :, i : i + self.tile_sample_min_size, j : j + self.tile_sample_min_size]
__lowerCamelCase = self.encoder(__UpperCAmelCase )
__lowerCamelCase = self.quant_conv(__UpperCAmelCase )
row.append(__UpperCAmelCase )
rows.append(__UpperCAmelCase )
__lowerCamelCase = []
for i, row in enumerate(__UpperCAmelCase ):
__lowerCamelCase = []
for j, tile in enumerate(__UpperCAmelCase ):
# 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] , __UpperCAmelCase , __UpperCAmelCase )
if j > 0:
__lowerCamelCase = self.blend_h(row[j - 1] , __UpperCAmelCase , __UpperCAmelCase )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(__UpperCAmelCase , dim=3 ) )
__lowerCamelCase = torch.cat(__UpperCAmelCase , dim=2 )
__lowerCamelCase = DiagonalGaussianDistribution(__UpperCAmelCase )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
__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] , __UpperCAmelCase ):
__lowerCamelCase = []
for j in range(0 , z.shape[3] , __UpperCAmelCase ):
__lowerCamelCase = z[:, :, i : i + self.tile_latent_min_size, j : j + self.tile_latent_min_size]
__lowerCamelCase = self.post_quant_conv(__UpperCAmelCase )
__lowerCamelCase = self.decoder(__UpperCAmelCase )
row.append(__UpperCAmelCase )
rows.append(__UpperCAmelCase )
__lowerCamelCase = []
for i, row in enumerate(__UpperCAmelCase ):
__lowerCamelCase = []
for j, tile in enumerate(__UpperCAmelCase ):
# 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] , __UpperCAmelCase , __UpperCAmelCase )
if j > 0:
__lowerCamelCase = self.blend_h(row[j - 1] , __UpperCAmelCase , __UpperCAmelCase )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(__UpperCAmelCase , dim=3 ) )
__lowerCamelCase = torch.cat(__UpperCAmelCase , dim=2 )
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False , __UpperCAmelCase = True , __UpperCAmelCase = None , ):
'''simple docstring'''
__lowerCamelCase = sample
__lowerCamelCase = self.encode(__UpperCAmelCase ).latent_dist
if sample_posterior:
__lowerCamelCase = posterior.sample(generator=__UpperCAmelCase )
else:
__lowerCamelCase = posterior.mode()
__lowerCamelCase = self.decode(__UpperCAmelCase ).sample
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
| 330 | 1 |
from collections import defaultdict
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
__lowerCamelCase = first_str.lower().strip()
__lowerCamelCase = second_str.lower().strip()
# Remove whitespace
__lowerCamelCase = first_str.replace(''' ''' ,'''''' )
__lowerCamelCase = second_str.replace(''' ''' ,'''''' )
# Strings of different lengths are not anagrams
if len(_UpperCamelCase ) != len(_UpperCamelCase ):
return False
# Default values for count should be 0
__lowerCamelCase = defaultdict(_UpperCamelCase )
# For each character in input strings,
# increment count in the corresponding
for i in range(len(_UpperCamelCase ) ):
count[first_str[i]] += 1
count[second_str[i]] -= 1
return all(_count == 0 for _count in count.values() )
if __name__ == "__main__":
from doctest import testmod
testmod()
a_ = input("""Enter the first string """).strip()
a_ = input("""Enter the second string """).strip()
a_ = check_anagrams(input_a, input_b)
print(f"{input_a} and {input_b} are {'' if status else 'not '}anagrams.")
| 330 |
import argparse
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import BigBirdPegasusConfig, BigBirdPegasusForConditionalGeneration
a_ = [
# tf -> hf
("""/""", """."""),
("""layer_""", """layers."""),
("""kernel""", """weight"""),
("""beta""", """bias"""),
("""gamma""", """weight"""),
("""pegasus""", """model"""),
]
a_ = [
(""".output.dense""", """.fc2"""),
("""intermediate.LayerNorm""", """final_layer_norm"""),
("""intermediate.dense""", """fc1"""),
]
a_ = (
INIT_COMMON
+ [
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.out_proj"""),
("""attention.self""", """self_attn"""),
("""attention.encdec.LayerNorm""", """encoder_attn_layer_norm"""),
("""attention.encdec_output.dense""", """encoder_attn.out_proj"""),
("""attention.encdec""", """encoder_attn"""),
("""key""", """k_proj"""),
("""value""", """v_proj"""),
("""query""", """q_proj"""),
("""decoder.LayerNorm""", """decoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = (
INIT_COMMON
+ [
("""embeddings.word_embeddings""", """shared.weight"""),
("""embeddings.position_embeddings""", """embed_positions.weight"""),
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.output"""),
("""attention.self""", """self_attn.self"""),
("""encoder.LayerNorm""", """encoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = [
"""encdec/key/bias""",
"""encdec/query/bias""",
"""encdec/value/bias""",
"""self/key/bias""",
"""self/query/bias""",
"""self/value/bias""",
"""encdec_output/dense/bias""",
"""attention/output/dense/bias""",
]
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : Optional[Any] ):
for tf_name, hf_name in patterns:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = BigBirdPegasusConfig(**_UpperCamelCase )
__lowerCamelCase = BigBirdPegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
__lowerCamelCase = {}
# separating decoder weights
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if k.startswith('''pegasus/decoder''' )}
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if not k.startswith('''pegasus/decoder''' )}
for k, v in tqdm(decoder_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = DECODER_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
for k, v in tqdm(remaining_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = REMAINING_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict and k != "pegasus/embeddings/position_embeddings":
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
if k != "pegasus/embeddings/position_embeddings":
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
__lowerCamelCase = mapping['''model.embed_positions.weight''']
__lowerCamelCase = mapping.pop('''model.embed_positions.weight''' )
__lowerCamelCase ,__lowerCamelCase = torch_model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k
for k in missing
if k
not in [
'''final_logits_bias''',
'''model.encoder.embed_tokens.weight''',
'''model.decoder.embed_tokens.weight''',
'''lm_head.weight''',
]
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ,_UpperCamelCase : dict ):
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = convert_bigbird_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""--save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
a_ = {}
convert_bigbird_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir, config_update=config_update)
| 330 | 1 |
from . import (
albert,
align,
altclip,
audio_spectrogram_transformer,
auto,
autoformer,
bark,
bart,
barthez,
bartpho,
beit,
bert,
bert_generation,
bert_japanese,
bertweet,
big_bird,
bigbird_pegasus,
biogpt,
bit,
blenderbot,
blenderbot_small,
blip,
blip_a,
bloom,
bridgetower,
byta,
camembert,
canine,
chinese_clip,
clap,
clip,
clipseg,
codegen,
conditional_detr,
convbert,
convnext,
convnextva,
cpm,
cpmant,
ctrl,
cvt,
dataavec,
deberta,
deberta_va,
decision_transformer,
deformable_detr,
deit,
deprecated,
deta,
detr,
dialogpt,
dinat,
distilbert,
dit,
donut,
dpr,
dpt,
efficientformer,
efficientnet,
electra,
encodec,
encoder_decoder,
ernie,
ernie_m,
esm,
falcon,
flaubert,
flava,
fnet,
focalnet,
fsmt,
funnel,
git,
glpn,
gpta,
gpt_bigcode,
gpt_neo,
gpt_neox,
gpt_neox_japanese,
gpt_swa,
gptj,
gptsan_japanese,
graphormer,
groupvit,
herbert,
hubert,
ibert,
imagegpt,
informer,
instructblip,
jukebox,
layoutlm,
layoutlmva,
layoutlmva,
layoutxlm,
led,
levit,
lilt,
llama,
longformer,
longta,
luke,
lxmert,
mam_aaa,
marian,
markuplm,
maskaformer,
maskformer,
mbart,
mbartaa,
mega,
megatron_bert,
megatron_gpta,
mgp_str,
mluke,
mobilebert,
mobilenet_va,
mobilenet_va,
mobilevit,
mobilevitva,
mpnet,
mra,
mta,
musicgen,
mvp,
nat,
nezha,
nllb,
nllb_moe,
nystromformer,
oneformer,
open_llama,
openai,
opt,
owlvit,
pegasus,
pegasus_x,
perceiver,
phobert,
pixastruct,
plbart,
poolformer,
prophetnet,
qdqbert,
rag,
realm,
reformer,
regnet,
rembert,
resnet,
roberta,
roberta_prelayernorm,
roc_bert,
roformer,
rwkv,
sam,
segformer,
sew,
sew_d,
speech_encoder_decoder,
speech_to_text,
speech_to_text_a,
speechta,
splinter,
squeezebert,
swiftformer,
swin,
swinasr,
swinva,
switch_transformers,
ta,
table_transformer,
tapas,
time_series_transformer,
timesformer,
timm_backbone,
transfo_xl,
trocr,
tvlt,
umta,
unispeech,
unispeech_sat,
upernet,
videomae,
vilt,
vision_encoder_decoder,
vision_text_dual_encoder,
visual_bert,
vit,
vit_hybrid,
vit_mae,
vit_msn,
vivit,
wavaveca,
wavaveca_conformer,
wavaveca_phoneme,
wavaveca_with_lm,
wavlm,
whisper,
x_clip,
xglm,
xlm,
xlm_prophetnet,
xlm_roberta,
xlm_roberta_xl,
xlnet,
xmod,
yolos,
yoso,
)
| 330 |
import uuid
from typing import Any, Dict, List, Optional, Union
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
if is_torch_available():
import torch
a_ = logging.get_logger(__name__)
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=None , __UpperCAmelCase=None ):
'''simple docstring'''
if not conversation_id:
__lowerCamelCase = uuid.uuida()
if past_user_inputs is None:
__lowerCamelCase = []
if generated_responses is None:
__lowerCamelCase = []
__lowerCamelCase = conversation_id
__lowerCamelCase = past_user_inputs
__lowerCamelCase = generated_responses
__lowerCamelCase = text
def __eq__( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
return False
if self.uuid == other.uuid:
return True
return (
self.new_user_input == other.new_user_input
and self.past_user_inputs == other.past_user_inputs
and self.generated_responses == other.generated_responses
)
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False ):
'''simple docstring'''
if self.new_user_input:
if overwrite:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" was overwritten """
F"""with: \"{text}\".""" )
__lowerCamelCase = text
else:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" new input """
F"""ignored: \"{text}\". Set `overwrite` to True to overwrite unprocessed user input""" )
else:
__lowerCamelCase = text
def lowerCamelCase ( self ):
'''simple docstring'''
if self.new_user_input:
self.past_user_inputs.append(self.new_user_input )
__lowerCamelCase = None
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
self.generated_responses.append(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
for user_input, generated_response in zip(self.past_user_inputs , self.generated_responses ):
yield True, user_input
yield False, generated_response
if self.new_user_input:
yield True, self.new_user_input
def __repr__( self ):
'''simple docstring'''
__lowerCamelCase = F"""Conversation id: {self.uuid} \n"""
for is_user, text in self.iter_texts():
__lowerCamelCase = '''user''' if is_user else '''bot'''
output += F"""{name} >> {text} \n"""
return output
@add_end_docstrings(
lowerCAmelCase__ , r"""
min_length_for_response (`int`, *optional*, defaults to 32):
The minimum length (in number of tokens) for a response.
minimum_tokens (`int`, *optional*, defaults to 10):
The minimum length of tokens to leave for a response.
""" , )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
if self.tokenizer.pad_token_id is None:
__lowerCamelCase = self.tokenizer.eos_token
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = {}
__lowerCamelCase = {}
__lowerCamelCase = {}
if min_length_for_response is not None:
__lowerCamelCase = min_length_for_response
if minimum_tokens is not None:
__lowerCamelCase = minimum_tokens
if "max_length" in generate_kwargs:
__lowerCamelCase = generate_kwargs['''max_length''']
# self.max_length = generate_kwargs.get("max_length", self.model.config.max_length)
if clean_up_tokenization_spaces is not None:
__lowerCamelCase = clean_up_tokenization_spaces
if generate_kwargs:
forward_params.update(__UpperCAmelCase )
return preprocess_params, forward_params, postprocess_params
def __call__( self , __UpperCAmelCase , __UpperCAmelCase=0 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = super().__call__(__UpperCAmelCase , num_workers=__UpperCAmelCase , **__UpperCAmelCase )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) == 1:
return outputs[0]
return outputs
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=32 ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError('''ConversationalPipeline, expects Conversation as inputs''' )
if conversation.new_user_input is None:
raise ValueError(
F"""Conversation with UUID {type(conversation.uuid )} does not contain new user input to process. """
'''Add user inputs with the conversation\'s `add_user_input` method''' )
if hasattr(self.tokenizer , '''_build_conversation_input_ids''' ):
__lowerCamelCase = self.tokenizer._build_conversation_input_ids(__UpperCAmelCase )
else:
# If the tokenizer cannot handle conversations, we default to only the old version
__lowerCamelCase = self._legacy_parse_and_tokenize(__UpperCAmelCase )
if self.framework == "pt":
__lowerCamelCase = torch.LongTensor([input_ids] )
elif self.framework == "tf":
__lowerCamelCase = tf.constant([input_ids] )
return {"input_ids": input_ids, "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=10 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = generate_kwargs.get('''max_length''' , self.model.config.max_length )
__lowerCamelCase = model_inputs['''input_ids'''].shape[1]
if max_length - minimum_tokens < n:
logger.warning(F"""Conversation input is to long ({n}), trimming it to ({max_length} - {minimum_tokens})""" )
__lowerCamelCase = max_length - minimum_tokens
__lowerCamelCase = model_inputs['''input_ids'''][:, -trim:]
if "attention_mask" in model_inputs:
__lowerCamelCase = model_inputs['''attention_mask'''][:, -trim:]
__lowerCamelCase = model_inputs.pop('''conversation''' )
__lowerCamelCase = max_length
__lowerCamelCase = self.model.generate(**__UpperCAmelCase , **__UpperCAmelCase )
if self.model.config.is_encoder_decoder:
__lowerCamelCase = 1
else:
__lowerCamelCase = n
return {"output_ids": output_ids[:, start_position:], "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = model_outputs['''output_ids''']
__lowerCamelCase = self.tokenizer.decode(
output_ids[0] , skip_special_tokens=__UpperCAmelCase , clean_up_tokenization_spaces=__UpperCAmelCase , )
__lowerCamelCase = model_outputs['''conversation''']
conversation.mark_processed()
conversation.append_response(__UpperCAmelCase )
return conversation
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer.eos_token_id
__lowerCamelCase = []
for is_user, text in conversation.iter_texts():
if eos_token_id is not None:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) + [eos_token_id] )
else:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) )
if len(__UpperCAmelCase ) > self.tokenizer.model_max_length:
__lowerCamelCase = input_ids[-self.tokenizer.model_max_length :]
return input_ids
| 330 | 1 |
import argparse
import os
# New Code #
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
from accelerate.utils import find_executable_batch_size
########################################################################
# This is a fully working simple example to use Accelerate,
# specifically showcasing how to ensure out-of-memory errors never
# interrupt training, and builds off the `nlp_example.py` script.
#
# This example trains a Bert base model on GLUE MRPC
# in any of the following settings (with the same script):
# - single CPU or single GPU
# - multi GPUS (using PyTorch distributed mode)
# - (multi) TPUs
# - fp16 (mixed-precision) or fp32 (normal precision)
#
# New additions from the base script can be found quickly by
# looking for the # New Code # tags
#
# To run it in each of these various modes, follow the instructions
# in the readme for examples:
# https://github.com/huggingface/accelerate/tree/main/examples
#
########################################################################
a_ = 16
a_ = 32
def a__ ( _UpperCamelCase : Accelerator ,_UpperCamelCase : int = 16 ):
__lowerCamelCase = AutoTokenizer.from_pretrained('''bert-base-cased''' )
__lowerCamelCase = load_dataset('''glue''' ,'''mrpc''' )
def tokenize_function(_UpperCamelCase : Optional[Any] ):
# max_length=None => use the model max length (it's actually the default)
__lowerCamelCase = tokenizer(examples['''sentence1'''] ,examples['''sentence2'''] ,truncation=_UpperCamelCase ,max_length=_UpperCamelCase )
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(
_UpperCamelCase ,batched=_UpperCamelCase ,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(_UpperCamelCase : Any ):
# On TPU it's best to pad everything to the same length or training will be very slow.
__lowerCamelCase = 1_28 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(
_UpperCamelCase ,padding='''longest''' ,max_length=_UpperCamelCase ,pad_to_multiple_of=_UpperCamelCase ,return_tensors='''pt''' ,)
# Instantiate dataloaders.
__lowerCamelCase = DataLoader(
tokenized_datasets['''train'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
__lowerCamelCase = DataLoader(
tokenized_datasets['''validation'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
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
a_ = mocked_dataloaders # noqa: F811
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# For testing only
if os.environ.get('''TESTING_MOCKED_DATALOADERS''' ,_UpperCamelCase ) == "1":
__lowerCamelCase = 2
# Initialize accelerator
__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'''] )
__lowerCamelCase = evaluate.load('''glue''' ,'''mrpc''' )
# New Code #
# We now can define an inner training loop function. It should take a batch size as the only parameter,
# and build the dataloaders in there.
# It also gets our decorator
@find_executable_batch_size(starting_batch_size=_UpperCamelCase )
def inner_training_loop(_UpperCamelCase : Union[str, Any] ):
# And now just move everything below under this function
# We need to bring in the Accelerator object from earlier
nonlocal accelerator
# And reset all of its attributes that could hold onto any memory:
accelerator.free_memory()
# Then we can declare the model, optimizer, and everything else:
set_seed(_UpperCamelCase )
# 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=_UpperCamelCase )
# 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=_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = get_dataloaders(_UpperCamelCase ,_UpperCamelCase )
# Instantiate scheduler
__lowerCamelCase = get_linear_schedule_with_warmup(
optimizer=_UpperCamelCase ,num_warmup_steps=1_00 ,num_training_steps=(len(_UpperCamelCase ) * num_epochs) ,)
# Prepare everything
# There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the
# prepare method.
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Now we train the model
for epoch in range(_UpperCamelCase ):
model.train()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.loss
accelerator.backward(_UpperCamelCase )
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
model.eval()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
with torch.no_grad():
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.logits.argmax(dim=-1 )
__lowerCamelCase ,__lowerCamelCase = accelerator.gather_for_metrics((predictions, batch['''labels''']) )
metric.add_batch(
predictions=_UpperCamelCase ,references=_UpperCamelCase ,)
__lowerCamelCase = metric.compute()
# Use accelerator.print to print only on the main process.
accelerator.print(F"""epoch {epoch}:""" ,_UpperCamelCase )
# New Code #
# And call it at the end with no arguments
# Note: You could also refactor this outside of your training loop function
inner_training_loop()
def a__ ( ):
__lowerCamelCase = argparse.ArgumentParser(description='''Simple example of training script.''' )
parser.add_argument(
'''--mixed_precision''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,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.''' )
__lowerCamelCase = parser.parse_args()
__lowerCamelCase = {'''lr''': 2e-5, '''num_epochs''': 3, '''seed''': 42, '''batch_size''': 16}
training_function(_UpperCamelCase ,_UpperCamelCase )
if __name__ == "__main__":
main()
| 330 |
import argparse
import os
from pathlib import Path
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import PegasusConfig, PegasusForConditionalGeneration, PegasusTokenizer
from transformers.models.pegasus.configuration_pegasus import DEFAULTS, task_specific_params
a_ = [
# replace left string with right string to get the relevant state_dict key (identical state dict to bart)
["""memory_attention""", """encoder_attn"""],
["""attention""", """attn"""],
["""/""", """."""],
[""".LayerNorm.gamma""", """_layer_norm.weight"""],
[""".LayerNorm.beta""", """_layer_norm.bias"""],
["""r.layer_""", """r.layers."""],
["""output_proj""", """out_proj"""],
["""ffn.dense_1.""", """fc2."""],
["""ffn.dense.""", """fc1."""],
["""ffn_layer_norm""", """final_layer_norm"""],
["""kernel""", """weight"""],
["""encoder_layer_norm.""", """encoder.layer_norm."""],
["""decoder_layer_norm.""", """decoder.layer_norm."""],
["""embeddings.weights""", """shared.weight"""],
]
def a__ ( _UpperCamelCase : int ):
for pegasus_name, hf_name in PATTERNS:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = DEFAULTS.copy()
cfg_kwargs.update(_UpperCamelCase )
__lowerCamelCase = PegasusConfig(**_UpperCamelCase )
__lowerCamelCase = PegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.model.state_dict()
__lowerCamelCase = {}
for k, v in tf_weights.items():
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase )
if new_k not in sd:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if "dense" in k or "proj" in new_k:
__lowerCamelCase = v.T
__lowerCamelCase = torch.tensor(_UpperCamelCase ,dtype=sd[new_k].dtype )
assert v.shape == sd[new_k].shape, F"""{new_k}, {k}, {v.shape}, {sd[new_k].shape}"""
# make sure embedding.padding_idx is respected
__lowerCamelCase = torch.zeros_like(mapping['''shared.weight'''][cfg.pad_token_id + 1] )
__lowerCamelCase = mapping['''shared.weight''']
__lowerCamelCase = mapping['''shared.weight''']
__lowerCamelCase = {k: torch.zeros_like(_UpperCamelCase ) for k, v in sd.items() if k.endswith('''bias''' ) and k not in mapping}
mapping.update(**_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = torch_model.model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k for k in missing if k not in ['''encoder.embed_positions.weight''', '''decoder.embed_positions.weight''']
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : str="./ckpt/aeslc/model.ckpt-32000" ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''Adafactor''', '''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# save tokenizer first
__lowerCamelCase = Path(_UpperCamelCase ).parent.name
__lowerCamelCase = task_specific_params[F"""summarization_{dataset}"""]['''max_position_embeddings''']
__lowerCamelCase = PegasusTokenizer.from_pretrained('''sshleifer/pegasus''' ,model_max_length=_UpperCamelCase )
assert tok.model_max_length == desired_max_model_length
tok.save_pretrained(_UpperCamelCase )
# convert model
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = task_specific_params[F"""summarization_{dataset}"""]
if dataset == "large":
__lowerCamelCase = task_specific_params
__lowerCamelCase = convert_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
sd.pop('''model.decoder.embed_positions.weight''' )
sd.pop('''model.encoder.embed_positions.weight''' )
torch.save(_UpperCamelCase ,Path(_UpperCamelCase ) / '''pytorch_model.bin''' )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument("""tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
if args.save_dir is None:
a_ = Path(args.tf_ckpt_path).parent.name
a_ = os.path.join("""pegasus""", dataset)
convert_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir)
| 330 | 1 |
from collections import OrderedDict
from typing import Mapping
from packaging import version
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...utils import logging
a_ = logging.get_logger(__name__)
a_ = {
"""hustvl/yolos-small""": """https://huggingface.co/hustvl/yolos-small/resolve/main/config.json""",
# See all YOLOS models at https://huggingface.co/models?filter=yolos
}
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """yolos"""
def __init__( self , __UpperCAmelCase=768 , __UpperCAmelCase=12 , __UpperCAmelCase=12 , __UpperCAmelCase=3072 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.02 , __UpperCAmelCase=1E-1_2 , __UpperCAmelCase=[512, 864] , __UpperCAmelCase=16 , __UpperCAmelCase=3 , __UpperCAmelCase=True , __UpperCAmelCase=100 , __UpperCAmelCase=True , __UpperCAmelCase=False , __UpperCAmelCase=1 , __UpperCAmelCase=5 , __UpperCAmelCase=2 , __UpperCAmelCase=5 , __UpperCAmelCase=2 , __UpperCAmelCase=0.1 , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__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 = image_size
__lowerCamelCase = patch_size
__lowerCamelCase = num_channels
__lowerCamelCase = qkv_bias
__lowerCamelCase = num_detection_tokens
__lowerCamelCase = use_mid_position_embeddings
__lowerCamelCase = auxiliary_loss
# Hungarian matcher
__lowerCamelCase = class_cost
__lowerCamelCase = bbox_cost
__lowerCamelCase = giou_cost
# Loss coefficients
__lowerCamelCase = bbox_loss_coefficient
__lowerCamelCase = giou_loss_coefficient
__lowerCamelCase = eos_coefficient
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = version.parse("""1.11""" )
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return OrderedDict(
[
('''pixel_values''', {0: '''batch''', 1: '''num_channels''', 2: '''height''', 3: '''width'''}),
] )
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return 1E-4
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return 12
| 330 |
import argparse
import fairseq
import torch
from transformers import UniSpeechSatConfig, UniSpeechSatForCTC, UniSpeechSatForPreTraining, logging
logging.set_verbosity_info()
a_ = logging.get_logger(__name__)
a_ = {
"""post_extract_proj""": """feature_projection.projection""",
"""encoder.pos_conv.0""": """encoder.pos_conv_embed.conv""",
"""self_attn.k_proj""": """encoder.layers.*.attention.k_proj""",
"""self_attn.v_proj""": """encoder.layers.*.attention.v_proj""",
"""self_attn.q_proj""": """encoder.layers.*.attention.q_proj""",
"""self_attn.out_proj""": """encoder.layers.*.attention.out_proj""",
"""self_attn_layer_norm""": """encoder.layers.*.layer_norm""",
"""fc1""": """encoder.layers.*.feed_forward.intermediate_dense""",
"""fc2""": """encoder.layers.*.feed_forward.output_dense""",
"""final_layer_norm""": """encoder.layers.*.final_layer_norm""",
"""encoder.layer_norm""": """encoder.layer_norm""",
"""encoder.layer_norm_for_extract""": """layer_norm_for_extract""",
"""w2v_model.layer_norm""": """feature_projection.layer_norm""",
"""quantizer.weight_proj""": """quantizer.weight_proj""",
"""quantizer.vars""": """quantizer.codevectors""",
"""project_q""": """project_q""",
"""final_proj""": """project_hid""",
"""w2v_encoder.proj""": """lm_head""",
"""label_embs_concat""": """label_embeddings_concat""",
"""mask_emb""": """masked_spec_embed""",
"""spk_proj""": """speaker_proj""",
}
a_ = [
"""lm_head""",
"""quantizer.weight_proj""",
"""quantizer.codevectors""",
"""project_q""",
"""project_hid""",
"""label_embeddings_concat""",
"""speaker_proj""",
"""layer_norm_for_extract""",
]
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : Tuple ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Dict ):
for attribute in key.split('''.''' ):
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if weight_type is not None:
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase ).shape
else:
__lowerCamelCase = hf_pointer.shape
if hf_shape != value.shape:
raise ValueError(
F"""Shape of hf {key + "." + weight_type if weight_type is not None else ""} is {hf_shape}, but should be"""
F""" {value.shape} for {full_name}""" )
if weight_type == "weight":
__lowerCamelCase = value
elif weight_type == "weight_g":
__lowerCamelCase = value
elif weight_type == "weight_v":
__lowerCamelCase = value
elif weight_type == "bias":
__lowerCamelCase = value
else:
__lowerCamelCase = value
logger.info(F"""{key + "." + weight_type if weight_type is not None else ""} was initialized from {full_name}.""" )
def a__ ( _UpperCamelCase : Any ,_UpperCamelCase : Any ):
__lowerCamelCase = []
__lowerCamelCase = fairseq_model.state_dict()
__lowerCamelCase = hf_model.unispeech_sat.feature_extractor
for name, value in fairseq_dict.items():
__lowerCamelCase = False
if "conv_layers" in name:
load_conv_layer(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,hf_model.config.feat_extract_norm == '''group''' ,)
__lowerCamelCase = True
else:
for key, mapped_key in MAPPING.items():
__lowerCamelCase = '''unispeech_sat.''' + mapped_key if mapped_key not in TOP_LEVEL_KEYS else mapped_key
if key in name or key.split('''w2v_model.''' )[-1] == name.split('''.''' )[0]:
if "layer_norm_for_extract" in name and (".".join(name.split('''.''' )[:-1] ) != key):
# special case since naming is very similar
continue
__lowerCamelCase = True
if "*" in mapped_key:
__lowerCamelCase = name.split(_UpperCamelCase )[0].split('''.''' )[-2]
__lowerCamelCase = mapped_key.replace('''*''' ,_UpperCamelCase )
if "weight_g" in name:
__lowerCamelCase = '''weight_g'''
elif "weight_v" in name:
__lowerCamelCase = '''weight_v'''
elif "bias" in name:
__lowerCamelCase = '''bias'''
elif "weight" in name:
# TODO: don't match quantizer.weight_proj
__lowerCamelCase = '''weight'''
else:
__lowerCamelCase = None
set_recursively(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
continue
if not is_used:
unused_weights.append(_UpperCamelCase )
logger.warning(F"""Unused weights: {unused_weights}""" )
def a__ ( _UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Dict ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = full_name.split('''conv_layers.''' )[-1]
__lowerCamelCase = name.split('''.''' )
__lowerCamelCase = int(items[0] )
__lowerCamelCase = int(items[1] )
if type_id == 0:
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif (type_id == 2 and not use_group_norm) or (type_id == 2 and layer_id == 0 and use_group_norm):
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(_UpperCamelCase )
@torch.no_grad()
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : List[str] ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : List[Any]=True ):
if config_path is not None:
__lowerCamelCase = UniSpeechSatConfig.from_pretrained(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatConfig()
__lowerCamelCase = ''''''
if is_finetuned:
__lowerCamelCase = UniSpeechSatForCTC(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatForPreTraining(_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
__lowerCamelCase = model[0].eval()
recursively_load_weights(_UpperCamelCase ,_UpperCamelCase )
hf_wavavec.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument("""--checkpoint_path""", default=None, type=str, help="""Path to fairseq checkpoint""")
parser.add_argument("""--dict_path""", default=None, type=str, help="""Path to dict of fine-tuned model""")
parser.add_argument("""--config_path""", default=None, type=str, help="""Path to hf config.json of model to convert""")
parser.add_argument(
"""--not_finetuned""", action="""store_true""", help="""Whether the model to convert is a fine-tuned model or not"""
)
a_ = parser.parse_args()
convert_unispeech_sat_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 330 | 1 |
import pytest
import datasets.config
from datasets.utils.info_utils import is_small_dataset
@pytest.mark.parametrize('''dataset_size''' ,[None, 4_00 * 2**20, 6_00 * 2**20] )
@pytest.mark.parametrize('''input_in_memory_max_size''' ,['''default''', 0, 1_00 * 2**20, 9_00 * 2**20] )
def a__ ( _UpperCamelCase : List[str] ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[int] ):
if input_in_memory_max_size != "default":
monkeypatch.setattr(datasets.config ,'''IN_MEMORY_MAX_SIZE''' ,_UpperCamelCase )
__lowerCamelCase = datasets.config.IN_MEMORY_MAX_SIZE
if input_in_memory_max_size == "default":
assert in_memory_max_size == 0
else:
assert in_memory_max_size == input_in_memory_max_size
if dataset_size and in_memory_max_size:
__lowerCamelCase = dataset_size < in_memory_max_size
else:
__lowerCamelCase = False
__lowerCamelCase = is_small_dataset(_UpperCamelCase )
assert result == expected
| 330 |
from typing import List, Union
import numpy as np
from ..utils import add_end_docstrings, is_torch_available, is_vision_available, logging, requires_backends
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_torch_available():
import torch
from ..models.auto.modeling_auto import MODEL_FOR_DEPTH_ESTIMATION_MAPPING
a_ = logging.get_logger(__name__)
@add_end_docstrings(lowerCAmelCase__ )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
requires_backends(self , '''vision''' )
self.check_model_type(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return super().__call__(__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
return {}, {}, {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = load_image(__UpperCAmelCase )
__lowerCamelCase = image.size
__lowerCamelCase = self.image_processor(images=__UpperCAmelCase , return_tensors=self.framework )
return model_inputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.model(**__UpperCAmelCase )
return model_outputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = model_outputs.predicted_depth
__lowerCamelCase = torch.nn.functional.interpolate(
predicted_depth.unsqueeze(1 ) , size=self.image_size[::-1] , mode='''bicubic''' , align_corners=__UpperCAmelCase )
__lowerCamelCase = prediction.squeeze().cpu().numpy()
__lowerCamelCase = (output * 255 / np.max(__UpperCAmelCase )).astype('''uint8''' )
__lowerCamelCase = Image.fromarray(__UpperCAmelCase )
__lowerCamelCase = {}
__lowerCamelCase = predicted_depth
__lowerCamelCase = depth
return output_dict
| 330 | 1 |
import pytest
from datasets import inspect_metric, list_metrics, load_metric
@pytest.fixture
def a__ ( _UpperCamelCase : List[str] ):
monkeypatch.setattr('''datasets.utils.deprecation_utils._emitted_deprecation_warnings''' ,set() )
@pytest.fixture
def a__ ( _UpperCamelCase : Tuple ):
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = metric_id
class __lowerCAmelCase :
lowerCAmelCase__ = [MetricMock(lowerCAmelCase__ ) for metric_id in ["""accuracy""", """mse""", """precision""", """codeparrot/apps_metric"""]]
def lowerCamelCase ( self ):
'''simple docstring'''
return self._metrics
monkeypatch.setattr('''datasets.inspect.huggingface_hub''' ,HfhMock() )
@pytest.mark.parametrize(
'''func, args''' ,[(load_metric, ('''metrics/mse''',)), (list_metrics, ()), (inspect_metric, ('''metrics/mse''', '''tmp_path'''))] )
def a__ ( _UpperCamelCase : int ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : List[str] ,_UpperCamelCase : str ,_UpperCamelCase : List[str] ):
if "tmp_path" in args:
__lowerCamelCase = tuple(arg if arg != '''tmp_path''' else tmp_path for arg in args )
with pytest.warns(_UpperCamelCase ,match='''https://huggingface.co/docs/evaluate''' ):
func(*_UpperCamelCase )
| 330 |
from typing import Dict, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
convert_to_rgb,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
OPENAI_CLIP_MEAN,
OPENAI_CLIP_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_vision_available, logging
a_ = logging.get_logger(__name__)
if is_vision_available():
import PIL
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""pixel_values"""]
def __init__( self , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = True , __UpperCAmelCase = 1 / 255 , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = True , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = size if size is not None else {'''shortest_edge''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
__lowerCamelCase = crop_size if crop_size is not None else {'''height''': 224, '''width''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase , 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 OPENAI_CLIP_MEAN
__lowerCamelCase = image_std if image_std is not None else OPENAI_CLIP_STD
__lowerCamelCase = do_convert_rgb
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
if "shortest_edge" not in size:
raise ValueError(F"""The `size` parameter must contain the key `shortest_edge`. Got {size.keys()}""" )
__lowerCamelCase = get_resize_output_image_size(__UpperCAmelCase , size=size['''shortest_edge'''] , default_to_square=__UpperCAmelCase )
return resize(__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase )
if "height" not in size or "width" not in size:
raise ValueError(F"""The `size` parameter must contain the keys (height, width). Got {size.keys()}""" )
return center_crop(__UpperCAmelCase , size=(size['''height'''], size['''width''']) , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return rescale(__UpperCAmelCase , scale=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return normalize(__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = ChannelDimension.FIRST , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = do_resize if do_resize is not None else self.do_resize
__lowerCamelCase = size if size is not None else self.size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''size''' , default_to_square=__UpperCAmelCase )
__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 = crop_size if crop_size is not None else self.crop_size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''crop_size''' , default_to_square=__UpperCAmelCase )
__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 = do_convert_rgb if do_convert_rgb is not None else self.do_convert_rgb
__lowerCamelCase = make_list_of_images(__UpperCAmelCase )
if not valid_images(__UpperCAmelCase ):
raise ValueError(
'''Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '''
'''torch.Tensor, tf.Tensor or jax.ndarray.''' )
if do_resize and size is None:
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.''' )
# PIL RGBA images are converted to RGB
if do_convert_rgb:
__lowerCamelCase = [convert_to_rgb(__UpperCAmelCase ) for image in images]
# All transformations expect numpy arrays.
__lowerCamelCase = [to_numpy_array(__UpperCAmelCase ) for image in images]
if do_resize:
__lowerCamelCase = [self.resize(image=__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase ) for image in images]
if do_center_crop:
__lowerCamelCase = [self.center_crop(image=__UpperCAmelCase , size=__UpperCAmelCase ) for image in images]
if do_rescale:
__lowerCamelCase = [self.rescale(image=__UpperCAmelCase , scale=__UpperCAmelCase ) for image in images]
if do_normalize:
__lowerCamelCase = [self.normalize(image=__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase ) for image in images]
__lowerCamelCase = [to_channel_dimension_format(__UpperCAmelCase , __UpperCAmelCase ) for image in images]
__lowerCamelCase = {'''pixel_values''': images}
return BatchFeature(data=__UpperCAmelCase , tensor_type=__UpperCAmelCase )
| 330 | 1 |
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = n ** (1 / 3)
return (val * val * val) == n
if __name__ == "__main__":
print(perfect_cube(27))
print(perfect_cube(4))
| 330 |
from __future__ import annotations
from typing import Generic, TypeVar
a_ = TypeVar("""T""")
class __lowerCAmelCase ( Generic[T] ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = data
__lowerCamelCase = self
__lowerCamelCase = 0
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# map from node name to the node object
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# create a new set with x as its member
__lowerCamelCase = DisjointSetTreeNode(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# find the set x belongs to (with path-compression)
__lowerCamelCase = self.map[data]
if elem_ref != elem_ref.parent:
__lowerCamelCase = self.find_set(elem_ref.parent.data )
return elem_ref.parent
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# helper function for union operation
if nodea.rank > nodea.rank:
__lowerCamelCase = nodea
else:
__lowerCamelCase = nodea
if nodea.rank == nodea.rank:
nodea.rank += 1
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# merge 2 disjoint sets
self.link(self.find_set(__UpperCAmelCase ) , self.find_set(__UpperCAmelCase ) )
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# connections: map from the node to the neighbouring nodes (with weights)
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# add a node ONLY if its not present in the graph
if node not in self.connections:
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# add an edge with the given weight
self.add_node(__UpperCAmelCase )
self.add_node(__UpperCAmelCase )
__lowerCamelCase = weight
__lowerCamelCase = weight
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = []
__lowerCamelCase = set()
for start in self.connections:
for end in self.connections[start]:
if (start, end) not in seen:
seen.add((end, start) )
edges.append((start, end, self.connections[start][end]) )
edges.sort(key=lambda __UpperCAmelCase : x[2] )
# creating the disjoint set
__lowerCamelCase = DisjointSetTree[T]()
for node in self.connections:
disjoint_set.make_set(__UpperCAmelCase )
# MST generation
__lowerCamelCase = 0
__lowerCamelCase = 0
__lowerCamelCase = GraphUndirectedWeighted[T]()
while num_edges < len(self.connections ) - 1:
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = edges[index]
index += 1
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
if parent_u != parent_v:
num_edges += 1
graph.add_edge(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
disjoint_set.union(__UpperCAmelCase , __UpperCAmelCase )
return graph
| 330 | 1 |
import functools
import operator
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a_ = logging.get_logger(__name__)
a_ = {
"""microsoft/unispeech-sat-base-100h-libri-ft""": (
"""https://huggingface.co/microsoft/unispeech-sat-base-100h-libri-ft/resolve/main/config.json"""
),
# See all UniSpeechSat models at https://huggingface.co/models?filter=unispeech_sat
}
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """unispeech-sat"""
def __init__( self , __UpperCAmelCase=32 , __UpperCAmelCase=768 , __UpperCAmelCase=12 , __UpperCAmelCase=12 , __UpperCAmelCase=3072 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.02 , __UpperCAmelCase=1E-5 , __UpperCAmelCase="group" , __UpperCAmelCase="gelu" , __UpperCAmelCase=(512, 512, 512, 512, 512, 512, 512) , __UpperCAmelCase=(5, 2, 2, 2, 2, 2, 2) , __UpperCAmelCase=(10, 3, 3, 3, 3, 2, 2) , __UpperCAmelCase=False , __UpperCAmelCase=128 , __UpperCAmelCase=16 , __UpperCAmelCase=False , __UpperCAmelCase=True , __UpperCAmelCase=0.05 , __UpperCAmelCase=10 , __UpperCAmelCase=2 , __UpperCAmelCase=0.0 , __UpperCAmelCase=10 , __UpperCAmelCase=0 , __UpperCAmelCase=320 , __UpperCAmelCase=2 , __UpperCAmelCase=0.1 , __UpperCAmelCase=100 , __UpperCAmelCase=256 , __UpperCAmelCase=256 , __UpperCAmelCase=0.1 , __UpperCAmelCase="mean" , __UpperCAmelCase=False , __UpperCAmelCase=False , __UpperCAmelCase=256 , __UpperCAmelCase=(512, 512, 512, 512, 1500) , __UpperCAmelCase=(5, 3, 3, 1, 1) , __UpperCAmelCase=(1, 2, 3, 1, 1) , __UpperCAmelCase=512 , __UpperCAmelCase=0 , __UpperCAmelCase=1 , __UpperCAmelCase=2 , __UpperCAmelCase=504 , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase , pad_token_id=__UpperCAmelCase , bos_token_id=__UpperCAmelCase , eos_token_id=__UpperCAmelCase )
__lowerCamelCase = hidden_size
__lowerCamelCase = feat_extract_norm
__lowerCamelCase = feat_extract_activation
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = conv_bias
__lowerCamelCase = num_conv_pos_embeddings
__lowerCamelCase = num_conv_pos_embedding_groups
__lowerCamelCase = len(self.conv_dim )
__lowerCamelCase = num_hidden_layers
__lowerCamelCase = intermediate_size
__lowerCamelCase = hidden_act
__lowerCamelCase = num_attention_heads
__lowerCamelCase = hidden_dropout
__lowerCamelCase = attention_dropout
__lowerCamelCase = activation_dropout
__lowerCamelCase = feat_proj_dropout
__lowerCamelCase = final_dropout
__lowerCamelCase = layerdrop
__lowerCamelCase = layer_norm_eps
__lowerCamelCase = initializer_range
__lowerCamelCase = vocab_size
__lowerCamelCase = num_clusters
__lowerCamelCase = do_stable_layer_norm
__lowerCamelCase = use_weighted_layer_sum
if (
(len(self.conv_stride ) != self.num_feat_extract_layers)
or (len(self.conv_kernel ) != self.num_feat_extract_layers)
or (len(self.conv_dim ) != self.num_feat_extract_layers)
):
raise ValueError(
'''Configuration for convolutional layers is incorrect. It is required that `len(config.conv_dim)` =='''
''' `len(config.conv_stride)` == `len(config.conv_kernel)`, but is `len(config.conv_dim) ='''
F""" {len(self.conv_dim )}`, `len(config.conv_stride) = {len(self.conv_stride )}`,"""
F""" `len(config.conv_kernel) = {len(self.conv_kernel )}`.""" )
# fine-tuning config parameters for SpecAugment: https://arxiv.org/abs/1904.08779
__lowerCamelCase = apply_spec_augment
__lowerCamelCase = mask_time_prob
__lowerCamelCase = mask_time_length
__lowerCamelCase = mask_time_min_masks
__lowerCamelCase = mask_feature_prob
__lowerCamelCase = mask_feature_length
__lowerCamelCase = mask_feature_min_masks
# parameters for pretraining with codevector quantized representations
__lowerCamelCase = num_codevectors_per_group
__lowerCamelCase = num_codevector_groups
__lowerCamelCase = contrastive_logits_temperature
__lowerCamelCase = feat_quantizer_dropout
__lowerCamelCase = num_negatives
__lowerCamelCase = codevector_dim
__lowerCamelCase = proj_codevector_dim
__lowerCamelCase = diversity_loss_weight
# ctc loss
__lowerCamelCase = ctc_loss_reduction
__lowerCamelCase = ctc_zero_infinity
# SequenceClassification-specific parameter. Feel free to ignore for other classes.
__lowerCamelCase = classifier_proj_size
# XVector-specific parameters. Feel free to ignore for other classes.
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = xvector_output_dim
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return functools.reduce(operator.mul , self.conv_stride , 1 )
| 330 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=2 , __UpperCAmelCase=8 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=99 , __UpperCAmelCase=16 , __UpperCAmelCase=5 , __UpperCAmelCase=2 , __UpperCAmelCase=36 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.0 , __UpperCAmelCase=512 , __UpperCAmelCase=16 , __UpperCAmelCase=2 , __UpperCAmelCase=0.02 , __UpperCAmelCase=3 , __UpperCAmelCase=4 , __UpperCAmelCase=None , ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
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=__UpperCAmelCase , initializer_range=self.initializer_range , )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_config()
__lowerCamelCase = 300
return config
def lowerCamelCase ( self ):
'''simple docstring'''
(
(
__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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = True
__lowerCamelCase = MraModel(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , )
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , )
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForQuestionAnswering(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , start_positions=__UpperCAmelCase , end_positions=__UpperCAmelCase , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForSequenceClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForTokenClassification(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_choices
__lowerCamelCase = MraForMultipleChoice(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
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(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowerCamelCase ( self ):
'''simple docstring'''
__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 __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = ()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModelTester(self )
__lowerCamelCase = ConfigTester(self , config_class=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__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(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = MraModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
@unittest.skip(reason='''MRA does not output attentions''' )
def lowerCamelCase ( self ):
'''simple docstring'''
return
@require_torch
class __lowerCAmelCase ( unittest.TestCase ):
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = torch.Size((1, 256, 768) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[-0.0_140, 0.0_830, -0.0_381], [0.1_546, 0.1_402, 0.0_220], [0.1_162, 0.0_851, 0.0_165]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 256, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[9.2_595, -3.6_038, 11.8_819], [9.3_869, -3.2_693, 11.0_956], [11.8_524, -3.4_938, 13.1_210]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' )
__lowerCamelCase = torch.arange(4096 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 4096, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[5.4_789, -2.3_564, 7.5_064], [7.9_067, -1.3_369, 9.9_668], [9.0_712, -1.8_106, 7.0_380]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
import copy
import os
import cva
import numpy as np
from matplotlib import pyplot as plt
class __lowerCAmelCase :
def __init__( self ):
'''simple docstring'''
__lowerCamelCase = ''''''
__lowerCamelCase = ''''''
__lowerCamelCase = []
__lowerCamelCase = 0
__lowerCamelCase = 256
__lowerCamelCase = 0
__lowerCamelCase = 0
__lowerCamelCase = 0
__lowerCamelCase = 0
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = cva.imread(__UpperCAmelCase , 0 )
__lowerCamelCase = copy.deepcopy(self.img )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = plt.hist(self.img.ravel() , 256 , [0, 256] , label='''x''' )
__lowerCamelCase = np.sum(__UpperCAmelCase )
for i in range(len(__UpperCAmelCase ) ):
__lowerCamelCase = x[i] / self.k
self.sk += prk
__lowerCamelCase = (self.L - 1) * self.sk
if self.rem != 0:
__lowerCamelCase = int(last % last )
__lowerCamelCase = int(last + 1 if self.rem >= 0.5 else last )
self.last_list.append(__UpperCAmelCase )
__lowerCamelCase = int(np.ma.count(self.img ) / self.img[1].size )
__lowerCamelCase = self.img[1].size
for i in range(self.number_of_cols ):
for j in range(self.number_of_rows ):
__lowerCamelCase = self.img[j][i]
if num != self.last_list[num]:
__lowerCamelCase = self.last_list[num]
cva.imwrite('''output_data/output.jpg''' , self.img )
def lowerCamelCase ( self ):
'''simple docstring'''
plt.hist(self.img.ravel() , 256 , [0, 256] )
def lowerCamelCase ( self ):
'''simple docstring'''
cva.imshow('''Output-Image''' , self.img )
cva.imshow('''Input-Image''' , self.original_image )
cva.waitKey(5000 )
cva.destroyAllWindows()
if __name__ == "__main__":
a_ = os.path.join(os.path.basename(__file__), """image_data/input.jpg""")
a_ = ConstantStretch()
stretcher.stretch(file_path)
stretcher.plot_histogram()
stretcher.show_image()
| 330 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_torch_available,
)
a_ = {"""configuration_encoder_decoder""": ["""EncoderDecoderConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""EncoderDecoderModel"""]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TFEncoderDecoderModel"""]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""FlaxEncoderDecoderModel"""]
if TYPE_CHECKING:
from .configuration_encoder_decoder import EncoderDecoderConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_encoder_decoder import EncoderDecoderModel
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_encoder_decoder import TFEncoderDecoderModel
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_encoder_decoder import FlaxEncoderDecoderModel
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
from typing import List, Union
import numpy as np
from ..utils import add_end_docstrings, is_torch_available, is_vision_available, logging, requires_backends
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_torch_available():
import torch
from ..models.auto.modeling_auto import MODEL_FOR_DEPTH_ESTIMATION_MAPPING
a_ = logging.get_logger(__name__)
@add_end_docstrings(lowerCAmelCase__ )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
requires_backends(self , '''vision''' )
self.check_model_type(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return super().__call__(__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
return {}, {}, {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = load_image(__UpperCAmelCase )
__lowerCamelCase = image.size
__lowerCamelCase = self.image_processor(images=__UpperCAmelCase , return_tensors=self.framework )
return model_inputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.model(**__UpperCAmelCase )
return model_outputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = model_outputs.predicted_depth
__lowerCamelCase = torch.nn.functional.interpolate(
predicted_depth.unsqueeze(1 ) , size=self.image_size[::-1] , mode='''bicubic''' , align_corners=__UpperCAmelCase )
__lowerCamelCase = prediction.squeeze().cpu().numpy()
__lowerCamelCase = (output * 255 / np.max(__UpperCAmelCase )).astype('''uint8''' )
__lowerCamelCase = Image.fromarray(__UpperCAmelCase )
__lowerCamelCase = {}
__lowerCamelCase = predicted_depth
__lowerCamelCase = depth
return output_dict
| 330 |
from string import ascii_lowercase, ascii_uppercase
def a__ ( _UpperCamelCase : str ):
if not sentence:
return ""
__lowerCamelCase = dict(zip(_UpperCamelCase ,_UpperCamelCase ) )
return lower_to_upper.get(sentence[0] ,sentence[0] ) + sentence[1:]
if __name__ == "__main__":
from doctest import testmod
testmod()
| 330 | 1 |
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = None
__lowerCamelCase = None
__lowerCamelCase = graph
self._normalize_graph(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = len(__UpperCAmelCase )
__lowerCamelCase = None
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
if sources is int:
__lowerCamelCase = [sources]
if sinks is int:
__lowerCamelCase = [sinks]
if len(__UpperCAmelCase ) == 0 or len(__UpperCAmelCase ) == 0:
return
__lowerCamelCase = sources[0]
__lowerCamelCase = sinks[0]
# make fake vertex if there are more
# than one source or sink
if len(__UpperCAmelCase ) > 1 or len(__UpperCAmelCase ) > 1:
__lowerCamelCase = 0
for i in sources:
max_input_flow += sum(self.graph[i] )
__lowerCamelCase = len(self.graph ) + 1
for room in self.graph:
room.insert(0 , 0 )
self.graph.insert(0 , [0] * size )
for i in sources:
__lowerCamelCase = max_input_flow
__lowerCamelCase = 0
__lowerCamelCase = len(self.graph ) + 1
for room in self.graph:
room.append(0 )
self.graph.append([0] * size )
for i in sinks:
__lowerCamelCase = max_input_flow
__lowerCamelCase = size - 1
def lowerCamelCase ( self ):
'''simple docstring'''
if self.maximum_flow_algorithm is None:
raise Exception('''You need to set maximum flow algorithm before.''' )
if self.source_index is None or self.sink_index is None:
return 0
self.maximum_flow_algorithm.execute()
return self.maximum_flow_algorithm.getMaximumFlow()
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = algorithm(self )
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = flow_network
__lowerCamelCase = flow_network.verticesCount
__lowerCamelCase = flow_network.sourceIndex
__lowerCamelCase = flow_network.sinkIndex
# it's just a reference, so you shouldn't change
# it in your algorithms, use deep copy before doing that
__lowerCamelCase = flow_network.graph
__lowerCamelCase = False
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.executed:
self._algorithm()
__lowerCamelCase = True
def lowerCamelCase ( self ):
'''simple docstring'''
pass
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
super().__init__(__UpperCAmelCase )
# use this to save your result
__lowerCamelCase = -1
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.executed:
raise Exception('''You should execute algorithm before using its result!''' )
return self.maximum_flow
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
super().__init__(__UpperCAmelCase )
__lowerCamelCase = [[0] * self.verticies_count for i in range(self.verticies_count )]
__lowerCamelCase = [0] * self.verticies_count
__lowerCamelCase = [0] * self.verticies_count
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.verticies_count
# push some substance to graph
for nextvertex_index, bandwidth in enumerate(self.graph[self.source_index] ):
self.preflow[self.source_index][nextvertex_index] += bandwidth
self.preflow[nextvertex_index][self.source_index] -= bandwidth
self.excesses[nextvertex_index] += bandwidth
# Relabel-to-front selection rule
__lowerCamelCase = [
i
for i in range(self.verticies_count )
if i != self.source_index and i != self.sink_index
]
# move through list
__lowerCamelCase = 0
while i < len(__UpperCAmelCase ):
__lowerCamelCase = vertices_list[i]
__lowerCamelCase = self.heights[vertex_index]
self.process_vertex(__UpperCAmelCase )
if self.heights[vertex_index] > previous_height:
# if it was relabeled, swap elements
# and start from 0 index
vertices_list.insert(0 , vertices_list.pop(__UpperCAmelCase ) )
__lowerCamelCase = 0
else:
i += 1
__lowerCamelCase = sum(self.preflow[self.source_index] )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
while self.excesses[vertex_index] > 0:
for neighbour_index in range(self.verticies_count ):
# if it's neighbour and current vertex is higher
if (
self.graph[vertex_index][neighbour_index]
- self.preflow[vertex_index][neighbour_index]
> 0
and self.heights[vertex_index] > self.heights[neighbour_index]
):
self.push(__UpperCAmelCase , __UpperCAmelCase )
self.relabel(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(
self.excesses[from_index] , self.graph[from_index][to_index] - self.preflow[from_index][to_index] , )
self.preflow[from_index][to_index] += preflow_delta
self.preflow[to_index][from_index] -= preflow_delta
self.excesses[from_index] -= preflow_delta
self.excesses[to_index] += preflow_delta
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = None
for to_index in range(self.verticies_count ):
if (
self.graph[vertex_index][to_index]
- self.preflow[vertex_index][to_index]
> 0
) and (min_height is None or self.heights[to_index] < min_height):
__lowerCamelCase = self.heights[to_index]
if min_height is not None:
__lowerCamelCase = min_height + 1
if __name__ == "__main__":
a_ = [0]
a_ = [3]
# graph = [
# [0, 0, 4, 6, 0, 0],
# [0, 0, 5, 2, 0, 0],
# [0, 0, 0, 0, 4, 4],
# [0, 0, 0, 0, 6, 6],
# [0, 0, 0, 0, 0, 0],
# [0, 0, 0, 0, 0, 0],
# ]
a_ = [[0, 7, 0, 0], [0, 0, 6, 0], [0, 0, 0, 8], [9, 0, 0, 0]]
# prepare our network
a_ = FlowNetwork(graph, entrances, exits)
# set algorithm
flow_network.set_maximum_flow_algorithm(PushRelabelExecutor)
# and calculate
a_ = flow_network.find_maximum_flow()
print(f"maximum flow is {maximum_flow}")
| 330 |
from transformers import BertTokenizer, EncoderDecoderModel, SeqaSeqTrainer, SeqaSeqTrainingArguments
from transformers.testing_utils import TestCasePlus, require_torch, slow
from transformers.utils import is_datasets_available
if is_datasets_available():
import datasets
class __lowerCAmelCase ( lowerCAmelCase__ ):
@slow
@require_torch
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = EncoderDecoderModel.from_encoder_decoder_pretrained('''prajjwal1/bert-tiny''' , '''prajjwal1/bert-tiny''' )
__lowerCamelCase = BertTokenizer.from_pretrained('''bert-base-uncased''' )
__lowerCamelCase = bertabert.config.encoder.vocab_size
__lowerCamelCase = tokenizer.sep_token_id
__lowerCamelCase = tokenizer.cls_token_id
__lowerCamelCase = 128
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''train[:1%]''' )
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''validation[:1%]''' )
__lowerCamelCase = train_dataset.select(range(32 ) )
__lowerCamelCase = val_dataset.select(range(16 ) )
__lowerCamelCase = 4
def _map_to_encoder_decoder_inputs(__UpperCAmelCase ):
# Tokenizer will automatically set [BOS] <text> [EOS]
__lowerCamelCase = tokenizer(batch['''article'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=512 )
__lowerCamelCase = tokenizer(batch['''highlights'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=128 )
__lowerCamelCase = inputs.input_ids
__lowerCamelCase = inputs.attention_mask
__lowerCamelCase = outputs.input_ids
__lowerCamelCase = outputs.input_ids.copy()
__lowerCamelCase = [
[-100 if token == tokenizer.pad_token_id else token for token in labels] for labels in batch['''labels''']
]
__lowerCamelCase = outputs.attention_mask
assert all(len(__UpperCAmelCase ) == 512 for x in inputs.input_ids )
assert all(len(__UpperCAmelCase ) == 128 for x in outputs.input_ids )
return batch
def _compute_metrics(__UpperCAmelCase ):
__lowerCamelCase = pred.label_ids
__lowerCamelCase = pred.predictions
# all unnecessary tokens are removed
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = sum([int(pred_str[i] == label_str[i] ) for i in range(len(__UpperCAmelCase ) )] ) / len(__UpperCAmelCase )
return {"accuracy": accuracy}
# map train dataset
__lowerCamelCase = train_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
train_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
# same for validation dataset
__lowerCamelCase = val_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
val_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
__lowerCamelCase = self.get_auto_remove_tmp_dir()
__lowerCamelCase = SeqaSeqTrainingArguments(
output_dir=__UpperCAmelCase , per_device_train_batch_size=__UpperCAmelCase , per_device_eval_batch_size=__UpperCAmelCase , predict_with_generate=__UpperCAmelCase , evaluation_strategy='''steps''' , do_train=__UpperCAmelCase , do_eval=__UpperCAmelCase , warmup_steps=0 , eval_steps=2 , logging_steps=2 , )
# instantiate trainer
__lowerCamelCase = SeqaSeqTrainer(
model=__UpperCAmelCase , args=__UpperCAmelCase , compute_metrics=_compute_metrics , train_dataset=__UpperCAmelCase , eval_dataset=__UpperCAmelCase , tokenizer=__UpperCAmelCase , )
# start training
trainer.train()
| 330 | 1 |
import os
import sys
import unittest
a_ = os.path.abspath(os.path.dirname(os.path.dirname(os.path.dirname(__file__))))
sys.path.append(os.path.join(git_repo_path, """utils"""))
import check_dummies # noqa: E402
from check_dummies import create_dummy_files, create_dummy_object, find_backend, read_init # noqa: E402
# Align TRANSFORMERS_PATH in check_dummies with the current path
a_ = os.path.join(git_repo_path, """src""", """transformers""")
a_ = """
{0} = None
"""
a_ = """
class {0}(metaclass=DummyObject):
_backends = {1}
def __init__(self, *args, **kwargs):
requires_backends(self, {1})
"""
a_ = """
def {0}(*args, **kwargs):
requires_backends({0}, {1})
"""
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = find_backend(''' _import_structure["models.albert"].append("AlbertTokenizerFast")''' )
self.assertIsNone(__UpperCAmelCase )
__lowerCamelCase = find_backend(''' if not is_tokenizers_available():''' )
self.assertEqual(__UpperCAmelCase , '''tokenizers''' )
__lowerCamelCase = find_backend(''' if not is_tensorflow_text_available():''' )
self.assertEqual(__UpperCAmelCase , '''tensorflow_text''' )
__lowerCamelCase = find_backend(''' if not (is_sentencepiece_available() and is_tokenizers_available()):''' )
self.assertEqual(__UpperCAmelCase , '''sentencepiece_and_tokenizers''' )
__lowerCamelCase = find_backend(
''' if not (is_sentencepiece_available() and is_tensorflow_text_available()):''' )
self.assertEqual(__UpperCAmelCase , '''sentencepiece_and_tensorflow_text''' )
__lowerCamelCase = find_backend(
''' if not (is_sentencepiece_available() and is_tokenizers_available() and is_vision_available()):''' )
self.assertEqual(__UpperCAmelCase , '''sentencepiece_and_tokenizers_and_vision''' )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = read_init()
# We don't assert on the exact list of keys to allow for smooth grow of backend-specific objects
self.assertIn('''torch''' , __UpperCAmelCase )
self.assertIn('''tensorflow_text''' , __UpperCAmelCase )
self.assertIn('''sentencepiece_and_tokenizers''' , __UpperCAmelCase )
# Likewise, we can't assert on the exact content of a key
self.assertIn('''BertModel''' , objects['''torch'''] )
self.assertIn('''TFBertModel''' , objects['''tf'''] )
self.assertIn('''FlaxBertModel''' , objects['''flax'''] )
self.assertIn('''BertModel''' , objects['''torch'''] )
self.assertIn('''TFBertTokenizer''' , objects['''tensorflow_text'''] )
self.assertIn('''convert_slow_tokenizer''' , objects['''sentencepiece_and_tokenizers'''] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = create_dummy_object('''CONSTANT''' , '''\'torch\'''' )
self.assertEqual(__UpperCAmelCase , '''\nCONSTANT = None\n''' )
__lowerCamelCase = create_dummy_object('''function''' , '''\'torch\'''' )
self.assertEqual(
__UpperCAmelCase , '''\ndef function(*args, **kwargs):\n requires_backends(function, \'torch\')\n''' )
__lowerCamelCase = '''
class FakeClass(metaclass=DummyObject):
_backends = \'torch\'
def __init__(self, *args, **kwargs):
requires_backends(self, \'torch\')
'''
__lowerCamelCase = create_dummy_object('''FakeClass''' , '''\'torch\'''' )
self.assertEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''# This file is autogenerated by the command `make fix-copies`, do not edit.
from ..utils import DummyObject, requires_backends
CONSTANT = None
def function(*args, **kwargs):
requires_backends(function, ["torch"])
class FakeClass(metaclass=DummyObject):
_backends = ["torch"]
def __init__(self, *args, **kwargs):
requires_backends(self, ["torch"])
'''
__lowerCamelCase = create_dummy_files({'''torch''': ['''CONSTANT''', '''function''', '''FakeClass''']} )
self.assertEqual(dummy_files['''torch'''] , __UpperCAmelCase )
| 330 |
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_timm_backbone""": ["""TimmBackboneConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TimmBackbone"""]
if TYPE_CHECKING:
from .configuration_timm_backbone import TimmBackboneConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_timm_backbone import TimmBackbone
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
import gc
import random
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTextModelWithProjection, CLIPTokenizer
from diffusers import (
AutoencoderKL,
DiffusionPipeline,
EulerDiscreteScheduler,
StableDiffusionXLImgaImgPipeline,
UNetaDConditionModel,
)
from diffusers.utils import floats_tensor, slow, torch_device
from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu
from ..pipeline_params import (
IMAGE_TO_IMAGE_IMAGE_PARAMS,
TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS,
TEXT_GUIDED_IMAGE_VARIATION_PARAMS,
)
from ..test_pipelines_common import PipelineLatentTesterMixin, PipelineTesterMixin
enable_full_determinism()
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = StableDiffusionXLImgaImgPipeline
lowerCAmelCase__ = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {"""height""", """width"""}
lowerCAmelCase__ = PipelineTesterMixin.required_optional_params - {"""latents"""}
lowerCAmelCase__ = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS
lowerCAmelCase__ = IMAGE_TO_IMAGE_IMAGE_PARAMS
lowerCAmelCase__ = IMAGE_TO_IMAGE_IMAGE_PARAMS
def lowerCamelCase ( self ):
'''simple docstring'''
torch.manual_seed(0 )
__lowerCamelCase = UNetaDConditionModel(
block_out_channels=(32, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=('''DownBlock2D''', '''CrossAttnDownBlock2D''') , up_block_types=('''CrossAttnUpBlock2D''', '''UpBlock2D''') , attention_head_dim=(2, 4) , use_linear_projection=__UpperCAmelCase , addition_embed_type='''text_time''' , addition_time_embed_dim=8 , transformer_layers_per_block=(1, 2) , projection_class_embeddings_input_dim=80 , cross_attention_dim=64 , )
__lowerCamelCase = EulerDiscreteScheduler(
beta_start=0.00_085 , beta_end=0.012 , steps_offset=1 , beta_schedule='''scaled_linear''' , timestep_spacing='''leading''' , )
torch.manual_seed(0 )
__lowerCamelCase = AutoencoderKL(
block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=['''DownEncoderBlock2D''', '''DownEncoderBlock2D'''] , up_block_types=['''UpDecoderBlock2D''', '''UpDecoderBlock2D'''] , latent_channels=4 , sample_size=128 , )
torch.manual_seed(0 )
__lowerCamelCase = CLIPTextConfig(
bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-0_5 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , hidden_act='''gelu''' , projection_dim=32 , )
__lowerCamelCase = CLIPTextModel(__UpperCAmelCase )
__lowerCamelCase = CLIPTokenizer.from_pretrained('''hf-internal-testing/tiny-random-clip''' , local_files_only=__UpperCAmelCase )
__lowerCamelCase = CLIPTextModelWithProjection(__UpperCAmelCase )
__lowerCamelCase = CLIPTokenizer.from_pretrained('''hf-internal-testing/tiny-random-clip''' , local_files_only=__UpperCAmelCase )
__lowerCamelCase = {
'''unet''': unet,
'''scheduler''': scheduler,
'''vae''': vae,
'''text_encoder''': text_encoder,
'''tokenizer''': tokenizer,
'''text_encoder_2''': text_encoder_a,
'''tokenizer_2''': tokenizer_a,
# "safety_checker": None,
# "feature_extractor": None,
}
return components
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=0 ):
'''simple docstring'''
__lowerCamelCase = floats_tensor((1, 3, 32, 32) , rng=random.Random(__UpperCAmelCase ) ).to(__UpperCAmelCase )
__lowerCamelCase = image / 2 + 0.5
if str(__UpperCAmelCase ).startswith('''mps''' ):
__lowerCamelCase = torch.manual_seed(__UpperCAmelCase )
else:
__lowerCamelCase = torch.Generator(device=__UpperCAmelCase ).manual_seed(__UpperCAmelCase )
__lowerCamelCase = {
'''prompt''': '''A painting of a squirrel eating a burger''',
'''image''': image,
'''generator''': generator,
'''num_inference_steps''': 2,
'''guidance_scale''': 5.0,
'''output_type''': '''numpy''',
'''strength''': 0.75,
}
return inputs
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''cpu''' # ensure determinism for the device-dependent torch.Generator
__lowerCamelCase = self.get_dummy_components()
__lowerCamelCase = StableDiffusionXLImgaImgPipeline(**__UpperCAmelCase )
__lowerCamelCase = sd_pipe.to(__UpperCAmelCase )
sd_pipe.set_progress_bar_config(disable=__UpperCAmelCase )
__lowerCamelCase = self.get_dummy_inputs(__UpperCAmelCase )
__lowerCamelCase = sd_pipe(**__UpperCAmelCase ).images
__lowerCamelCase = image[0, -3:, -3:, -1]
assert image.shape == (1, 32, 32, 3)
__lowerCamelCase = np.array([0.4_656, 0.4_840, 0.4_439, 0.6_698, 0.5_574, 0.4_524, 0.5_799, 0.5_943, 0.5_165] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
def lowerCamelCase ( self ):
'''simple docstring'''
super().test_attention_slicing_forward_pass(expected_max_diff=3E-3 )
def lowerCamelCase ( self ):
'''simple docstring'''
super().test_inference_batch_single_identical(expected_max_diff=3E-3 )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_dummy_components()
__lowerCamelCase = StableDiffusionXLImgaImgPipeline(**__UpperCAmelCase )
__lowerCamelCase = sd_pipe.to(__UpperCAmelCase )
__lowerCamelCase = sd_pipe.to(__UpperCAmelCase )
sd_pipe.set_progress_bar_config(disable=__UpperCAmelCase )
# forward without prompt embeds
__lowerCamelCase = self.get_dummy_inputs(__UpperCAmelCase )
__lowerCamelCase = 3 * ['''this is a negative prompt''']
__lowerCamelCase = negative_prompt
__lowerCamelCase = 3 * [inputs['''prompt''']]
__lowerCamelCase = sd_pipe(**__UpperCAmelCase )
__lowerCamelCase = output.images[0, -3:, -3:, -1]
# forward with prompt embeds
__lowerCamelCase = self.get_dummy_inputs(__UpperCAmelCase )
__lowerCamelCase = 3 * ['''this is a negative prompt''']
__lowerCamelCase = 3 * [inputs.pop('''prompt''' )]
(
(
__lowerCamelCase
) ,(
__lowerCamelCase
) ,(
__lowerCamelCase
) ,(
__lowerCamelCase
) ,
) = sd_pipe.encode_prompt(__UpperCAmelCase , negative_prompt=__UpperCAmelCase )
__lowerCamelCase = sd_pipe(
**__UpperCAmelCase , prompt_embeds=__UpperCAmelCase , negative_prompt_embeds=__UpperCAmelCase , pooled_prompt_embeds=__UpperCAmelCase , negative_pooled_prompt_embeds=__UpperCAmelCase , )
__lowerCamelCase = output.images[0, -3:, -3:, -1]
# make sure that it's equal
assert np.abs(image_slice_a.flatten() - image_slice_a.flatten() ).max() < 1E-4
@slow
@require_torch_gpu
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase="cpu" , __UpperCAmelCase=torch.floataa , __UpperCAmelCase=0 ):
'''simple docstring'''
__lowerCamelCase = torch.Generator(device=__UpperCAmelCase ).manual_seed(__UpperCAmelCase )
__lowerCamelCase = np.random.RandomState(__UpperCAmelCase ).standard_normal((1, 4, 64, 64) )
__lowerCamelCase = torch.from_numpy(__UpperCAmelCase ).to(device=__UpperCAmelCase , dtype=__UpperCAmelCase )
__lowerCamelCase = {
'''prompt''': '''a photograph of an astronaut riding a horse''',
'''latents''': latents,
'''generator''': generator,
'''num_inference_steps''': 3,
'''guidance_scale''': 7.5,
'''output_type''': '''numpy''',
}
return inputs
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DiffusionPipeline.from_pretrained('''stabilityai/stable-diffusion-2-base''' )
pipe.to(__UpperCAmelCase )
pipe.set_progress_bar_config(disable=__UpperCAmelCase )
__lowerCamelCase = self.get_inputs(__UpperCAmelCase )
__lowerCamelCase = pipe(**__UpperCAmelCase ).images
__lowerCamelCase = image[0, -3:, -3:, -1].flatten()
assert image.shape == (1, 512, 512, 3)
__lowerCamelCase = np.array([0.49_493, 0.47_896, 0.40_798, 0.54_214, 0.53_212, 0.48_202, 0.47_656, 0.46_329, 0.48_506] )
assert np.abs(image_slice - expected_slice ).max() < 7E-3
| 330 |
import random
import unittest
from torch.utils.data import BatchSampler, DataLoader, IterableDataset
from accelerate import Accelerator
from accelerate.data_loader import (
BatchSamplerShard,
DataLoaderDispatcher,
DataLoaderShard,
IterableDatasetShard,
SkipBatchSampler,
SkipDataLoader,
skip_first_batches,
)
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase=0.01 , __UpperCAmelCase=1000 ):
'''simple docstring'''
__lowerCamelCase = p_stop
__lowerCamelCase = max_length
def __iter__( self ):
'''simple docstring'''
__lowerCamelCase = 0
__lowerCamelCase = False
while not stop and count < self.max_length:
yield count
count += 1
__lowerCamelCase = random.random() < self.p_stop
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = [
BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
for i in range(2 )
]
__lowerCamelCase = [list(__UpperCAmelCase ) for batch_sampler_shard in batch_sampler_shards]
if not split_batches:
self.assertListEqual([len(__UpperCAmelCase ) for shard in batch_sampler_shards] , [len(__UpperCAmelCase ) for e in expected] )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [0, 1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 0]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [1, 2, 3]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1, 0]], [[1, 0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 0]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], [[0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = [[0, 1, 2], [3, 4], [5, 6, 7, 8], [9, 10, 11], [12, 13]]
__lowerCamelCase = [BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , even_batches=__UpperCAmelCase ) for i in range(2 )]
self.assertEqual(len(batch_sampler_shards[0] ) , 3 )
self.assertEqual(len(batch_sampler_shards[1] ) , 2 )
self.assertListEqual(list(batch_sampler_shards[0] ) , [[0, 1, 2], [5, 6, 7, 8], [12, 13]] )
self.assertListEqual(list(batch_sampler_shards[1] ) , [[3, 4], [9, 10, 11]] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=2 , __UpperCAmelCase=False ):
'''simple docstring'''
random.seed(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = [
IterableDatasetShard(
__UpperCAmelCase , batch_size=__UpperCAmelCase , drop_last=__UpperCAmelCase , num_processes=__UpperCAmelCase , process_index=__UpperCAmelCase , split_batches=__UpperCAmelCase , )
for i in range(__UpperCAmelCase )
]
__lowerCamelCase = []
for iterable_dataset_shard in iterable_dataset_shards:
# Since our random iterable dataset will be... random... we need to use a seed to get reproducible results.
random.seed(__UpperCAmelCase )
iterable_dataset_lists.append(list(__UpperCAmelCase ) )
__lowerCamelCase = batch_size // num_processes if split_batches else batch_size
# All iterable dataset shard should have the same length, a round multiple of shard_batch_size
__lowerCamelCase = iterable_dataset_lists[0]
for l in iterable_dataset_lists[1:]:
self.assertEqual(len(__UpperCAmelCase ) , len(__UpperCAmelCase ) )
self.assertTrue(len(__UpperCAmelCase ) % shard_batch_size == 0 )
__lowerCamelCase = []
for idx in range(0 , len(__UpperCAmelCase ) , __UpperCAmelCase ):
for l in iterable_dataset_lists:
observed += l[idx : idx + shard_batch_size]
if not drop_last:
while len(__UpperCAmelCase ) < len(__UpperCAmelCase ):
reference += reference
self.assertListEqual(__UpperCAmelCase , reference[: len(__UpperCAmelCase )] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = 42
__lowerCamelCase = RandomIterableDataset()
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
# Edge case with a very small dataset
__lowerCamelCase = RandomIterableDataset(max_length=2 )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BatchSampler(range(16 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = SkipBatchSampler(__UpperCAmelCase , 2 )
self.assertListEqual(list(__UpperCAmelCase ) , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = SkipDataLoader(list(range(16 ) ) , batch_size=4 , skip_batches=2 )
self.assertListEqual([t.tolist() for t in dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoader(list(range(16 ) ) , batch_size=4 )
__lowerCamelCase = skip_first_batches(__UpperCAmelCase , num_batches=2 )
self.assertListEqual([t.tolist() for t in new_dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoaderShard(list(range(16 ) ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
def lowerCamelCase ( self ):
'''simple docstring'''
Accelerator()
__lowerCamelCase = DataLoaderDispatcher(range(16 ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
| 330 | 1 |
from unittest.mock import patch
import pyspark
from datasets.packaged_modules.spark.spark import (
Spark,
SparkExamplesIterable,
_generate_iterable_examples,
)
from ..utils import (
require_dill_gt_0_3_2,
require_not_windows,
)
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : Dict ):
__lowerCamelCase = []
for part_id in partition_order:
__lowerCamelCase = df.where(F"""SPARK_PARTITION_ID() = {part_id}""" ).collect()
for row_idx, row in enumerate(_UpperCamelCase ):
expected_row_ids_and_row_dicts.append((F"""{part_id}_{row_idx}""", row.asDict()) )
return expected_row_ids_and_row_dicts
@require_not_windows
@require_dill_gt_0_3_2
def a__ ( ):
__lowerCamelCase = pyspark.sql.SparkSession.builder.master('''local[*]''' ).appName('''pyspark''' ).getOrCreate()
__lowerCamelCase = spark.range(1_00 ).repartition(1 )
__lowerCamelCase = Spark(_UpperCamelCase )
# The id ints will be converted to Pyarrow int64s, so each row will be 8 bytes. Setting a max_shard_size of 16 means
# that each partition can hold 2 rows.
spark_builder._repartition_df_if_needed(max_shard_size=16 )
# Given that the dataframe has 100 rows and each partition has 2 rows, we expect 50 partitions.
assert spark_builder.df.rdd.getNumPartitions() == 50
@require_not_windows
@require_dill_gt_0_3_2
def a__ ( ):
__lowerCamelCase = pyspark.sql.SparkSession.builder.master('''local[*]''' ).appName('''pyspark''' ).getOrCreate()
__lowerCamelCase = spark.range(10 ).repartition(2 )
__lowerCamelCase = [1, 0]
__lowerCamelCase = _generate_iterable_examples(_UpperCamelCase ,_UpperCamelCase ) # Reverse the partitions.
__lowerCamelCase = _get_expected_row_ids_and_row_dicts_for_partition_order(_UpperCamelCase ,_UpperCamelCase )
for i, (row_id, row_dict) in enumerate(generate_fn() ):
__lowerCamelCase ,__lowerCamelCase = expected_row_ids_and_row_dicts[i]
assert row_id == expected_row_id
assert row_dict == expected_row_dict
@require_not_windows
@require_dill_gt_0_3_2
def a__ ( ):
__lowerCamelCase = pyspark.sql.SparkSession.builder.master('''local[*]''' ).appName('''pyspark''' ).getOrCreate()
__lowerCamelCase = spark.range(10 ).repartition(1 )
__lowerCamelCase = SparkExamplesIterable(_UpperCamelCase )
assert it.n_shards == 1
for i, (row_id, row_dict) in enumerate(_UpperCamelCase ):
assert row_id == F"""0_{i}"""
assert row_dict == {"id": i}
@require_not_windows
@require_dill_gt_0_3_2
def a__ ( ):
__lowerCamelCase = pyspark.sql.SparkSession.builder.master('''local[*]''' ).appName('''pyspark''' ).getOrCreate()
__lowerCamelCase = spark.range(30 ).repartition(3 )
# Mock the generator so that shuffle reverses the partition indices.
with patch('''numpy.random.Generator''' ) as generator_mock:
__lowerCamelCase = lambda _UpperCamelCase : x.reverse()
__lowerCamelCase = _get_expected_row_ids_and_row_dicts_for_partition_order(_UpperCamelCase ,[2, 1, 0] )
__lowerCamelCase = SparkExamplesIterable(_UpperCamelCase ).shuffle_data_sources(_UpperCamelCase )
assert shuffled_it.n_shards == 3
for i, (row_id, row_dict) in enumerate(_UpperCamelCase ):
__lowerCamelCase ,__lowerCamelCase = expected_row_ids_and_row_dicts[i]
assert row_id == expected_row_id
assert row_dict == expected_row_dict
@require_not_windows
@require_dill_gt_0_3_2
def a__ ( ):
__lowerCamelCase = pyspark.sql.SparkSession.builder.master('''local[*]''' ).appName('''pyspark''' ).getOrCreate()
__lowerCamelCase = spark.range(20 ).repartition(4 )
# Partitions 0 and 2
__lowerCamelCase = SparkExamplesIterable(_UpperCamelCase ).shard_data_sources(worker_id=0 ,num_workers=2 )
assert shard_it_a.n_shards == 2
__lowerCamelCase = _get_expected_row_ids_and_row_dicts_for_partition_order(_UpperCamelCase ,[0, 2] )
for i, (row_id, row_dict) in enumerate(_UpperCamelCase ):
__lowerCamelCase ,__lowerCamelCase = expected_row_ids_and_row_dicts_a[i]
assert row_id == expected_row_id
assert row_dict == expected_row_dict
# Partitions 1 and 3
__lowerCamelCase = SparkExamplesIterable(_UpperCamelCase ).shard_data_sources(worker_id=1 ,num_workers=2 )
assert shard_it_a.n_shards == 2
__lowerCamelCase = _get_expected_row_ids_and_row_dicts_for_partition_order(_UpperCamelCase ,[1, 3] )
for i, (row_id, row_dict) in enumerate(_UpperCamelCase ):
__lowerCamelCase ,__lowerCamelCase = expected_row_ids_and_row_dicts_a[i]
assert row_id == expected_row_id
assert row_dict == expected_row_dict
@require_not_windows
@require_dill_gt_0_3_2
def a__ ( ):
__lowerCamelCase = pyspark.sql.SparkSession.builder.master('''local[*]''' ).appName('''pyspark''' ).getOrCreate()
__lowerCamelCase = spark.range(1_00 ).repartition(1 )
__lowerCamelCase = Spark(_UpperCamelCase )
# Choose a small max_shard_size for maximum partitioning.
spark_builder._repartition_df_if_needed(max_shard_size=1 )
# The new number of partitions should not be greater than the number of rows.
assert spark_builder.df.rdd.getNumPartitions() == 1_00
| 330 |
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = n ** (1 / 3)
return (val * val * val) == n
if __name__ == "__main__":
print(perfect_cube(27))
print(perfect_cube(4))
| 330 | 1 |
import importlib
import os
from dataclasses import dataclass
from enum import Enum
from typing import Any, Dict, Optional, Union
import torch
from ..utils import BaseOutput
a_ = """scheduler_config.json"""
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 1
lowerCAmelCase__ = 2
lowerCAmelCase__ = 3
lowerCAmelCase__ = 4
lowerCAmelCase__ = 5
lowerCAmelCase__ = 6
lowerCAmelCase__ = 7
lowerCAmelCase__ = 8
lowerCAmelCase__ = 9
lowerCAmelCase__ = 1_0
lowerCAmelCase__ = 1_1
lowerCAmelCase__ = 1_2
lowerCAmelCase__ = 1_3
lowerCAmelCase__ = 1_4
@dataclass
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 42
class __lowerCAmelCase :
lowerCAmelCase__ = SCHEDULER_CONFIG_NAME
lowerCAmelCase__ = []
lowerCAmelCase__ = True
@classmethod
def lowerCamelCase ( cls , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=False , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = cls.load_config(
pretrained_model_name_or_path=__UpperCAmelCase , subfolder=__UpperCAmelCase , return_unused_kwargs=__UpperCAmelCase , return_commit_hash=__UpperCAmelCase , **__UpperCAmelCase , )
return cls.from_config(__UpperCAmelCase , return_unused_kwargs=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False , **__UpperCAmelCase ):
'''simple docstring'''
self.save_config(save_directory=__UpperCAmelCase , push_to_hub=__UpperCAmelCase , **__UpperCAmelCase )
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._get_compatibles()
@classmethod
def lowerCamelCase ( cls ):
'''simple docstring'''
__lowerCamelCase = list(set([cls.__name__] + cls._compatibles ) )
__lowerCamelCase = importlib.import_module(__name__.split('''.''' )[0] )
__lowerCamelCase = [
getattr(__UpperCAmelCase , __UpperCAmelCase ) for c in compatible_classes_str if hasattr(__UpperCAmelCase , __UpperCAmelCase )
]
return compatible_classes
| 330 |
import argparse
import os
# New Code #
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
from accelerate.utils import find_executable_batch_size
########################################################################
# This is a fully working simple example to use Accelerate,
# specifically showcasing how to ensure out-of-memory errors never
# interrupt training, and builds off the `nlp_example.py` script.
#
# This example trains a Bert base model on GLUE MRPC
# in any of the following settings (with the same script):
# - single CPU or single GPU
# - multi GPUS (using PyTorch distributed mode)
# - (multi) TPUs
# - fp16 (mixed-precision) or fp32 (normal precision)
#
# New additions from the base script can be found quickly by
# looking for the # New Code # tags
#
# To run it in each of these various modes, follow the instructions
# in the readme for examples:
# https://github.com/huggingface/accelerate/tree/main/examples
#
########################################################################
a_ = 16
a_ = 32
def a__ ( _UpperCamelCase : Accelerator ,_UpperCamelCase : int = 16 ):
__lowerCamelCase = AutoTokenizer.from_pretrained('''bert-base-cased''' )
__lowerCamelCase = load_dataset('''glue''' ,'''mrpc''' )
def tokenize_function(_UpperCamelCase : Optional[Any] ):
# max_length=None => use the model max length (it's actually the default)
__lowerCamelCase = tokenizer(examples['''sentence1'''] ,examples['''sentence2'''] ,truncation=_UpperCamelCase ,max_length=_UpperCamelCase )
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(
_UpperCamelCase ,batched=_UpperCamelCase ,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(_UpperCamelCase : Any ):
# On TPU it's best to pad everything to the same length or training will be very slow.
__lowerCamelCase = 1_28 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(
_UpperCamelCase ,padding='''longest''' ,max_length=_UpperCamelCase ,pad_to_multiple_of=_UpperCamelCase ,return_tensors='''pt''' ,)
# Instantiate dataloaders.
__lowerCamelCase = DataLoader(
tokenized_datasets['''train'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
__lowerCamelCase = DataLoader(
tokenized_datasets['''validation'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
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
a_ = mocked_dataloaders # noqa: F811
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# For testing only
if os.environ.get('''TESTING_MOCKED_DATALOADERS''' ,_UpperCamelCase ) == "1":
__lowerCamelCase = 2
# Initialize accelerator
__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'''] )
__lowerCamelCase = evaluate.load('''glue''' ,'''mrpc''' )
# New Code #
# We now can define an inner training loop function. It should take a batch size as the only parameter,
# and build the dataloaders in there.
# It also gets our decorator
@find_executable_batch_size(starting_batch_size=_UpperCamelCase )
def inner_training_loop(_UpperCamelCase : Union[str, Any] ):
# And now just move everything below under this function
# We need to bring in the Accelerator object from earlier
nonlocal accelerator
# And reset all of its attributes that could hold onto any memory:
accelerator.free_memory()
# Then we can declare the model, optimizer, and everything else:
set_seed(_UpperCamelCase )
# 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=_UpperCamelCase )
# 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=_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = get_dataloaders(_UpperCamelCase ,_UpperCamelCase )
# Instantiate scheduler
__lowerCamelCase = get_linear_schedule_with_warmup(
optimizer=_UpperCamelCase ,num_warmup_steps=1_00 ,num_training_steps=(len(_UpperCamelCase ) * num_epochs) ,)
# Prepare everything
# There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the
# prepare method.
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Now we train the model
for epoch in range(_UpperCamelCase ):
model.train()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.loss
accelerator.backward(_UpperCamelCase )
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
model.eval()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
with torch.no_grad():
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.logits.argmax(dim=-1 )
__lowerCamelCase ,__lowerCamelCase = accelerator.gather_for_metrics((predictions, batch['''labels''']) )
metric.add_batch(
predictions=_UpperCamelCase ,references=_UpperCamelCase ,)
__lowerCamelCase = metric.compute()
# Use accelerator.print to print only on the main process.
accelerator.print(F"""epoch {epoch}:""" ,_UpperCamelCase )
# New Code #
# And call it at the end with no arguments
# Note: You could also refactor this outside of your training loop function
inner_training_loop()
def a__ ( ):
__lowerCamelCase = argparse.ArgumentParser(description='''Simple example of training script.''' )
parser.add_argument(
'''--mixed_precision''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,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.''' )
__lowerCamelCase = parser.parse_args()
__lowerCamelCase = {'''lr''': 2e-5, '''num_epochs''': 3, '''seed''': 42, '''batch_size''': 16}
training_function(_UpperCamelCase ,_UpperCamelCase )
if __name__ == "__main__":
main()
| 330 | 1 |
import math
from enum import Enum
from typing import Optional, Union
from torch.optim import Optimizer
from torch.optim.lr_scheduler import LambdaLR
from .utils import logging
a_ = logging.get_logger(__name__)
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """linear"""
lowerCAmelCase__ = """cosine"""
lowerCAmelCase__ = """cosine_with_restarts"""
lowerCAmelCase__ = """polynomial"""
lowerCAmelCase__ = """constant"""
lowerCAmelCase__ = """constant_with_warmup"""
lowerCAmelCase__ = """piecewise_constant"""
def a__ ( _UpperCamelCase : Optimizer ,_UpperCamelCase : int = -1 ):
return LambdaLR(_UpperCamelCase ,lambda _UpperCamelCase : 1 ,last_epoch=_UpperCamelCase )
def a__ ( _UpperCamelCase : Optimizer ,_UpperCamelCase : int ,_UpperCamelCase : int = -1 ):
def lr_lambda(_UpperCamelCase : int ):
if current_step < num_warmup_steps:
return float(_UpperCamelCase ) / float(max(1.0 ,_UpperCamelCase ) )
return 1.0
return LambdaLR(_UpperCamelCase ,_UpperCamelCase ,last_epoch=_UpperCamelCase )
def a__ ( _UpperCamelCase : Optimizer ,_UpperCamelCase : str ,_UpperCamelCase : int = -1 ):
__lowerCamelCase = {}
__lowerCamelCase = step_rules.split(''',''' )
for rule_str in rule_list[:-1]:
__lowerCamelCase ,__lowerCamelCase = rule_str.split(''':''' )
__lowerCamelCase = int(_UpperCamelCase )
__lowerCamelCase = float(_UpperCamelCase )
__lowerCamelCase = value
__lowerCamelCase = float(rule_list[-1] )
def create_rules_function(_UpperCamelCase : List[Any] ,_UpperCamelCase : Any ):
def rule_func(_UpperCamelCase : int ) -> float:
__lowerCamelCase = sorted(rules_dict.keys() )
for i, sorted_step in enumerate(_UpperCamelCase ):
if steps < sorted_step:
return rules_dict[sorted_steps[i]]
return last_lr_multiple
return rule_func
__lowerCamelCase = create_rules_function(_UpperCamelCase ,_UpperCamelCase )
return LambdaLR(_UpperCamelCase ,_UpperCamelCase ,last_epoch=_UpperCamelCase )
def a__ ( _UpperCamelCase : Optional[Any] ,_UpperCamelCase : List[Any] ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Dict=-1 ):
def lr_lambda(_UpperCamelCase : int ):
if current_step < num_warmup_steps:
return float(_UpperCamelCase ) / float(max(1 ,_UpperCamelCase ) )
return max(
0.0 ,float(num_training_steps - current_step ) / float(max(1 ,num_training_steps - num_warmup_steps ) ) )
return LambdaLR(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
def a__ ( _UpperCamelCase : Optimizer ,_UpperCamelCase : int ,_UpperCamelCase : int ,_UpperCamelCase : float = 0.5 ,_UpperCamelCase : int = -1 ):
def lr_lambda(_UpperCamelCase : int ):
if current_step < num_warmup_steps:
return float(_UpperCamelCase ) / float(max(1 ,_UpperCamelCase ) )
__lowerCamelCase = float(current_step - num_warmup_steps ) / float(max(1 ,num_training_steps - num_warmup_steps ) )
return max(0.0 ,0.5 * (1.0 + math.cos(math.pi * float(_UpperCamelCase ) * 2.0 * progress )) )
return LambdaLR(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
def a__ ( _UpperCamelCase : Optimizer ,_UpperCamelCase : int ,_UpperCamelCase : int ,_UpperCamelCase : int = 1 ,_UpperCamelCase : int = -1 ):
def lr_lambda(_UpperCamelCase : Union[str, Any] ):
if current_step < num_warmup_steps:
return float(_UpperCamelCase ) / float(max(1 ,_UpperCamelCase ) )
__lowerCamelCase = float(current_step - num_warmup_steps ) / float(max(1 ,num_training_steps - num_warmup_steps ) )
if progress >= 1.0:
return 0.0
return max(0.0 ,0.5 * (1.0 + math.cos(math.pi * ((float(_UpperCamelCase ) * progress) % 1.0) )) )
return LambdaLR(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : Dict ,_UpperCamelCase : Tuple ,_UpperCamelCase : Tuple=1e-7 ,_UpperCamelCase : List[str]=1.0 ,_UpperCamelCase : Tuple=-1 ):
__lowerCamelCase = optimizer.defaults['''lr''']
if not (lr_init > lr_end):
raise ValueError(F"""lr_end ({lr_end}) must be be smaller than initial lr ({lr_init})""" )
def lr_lambda(_UpperCamelCase : int ):
if current_step < num_warmup_steps:
return float(_UpperCamelCase ) / float(max(1 ,_UpperCamelCase ) )
elif current_step > num_training_steps:
return lr_end / lr_init # as LambdaLR multiplies by lr_init
else:
__lowerCamelCase = lr_init - lr_end
__lowerCamelCase = num_training_steps - num_warmup_steps
__lowerCamelCase = 1 - (current_step - num_warmup_steps) / decay_steps
__lowerCamelCase = lr_range * pct_remaining**power + lr_end
return decay / lr_init # as LambdaLR multiplies by lr_init
return LambdaLR(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
a_ = {
SchedulerType.LINEAR: get_linear_schedule_with_warmup,
SchedulerType.COSINE: get_cosine_schedule_with_warmup,
SchedulerType.COSINE_WITH_RESTARTS: get_cosine_with_hard_restarts_schedule_with_warmup,
SchedulerType.POLYNOMIAL: get_polynomial_decay_schedule_with_warmup,
SchedulerType.CONSTANT: get_constant_schedule,
SchedulerType.CONSTANT_WITH_WARMUP: get_constant_schedule_with_warmup,
SchedulerType.PIECEWISE_CONSTANT: get_piecewise_constant_schedule,
}
def a__ ( _UpperCamelCase : Union[str, SchedulerType] ,_UpperCamelCase : Optimizer ,_UpperCamelCase : Optional[str] = None ,_UpperCamelCase : Optional[int] = None ,_UpperCamelCase : Optional[int] = None ,_UpperCamelCase : int = 1 ,_UpperCamelCase : float = 1.0 ,_UpperCamelCase : int = -1 ,):
__lowerCamelCase = SchedulerType(_UpperCamelCase )
__lowerCamelCase = TYPE_TO_SCHEDULER_FUNCTION[name]
if name == SchedulerType.CONSTANT:
return schedule_func(_UpperCamelCase ,last_epoch=_UpperCamelCase )
if name == SchedulerType.PIECEWISE_CONSTANT:
return schedule_func(_UpperCamelCase ,step_rules=_UpperCamelCase ,last_epoch=_UpperCamelCase )
# All other schedulers require `num_warmup_steps`
if num_warmup_steps is None:
raise ValueError(F"""{name} requires `num_warmup_steps`, please provide that argument.""" )
if name == SchedulerType.CONSTANT_WITH_WARMUP:
return schedule_func(_UpperCamelCase ,num_warmup_steps=_UpperCamelCase ,last_epoch=_UpperCamelCase )
# All other schedulers require `num_training_steps`
if num_training_steps is None:
raise ValueError(F"""{name} requires `num_training_steps`, please provide that argument.""" )
if name == SchedulerType.COSINE_WITH_RESTARTS:
return schedule_func(
_UpperCamelCase ,num_warmup_steps=_UpperCamelCase ,num_training_steps=_UpperCamelCase ,num_cycles=_UpperCamelCase ,last_epoch=_UpperCamelCase ,)
if name == SchedulerType.POLYNOMIAL:
return schedule_func(
_UpperCamelCase ,num_warmup_steps=_UpperCamelCase ,num_training_steps=_UpperCamelCase ,power=_UpperCamelCase ,last_epoch=_UpperCamelCase ,)
return schedule_func(
_UpperCamelCase ,num_warmup_steps=_UpperCamelCase ,num_training_steps=_UpperCamelCase ,last_epoch=_UpperCamelCase )
| 330 |
import logging
import os
import threading
import time
try:
import warnings
except ImportError:
a_ = None
try:
import msvcrt
except ImportError:
a_ = None
try:
import fcntl
except ImportError:
a_ = None
# Backward compatibility
# ------------------------------------------------
try:
TimeoutError
except NameError:
a_ = OSError
# Data
# ------------------------------------------------
a_ = [
"""Timeout""",
"""BaseFileLock""",
"""WindowsFileLock""",
"""UnixFileLock""",
"""SoftFileLock""",
"""FileLock""",
]
a_ = """3.0.12"""
a_ = None
def a__ ( ):
global _logger
__lowerCamelCase = _logger or logging.getLogger(__name__ )
return _logger
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = lock_file
return None
def __str__( self ):
'''simple docstring'''
__lowerCamelCase = F"""The file lock '{self.lock_file}' could not be acquired."""
return temp
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = lock
return None
def __enter__( self ):
'''simple docstring'''
return self.lock
def __exit__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
self.lock.release()
return None
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
__lowerCamelCase = max_filename_length if max_filename_length is not None else 255
# Hash the filename if it's too long
__lowerCamelCase = self.hash_filename_if_too_long(__UpperCAmelCase , __UpperCAmelCase )
# The path to the lock file.
__lowerCamelCase = lock_file
# The file descriptor for the *_lock_file* as it is returned by the
# os.open() function.
# This file lock is only NOT None, if the object currently holds the
# lock.
__lowerCamelCase = None
# The default timeout value.
__lowerCamelCase = timeout
# We use this lock primarily for the lock counter.
__lowerCamelCase = threading.Lock()
# The lock counter is used for implementing the nested locking
# mechanism. Whenever the lock is acquired, the counter is increased and
# the lock is only released, when this value is 0 again.
__lowerCamelCase = 0
return None
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._lock_file
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._timeout
@timeout.setter
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = float(__UpperCAmelCase )
return None
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError()
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError()
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._lock_file_fd is not None
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=0.05 ):
'''simple docstring'''
# Use the default timeout, if no timeout is provided.
if timeout is None:
__lowerCamelCase = self.timeout
# Increment the number right at the beginning.
# We can still undo it, if something fails.
with self._thread_lock:
self._lock_counter += 1
__lowerCamelCase = id(self )
__lowerCamelCase = self._lock_file
__lowerCamelCase = time.time()
try:
while True:
with self._thread_lock:
if not self.is_locked:
logger().debug(F"""Attempting to acquire lock {lock_id} on {lock_filename}""" )
self._acquire()
if self.is_locked:
logger().debug(F"""Lock {lock_id} acquired on {lock_filename}""" )
break
elif timeout >= 0 and time.time() - start_time > timeout:
logger().debug(F"""Timeout on acquiring lock {lock_id} on {lock_filename}""" )
raise Timeout(self._lock_file )
else:
logger().debug(
F"""Lock {lock_id} not acquired on {lock_filename}, waiting {poll_intervall} seconds ...""" )
time.sleep(__UpperCAmelCase )
except: # noqa
# Something did go wrong, so decrement the counter.
with self._thread_lock:
__lowerCamelCase = max(0 , self._lock_counter - 1 )
raise
return _Acquire_ReturnProxy(lock=self )
def lowerCamelCase ( self , __UpperCAmelCase=False ):
'''simple docstring'''
with self._thread_lock:
if self.is_locked:
self._lock_counter -= 1
if self._lock_counter == 0 or force:
__lowerCamelCase = id(self )
__lowerCamelCase = self._lock_file
logger().debug(F"""Attempting to release lock {lock_id} on {lock_filename}""" )
self._release()
__lowerCamelCase = 0
logger().debug(F"""Lock {lock_id} released on {lock_filename}""" )
return None
def __enter__( self ):
'''simple docstring'''
self.acquire()
return self
def __exit__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
self.release()
return None
def __del__( self ):
'''simple docstring'''
self.release(force=__UpperCAmelCase )
return None
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = os.path.basename(__UpperCAmelCase )
if len(__UpperCAmelCase ) > max_length and max_length > 0:
__lowerCamelCase = os.path.dirname(__UpperCAmelCase )
__lowerCamelCase = str(hash(__UpperCAmelCase ) )
__lowerCamelCase = filename[: max_length - len(__UpperCAmelCase ) - 8] + '''...''' + hashed_filename + '''.lock'''
return os.path.join(__UpperCAmelCase , __UpperCAmelCase )
else:
return path
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
from .file_utils import relative_to_absolute_path
super().__init__(__UpperCAmelCase , timeout=__UpperCAmelCase , max_filename_length=__UpperCAmelCase )
__lowerCamelCase = '''\\\\?\\''' + relative_to_absolute_path(self.lock_file )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_RDWR | os.O_CREAT | os.O_TRUNC
try:
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
except OSError:
pass
else:
try:
msvcrt.locking(__UpperCAmelCase , msvcrt.LK_NBLCK , 1 )
except OSError:
os.close(__UpperCAmelCase )
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self._lock_file_fd
__lowerCamelCase = None
msvcrt.locking(__UpperCAmelCase , msvcrt.LK_UNLCK , 1 )
os.close(__UpperCAmelCase )
try:
os.remove(self._lock_file )
# Probably another instance of the application
# that acquired the file lock.
except OSError:
pass
return None
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
__lowerCamelCase = os.statvfs(os.path.dirname(__UpperCAmelCase ) ).f_namemax
super().__init__(__UpperCAmelCase , timeout=__UpperCAmelCase , max_filename_length=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_RDWR | os.O_CREAT | os.O_TRUNC
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
try:
fcntl.flock(__UpperCAmelCase , fcntl.LOCK_EX | fcntl.LOCK_NB )
except OSError:
os.close(__UpperCAmelCase )
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
# Do not remove the lockfile:
#
# https://github.com/benediktschmitt/py-filelock/issues/31
# https://stackoverflow.com/questions/17708885/flock-removing-locked-file-without-race-condition
__lowerCamelCase = self._lock_file_fd
__lowerCamelCase = None
fcntl.flock(__UpperCAmelCase , fcntl.LOCK_UN )
os.close(__UpperCAmelCase )
return None
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_WRONLY | os.O_CREAT | os.O_EXCL | os.O_TRUNC
try:
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
except OSError:
pass
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
os.close(self._lock_file_fd )
__lowerCamelCase = None
try:
os.remove(self._lock_file )
# The file is already deleted and that's what we want.
except OSError:
pass
return None
a_ = None
if msvcrt:
a_ = WindowsFileLock
elif fcntl:
a_ = UnixFileLock
else:
a_ = SoftFileLock
if warnings is not None:
warnings.warn("""only soft file lock is available""")
| 330 | 1 |
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
from ...utils.dataclasses import (
ComputeEnvironment,
DistributedType,
DynamoBackend,
PrecisionType,
SageMakerDistributedType,
)
from ..menu import BulletMenu
a_ = [
"""EAGER""",
"""AOT_EAGER""",
"""INDUCTOR""",
"""NVFUSER""",
"""AOT_NVFUSER""",
"""AOT_CUDAGRAPHS""",
"""OFI""",
"""FX2TRT""",
"""ONNXRT""",
"""IPEX""",
]
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : Union[str, Any]=None ,_UpperCamelCase : Union[str, Any]=None ,_UpperCamelCase : Union[str, Any]=None ):
__lowerCamelCase = True
while ask_again:
__lowerCamelCase = input(_UpperCamelCase )
try:
if default is not None and len(_UpperCamelCase ) == 0:
return default
return convert_value(_UpperCamelCase ) if convert_value is not None else result
except Exception:
if error_message is not None:
print(_UpperCamelCase )
def a__ ( _UpperCamelCase : Optional[Any] ,_UpperCamelCase : str=[] ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : Optional[int]=0 ):
__lowerCamelCase = BulletMenu(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = menu.run(default_choice=_UpperCamelCase )
return convert_value(_UpperCamelCase ) if convert_value is not None else result
def a__ ( _UpperCamelCase : Optional[int] ):
__lowerCamelCase = int(_UpperCamelCase )
return ComputeEnvironment(['''LOCAL_MACHINE''', '''AMAZON_SAGEMAKER'''][value] )
def a__ ( _UpperCamelCase : Any ):
__lowerCamelCase = int(_UpperCamelCase )
return DistributedType(['''NO''', '''MULTI_CPU''', '''MULTI_XPU''', '''MULTI_GPU''', '''MULTI_NPU''', '''TPU'''][value] )
def a__ ( _UpperCamelCase : Optional[int] ):
__lowerCamelCase = int(_UpperCamelCase )
return DynamoBackend(DYNAMO_BACKENDS[value] ).value
def a__ ( _UpperCamelCase : Tuple ):
__lowerCamelCase = int(_UpperCamelCase )
return PrecisionType(['''no''', '''fp16''', '''bf16''', '''fp8'''][value] )
def a__ ( _UpperCamelCase : List[str] ):
__lowerCamelCase = int(_UpperCamelCase )
return SageMakerDistributedType(['''NO''', '''DATA_PARALLEL''', '''MODEL_PARALLEL'''][value] )
def a__ ( _UpperCamelCase : Optional[int] ):
return {"yes": True, "no": False}[value.lower()]
class __lowerCAmelCase ( argparse.RawDescriptionHelpFormatter ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = super()._format_usage(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = usage.replace('''<command> [<args>] ''' , '''''' )
return usage
| 330 |
import copy
import inspect
import unittest
import numpy as np
from huggingface_hub import hf_hub_download
from transformers import TimesformerConfig
from transformers.models.auto import get_values
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 (
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING,
TimesformerForVideoClassification,
TimesformerModel,
)
from transformers.models.timesformer.modeling_timesformer import TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from transformers import VideoMAEImageProcessor
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=13 , __UpperCAmelCase=10 , __UpperCAmelCase=3 , __UpperCAmelCase=2 , __UpperCAmelCase=2 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=32 , __UpperCAmelCase=5 , __UpperCAmelCase=4 , __UpperCAmelCase=37 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=10 , __UpperCAmelCase=0.02 , __UpperCAmelCase="divided_space_time" , __UpperCAmelCase=None , ):
'''simple docstring'''
__lowerCamelCase = parent
__lowerCamelCase = batch_size
__lowerCamelCase = image_size
__lowerCamelCase = num_channels
__lowerCamelCase = patch_size
__lowerCamelCase = num_frames
__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 = attention_type
__lowerCamelCase = initializer_range
__lowerCamelCase = scope
__lowerCamelCase = num_labels
# in TimeSformer, the number of spatial tokens equals num_frames * num_patches per frame + 1 CLS token
__lowerCamelCase = (image_size // patch_size) ** 2
__lowerCamelCase = (num_frames) * self.num_patches_per_frame + 1
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = floats_tensor(
[self.batch_size, self.num_frames, 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 lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , num_frames=self.num_frames , 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 , initializer_range=self.initializer_range , attention_type=self.attention_type , )
__lowerCamelCase = self.num_labels
return config
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
# verify the logits shape
__lowerCamelCase = torch.Size((self.batch_size, self.num_labels) )
self.parent.assertEqual(result.logits.shape , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = config_and_inputs
__lowerCamelCase = {'''pixel_values''': pixel_values}
return config, inputs_dict
@require_torch
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (TimesformerModel, TimesformerForVideoClassification) if is_torch_available() else ()
lowerCAmelCase__ = (
{"""feature-extraction""": TimesformerModel, """video-classification""": TimesformerForVideoClassification}
if is_torch_available()
else {}
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerModelTester(self )
__lowerCamelCase = ConfigTester(
self , config_class=__UpperCAmelCase , has_text_modality=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = copy.deepcopy(__UpperCAmelCase )
if return_labels:
if model_class in get_values(__UpperCAmelCase ):
__lowerCamelCase = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=__UpperCAmelCase )
return inputs_dict
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
@unittest.skip(reason='''TimeSformer does not use inputs_embeds''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
__lowerCamelCase = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(__UpperCAmelCase , nn.Linear ) )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
__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] , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_video_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = TimesformerModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.has_attentions:
pass
else:
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCamelCase = True
for model_class in self.all_model_classes:
__lowerCamelCase = self.model_tester.seq_length
__lowerCamelCase = self.model_tester.num_frames
__lowerCamelCase = True
__lowerCamelCase = False
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
__lowerCamelCase = len(__UpperCAmelCase )
# Check attention is always last and order is fine
__lowerCamelCase = True
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
self.assertEqual(out_len + 1 , len(__UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
def lowerCamelCase ( self ):
'''simple docstring'''
def check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.hidden_states
__lowerCamelCase = self.model_tester.num_hidden_layers + 1
self.assertEqual(len(__UpperCAmelCase ) , __UpperCAmelCase )
__lowerCamelCase = self.model_tester.seq_length
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [seq_length, self.model_tester.hidden_size] , )
__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(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCamelCase = True
check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def a__ ( ):
__lowerCamelCase = hf_hub_download(
repo_id='''hf-internal-testing/spaghetti-video''' ,filename='''eating_spaghetti.npy''' ,repo_type='''dataset''' )
__lowerCamelCase = np.load(_UpperCamelCase )
return list(_UpperCamelCase )
@require_torch
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
@cached_property
def lowerCamelCase ( self ):
'''simple docstring'''
# logits were tested with a different mean and std, so we use the same here
return (
VideoMAEImageProcessor(image_mean=[0.5, 0.5, 0.5] , image_std=[0.5, 0.5, 0.5] )
if is_vision_available()
else None
)
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification.from_pretrained('''facebook/timesformer-base-finetuned-k400''' ).to(
__UpperCAmelCase )
__lowerCamelCase = self.default_image_processor
__lowerCamelCase = prepare_video()
__lowerCamelCase = image_processor(video[:8] , return_tensors='''pt''' ).to(__UpperCAmelCase )
# forward pass
with torch.no_grad():
__lowerCamelCase = model(**__UpperCAmelCase )
# verify the logits
__lowerCamelCase = torch.Size((1, 400) )
self.assertEqual(outputs.logits.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor([-0.3_016, -0.7_713, -0.4_205] ).to(__UpperCAmelCase )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
# NOTE: This file is deprecated and will be removed in a future version.
# It only exists so that temporarely `from diffusers.pipelines import DiffusionPipeline` works
from ...utils import deprecate
from ..controlnet.pipeline_flax_controlnet import FlaxStableDiffusionControlNetPipeline # noqa: F401
deprecate(
"""stable diffusion controlnet""",
"""0.22.0""",
"""Importing `FlaxStableDiffusionControlNetPipeline` from diffusers.pipelines.stable_diffusion.flax_pipeline_stable_diffusion_controlnet is deprecated. Please import `from diffusers import FlaxStableDiffusionControlNetPipeline` instead.""",
standard_warn=False,
stacklevel=3,
)
| 330 |
def a__ ( _UpperCamelCase : int ):
if not isinstance(_UpperCamelCase ,_UpperCamelCase ):
__lowerCamelCase = F"""Input value of [number={number}] must be an integer"""
raise TypeError(_UpperCamelCase )
if number < 0:
return False
__lowerCamelCase = number * number
while number > 0:
if number % 10 != number_square % 10:
return False
number //= 10
number_square //= 10
return True
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 | 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
a_ = False
class __lowerCAmelCase ( unittest.TestCase ):
pass
@nightly
@require_torch_gpu
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self ):
'''simple docstring'''
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = VersatileDiffusionTextToImagePipeline.from_pretrained('''shi-labs/versatile-diffusion''' )
# remove text_unet
pipe.remove_unused_weights()
pipe.to(__UpperCAmelCase )
pipe.set_progress_bar_config(disable=__UpperCAmelCase )
__lowerCamelCase = '''A painting of a squirrel eating a burger '''
__lowerCamelCase = torch.manual_seed(0 )
__lowerCamelCase = pipe(
prompt=__UpperCAmelCase , generator=__UpperCAmelCase , guidance_scale=7.5 , num_inference_steps=2 , output_type='''numpy''' ).images
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(__UpperCAmelCase )
__lowerCamelCase = VersatileDiffusionTextToImagePipeline.from_pretrained(__UpperCAmelCase )
pipe.to(__UpperCAmelCase )
pipe.set_progress_bar_config(disable=__UpperCAmelCase )
__lowerCamelCase = generator.manual_seed(0 )
__lowerCamelCase = pipe(
prompt=__UpperCAmelCase , generator=__UpperCAmelCase , 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 lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = VersatileDiffusionTextToImagePipeline.from_pretrained(
'''shi-labs/versatile-diffusion''' , torch_dtype=torch.floataa )
pipe.to(__UpperCAmelCase )
pipe.set_progress_bar_config(disable=__UpperCAmelCase )
__lowerCamelCase = '''A painting of a squirrel eating a burger '''
__lowerCamelCase = torch.manual_seed(0 )
__lowerCamelCase = pipe(
prompt=__UpperCAmelCase , generator=__UpperCAmelCase , guidance_scale=7.5 , num_inference_steps=50 , output_type='''numpy''' ).images
__lowerCamelCase = image[0, 253:256, 253:256, -1]
assert image.shape == (1, 512, 512, 3)
__lowerCamelCase = np.array([0.3_367, 0.3_169, 0.2_656, 0.3_870, 0.4_790, 0.3_796, 0.4_009, 0.4_878, 0.4_778] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
| 330 |
import gc
import unittest
from parameterized import parameterized
from diffusers import FlaxUNetaDConditionModel
from diffusers.utils import is_flax_available
from diffusers.utils.testing_utils import load_hf_numpy, require_flax, slow
if is_flax_available():
import jax
import jax.numpy as jnp
@slow
@require_flax
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
return F"""gaussian_noise_s={seed}_shape={"_".join([str(__UpperCAmelCase ) for s in shape] )}.npy"""
def lowerCamelCase ( self ):
'''simple docstring'''
# clean up the VRAM after each test
super().tearDown()
gc.collect()
def lowerCamelCase ( self , __UpperCAmelCase=0 , __UpperCAmelCase=(4, 4, 64, 64) , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = jnp.array(load_hf_numpy(self.get_file_format(__UpperCAmelCase , __UpperCAmelCase ) ) , dtype=__UpperCAmelCase )
return image
def lowerCamelCase ( self , __UpperCAmelCase=False , __UpperCAmelCase="CompVis/stable-diffusion-v1-4" ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = '''bf16''' if fpaa else None
__lowerCamelCase ,__lowerCamelCase = FlaxUNetaDConditionModel.from_pretrained(
__UpperCAmelCase , subfolder='''unet''' , dtype=__UpperCAmelCase , revision=__UpperCAmelCase )
return model, params
def lowerCamelCase ( self , __UpperCAmelCase=0 , __UpperCAmelCase=(4, 77, 768) , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = jnp.array(load_hf_numpy(self.get_file_format(__UpperCAmelCase , __UpperCAmelCase ) ) , dtype=__UpperCAmelCase )
return hidden_states
@parameterized.expand(
[
# fmt: off
[83, 4, [-0.2_323, -0.1_304, 0.0_813, -0.3_093, -0.0_919, -0.1_571, -0.1_125, -0.5_806]],
[17, 0.55, [-0.0_831, -0.2_443, 0.0_901, -0.0_919, 0.3_396, 0.0_103, -0.3_743, 0.0_701]],
[8, 0.89, [-0.4_863, 0.0_859, 0.0_875, -0.1_658, 0.9_199, -0.0_114, 0.4_839, 0.4_639]],
[3, 1000, [-0.5_649, 0.2_402, -0.5_518, 0.1_248, 1.1_328, -0.2_443, -0.0_325, -1.0_078]],
# fmt: on
] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_unet_model(model_id='''CompVis/stable-diffusion-v1-4''' , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_latents(__UpperCAmelCase , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_encoder_hidden_states(__UpperCAmelCase , fpaa=__UpperCAmelCase )
__lowerCamelCase = model.apply(
{'''params''': params} , __UpperCAmelCase , jnp.array(__UpperCAmelCase , dtype=jnp.intaa ) , encoder_hidden_states=__UpperCAmelCase , ).sample
assert sample.shape == latents.shape
__lowerCamelCase = jnp.asarray(jax.device_get((sample[-1, -2:, -2:, :2].flatten()) ) , dtype=jnp.floataa )
__lowerCamelCase = jnp.array(__UpperCAmelCase , dtype=jnp.floataa )
# Found torch (float16) and flax (bfloat16) outputs to be within this tolerance, in the same hardware
assert jnp.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-2 )
@parameterized.expand(
[
# fmt: off
[83, 4, [0.1_514, 0.0_807, 0.1_624, 0.1_016, -0.1_896, 0.0_263, 0.0_677, 0.2_310]],
[17, 0.55, [0.1_164, -0.0_216, 0.0_170, 0.1_589, -0.3_120, 0.1_005, -0.0_581, -0.1_458]],
[8, 0.89, [-0.1_758, -0.0_169, 0.1_004, -0.1_411, 0.1_312, 0.1_103, -0.1_996, 0.2_139]],
[3, 1000, [0.1_214, 0.0_352, -0.0_731, -0.1_562, -0.0_994, -0.0_906, -0.2_340, -0.0_539]],
# fmt: on
] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_unet_model(model_id='''stabilityai/stable-diffusion-2''' , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_latents(__UpperCAmelCase , shape=(4, 4, 96, 96) , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_encoder_hidden_states(__UpperCAmelCase , shape=(4, 77, 1024) , fpaa=__UpperCAmelCase )
__lowerCamelCase = model.apply(
{'''params''': params} , __UpperCAmelCase , jnp.array(__UpperCAmelCase , dtype=jnp.intaa ) , encoder_hidden_states=__UpperCAmelCase , ).sample
assert sample.shape == latents.shape
__lowerCamelCase = jnp.asarray(jax.device_get((sample[-1, -2:, -2:, :2].flatten()) ) , dtype=jnp.floataa )
__lowerCamelCase = jnp.array(__UpperCAmelCase , dtype=jnp.floataa )
# Found torch (float16) and flax (bfloat16) outputs to be within this tolerance, on the same hardware
assert jnp.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-2 )
| 330 | 1 |
a_ = {
"""A""": ["""B""", """C""", """E"""],
"""B""": ["""A""", """D""", """E"""],
"""C""": ["""A""", """F""", """G"""],
"""D""": ["""B"""],
"""E""": ["""A""", """B""", """D"""],
"""F""": ["""C"""],
"""G""": ["""C"""],
}
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : List[str] ,_UpperCamelCase : Any ):
__lowerCamelCase = set()
# keep track of all the paths to be checked
__lowerCamelCase = [[start]]
# return path if start is goal
if start == goal:
return [start]
# keeps looping until all possible paths have been checked
while queue:
# pop the first path from the queue
__lowerCamelCase = queue.pop(0 )
# get the last node from the path
__lowerCamelCase = path[-1]
if node not in explored:
__lowerCamelCase = graph[node]
# go through all neighbour nodes, construct a new path and
# push it into the queue
for neighbour in neighbours:
__lowerCamelCase = list(_UpperCamelCase )
new_path.append(_UpperCamelCase )
queue.append(_UpperCamelCase )
# return path if neighbour is goal
if neighbour == goal:
return new_path
# mark node as explored
explored.add(_UpperCamelCase )
# in case there's no path between the 2 nodes
return []
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : Dict ,_UpperCamelCase : Tuple ):
if not graph or start not in graph or target not in graph:
return -1
if start == target:
return 0
__lowerCamelCase = [start]
__lowerCamelCase = set(_UpperCamelCase )
# Keep tab on distances from `start` node.
__lowerCamelCase = {start: 0, target: -1}
while queue:
__lowerCamelCase = queue.pop(0 )
if node == target:
__lowerCamelCase = (
dist[node] if dist[target] == -1 else min(dist[target] ,dist[node] )
)
for adjacent in graph[node]:
if adjacent not in visited:
visited.add(_UpperCamelCase )
queue.append(_UpperCamelCase )
__lowerCamelCase = dist[node] + 1
return dist[target]
if __name__ == "__main__":
print(bfs_shortest_path(demo_graph, """G""", """D""")) # returns ['G', 'C', 'A', 'B', 'D']
print(bfs_shortest_path_distance(demo_graph, """G""", """D""")) # returns 4
| 330 |
from typing import TYPE_CHECKING
from ....utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_mmbt""": ["""MMBTConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""MMBTForClassification""", """MMBTModel""", """ModalEmbeddings"""]
if TYPE_CHECKING:
from .configuration_mmbt import MMBTConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_mmbt import MMBTForClassification, MMBTModel, ModalEmbeddings
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a_ = logging.get_logger(__name__)
a_ = {
"""microsoft/swinv2-tiny-patch4-window8-256""": (
"""https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256/resolve/main/config.json"""
),
}
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """swinv2"""
lowerCAmelCase__ = {
"""num_attention_heads""": """num_heads""",
"""num_hidden_layers""": """num_layers""",
}
def __init__( self , __UpperCAmelCase=224 , __UpperCAmelCase=4 , __UpperCAmelCase=3 , __UpperCAmelCase=96 , __UpperCAmelCase=[2, 2, 6, 2] , __UpperCAmelCase=[3, 6, 12, 24] , __UpperCAmelCase=7 , __UpperCAmelCase=4.0 , __UpperCAmelCase=True , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.1 , __UpperCAmelCase="gelu" , __UpperCAmelCase=False , __UpperCAmelCase=0.02 , __UpperCAmelCase=1E-5 , __UpperCAmelCase=32 , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = image_size
__lowerCamelCase = patch_size
__lowerCamelCase = num_channels
__lowerCamelCase = embed_dim
__lowerCamelCase = depths
__lowerCamelCase = len(__UpperCAmelCase )
__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 = encoder_stride
# we set the hidden_size attribute in order to make Swinv2 work with VisionEncoderDecoderModel
# this indicates the channel dimension after the last stage of the model
__lowerCamelCase = int(embed_dim * 2 ** (len(__UpperCAmelCase ) - 1) )
__lowerCamelCase = (0, 0, 0, 0)
| 330 |
import gc
import importlib.metadata
import tempfile
import unittest
from packaging import version
from transformers import (
AutoModel,
AutoModelForCausalLM,
AutoModelForSeqaSeqLM,
AutoModelForSequenceClassification,
AutoTokenizer,
BitsAndBytesConfig,
pipeline,
)
from transformers.testing_utils import (
is_torch_available,
require_accelerate,
require_bitsandbytes,
require_torch,
require_torch_gpu,
require_torch_multi_gpu,
slow,
)
def a__ ( _UpperCamelCase : Optional[int] ):
if model.config.model_type == "gpt2":
return model.transformer.h[0].mlp.c_fc
return model.transformer.h[0].mlp.dense_ah_to_h
if is_torch_available():
import torch
import torch.nn as nn
class __lowerCAmelCase ( nn.Module ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
super().__init__()
__lowerCamelCase = module
__lowerCamelCase = nn.Sequential(
nn.Linear(module.in_features , __UpperCAmelCase , bias=__UpperCAmelCase ) , nn.Linear(__UpperCAmelCase , module.out_features , bias=__UpperCAmelCase ) , )
__lowerCamelCase = (2.0 / (5 * min(module.in_features , module.out_features ))) ** 0.5
nn.init.normal_(self.adapter[0].weight , std=__UpperCAmelCase )
nn.init.zeros_(self.adapter[1].weight )
self.adapter.to(module.weight.device )
def lowerCamelCase ( self , __UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return self.module(__UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ) + self.adapter(__UpperCAmelCase )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class __lowerCAmelCase ( unittest.TestCase ):
# We keep the constants inside the init function and model loading inside setUp function
# We need to test on relatively large models (aka >1b parameters otherwise the quantiztion may not work as expected)
# Therefore here we use only bloom-1b3 to test our module
lowerCAmelCase__ = """bigscience/bloom-1b7"""
# Constant values
lowerCAmelCase__ = 2.1_09_65_95_52_69_25_74
lowerCAmelCase__ = """Hello my name is"""
lowerCAmelCase__ = set()
EXPECTED_OUTPUTS.add("""Hello my name is John and I am a professional photographer. I""" )
EXPECTED_OUTPUTS.add("""Hello my name is John.\nI am a friend of your father.\n""" )
EXPECTED_OUTPUTS.add("""Hello my name is John Doe, I am a student at the University""" )
lowerCAmelCase__ = 1_0
def lowerCamelCase ( self ):
'''simple docstring'''
# Models and tokenizer
__lowerCamelCase = AutoTokenizer.from_pretrained(self.model_name )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# Models and tokenizer
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , torch_dtype=torch.floataa , device_map='''auto''' )
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
def lowerCamelCase ( self ):
'''simple docstring'''
del self.model_fpaa
del self.model_abit
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_abit.config
self.assertTrue(hasattr(__UpperCAmelCase , '''quantization_config''' ) )
__lowerCamelCase = config.to_dict()
__lowerCamelCase = config.to_diff_dict()
__lowerCamelCase = config.to_json_string()
def lowerCamelCase ( self ):
'''simple docstring'''
from bitsandbytes.nn import Paramsabit
__lowerCamelCase = self.model_fpaa.get_memory_footprint()
__lowerCamelCase = self.model_abit.get_memory_footprint()
self.assertAlmostEqual(mem_fpaa / mem_abit , self.EXPECTED_RELATIVE_DIFFERENCE )
__lowerCamelCase = get_some_linear_layer(self.model_abit )
self.assertTrue(linear.weight.__class__ == Paramsabit )
def lowerCamelCase ( self ):
'''simple docstring'''
from transformers import TaPreTrainedModel
self.model_fpaa.get_memory_footprint()
self.model_abit.get_memory_footprint()
for name, module in self.model_abit.named_modules():
if isinstance(__UpperCAmelCase , torch.nn.Linear ):
if name not in ["lm_head"] + TaPreTrainedModel._keep_in_fpaa_modules:
# 4-bit parameters are packed in uint8 variables
self.assertTrue(module.weight.dtype == torch.uinta )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = self.model_abit.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_sequences[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitsAndBytesConfig()
__lowerCamelCase = True
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = model_abit_from_config.generate(
input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_sequences[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
def lowerCamelCase ( self ):
'''simple docstring'''
with self.assertRaises(__UpperCAmelCase ), tempfile.TemporaryDirectory() as tmpdirname:
self.model_abit.save_pretrained(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitsAndBytesConfig()
with self.assertRaises(__UpperCAmelCase ):
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=__UpperCAmelCase , load_in_abit=__UpperCAmelCase , device_map='''auto''' , bnb_abit_quant_type='''nf4''' , )
def lowerCamelCase ( self ):
'''simple docstring'''
with self.assertRaises(__UpperCAmelCase ):
# Tries with `str`
self.model_abit.to('''cpu''' )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `dtype``
self.model_abit.to(torch.floataa )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.to(torch.device('''cuda:0''' ) )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.float()
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.half()
# Test if we did not break anything
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = self.model_fpaa.to(torch.floataa )
__lowerCamelCase = self.model_fpaa.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.to('''cpu''' )
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.half()
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.float()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = AutoModelForSeqaSeqLM.from_pretrained('''t5-small''' , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
self.assertTrue(model.decoder.block[0].layer[2].DenseReluDense.wo.weight.dtype == torch.floataa )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class __lowerCAmelCase ( unittest.TestCase ):
@classmethod
def lowerCamelCase ( cls ):
'''simple docstring'''
__lowerCamelCase = '''t5-small'''
__lowerCamelCase = '''google/flan-t5-small''' # flan-t5 uses dense-act instead of dense-relu-dense
__lowerCamelCase = AutoTokenizer.from_pretrained(cls.model_name )
__lowerCamelCase = '''Translate in German: Hello, my dog is cute'''
def lowerCamelCase ( self ):
'''simple docstring'''
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
from transformers import TaForConditionalGeneration
__lowerCamelCase = TaForConditionalGeneration._keep_in_fpaa_modules
__lowerCamelCase = None
# test with `t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
# test with `flan-t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
__lowerCamelCase = modules
def lowerCamelCase ( self ):
'''simple docstring'''
import bitsandbytes as bnb
from transformers import TaForConditionalGeneration
# test with `t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# there was a bug with decoders - this test checks that it is fixed
self.assertTrue(isinstance(model.decoder.block[0].layer[0].SelfAttention.q , bnb.nn.Linearabit ) )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
# test with `flan-t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# model_name
__lowerCamelCase = '''bigscience/bloom-560m'''
__lowerCamelCase = '''t5-small'''
# Different types of model
__lowerCamelCase = AutoModel.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# Sequence classification model
__lowerCamelCase = AutoModelForSequenceClassification.from_pretrained(
self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# CausalLM model
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# Seq2seq model
__lowerCamelCase = AutoModelForSeqaSeqLM.from_pretrained(
self.seq_to_seq_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
def lowerCamelCase ( self ):
'''simple docstring'''
del self.base_model
del self.sequence_model
del self.model_abit
del self.seq_to_seq_model
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
from bitsandbytes.nn import Paramsabit
self.assertTrue(self.base_model.h[-1].mlp.dense_ah_to_h.weight.__class__ == Paramsabit )
# Other heads should be nn.Parameter
self.assertTrue(self.model_abit.lm_head.weight.__class__ == torch.nn.Parameter )
self.assertTrue(self.sequence_model.score.weight.__class__ == torch.nn.Parameter )
self.assertTrue(self.seq_to_seq_model.lm_head.weight.__class__ == torch.nn.Parameter )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
del self.pipe
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = pipeline(
'''text-generation''' , model=self.model_name , model_kwargs={'''device_map''': '''auto''', '''load_in_4bit''': True, '''torch_dtype''': torch.floataa} , max_new_tokens=self.MAX_NEW_TOKENS , )
# Real second forward pass
__lowerCamelCase = self.pipe(self.input_text )
self.assertIn(pipeline_output[0]['''generated_text'''] , self.EXPECTED_OUTPUTS )
@require_torch_multi_gpu
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , load_in_abit=__UpperCAmelCase , device_map='''balanced''' )
# Check correct device map
self.assertEqual(set(model_parallel.hf_device_map.values() ) , {0, 1} )
# Check that inference pass works on the model
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
# Second real batch
__lowerCamelCase = model_parallel.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_parallel[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''facebook/opt-350m'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
if version.parse(importlib.metadata.version('''bitsandbytes''' ) ) < version.parse('''0.37.0''' ):
return
# Step 1: freeze all parameters
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase )
self.assertEqual(set(model.hf_device_map.values() ) , {torch.cuda.current_device()} )
for param in model.parameters():
__lowerCamelCase = False # freeze the model - train adapters later
if param.ndim == 1:
# cast the small parameters (e.g. layernorm) to fp32 for stability
__lowerCamelCase = param.data.to(torch.floataa )
# Step 2: add adapters
for _, module in model.named_modules():
if "OPTAttention" in repr(type(__UpperCAmelCase ) ):
__lowerCamelCase = LoRALayer(module.q_proj , rank=16 )
__lowerCamelCase = LoRALayer(module.k_proj , rank=16 )
__lowerCamelCase = LoRALayer(module.v_proj , rank=16 )
# Step 3: dummy batch
__lowerCamelCase = self.tokenizer('''Test batch ''' , return_tensors='''pt''' ).to(0 )
# Step 4: Check if the gradient is not None
with torch.cuda.amp.autocast():
__lowerCamelCase = model.forward(**__UpperCAmelCase )
out.logits.norm().backward()
for module in model.modules():
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
self.assertTrue(module.adapter[1].weight.grad is not None )
self.assertTrue(module.adapter[1].weight.grad.norm().item() > 0 )
elif isinstance(__UpperCAmelCase , nn.Embedding ):
self.assertTrue(module.weight.grad is None )
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """gpt2-xl"""
lowerCAmelCase__ = 3.31_91_85_48_54_15_21_87
| 330 | 1 |
# Lint as: python3
import dataclasses
import re
from dataclasses import dataclass
from functools import total_ordering
from typing import Optional, Union
a_ = re.compile(R"""^(?P<major>\d+)""" R"""\.(?P<minor>\d+)""" R"""\.(?P<patch>\d+)$""")
@total_ordering
@dataclass
class __lowerCAmelCase :
lowerCAmelCase__ = 42
lowerCAmelCase__ = None
lowerCAmelCase__ = None
lowerCAmelCase__ = None
lowerCAmelCase__ = None
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = _str_to_version_tuple(self.version_str )
def __repr__( self ):
'''simple docstring'''
return F"""{self.tuple[0]}.{self.tuple[1]}.{self.tuple[2]}"""
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self.major, self.minor, self.patch
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
return Version(__UpperCAmelCase )
elif isinstance(__UpperCAmelCase , __UpperCAmelCase ):
return other
raise TypeError(F"""{other} (type {type(__UpperCAmelCase )}) cannot be compared to version.""" )
def __eq__( self , __UpperCAmelCase ):
'''simple docstring'''
try:
__lowerCamelCase = self._validate_operand(__UpperCAmelCase )
except (TypeError, ValueError):
return False
else:
return self.tuple == other.tuple
def __lt__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self._validate_operand(__UpperCAmelCase )
return self.tuple < other.tuple
def __hash__( self ):
'''simple docstring'''
return hash(_version_tuple_to_str(self.tuple ) )
@classmethod
def lowerCamelCase ( cls , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = {f.name for f in dataclasses.fields(cls )}
return cls(**{k: v for k, v in dic.items() if k in field_names} )
def lowerCamelCase ( self ):
'''simple docstring'''
return self.version_str
def a__ ( _UpperCamelCase : Optional[int] ):
__lowerCamelCase = _VERSION_REG.match(_UpperCamelCase )
if not res:
raise ValueError(F"""Invalid version '{version_str}'. Format should be x.y.z with {{x,y,z}} being digits.""" )
return tuple(int(_UpperCamelCase ) for v in [res.group('''major''' ), res.group('''minor''' ), res.group('''patch''' )] )
def a__ ( _UpperCamelCase : Optional[int] ):
return ".".join(str(_UpperCamelCase ) for v in version_tuple )
| 330 |
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 __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 42
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ ):
lowerCAmelCase__ = True
@register_to_config
def __init__( self , __UpperCAmelCase = 3 , __UpperCAmelCase = 3 , __UpperCAmelCase = ("DownEncoderBlock2D",) , __UpperCAmelCase = ("UpDecoderBlock2D",) , __UpperCAmelCase = (64,) , __UpperCAmelCase = 1 , __UpperCAmelCase = "silu" , __UpperCAmelCase = 4 , __UpperCAmelCase = 32 , __UpperCAmelCase = 32 , __UpperCAmelCase = 0.18_215 , ):
'''simple docstring'''
super().__init__()
# pass init params to Encoder
__lowerCamelCase = Encoder(
in_channels=__UpperCAmelCase , out_channels=__UpperCAmelCase , down_block_types=__UpperCAmelCase , block_out_channels=__UpperCAmelCase , layers_per_block=__UpperCAmelCase , act_fn=__UpperCAmelCase , norm_num_groups=__UpperCAmelCase , double_z=__UpperCAmelCase , )
# pass init params to Decoder
__lowerCamelCase = Decoder(
in_channels=__UpperCAmelCase , out_channels=__UpperCAmelCase , up_block_types=__UpperCAmelCase , block_out_channels=__UpperCAmelCase , layers_per_block=__UpperCAmelCase , norm_num_groups=__UpperCAmelCase , act_fn=__UpperCAmelCase , )
__lowerCamelCase = nn.Convad(2 * latent_channels , 2 * latent_channels , 1 )
__lowerCamelCase = nn.Convad(__UpperCAmelCase , __UpperCAmelCase , 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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
if isinstance(__UpperCAmelCase , (Encoder, Decoder) ):
__lowerCamelCase = value
def lowerCamelCase ( self , __UpperCAmelCase = True ):
'''simple docstring'''
__lowerCamelCase = use_tiling
def lowerCamelCase ( self ):
'''simple docstring'''
self.enable_tiling(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = True
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = False
@property
# Copied from diffusers.models.unet_2d_condition.UNet2DConditionModel.attn_processors
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = {}
def fn_recursive_add_processors(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
if hasattr(__UpperCAmelCase , '''set_processor''' ):
__lowerCamelCase = module.processor
for sub_name, child in module.named_children():
fn_recursive_add_processors(F"""{name}.{sub_name}""" , __UpperCAmelCase , __UpperCAmelCase )
return processors
for name, module in self.named_children():
fn_recursive_add_processors(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
return processors
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = len(self.attn_processors.keys() )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) != count:
raise ValueError(
F"""A dict of processors was passed, but the number of processors {len(__UpperCAmelCase )} does not match the"""
F""" number of attention layers: {count}. Please make sure to pass {count} processor classes.""" )
def fn_recursive_attn_processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
if hasattr(__UpperCAmelCase , '''set_processor''' ):
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
module.set_processor(__UpperCAmelCase )
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}""" , __UpperCAmelCase , __UpperCAmelCase )
for name, module in self.named_children():
fn_recursive_attn_processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
self.set_attn_processor(AttnProcessor() )
@apply_forward_hook
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
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(__UpperCAmelCase , return_dict=__UpperCAmelCase )
if self.use_slicing and x.shape[0] > 1:
__lowerCamelCase = [self.encoder(__UpperCAmelCase ) for x_slice in x.split(1 )]
__lowerCamelCase = torch.cat(__UpperCAmelCase )
else:
__lowerCamelCase = self.encoder(__UpperCAmelCase )
__lowerCamelCase = self.quant_conv(__UpperCAmelCase )
__lowerCamelCase = DiagonalGaussianDistribution(__UpperCAmelCase )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
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(__UpperCAmelCase , return_dict=__UpperCAmelCase )
__lowerCamelCase = self.post_quant_conv(__UpperCAmelCase )
__lowerCamelCase = self.decoder(__UpperCAmelCase )
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
@apply_forward_hook
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
if self.use_slicing and z.shape[0] > 1:
__lowerCamelCase = [self._decode(__UpperCAmelCase ).sample for z_slice in z.split(1 )]
__lowerCamelCase = torch.cat(__UpperCAmelCase )
else:
__lowerCamelCase = self._decode(__UpperCAmelCase ).sample
if not return_dict:
return (decoded,)
return DecoderOutput(sample=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(a.shape[2] , b.shape[2] , __UpperCAmelCase )
for y in range(__UpperCAmelCase ):
__lowerCamelCase = a[:, :, -blend_extent + y, :] * (1 - y / blend_extent) + b[:, :, y, :] * (y / blend_extent)
return b
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(a.shape[3] , b.shape[3] , __UpperCAmelCase )
for x in range(__UpperCAmelCase ):
__lowerCamelCase = a[:, :, :, -blend_extent + x] * (1 - x / blend_extent) + b[:, :, :, x] * (x / blend_extent)
return b
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
__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] , __UpperCAmelCase ):
__lowerCamelCase = []
for j in range(0 , x.shape[3] , __UpperCAmelCase ):
__lowerCamelCase = x[:, :, i : i + self.tile_sample_min_size, j : j + self.tile_sample_min_size]
__lowerCamelCase = self.encoder(__UpperCAmelCase )
__lowerCamelCase = self.quant_conv(__UpperCAmelCase )
row.append(__UpperCAmelCase )
rows.append(__UpperCAmelCase )
__lowerCamelCase = []
for i, row in enumerate(__UpperCAmelCase ):
__lowerCamelCase = []
for j, tile in enumerate(__UpperCAmelCase ):
# 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] , __UpperCAmelCase , __UpperCAmelCase )
if j > 0:
__lowerCamelCase = self.blend_h(row[j - 1] , __UpperCAmelCase , __UpperCAmelCase )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(__UpperCAmelCase , dim=3 ) )
__lowerCamelCase = torch.cat(__UpperCAmelCase , dim=2 )
__lowerCamelCase = DiagonalGaussianDistribution(__UpperCAmelCase )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
__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] , __UpperCAmelCase ):
__lowerCamelCase = []
for j in range(0 , z.shape[3] , __UpperCAmelCase ):
__lowerCamelCase = z[:, :, i : i + self.tile_latent_min_size, j : j + self.tile_latent_min_size]
__lowerCamelCase = self.post_quant_conv(__UpperCAmelCase )
__lowerCamelCase = self.decoder(__UpperCAmelCase )
row.append(__UpperCAmelCase )
rows.append(__UpperCAmelCase )
__lowerCamelCase = []
for i, row in enumerate(__UpperCAmelCase ):
__lowerCamelCase = []
for j, tile in enumerate(__UpperCAmelCase ):
# 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] , __UpperCAmelCase , __UpperCAmelCase )
if j > 0:
__lowerCamelCase = self.blend_h(row[j - 1] , __UpperCAmelCase , __UpperCAmelCase )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(__UpperCAmelCase , dim=3 ) )
__lowerCamelCase = torch.cat(__UpperCAmelCase , dim=2 )
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False , __UpperCAmelCase = True , __UpperCAmelCase = None , ):
'''simple docstring'''
__lowerCamelCase = sample
__lowerCamelCase = self.encode(__UpperCAmelCase ).latent_dist
if sample_posterior:
__lowerCamelCase = posterior.sample(generator=__UpperCAmelCase )
else:
__lowerCamelCase = posterior.mode()
__lowerCamelCase = self.decode(__UpperCAmelCase ).sample
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
| 330 | 1 |
def a__ ( _UpperCamelCase : float ,_UpperCamelCase : float ,_UpperCamelCase : int ):
if principal <= 0:
raise Exception('''Principal borrowed must be > 0''' )
if rate_per_annum < 0:
raise Exception('''Rate of interest must be >= 0''' )
if years_to_repay <= 0 or not isinstance(_UpperCamelCase ,_UpperCamelCase ):
raise Exception('''Years to repay must be an integer > 0''' )
# Yearly rate is divided by 12 to get monthly rate
__lowerCamelCase = rate_per_annum / 12
# Years to repay is multiplied by 12 to get number of payments as payment is monthly
__lowerCamelCase = years_to_repay * 12
return (
principal
* rate_per_month
* (1 + rate_per_month) ** number_of_payments
/ ((1 + rate_per_month) ** number_of_payments - 1)
)
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 |
import argparse
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import BigBirdPegasusConfig, BigBirdPegasusForConditionalGeneration
a_ = [
# tf -> hf
("""/""", """."""),
("""layer_""", """layers."""),
("""kernel""", """weight"""),
("""beta""", """bias"""),
("""gamma""", """weight"""),
("""pegasus""", """model"""),
]
a_ = [
(""".output.dense""", """.fc2"""),
("""intermediate.LayerNorm""", """final_layer_norm"""),
("""intermediate.dense""", """fc1"""),
]
a_ = (
INIT_COMMON
+ [
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.out_proj"""),
("""attention.self""", """self_attn"""),
("""attention.encdec.LayerNorm""", """encoder_attn_layer_norm"""),
("""attention.encdec_output.dense""", """encoder_attn.out_proj"""),
("""attention.encdec""", """encoder_attn"""),
("""key""", """k_proj"""),
("""value""", """v_proj"""),
("""query""", """q_proj"""),
("""decoder.LayerNorm""", """decoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = (
INIT_COMMON
+ [
("""embeddings.word_embeddings""", """shared.weight"""),
("""embeddings.position_embeddings""", """embed_positions.weight"""),
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.output"""),
("""attention.self""", """self_attn.self"""),
("""encoder.LayerNorm""", """encoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = [
"""encdec/key/bias""",
"""encdec/query/bias""",
"""encdec/value/bias""",
"""self/key/bias""",
"""self/query/bias""",
"""self/value/bias""",
"""encdec_output/dense/bias""",
"""attention/output/dense/bias""",
]
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : Optional[Any] ):
for tf_name, hf_name in patterns:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = BigBirdPegasusConfig(**_UpperCamelCase )
__lowerCamelCase = BigBirdPegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
__lowerCamelCase = {}
# separating decoder weights
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if k.startswith('''pegasus/decoder''' )}
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if not k.startswith('''pegasus/decoder''' )}
for k, v in tqdm(decoder_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = DECODER_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
for k, v in tqdm(remaining_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = REMAINING_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict and k != "pegasus/embeddings/position_embeddings":
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
if k != "pegasus/embeddings/position_embeddings":
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
__lowerCamelCase = mapping['''model.embed_positions.weight''']
__lowerCamelCase = mapping.pop('''model.embed_positions.weight''' )
__lowerCamelCase ,__lowerCamelCase = torch_model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k
for k in missing
if k
not in [
'''final_logits_bias''',
'''model.encoder.embed_tokens.weight''',
'''model.decoder.embed_tokens.weight''',
'''lm_head.weight''',
]
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ,_UpperCamelCase : dict ):
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = convert_bigbird_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""--save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
a_ = {}
convert_bigbird_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir, config_update=config_update)
| 330 | 1 |
import dataclasses
import json
import warnings
from dataclasses import dataclass, field
from time import time
from typing import List
from ..utils import logging
a_ = logging.get_logger(__name__)
def a__ ( _UpperCamelCase : Optional[int]=None ,_UpperCamelCase : str=None ):
return field(default_factory=lambda: default ,metadata=_UpperCamelCase )
@dataclass
class __lowerCAmelCase :
lowerCAmelCase__ = list_field(
default=[] , metadata={
"""help""": (
"""Model checkpoints to be provided to the AutoModel classes. Leave blank to benchmark the base version"""
""" of all available models"""
)
} , )
lowerCAmelCase__ = list_field(
default=[8] , metadata={"""help""": """List of batch sizes for which memory and time performance will be evaluated"""} )
lowerCAmelCase__ = list_field(
default=[8, 3_2, 1_2_8, 5_1_2] , metadata={"""help""": """List of sequence lengths for which memory and time performance will be evaluated"""} , )
lowerCAmelCase__ = field(
default=lowerCAmelCase__ , metadata={"""help""": """Whether to benchmark inference of model. Inference can be disabled via --no-inference."""} , )
lowerCAmelCase__ = field(
default=lowerCAmelCase__ , metadata={"""help""": """Whether to run on available cuda devices. Cuda can be disabled via --no-cuda."""} , )
lowerCAmelCase__ = field(
default=lowerCAmelCase__ , metadata={"""help""": """Whether to run on available tpu devices. TPU can be disabled via --no-tpu."""} )
lowerCAmelCase__ = field(default=lowerCAmelCase__ , metadata={"""help""": """Use FP16 to accelerate inference."""} )
lowerCAmelCase__ = field(default=lowerCAmelCase__ , metadata={"""help""": """Benchmark training of model"""} )
lowerCAmelCase__ = field(default=lowerCAmelCase__ , metadata={"""help""": """Verbose memory tracing"""} )
lowerCAmelCase__ = field(
default=lowerCAmelCase__ , metadata={"""help""": """Whether to perform speed measurements. Speed measurements can be disabled via --no-speed."""} , )
lowerCAmelCase__ = field(
default=lowerCAmelCase__ , metadata={
"""help""": """Whether to perform memory measurements. Memory measurements can be disabled via --no-memory"""
} , )
lowerCAmelCase__ = field(default=lowerCAmelCase__ , metadata={"""help""": """Trace memory line by line"""} )
lowerCAmelCase__ = field(default=lowerCAmelCase__ , metadata={"""help""": """Save result to a CSV file"""} )
lowerCAmelCase__ = field(default=lowerCAmelCase__ , metadata={"""help""": """Save all print statements in a log file"""} )
lowerCAmelCase__ = field(default=lowerCAmelCase__ , metadata={"""help""": """Whether to print environment information"""} )
lowerCAmelCase__ = field(
default=lowerCAmelCase__ , metadata={
"""help""": (
"""Whether to use multiprocessing for memory and speed measurement. It is highly recommended to use"""
""" multiprocessing for accurate CPU and GPU memory measurements. This option should only be disabled"""
""" for debugging / testing and on TPU."""
)
} , )
lowerCAmelCase__ = field(
default=f"inference_time_{round(time() )}.csv" , metadata={"""help""": """CSV filename used if saving time results to csv."""} , )
lowerCAmelCase__ = field(
default=f"inference_memory_{round(time() )}.csv" , metadata={"""help""": """CSV filename used if saving memory results to csv."""} , )
lowerCAmelCase__ = field(
default=f"train_time_{round(time() )}.csv" , metadata={"""help""": """CSV filename used if saving time results to csv for training."""} , )
lowerCAmelCase__ = field(
default=f"train_memory_{round(time() )}.csv" , metadata={"""help""": """CSV filename used if saving memory results to csv for training."""} , )
lowerCAmelCase__ = field(
default=f"env_info_{round(time() )}.csv" , metadata={"""help""": """CSV filename used if saving environment information."""} , )
lowerCAmelCase__ = field(
default=f"log_{round(time() )}.csv" , metadata={"""help""": """Log filename used if print statements are saved in log."""} , )
lowerCAmelCase__ = field(default=3 , metadata={"""help""": """Times an experiment will be run."""} )
lowerCAmelCase__ = field(
default=lowerCAmelCase__ , metadata={
"""help""": (
"""Instead of loading the model as defined in `config.architectures` if exists, just load the pretrain"""
""" model weights."""
)
} , )
def lowerCamelCase ( self ):
'''simple docstring'''
warnings.warn(
F"""The class {self.__class__} is deprecated. Hugging Face Benchmarking utils"""
''' are deprecated in general and it is advised to use external Benchmarking libraries '''
''' to benchmark Transformer models.''' , __UpperCAmelCase , )
def lowerCamelCase ( self ):
'''simple docstring'''
return json.dumps(dataclasses.asdict(self ) , indent=2 )
@property
def lowerCamelCase ( self ):
'''simple docstring'''
if len(self.models ) <= 0:
raise ValueError(
'''Please make sure you provide at least one model name / model identifier, *e.g.* `--models'''
''' bert-base-cased` or `args.models = [\'bert-base-cased\'].''' )
return self.models
@property
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.multi_process:
return False
elif self.is_tpu:
logger.info('''Multiprocessing is currently not possible on TPU.''' )
return False
else:
return True
| 330 |
import uuid
from typing import Any, Dict, List, Optional, Union
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
if is_torch_available():
import torch
a_ = logging.get_logger(__name__)
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=None , __UpperCAmelCase=None ):
'''simple docstring'''
if not conversation_id:
__lowerCamelCase = uuid.uuida()
if past_user_inputs is None:
__lowerCamelCase = []
if generated_responses is None:
__lowerCamelCase = []
__lowerCamelCase = conversation_id
__lowerCamelCase = past_user_inputs
__lowerCamelCase = generated_responses
__lowerCamelCase = text
def __eq__( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
return False
if self.uuid == other.uuid:
return True
return (
self.new_user_input == other.new_user_input
and self.past_user_inputs == other.past_user_inputs
and self.generated_responses == other.generated_responses
)
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False ):
'''simple docstring'''
if self.new_user_input:
if overwrite:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" was overwritten """
F"""with: \"{text}\".""" )
__lowerCamelCase = text
else:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" new input """
F"""ignored: \"{text}\". Set `overwrite` to True to overwrite unprocessed user input""" )
else:
__lowerCamelCase = text
def lowerCamelCase ( self ):
'''simple docstring'''
if self.new_user_input:
self.past_user_inputs.append(self.new_user_input )
__lowerCamelCase = None
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
self.generated_responses.append(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
for user_input, generated_response in zip(self.past_user_inputs , self.generated_responses ):
yield True, user_input
yield False, generated_response
if self.new_user_input:
yield True, self.new_user_input
def __repr__( self ):
'''simple docstring'''
__lowerCamelCase = F"""Conversation id: {self.uuid} \n"""
for is_user, text in self.iter_texts():
__lowerCamelCase = '''user''' if is_user else '''bot'''
output += F"""{name} >> {text} \n"""
return output
@add_end_docstrings(
lowerCAmelCase__ , r"""
min_length_for_response (`int`, *optional*, defaults to 32):
The minimum length (in number of tokens) for a response.
minimum_tokens (`int`, *optional*, defaults to 10):
The minimum length of tokens to leave for a response.
""" , )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
if self.tokenizer.pad_token_id is None:
__lowerCamelCase = self.tokenizer.eos_token
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = {}
__lowerCamelCase = {}
__lowerCamelCase = {}
if min_length_for_response is not None:
__lowerCamelCase = min_length_for_response
if minimum_tokens is not None:
__lowerCamelCase = minimum_tokens
if "max_length" in generate_kwargs:
__lowerCamelCase = generate_kwargs['''max_length''']
# self.max_length = generate_kwargs.get("max_length", self.model.config.max_length)
if clean_up_tokenization_spaces is not None:
__lowerCamelCase = clean_up_tokenization_spaces
if generate_kwargs:
forward_params.update(__UpperCAmelCase )
return preprocess_params, forward_params, postprocess_params
def __call__( self , __UpperCAmelCase , __UpperCAmelCase=0 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = super().__call__(__UpperCAmelCase , num_workers=__UpperCAmelCase , **__UpperCAmelCase )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) == 1:
return outputs[0]
return outputs
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=32 ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError('''ConversationalPipeline, expects Conversation as inputs''' )
if conversation.new_user_input is None:
raise ValueError(
F"""Conversation with UUID {type(conversation.uuid )} does not contain new user input to process. """
'''Add user inputs with the conversation\'s `add_user_input` method''' )
if hasattr(self.tokenizer , '''_build_conversation_input_ids''' ):
__lowerCamelCase = self.tokenizer._build_conversation_input_ids(__UpperCAmelCase )
else:
# If the tokenizer cannot handle conversations, we default to only the old version
__lowerCamelCase = self._legacy_parse_and_tokenize(__UpperCAmelCase )
if self.framework == "pt":
__lowerCamelCase = torch.LongTensor([input_ids] )
elif self.framework == "tf":
__lowerCamelCase = tf.constant([input_ids] )
return {"input_ids": input_ids, "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=10 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = generate_kwargs.get('''max_length''' , self.model.config.max_length )
__lowerCamelCase = model_inputs['''input_ids'''].shape[1]
if max_length - minimum_tokens < n:
logger.warning(F"""Conversation input is to long ({n}), trimming it to ({max_length} - {minimum_tokens})""" )
__lowerCamelCase = max_length - minimum_tokens
__lowerCamelCase = model_inputs['''input_ids'''][:, -trim:]
if "attention_mask" in model_inputs:
__lowerCamelCase = model_inputs['''attention_mask'''][:, -trim:]
__lowerCamelCase = model_inputs.pop('''conversation''' )
__lowerCamelCase = max_length
__lowerCamelCase = self.model.generate(**__UpperCAmelCase , **__UpperCAmelCase )
if self.model.config.is_encoder_decoder:
__lowerCamelCase = 1
else:
__lowerCamelCase = n
return {"output_ids": output_ids[:, start_position:], "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = model_outputs['''output_ids''']
__lowerCamelCase = self.tokenizer.decode(
output_ids[0] , skip_special_tokens=__UpperCAmelCase , clean_up_tokenization_spaces=__UpperCAmelCase , )
__lowerCamelCase = model_outputs['''conversation''']
conversation.mark_processed()
conversation.append_response(__UpperCAmelCase )
return conversation
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer.eos_token_id
__lowerCamelCase = []
for is_user, text in conversation.iter_texts():
if eos_token_id is not None:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) + [eos_token_id] )
else:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) )
if len(__UpperCAmelCase ) > self.tokenizer.model_max_length:
__lowerCamelCase = input_ids[-self.tokenizer.model_max_length :]
return input_ids
| 330 | 1 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
a_ = {
"""configuration_lxmert""": ["""LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP""", """LxmertConfig"""],
"""tokenization_lxmert""": ["""LxmertTokenizer"""],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""LxmertTokenizerFast"""]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = [
"""LxmertEncoder""",
"""LxmertForPreTraining""",
"""LxmertForQuestionAnswering""",
"""LxmertModel""",
"""LxmertPreTrainedModel""",
"""LxmertVisualFeatureEncoder""",
"""LxmertXLayer""",
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = [
"""TF_LXMERT_PRETRAINED_MODEL_ARCHIVE_LIST""",
"""TFLxmertForPreTraining""",
"""TFLxmertMainLayer""",
"""TFLxmertModel""",
"""TFLxmertPreTrainedModel""",
"""TFLxmertVisualFeatureEncoder""",
]
if TYPE_CHECKING:
from .configuration_lxmert import LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP, LxmertConfig
from .tokenization_lxmert import LxmertTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_lxmert_fast import LxmertTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_lxmert import (
LxmertEncoder,
LxmertForPreTraining,
LxmertForQuestionAnswering,
LxmertModel,
LxmertPreTrainedModel,
LxmertVisualFeatureEncoder,
LxmertXLayer,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_lxmert import (
TF_LXMERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLxmertForPreTraining,
TFLxmertMainLayer,
TFLxmertModel,
TFLxmertPreTrainedModel,
TFLxmertVisualFeatureEncoder,
)
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 |
import argparse
import os
from pathlib import Path
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import PegasusConfig, PegasusForConditionalGeneration, PegasusTokenizer
from transformers.models.pegasus.configuration_pegasus import DEFAULTS, task_specific_params
a_ = [
# replace left string with right string to get the relevant state_dict key (identical state dict to bart)
["""memory_attention""", """encoder_attn"""],
["""attention""", """attn"""],
["""/""", """."""],
[""".LayerNorm.gamma""", """_layer_norm.weight"""],
[""".LayerNorm.beta""", """_layer_norm.bias"""],
["""r.layer_""", """r.layers."""],
["""output_proj""", """out_proj"""],
["""ffn.dense_1.""", """fc2."""],
["""ffn.dense.""", """fc1."""],
["""ffn_layer_norm""", """final_layer_norm"""],
["""kernel""", """weight"""],
["""encoder_layer_norm.""", """encoder.layer_norm."""],
["""decoder_layer_norm.""", """decoder.layer_norm."""],
["""embeddings.weights""", """shared.weight"""],
]
def a__ ( _UpperCamelCase : int ):
for pegasus_name, hf_name in PATTERNS:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = DEFAULTS.copy()
cfg_kwargs.update(_UpperCamelCase )
__lowerCamelCase = PegasusConfig(**_UpperCamelCase )
__lowerCamelCase = PegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.model.state_dict()
__lowerCamelCase = {}
for k, v in tf_weights.items():
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase )
if new_k not in sd:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if "dense" in k or "proj" in new_k:
__lowerCamelCase = v.T
__lowerCamelCase = torch.tensor(_UpperCamelCase ,dtype=sd[new_k].dtype )
assert v.shape == sd[new_k].shape, F"""{new_k}, {k}, {v.shape}, {sd[new_k].shape}"""
# make sure embedding.padding_idx is respected
__lowerCamelCase = torch.zeros_like(mapping['''shared.weight'''][cfg.pad_token_id + 1] )
__lowerCamelCase = mapping['''shared.weight''']
__lowerCamelCase = mapping['''shared.weight''']
__lowerCamelCase = {k: torch.zeros_like(_UpperCamelCase ) for k, v in sd.items() if k.endswith('''bias''' ) and k not in mapping}
mapping.update(**_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = torch_model.model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k for k in missing if k not in ['''encoder.embed_positions.weight''', '''decoder.embed_positions.weight''']
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : str="./ckpt/aeslc/model.ckpt-32000" ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''Adafactor''', '''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# save tokenizer first
__lowerCamelCase = Path(_UpperCamelCase ).parent.name
__lowerCamelCase = task_specific_params[F"""summarization_{dataset}"""]['''max_position_embeddings''']
__lowerCamelCase = PegasusTokenizer.from_pretrained('''sshleifer/pegasus''' ,model_max_length=_UpperCamelCase )
assert tok.model_max_length == desired_max_model_length
tok.save_pretrained(_UpperCamelCase )
# convert model
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = task_specific_params[F"""summarization_{dataset}"""]
if dataset == "large":
__lowerCamelCase = task_specific_params
__lowerCamelCase = convert_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
sd.pop('''model.decoder.embed_positions.weight''' )
sd.pop('''model.encoder.embed_positions.weight''' )
torch.save(_UpperCamelCase ,Path(_UpperCamelCase ) / '''pytorch_model.bin''' )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument("""tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
if args.save_dir is None:
a_ = Path(args.tf_ckpt_path).parent.name
a_ = os.path.join("""pegasus""", dataset)
convert_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir)
| 330 | 1 |
import argparse
import collections
import numpy as np
import torch
from flax import traverse_util
from tax import checkpoints
from transformers import MTaConfig, UMTaEncoderModel, UMTaForConditionalGeneration
from transformers.utils import logging
logging.set_verbosity_info()
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : str ,_UpperCamelCase : List[str] ):
return params[F"""{prefix}/{prefix}/relpos_bias/rel_embedding"""][:, i, :]
def a__ ( _UpperCamelCase : int ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Tuple ,_UpperCamelCase : Any="attention" ):
__lowerCamelCase = __lowerCamelCase = np.ascontiguousarray(params[F"""{prefix}/{prefix}/{layer_name}/key/kernel"""][:, i, :, :] )
__lowerCamelCase = k_tmp.reshape(k_tmp.shape[0] ,k_tmp.shape[1] * k_tmp.shape[2] )
__lowerCamelCase = np.ascontiguousarray(params[F"""{prefix}/{prefix}/{layer_name}/out/kernel"""][:, i, :, :] )
__lowerCamelCase = o_tmp.reshape(o_tmp.shape[0] * o_tmp.shape[1] ,o_tmp.shape[2] )
__lowerCamelCase = np.ascontiguousarray(params[F"""{prefix}/{prefix}/{layer_name}/query/kernel"""][:, i, :, :] )
__lowerCamelCase = q_tmp.reshape(q_tmp.shape[0] ,q_tmp.shape[1] * q_tmp.shape[2] )
__lowerCamelCase = np.ascontiguousarray(params[F"""{prefix}/{prefix}/{layer_name}/value/kernel"""][:, i, :, :] )
__lowerCamelCase = v_tmp.reshape(v_tmp.shape[0] ,v_tmp.shape[1] * v_tmp.shape[2] )
return k, o, q, v
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : List[Any] ,_UpperCamelCase : List[Any] ,_UpperCamelCase : Optional[int]=False ):
if split_mlp_wi:
__lowerCamelCase = params[F"""{prefix}/{prefix}/mlp/wi_0/kernel"""][:, i, :]
__lowerCamelCase = params[F"""{prefix}/{prefix}/mlp/wi_1/kernel"""][:, i, :]
__lowerCamelCase = (wi_a, wi_a)
else:
__lowerCamelCase = params[F"""{prefix}/{prefix}/mlp/wi/kernel"""][:, i, :]
__lowerCamelCase = params[F"""{prefix}/{prefix}/mlp/wo/kernel"""][:, i, :]
return wi, wo
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : Any ,_UpperCamelCase : List[str] ,_UpperCamelCase : Optional[int] ):
return params[F"""{prefix}/{prefix}/{layer_name}/scale"""][:, i]
def a__ ( _UpperCamelCase : dict ,*, _UpperCamelCase : int ,_UpperCamelCase : bool ,_UpperCamelCase : bool = False ):
__lowerCamelCase = traverse_util.flatten_dict(variables['''target'''] )
__lowerCamelCase = {'''/'''.join(_UpperCamelCase ): v for k, v in old.items()}
# v1.1 models have a gated GeLU with wi_0 and wi_1 instead of wi
__lowerCamelCase = '''encoder/encoder/mlp/wi_0/kernel''' in old
print('''Split MLP:''' ,_UpperCamelCase )
__lowerCamelCase = collections.OrderedDict()
# Shared embeddings.
__lowerCamelCase = old['''token_embedder/embedding''']
# Encoder.
for i in range(_UpperCamelCase ):
# Block i, layer 0 (Self Attention).
__lowerCamelCase = tax_layer_norm_lookup(_UpperCamelCase ,_UpperCamelCase ,'''encoder''' ,'''pre_attention_layer_norm''' )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = tax_attention_lookup(_UpperCamelCase ,_UpperCamelCase ,'''encoder''' ,'''attention''' )
__lowerCamelCase = layer_norm
__lowerCamelCase = k.T
__lowerCamelCase = o.T
__lowerCamelCase = q.T
__lowerCamelCase = v.T
# Block i, layer 1 (MLP).
__lowerCamelCase = tax_layer_norm_lookup(_UpperCamelCase ,_UpperCamelCase ,'''encoder''' ,'''pre_mlp_layer_norm''' )
__lowerCamelCase ,__lowerCamelCase = tax_mlp_lookup(_UpperCamelCase ,_UpperCamelCase ,'''encoder''' ,_UpperCamelCase )
__lowerCamelCase = layer_norm
if split_mlp_wi:
__lowerCamelCase = wi[0].T
__lowerCamelCase = wi[1].T
else:
__lowerCamelCase = wi.T
__lowerCamelCase = wo.T
if scalable_attention:
# convert the rel_embedding of each layer
__lowerCamelCase = tax_relpos_bias_lookup(
_UpperCamelCase ,_UpperCamelCase ,'''encoder''' ).T
__lowerCamelCase = old['''encoder/encoder_norm/scale''']
if not scalable_attention:
__lowerCamelCase = tax_relpos_bias_lookup(
_UpperCamelCase ,0 ,'''encoder''' ).T
__lowerCamelCase = tax_relpos_bias_lookup(
_UpperCamelCase ,0 ,'''decoder''' ).T
if not is_encoder_only:
# Decoder.
for i in range(_UpperCamelCase ):
# Block i, layer 0 (Self Attention).
__lowerCamelCase = tax_layer_norm_lookup(_UpperCamelCase ,_UpperCamelCase ,'''decoder''' ,'''pre_self_attention_layer_norm''' )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = tax_attention_lookup(_UpperCamelCase ,_UpperCamelCase ,'''decoder''' ,'''self_attention''' )
__lowerCamelCase = layer_norm
__lowerCamelCase = k.T
__lowerCamelCase = o.T
__lowerCamelCase = q.T
__lowerCamelCase = v.T
# Block i, layer 1 (Cross Attention).
__lowerCamelCase = tax_layer_norm_lookup(_UpperCamelCase ,_UpperCamelCase ,'''decoder''' ,'''pre_cross_attention_layer_norm''' )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = tax_attention_lookup(_UpperCamelCase ,_UpperCamelCase ,'''decoder''' ,'''encoder_decoder_attention''' )
__lowerCamelCase = layer_norm
__lowerCamelCase = k.T
__lowerCamelCase = o.T
__lowerCamelCase = q.T
__lowerCamelCase = v.T
# Block i, layer 2 (MLP).
__lowerCamelCase = tax_layer_norm_lookup(_UpperCamelCase ,_UpperCamelCase ,'''decoder''' ,'''pre_mlp_layer_norm''' )
__lowerCamelCase ,__lowerCamelCase = tax_mlp_lookup(_UpperCamelCase ,_UpperCamelCase ,'''decoder''' ,_UpperCamelCase )
__lowerCamelCase = layer_norm
if split_mlp_wi:
__lowerCamelCase = wi[0].T
__lowerCamelCase = wi[1].T
else:
__lowerCamelCase = wi.T
__lowerCamelCase = wo.T
if scalable_attention:
# convert the rel_embedding of each layer
__lowerCamelCase = tax_relpos_bias_lookup(_UpperCamelCase ,_UpperCamelCase ,'''decoder''' ).T
__lowerCamelCase = old['''decoder/decoder_norm/scale''']
# LM Head (only in v1.1 checkpoints, in v1.0 embeddings are used instead)
if "decoder/logits_dense/kernel" in old:
__lowerCamelCase = old['''decoder/logits_dense/kernel'''].T
return new
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : bool ):
__lowerCamelCase = collections.OrderedDict([(k, torch.from_numpy(v.copy() )) for (k, v) in converted_params.items()] )
# Add what is missing.
if "encoder.embed_tokens.weight" not in state_dict:
__lowerCamelCase = state_dict['''shared.weight''']
if not is_encoder_only:
if "decoder.embed_tokens.weight" not in state_dict:
__lowerCamelCase = state_dict['''shared.weight''']
if "lm_head.weight" not in state_dict: # For old 1.0 models.
print('''Using shared word embeddings as lm_head.''' )
__lowerCamelCase = state_dict['''shared.weight''']
return state_dict
def a__ ( _UpperCamelCase : int ,_UpperCamelCase : str ,_UpperCamelCase : Any ,_UpperCamelCase : int ,_UpperCamelCase : str ):
__lowerCamelCase = checkpoints.load_tax_checkpoint(_UpperCamelCase )
__lowerCamelCase = convert_tax_to_pytorch(
_UpperCamelCase ,num_layers=config.num_layers ,is_encoder_only=_UpperCamelCase ,scalable_attention=_UpperCamelCase )
__lowerCamelCase = make_state_dict(_UpperCamelCase ,_UpperCamelCase )
model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
def a__ ( _UpperCamelCase : int ,_UpperCamelCase : str ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : bool = False ,_UpperCamelCase : bool = False ,):
__lowerCamelCase = MTaConfig.from_json_file(_UpperCamelCase )
print(F"""Building PyTorch model from configuration: {config}""" )
# Non-v1.1 checkpoints could also use T5Model, but this works for all.
# The v1.0 checkpoints will simply have an LM head that is the word embeddings.
if is_encoder_only:
__lowerCamelCase = UMTaEncoderModel(_UpperCamelCase )
else:
__lowerCamelCase = UMTaForConditionalGeneration(_UpperCamelCase )
# Load weights from tf checkpoint
load_tax_weights_in_ta(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Save pytorch-model
print(F"""Save PyTorch model to {pytorch_dump_path}""" )
model.save_pretrained(_UpperCamelCase )
# Verify that we can load the checkpoint.
model.from_pretrained(_UpperCamelCase )
print('''Done''' )
if __name__ == "__main__":
a_ = argparse.ArgumentParser(description="""Converts a native T5X checkpoint into a PyTorch checkpoint.""")
# Required parameters
parser.add_argument(
"""--t5x_checkpoint_path""", default=None, type=str, required=True, help="""Path to the T5X checkpoint."""
)
parser.add_argument(
"""--config_file""",
default=None,
type=str,
required=True,
help="""The config json file corresponding to the pre-trained T5 model.\nThis specifies the model architecture.""",
)
parser.add_argument(
"""--pytorch_dump_path""", default=None, type=str, required=True, help="""Path to the output PyTorch model."""
)
parser.add_argument(
"""--is_encoder_only""", action="""store_true""", help="""Check if the model is encoder-decoder model""", default=False
)
parser.add_argument(
"""--scalable_attention""",
action="""store_true""",
help="""Whether the model uses scaled attention (umt5 model)""",
default=False,
)
a_ = parser.parse_args()
convert_tax_checkpoint_to_pytorch(
args.tax_checkpoint_path,
args.config_file,
args.pytorch_dump_path,
args.is_encoder_only,
args.scalable_attention,
)
| 330 |
import argparse
import fairseq
import torch
from transformers import UniSpeechSatConfig, UniSpeechSatForCTC, UniSpeechSatForPreTraining, logging
logging.set_verbosity_info()
a_ = logging.get_logger(__name__)
a_ = {
"""post_extract_proj""": """feature_projection.projection""",
"""encoder.pos_conv.0""": """encoder.pos_conv_embed.conv""",
"""self_attn.k_proj""": """encoder.layers.*.attention.k_proj""",
"""self_attn.v_proj""": """encoder.layers.*.attention.v_proj""",
"""self_attn.q_proj""": """encoder.layers.*.attention.q_proj""",
"""self_attn.out_proj""": """encoder.layers.*.attention.out_proj""",
"""self_attn_layer_norm""": """encoder.layers.*.layer_norm""",
"""fc1""": """encoder.layers.*.feed_forward.intermediate_dense""",
"""fc2""": """encoder.layers.*.feed_forward.output_dense""",
"""final_layer_norm""": """encoder.layers.*.final_layer_norm""",
"""encoder.layer_norm""": """encoder.layer_norm""",
"""encoder.layer_norm_for_extract""": """layer_norm_for_extract""",
"""w2v_model.layer_norm""": """feature_projection.layer_norm""",
"""quantizer.weight_proj""": """quantizer.weight_proj""",
"""quantizer.vars""": """quantizer.codevectors""",
"""project_q""": """project_q""",
"""final_proj""": """project_hid""",
"""w2v_encoder.proj""": """lm_head""",
"""label_embs_concat""": """label_embeddings_concat""",
"""mask_emb""": """masked_spec_embed""",
"""spk_proj""": """speaker_proj""",
}
a_ = [
"""lm_head""",
"""quantizer.weight_proj""",
"""quantizer.codevectors""",
"""project_q""",
"""project_hid""",
"""label_embeddings_concat""",
"""speaker_proj""",
"""layer_norm_for_extract""",
]
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : Tuple ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Dict ):
for attribute in key.split('''.''' ):
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if weight_type is not None:
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase ).shape
else:
__lowerCamelCase = hf_pointer.shape
if hf_shape != value.shape:
raise ValueError(
F"""Shape of hf {key + "." + weight_type if weight_type is not None else ""} is {hf_shape}, but should be"""
F""" {value.shape} for {full_name}""" )
if weight_type == "weight":
__lowerCamelCase = value
elif weight_type == "weight_g":
__lowerCamelCase = value
elif weight_type == "weight_v":
__lowerCamelCase = value
elif weight_type == "bias":
__lowerCamelCase = value
else:
__lowerCamelCase = value
logger.info(F"""{key + "." + weight_type if weight_type is not None else ""} was initialized from {full_name}.""" )
def a__ ( _UpperCamelCase : Any ,_UpperCamelCase : Any ):
__lowerCamelCase = []
__lowerCamelCase = fairseq_model.state_dict()
__lowerCamelCase = hf_model.unispeech_sat.feature_extractor
for name, value in fairseq_dict.items():
__lowerCamelCase = False
if "conv_layers" in name:
load_conv_layer(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,hf_model.config.feat_extract_norm == '''group''' ,)
__lowerCamelCase = True
else:
for key, mapped_key in MAPPING.items():
__lowerCamelCase = '''unispeech_sat.''' + mapped_key if mapped_key not in TOP_LEVEL_KEYS else mapped_key
if key in name or key.split('''w2v_model.''' )[-1] == name.split('''.''' )[0]:
if "layer_norm_for_extract" in name and (".".join(name.split('''.''' )[:-1] ) != key):
# special case since naming is very similar
continue
__lowerCamelCase = True
if "*" in mapped_key:
__lowerCamelCase = name.split(_UpperCamelCase )[0].split('''.''' )[-2]
__lowerCamelCase = mapped_key.replace('''*''' ,_UpperCamelCase )
if "weight_g" in name:
__lowerCamelCase = '''weight_g'''
elif "weight_v" in name:
__lowerCamelCase = '''weight_v'''
elif "bias" in name:
__lowerCamelCase = '''bias'''
elif "weight" in name:
# TODO: don't match quantizer.weight_proj
__lowerCamelCase = '''weight'''
else:
__lowerCamelCase = None
set_recursively(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
continue
if not is_used:
unused_weights.append(_UpperCamelCase )
logger.warning(F"""Unused weights: {unused_weights}""" )
def a__ ( _UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Dict ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = full_name.split('''conv_layers.''' )[-1]
__lowerCamelCase = name.split('''.''' )
__lowerCamelCase = int(items[0] )
__lowerCamelCase = int(items[1] )
if type_id == 0:
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif (type_id == 2 and not use_group_norm) or (type_id == 2 and layer_id == 0 and use_group_norm):
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(_UpperCamelCase )
@torch.no_grad()
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : List[str] ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : List[Any]=True ):
if config_path is not None:
__lowerCamelCase = UniSpeechSatConfig.from_pretrained(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatConfig()
__lowerCamelCase = ''''''
if is_finetuned:
__lowerCamelCase = UniSpeechSatForCTC(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatForPreTraining(_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
__lowerCamelCase = model[0].eval()
recursively_load_weights(_UpperCamelCase ,_UpperCamelCase )
hf_wavavec.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument("""--checkpoint_path""", default=None, type=str, help="""Path to fairseq checkpoint""")
parser.add_argument("""--dict_path""", default=None, type=str, help="""Path to dict of fine-tuned model""")
parser.add_argument("""--config_path""", default=None, type=str, help="""Path to hf config.json of model to convert""")
parser.add_argument(
"""--not_finetuned""", action="""store_true""", help="""Whether the model to convert is a fine-tuned model or not"""
)
a_ = parser.parse_args()
convert_unispeech_sat_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 330 | 1 |
import os
import time
import pytest
from datasets.utils.filelock import FileLock, Timeout
def a__ ( _UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = FileLock(str(tmpdir / '''foo.lock''' ) )
__lowerCamelCase = FileLock(str(tmpdir / '''foo.lock''' ) )
__lowerCamelCase = 0.01
with locka.acquire():
with pytest.raises(_UpperCamelCase ):
__lowerCamelCase = time.time()
locka.acquire(_UpperCamelCase )
assert time.time() - _start > timeout
def a__ ( _UpperCamelCase : Tuple ):
__lowerCamelCase = '''a''' * 10_00 + '''.lock'''
__lowerCamelCase = FileLock(str(tmpdir / filename ) )
assert locka._lock_file.endswith('''.lock''' )
assert not locka._lock_file.endswith(_UpperCamelCase )
assert len(os.path.basename(locka._lock_file ) ) <= 2_55
__lowerCamelCase = FileLock(tmpdir / filename )
with locka.acquire():
with pytest.raises(_UpperCamelCase ):
locka.acquire(0 )
| 330 |
from typing import List, Union
import numpy as np
from ..utils import add_end_docstrings, is_torch_available, is_vision_available, logging, requires_backends
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_torch_available():
import torch
from ..models.auto.modeling_auto import MODEL_FOR_DEPTH_ESTIMATION_MAPPING
a_ = logging.get_logger(__name__)
@add_end_docstrings(lowerCAmelCase__ )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
requires_backends(self , '''vision''' )
self.check_model_type(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return super().__call__(__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
return {}, {}, {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = load_image(__UpperCAmelCase )
__lowerCamelCase = image.size
__lowerCamelCase = self.image_processor(images=__UpperCAmelCase , return_tensors=self.framework )
return model_inputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.model(**__UpperCAmelCase )
return model_outputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = model_outputs.predicted_depth
__lowerCamelCase = torch.nn.functional.interpolate(
predicted_depth.unsqueeze(1 ) , size=self.image_size[::-1] , mode='''bicubic''' , align_corners=__UpperCAmelCase )
__lowerCamelCase = prediction.squeeze().cpu().numpy()
__lowerCamelCase = (output * 255 / np.max(__UpperCAmelCase )).astype('''uint8''' )
__lowerCamelCase = Image.fromarray(__UpperCAmelCase )
__lowerCamelCase = {}
__lowerCamelCase = predicted_depth
__lowerCamelCase = depth
return output_dict
| 330 | 1 |
from string import ascii_lowercase, ascii_uppercase
def a__ ( _UpperCamelCase : str ):
if not sentence:
return ""
__lowerCamelCase = dict(zip(_UpperCamelCase ,_UpperCamelCase ) )
return lower_to_upper.get(sentence[0] ,sentence[0] ) + sentence[1:]
if __name__ == "__main__":
from doctest import testmod
testmod()
| 330 |
from typing import Dict, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
convert_to_rgb,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
OPENAI_CLIP_MEAN,
OPENAI_CLIP_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_vision_available, logging
a_ = logging.get_logger(__name__)
if is_vision_available():
import PIL
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""pixel_values"""]
def __init__( self , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = True , __UpperCAmelCase = 1 / 255 , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = True , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = size if size is not None else {'''shortest_edge''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
__lowerCamelCase = crop_size if crop_size is not None else {'''height''': 224, '''width''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase , 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 OPENAI_CLIP_MEAN
__lowerCamelCase = image_std if image_std is not None else OPENAI_CLIP_STD
__lowerCamelCase = do_convert_rgb
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
if "shortest_edge" not in size:
raise ValueError(F"""The `size` parameter must contain the key `shortest_edge`. Got {size.keys()}""" )
__lowerCamelCase = get_resize_output_image_size(__UpperCAmelCase , size=size['''shortest_edge'''] , default_to_square=__UpperCAmelCase )
return resize(__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase )
if "height" not in size or "width" not in size:
raise ValueError(F"""The `size` parameter must contain the keys (height, width). Got {size.keys()}""" )
return center_crop(__UpperCAmelCase , size=(size['''height'''], size['''width''']) , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return rescale(__UpperCAmelCase , scale=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return normalize(__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = ChannelDimension.FIRST , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = do_resize if do_resize is not None else self.do_resize
__lowerCamelCase = size if size is not None else self.size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''size''' , default_to_square=__UpperCAmelCase )
__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 = crop_size if crop_size is not None else self.crop_size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''crop_size''' , default_to_square=__UpperCAmelCase )
__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 = do_convert_rgb if do_convert_rgb is not None else self.do_convert_rgb
__lowerCamelCase = make_list_of_images(__UpperCAmelCase )
if not valid_images(__UpperCAmelCase ):
raise ValueError(
'''Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '''
'''torch.Tensor, tf.Tensor or jax.ndarray.''' )
if do_resize and size is None:
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.''' )
# PIL RGBA images are converted to RGB
if do_convert_rgb:
__lowerCamelCase = [convert_to_rgb(__UpperCAmelCase ) for image in images]
# All transformations expect numpy arrays.
__lowerCamelCase = [to_numpy_array(__UpperCAmelCase ) for image in images]
if do_resize:
__lowerCamelCase = [self.resize(image=__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase ) for image in images]
if do_center_crop:
__lowerCamelCase = [self.center_crop(image=__UpperCAmelCase , size=__UpperCAmelCase ) for image in images]
if do_rescale:
__lowerCamelCase = [self.rescale(image=__UpperCAmelCase , scale=__UpperCAmelCase ) for image in images]
if do_normalize:
__lowerCamelCase = [self.normalize(image=__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase ) for image in images]
__lowerCamelCase = [to_channel_dimension_format(__UpperCAmelCase , __UpperCAmelCase ) for image in images]
__lowerCamelCase = {'''pixel_values''': images}
return BatchFeature(data=__UpperCAmelCase , tensor_type=__UpperCAmelCase )
| 330 | 1 |
from .glue import GlueDataset, GlueDataTrainingArguments
from .language_modeling import (
LineByLineTextDataset,
LineByLineWithRefDataset,
LineByLineWithSOPTextDataset,
TextDataset,
TextDatasetForNextSentencePrediction,
)
from .squad import SquadDataset, SquadDataTrainingArguments
| 330 |
from __future__ import annotations
from typing import Generic, TypeVar
a_ = TypeVar("""T""")
class __lowerCAmelCase ( Generic[T] ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = data
__lowerCamelCase = self
__lowerCamelCase = 0
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# map from node name to the node object
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# create a new set with x as its member
__lowerCamelCase = DisjointSetTreeNode(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# find the set x belongs to (with path-compression)
__lowerCamelCase = self.map[data]
if elem_ref != elem_ref.parent:
__lowerCamelCase = self.find_set(elem_ref.parent.data )
return elem_ref.parent
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# helper function for union operation
if nodea.rank > nodea.rank:
__lowerCamelCase = nodea
else:
__lowerCamelCase = nodea
if nodea.rank == nodea.rank:
nodea.rank += 1
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# merge 2 disjoint sets
self.link(self.find_set(__UpperCAmelCase ) , self.find_set(__UpperCAmelCase ) )
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# connections: map from the node to the neighbouring nodes (with weights)
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# add a node ONLY if its not present in the graph
if node not in self.connections:
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# add an edge with the given weight
self.add_node(__UpperCAmelCase )
self.add_node(__UpperCAmelCase )
__lowerCamelCase = weight
__lowerCamelCase = weight
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = []
__lowerCamelCase = set()
for start in self.connections:
for end in self.connections[start]:
if (start, end) not in seen:
seen.add((end, start) )
edges.append((start, end, self.connections[start][end]) )
edges.sort(key=lambda __UpperCAmelCase : x[2] )
# creating the disjoint set
__lowerCamelCase = DisjointSetTree[T]()
for node in self.connections:
disjoint_set.make_set(__UpperCAmelCase )
# MST generation
__lowerCamelCase = 0
__lowerCamelCase = 0
__lowerCamelCase = GraphUndirectedWeighted[T]()
while num_edges < len(self.connections ) - 1:
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = edges[index]
index += 1
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
if parent_u != parent_v:
num_edges += 1
graph.add_edge(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
disjoint_set.union(__UpperCAmelCase , __UpperCAmelCase )
return graph
| 330 | 1 |
import numpy as np
import pandas as pd
from sklearn.preprocessing import MinMaxScaler
from tensorflow.keras.layers import LSTM, Dense
from tensorflow.keras.models import Sequential
if __name__ == "__main__":
a_ = pd.read_csv("""sample_data.csv""", header=None)
a_ = df.shape[:1][0]
# If you're using some other dataset input the target column
a_ = df.iloc[:, 1:2]
a_ = actual_data.values.reshape(len_data, 1)
a_ = MinMaxScaler().fit_transform(actual_data)
a_ = 10
a_ = 5
a_ = 20
a_ = len_data - periods * look_back
a_ = actual_data[:division]
a_ = actual_data[division - look_back :]
a_ , a_ = [], []
a_ , a_ = [], []
for i in range(0, len(train_data) - forward_days - look_back + 1):
train_x.append(train_data[i : i + look_back])
train_y.append(train_data[i + look_back : i + look_back + forward_days])
for i in range(0, len(test_data) - forward_days - look_back + 1):
test_x.append(test_data[i : i + look_back])
test_y.append(test_data[i + look_back : i + look_back + forward_days])
a_ = np.array(train_x)
a_ = np.array(test_x)
a_ = np.array([list(i.ravel()) for i in train_y])
a_ = np.array([list(i.ravel()) for i in test_y])
a_ = Sequential()
model.add(LSTM(128, input_shape=(look_back, 1), return_sequences=True))
model.add(LSTM(64, input_shape=(128, 1)))
model.add(Dense(forward_days))
model.compile(loss="""mean_squared_error""", optimizer="""adam""")
a_ = model.fit(
x_train, y_train, epochs=150, verbose=1, shuffle=True, batch_size=4
)
a_ = model.predict(x_test)
| 330 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=2 , __UpperCAmelCase=8 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=99 , __UpperCAmelCase=16 , __UpperCAmelCase=5 , __UpperCAmelCase=2 , __UpperCAmelCase=36 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.0 , __UpperCAmelCase=512 , __UpperCAmelCase=16 , __UpperCAmelCase=2 , __UpperCAmelCase=0.02 , __UpperCAmelCase=3 , __UpperCAmelCase=4 , __UpperCAmelCase=None , ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
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=__UpperCAmelCase , initializer_range=self.initializer_range , )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_config()
__lowerCamelCase = 300
return config
def lowerCamelCase ( self ):
'''simple docstring'''
(
(
__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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = True
__lowerCamelCase = MraModel(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , )
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , )
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForQuestionAnswering(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , start_positions=__UpperCAmelCase , end_positions=__UpperCAmelCase , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForSequenceClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForTokenClassification(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_choices
__lowerCamelCase = MraForMultipleChoice(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
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(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowerCamelCase ( self ):
'''simple docstring'''
__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 __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = ()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModelTester(self )
__lowerCamelCase = ConfigTester(self , config_class=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__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(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = MraModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
@unittest.skip(reason='''MRA does not output attentions''' )
def lowerCamelCase ( self ):
'''simple docstring'''
return
@require_torch
class __lowerCAmelCase ( unittest.TestCase ):
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = torch.Size((1, 256, 768) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[-0.0_140, 0.0_830, -0.0_381], [0.1_546, 0.1_402, 0.0_220], [0.1_162, 0.0_851, 0.0_165]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 256, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[9.2_595, -3.6_038, 11.8_819], [9.3_869, -3.2_693, 11.0_956], [11.8_524, -3.4_938, 13.1_210]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' )
__lowerCamelCase = torch.arange(4096 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 4096, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[5.4_789, -2.3_564, 7.5_064], [7.9_067, -1.3_369, 9.9_668], [9.0_712, -1.8_106, 7.0_380]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
import random
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import (
AutoencoderKL,
DDIMScheduler,
UNetaDConditionModel,
VideoToVideoSDPipeline,
)
from diffusers.utils import floats_tensor, is_xformers_available, skip_mps
from diffusers.utils.testing_utils import enable_full_determinism, slow, torch_device
from ..pipeline_params import (
TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS,
TEXT_GUIDED_IMAGE_VARIATION_PARAMS,
)
from ..test_pipelines_common import PipelineTesterMixin
enable_full_determinism()
@skip_mps
class __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = VideoToVideoSDPipeline
lowerCAmelCase__ = TEXT_GUIDED_IMAGE_VARIATION_PARAMS.union({"""video"""} ) - {"""image""", """width""", """height"""}
lowerCAmelCase__ = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS.union({"""video"""} ) - {"""image"""}
lowerCAmelCase__ = PipelineTesterMixin.required_optional_params - {"""latents"""}
lowerCAmelCase__ = False
# No `output_type`.
lowerCAmelCase__ = frozenset(
[
"""num_inference_steps""",
"""generator""",
"""latents""",
"""return_dict""",
"""callback""",
"""callback_steps""",
] )
def lowerCamelCase ( self ):
'''simple docstring'''
torch.manual_seed(0 )
__lowerCamelCase = UNetaDConditionModel(
block_out_channels=(32, 64, 64, 64) , layers_per_block=2 , sample_size=32 , in_channels=4 , out_channels=4 , down_block_types=('''CrossAttnDownBlock3D''', '''CrossAttnDownBlock3D''', '''CrossAttnDownBlock3D''', '''DownBlock3D''') , up_block_types=('''UpBlock3D''', '''CrossAttnUpBlock3D''', '''CrossAttnUpBlock3D''', '''CrossAttnUpBlock3D''') , cross_attention_dim=32 , attention_head_dim=4 , )
__lowerCamelCase = DDIMScheduler(
beta_start=0.00_085 , beta_end=0.012 , beta_schedule='''scaled_linear''' , clip_sample=__UpperCAmelCase , set_alpha_to_one=__UpperCAmelCase , )
torch.manual_seed(0 )
__lowerCamelCase = AutoencoderKL(
block_out_channels=[32, 64] , in_channels=3 , out_channels=3 , down_block_types=['''DownEncoderBlock2D''', '''DownEncoderBlock2D'''] , up_block_types=['''UpDecoderBlock2D''', '''UpDecoderBlock2D'''] , latent_channels=4 , sample_size=128 , )
torch.manual_seed(0 )
__lowerCamelCase = CLIPTextConfig(
bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-0_5 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , hidden_act='''gelu''' , projection_dim=512 , )
__lowerCamelCase = CLIPTextModel(__UpperCAmelCase )
__lowerCamelCase = CLIPTokenizer.from_pretrained('''hf-internal-testing/tiny-random-clip''' )
__lowerCamelCase = {
'''unet''': unet,
'''scheduler''': scheduler,
'''vae''': vae,
'''text_encoder''': text_encoder,
'''tokenizer''': tokenizer,
}
return components
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=0 ):
'''simple docstring'''
# 3 frames
__lowerCamelCase = floats_tensor((1, 3, 3, 32, 32) , rng=random.Random(__UpperCAmelCase ) ).to(__UpperCAmelCase )
if str(__UpperCAmelCase ).startswith('''mps''' ):
__lowerCamelCase = torch.manual_seed(__UpperCAmelCase )
else:
__lowerCamelCase = torch.Generator(device=__UpperCAmelCase ).manual_seed(__UpperCAmelCase )
__lowerCamelCase = {
'''prompt''': '''A painting of a squirrel eating a burger''',
'''video''': video,
'''generator''': generator,
'''num_inference_steps''': 2,
'''guidance_scale''': 6.0,
'''output_type''': '''pt''',
}
return inputs
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''cpu''' # ensure determinism for the device-dependent torch.Generator
__lowerCamelCase = self.get_dummy_components()
__lowerCamelCase = VideoToVideoSDPipeline(**__UpperCAmelCase )
__lowerCamelCase = sd_pipe.to(__UpperCAmelCase )
sd_pipe.set_progress_bar_config(disable=__UpperCAmelCase )
__lowerCamelCase = self.get_dummy_inputs(__UpperCAmelCase )
__lowerCamelCase = '''np'''
__lowerCamelCase = sd_pipe(**__UpperCAmelCase ).frames
__lowerCamelCase = frames[0][-3:, -3:, -1]
assert frames[0].shape == (32, 32, 3)
__lowerCamelCase = np.array([106, 117, 113, 174, 137, 112, 148, 151, 131] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
@unittest.skipIf(
torch_device != '''cuda''' or not is_xformers_available() , reason='''XFormers attention is only available with CUDA and `xformers` installed''' , )
def lowerCamelCase ( self ):
'''simple docstring'''
self._test_xformers_attention_forwardGenerator_pass(test_mean_pixel_difference=__UpperCAmelCase , expected_max_diff=5E-3 )
@unittest.skip(reason='''Batching needs to be properly figured out first for this pipeline.''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
@unittest.skip(reason='''Batching needs to be properly figured out first for this pipeline.''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
@unittest.skip(reason='''`num_images_per_prompt` argument is not supported for this pipeline.''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
return super().test_progress_bar()
@slow
@skip_mps
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = VideoToVideoSDPipeline.from_pretrained('''cerspense/zeroscope_v2_XL''' , torch_dtype=torch.floataa )
pipe.enable_model_cpu_offload()
# 10 frames
__lowerCamelCase = torch.Generator(device='''cpu''' ).manual_seed(0 )
__lowerCamelCase = torch.randn((1, 10, 3, 1024, 576) , generator=__UpperCAmelCase )
__lowerCamelCase = video.to('''cuda''' )
__lowerCamelCase = '''Spiderman is surfing'''
__lowerCamelCase = pipe(__UpperCAmelCase , video=__UpperCAmelCase , generator=__UpperCAmelCase , num_inference_steps=3 , output_type='''pt''' ).frames
__lowerCamelCase = np.array([-1.0_458_984, -1.1_279_297, -0.9_663_086, -0.91_503_906, -0.75_097_656] )
assert np.abs(video_frames.cpu().numpy()[0, 0, 0, 0, -5:] - expected_array ).sum() < 1E-2
| 330 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_torch_available,
)
a_ = {"""configuration_encoder_decoder""": ["""EncoderDecoderConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""EncoderDecoderModel"""]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TFEncoderDecoderModel"""]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""FlaxEncoderDecoderModel"""]
if TYPE_CHECKING:
from .configuration_encoder_decoder import EncoderDecoderConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_encoder_decoder import EncoderDecoderModel
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_encoder_decoder import TFEncoderDecoderModel
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_encoder_decoder import FlaxEncoderDecoderModel
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
from google.protobuf import descriptor as _descriptor
from google.protobuf import descriptor_pool as _descriptor_pool
from google.protobuf import symbol_database as _symbol_database
from google.protobuf.internal import builder as _builder
# @@protoc_insertion_point(imports)
a_ = _symbol_database.Default()
a_ = _descriptor_pool.Default().AddSerializedFile(
b"""\n\x19sentencepiece_model.proto\x12\rsentencepiece\"\x80\x0c\n\x0bTrainerSpec\x12\r\n\x05input\x18\x01 \x03(\t\x12\x14\n\x0cinput_format\x18\x07 \x01(\t\x12\x14\n\x0cmodel_prefix\x18\x02 \x01(\t\x12\x41\n\nmodel_type\x18\x03 \x01(\x0e\x32$.sentencepiece.TrainerSpec.ModelType:\x07UNIGRAM\x12\x18\n\nvocab_size\x18\x04 \x01(\x05:\x04\x38\x30\x30\x30\x12\x17\n\x0f\x61\x63\x63\x65pt_language\x18\x05 \x03(\t\x12 \n\x15self_test_sample_size\x18\x06 \x01(\x05:\x01\x30\x12*\n\x1b\x65nable_differential_privacy\x18\x32 \x01(\x08:\x05\x66\x61lse\x12+\n differential_privacy_noise_level\x18\x33 \x01(\x02:\x01\x30\x12\x32\n\'differential_privacy_clipping_threshold\x18\x34 \x01(\x04:\x01\x30\x12\"\n\x12\x63haracter_coverage\x18\n \x01(\x02:\x06\x30.9995\x12\x1e\n\x13input_sentence_size\x18\x0b \x01(\x04:\x01\x30\x12$\n\x16shuffle_input_sentence\x18\x13 \x01(\x08:\x04true\x12 \n\x14mining_sentence_size\x18\x0c \x01(\x05\x42\x02\x18\x01\x12\"\n\x16training_sentence_size\x18\r \x01(\x05\x42\x02\x18\x01\x12(\n\x17seed_sentencepiece_size\x18\x0e \x01(\x05:\x07\x31\x30\x30\x30\x30\x30\x30\x12\x1e\n\x10shrinking_factor\x18\x0f \x01(\x02:\x04\x30.75\x12!\n\x13max_sentence_length\x18\x12 \x01(\x05:\x04\x34\x31\x39\x32\x12\x17\n\x0bnum_threads\x18\x10 \x01(\x05:\x02\x31\x36\x12\x1d\n\x12num_sub_iterations\x18\x11 \x01(\x05:\x01\x32\x12$\n\x18max_sentencepiece_length\x18\x14 \x01(\x05:\x02\x31\x36\x12%\n\x17split_by_unicode_script\x18\x15 \x01(\x08:\x04true\x12\x1d\n\x0fsplit_by_number\x18\x17 \x01(\x08:\x04true\x12!\n\x13split_by_whitespace\x18\x16 \x01(\x08:\x04true\x12)\n\x1atreat_whitespace_as_suffix\x18\x18 \x01(\x08:\x05\x66\x61lse\x12+\n\x1c\x61llow_whitespace_only_pieces\x18\x1a \x01(\x08:\x05\x66\x61lse\x12\x1b\n\x0csplit_digits\x18\x19 \x01(\x08:\x05\x66\x61lse\x12#\n\x19pretokenization_delimiter\x18\x35 \x01(\t:\x00\x12\x17\n\x0f\x63ontrol_symbols\x18\x1e \x03(\t\x12\x1c\n\x14user_defined_symbols\x18\x1f \x03(\t\x12\x16\n\x0erequired_chars\x18$ \x01(\t\x12\x1c\n\rbyte_fallback\x18# \x01(\x08:\x05\x66\x61lse\x12+\n\x1dvocabulary_output_piece_score\x18 \x01(\x08:\x04true\x12\x1e\n\x10hard_vocab_limit\x18! \x01(\x08:\x04true\x12\x1c\n\ruse_all_vocab\x18\" \x01(\x08:\x05\x66\x61lse\x12\x11\n\x06unk_id\x18( \x01(\x05:\x01\x30\x12\x11\n\x06\x62os_id\x18) \x01(\x05:\x01\x31\x12\x11\n\x06\x65os_id\x18* \x01(\x05:\x01\x32\x12\x12\n\x06pad_id\x18+ \x01(\x05:\x02-1\x12\x18\n\tunk_piece\x18- \x01(\t:\x05<unk>\x12\x16\n\tbos_piece\x18. \x01(\t:\x03<s>\x12\x17\n\teos_piece\x18/ \x01(\t:\x04</s>\x12\x18\n\tpad_piece\x18\x30 \x01(\t:\x05<pad>\x12\x1a\n\x0bunk_surface\x18, \x01(\t:\x05 \xe2\x81\x87 \x12+\n\x1ctrain_extremely_large_corpus\x18\x31 \x01(\x08:\x05\x66\x61lse\"5\n\tModelType\x12\x0b\n\x07UNIGRAM\x10\x01\x12\x07\n\x03\x42PE\x10\x02\x12\x08\n\x04WORD\x10\x03\x12\x08\n\x04\x43HAR\x10\x04*\t\x08\xc8\x01\x10\x80\x80\x80\x80\x02\"\xd1\x01\n\x0eNormalizerSpec\x12\x0c\n\x04name\x18\x01 \x01(\t\x12\x1c\n\x14precompiled_charsmap\x18\x02 \x01(\x0c\x12\x1e\n\x10\x61\x64\x64_dummy_prefix\x18\x03 \x01(\x08:\x04true\x12&\n\x18remove_extra_whitespaces\x18\x04 \x01(\x08:\x04true\x12 \n\x12\x65scape_whitespaces\x18\x05 \x01(\x08:\x04true\x12\x1e\n\x16normalization_rule_tsv\x18\x06 \x01(\t*\t\x08\xc8\x01\x10\x80\x80\x80\x80\x02\"y\n\x0cSelfTestData\x12\x33\n\x07samples\x18\x01 \x03(\x0b\x32\".sentencepiece.SelfTestData.Sample\x1a)\n\x06Sample\x12\r\n\x05input\x18\x01 \x01(\t\x12\x10\n\x08\x65xpected\x18\x02 \x01(\t*\t\x08\xc8\x01\x10\x80\x80\x80\x80\x02\"\xfe\x03\n\nModelProto\x12\x37\n\x06pieces\x18\x01 \x03(\x0b\x32\'.sentencepiece.ModelProto.SentencePiece\x12\x30\n\x0ctrainer_spec\x18\x02 \x01(\x0b\x32\x1a.sentencepiece.TrainerSpec\x12\x36\n\x0fnormalizer_spec\x18\x03 \x01(\x0b\x32\x1d.sentencepiece.NormalizerSpec\x12\x33\n\x0eself_test_data\x18\x04 \x01(\x0b\x32\x1b.sentencepiece.SelfTestData\x12\x38\n\x11\x64\x65normalizer_spec\x18\x05 \x01(\x0b\x32\x1d.sentencepiece.NormalizerSpec\x1a\xd2\x01\n\rSentencePiece\x12\r\n\x05piece\x18\x01 \x01(\t\x12\r\n\x05score\x18\x02 \x01(\x02\x12\x42\n\x04type\x18\x03 \x01(\x0e\x32,.sentencepiece.ModelProto.SentencePiece.Type:\x06NORMAL\"T\n\x04Type\x12\n\n\x06NORMAL\x10\x01\x12\x0b\n\x07UNKNOWN\x10\x02\x12\x0b\n\x07\x43ONTROL\x10\x03\x12\x10\n\x0cUSER_DEFINED\x10\x04\x12\x08\n\x04\x42YTE\x10\x06\x12\n\n\x06UNUSED\x10\x05*\t\x08\xc8\x01\x10\x80\x80\x80\x80\x02*\t\x08\xc8\x01\x10\x80\x80\x80\x80\x02\x42\x02H\x03"""
)
a_ = globals()
_builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
_builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, """sentencepiece_model_pb2""", _globals)
if _descriptor._USE_C_DESCRIPTORS is False:
a_ = None
a_ = b"""H\003"""
# (generated by protobuf compiler, but `_TRAINERSPEC` is not defined)
# _TRAINERSPEC.fields_by_name["mining_sentence_size"]._options = None
# _TRAINERSPEC.fields_by_name["mining_sentence_size"]._serialized_options = b"\030\001"
# _TRAINERSPEC.fields_by_name["training_sentence_size"]._options = None
# _TRAINERSPEC.fields_by_name["training_sentence_size"]._serialized_options = b"\030\001"
a_ = 45
a_ = 1_581
a_ = 1_517
a_ = 1_570
a_ = 1_584
a_ = 1_793
a_ = 1_795
a_ = 1_916
a_ = 1_864
a_ = 1_905
a_ = 1_919
a_ = 2_429
a_ = 2_208
a_ = 2_418
a_ = 2_323
a_ = 2_407
# @@protoc_insertion_point(module_scope)
| 330 |
from string import ascii_lowercase, ascii_uppercase
def a__ ( _UpperCamelCase : str ):
if not sentence:
return ""
__lowerCamelCase = dict(zip(_UpperCamelCase ,_UpperCamelCase ) )
return lower_to_upper.get(sentence[0] ,sentence[0] ) + sentence[1:]
if __name__ == "__main__":
from doctest import testmod
testmod()
| 330 | 1 |
from typing import Optional, Tuple, Union
import tensorflow as tf
from ...activations_tf import ACTaFN
from ...file_utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward
from ...modeling_tf_outputs import (
TFBaseModelOutputWithNoAttention,
TFBaseModelOutputWithPoolingAndNoAttention,
TFSequenceClassifierOutput,
)
from ...modeling_tf_utils import TFPreTrainedModel, TFSequenceClassificationLoss, keras_serializable, unpack_inputs
from ...tf_utils import shape_list
from ...utils import logging
from .configuration_regnet import RegNetConfig
a_ = logging.get_logger(__name__)
# General docstring
a_ = """RegNetConfig"""
# Base docstring
a_ = """facebook/regnet-y-040"""
a_ = [1, 1_088, 7, 7]
# Image classification docstring
a_ = """facebook/regnet-y-040"""
a_ = """tabby, tabby cat"""
a_ = [
"""facebook/regnet-y-040""",
# See all regnet models at https://huggingface.co/models?filter=regnet
]
class __lowerCAmelCase ( tf.keras.layers.Layer ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase = 3 , __UpperCAmelCase = 1 , __UpperCAmelCase = 1 , __UpperCAmelCase = "relu" , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
# The padding and conv has been verified in
# https://colab.research.google.com/gist/sayakpaul/854bc10eeaf21c9ee2119e0b9f3841a7/scratchpad.ipynb
__lowerCamelCase = tf.keras.layers.ZeroPaddingaD(padding=kernel_size // 2 )
__lowerCamelCase = tf.keras.layers.ConvaD(
filters=__UpperCAmelCase , kernel_size=__UpperCAmelCase , strides=__UpperCAmelCase , padding='''VALID''' , groups=__UpperCAmelCase , use_bias=__UpperCAmelCase , name='''convolution''' , )
__lowerCamelCase = tf.keras.layers.BatchNormalization(epsilon=1E-5 , momentum=0.9 , name='''normalization''' )
__lowerCamelCase = ACTaFN[activation] if activation is not None else tf.identity
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.convolution(self.padding(__UpperCAmelCase ) )
__lowerCamelCase = self.normalization(__UpperCAmelCase )
__lowerCamelCase = self.activation(__UpperCAmelCase )
return hidden_state
class __lowerCAmelCase ( tf.keras.layers.Layer ):
def __init__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = config.num_channels
__lowerCamelCase = TFRegNetConvLayer(
out_channels=config.embedding_size , kernel_size=3 , stride=2 , activation=config.hidden_act , name='''embedder''' , )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = shape_list(__UpperCAmelCase )[1]
if tf.executing_eagerly() and num_channels != self.num_channels:
raise ValueError(
'''Make sure that the channel dimension of the pixel values match with the one set in the configuration.''' )
# When running on CPU, `tf.keras.layers.Conv2D` doesn't support `NCHW` format.
# So change the input format from `NCHW` to `NHWC`.
# shape = (batch_size, in_height, in_width, in_channels=num_channels)
__lowerCamelCase = tf.transpose(__UpperCAmelCase , perm=(0, 2, 3, 1) )
__lowerCamelCase = self.embedder(__UpperCAmelCase )
return hidden_state
class __lowerCAmelCase ( tf.keras.layers.Layer ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase = 2 , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = tf.keras.layers.ConvaD(
filters=__UpperCAmelCase , kernel_size=1 , strides=__UpperCAmelCase , use_bias=__UpperCAmelCase , name='''convolution''' )
__lowerCamelCase = tf.keras.layers.BatchNormalization(epsilon=1E-5 , momentum=0.9 , name='''normalization''' )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False ):
'''simple docstring'''
return self.normalization(self.convolution(__UpperCAmelCase ) , training=__UpperCAmelCase )
class __lowerCAmelCase ( tf.keras.layers.Layer ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = tf.keras.layers.GlobalAveragePoolingaD(keepdims=__UpperCAmelCase , name='''pooler''' )
__lowerCamelCase = [
tf.keras.layers.ConvaD(filters=__UpperCAmelCase , kernel_size=1 , activation='''relu''' , name='''attention.0''' ),
tf.keras.layers.ConvaD(filters=__UpperCAmelCase , kernel_size=1 , activation='''sigmoid''' , name='''attention.2''' ),
]
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# [batch_size, h, w, num_channels] -> [batch_size, 1, 1, num_channels]
__lowerCamelCase = self.pooler(__UpperCAmelCase )
for layer_module in self.attention:
__lowerCamelCase = layer_module(__UpperCAmelCase )
__lowerCamelCase = hidden_state * pooled
return hidden_state
class __lowerCAmelCase ( tf.keras.layers.Layer ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = 1 , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = in_channels != out_channels or stride != 1
__lowerCamelCase = max(1 , out_channels // config.groups_width )
__lowerCamelCase = (
TFRegNetShortCut(__UpperCAmelCase , stride=__UpperCAmelCase , name='''shortcut''' )
if should_apply_shortcut
else tf.keras.layers.Activation('''linear''' , name='''shortcut''' )
)
# `self.layers` instead of `self.layer` because that is a reserved argument.
__lowerCamelCase = [
TFRegNetConvLayer(__UpperCAmelCase , kernel_size=1 , activation=config.hidden_act , name='''layer.0''' ),
TFRegNetConvLayer(
__UpperCAmelCase , stride=__UpperCAmelCase , groups=__UpperCAmelCase , activation=config.hidden_act , name='''layer.1''' ),
TFRegNetConvLayer(__UpperCAmelCase , kernel_size=1 , activation=__UpperCAmelCase , name='''layer.2''' ),
]
__lowerCamelCase = ACTaFN[config.hidden_act]
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = hidden_state
for layer_module in self.layers:
__lowerCamelCase = layer_module(__UpperCAmelCase )
__lowerCamelCase = self.shortcut(__UpperCAmelCase )
hidden_state += residual
__lowerCamelCase = self.activation(__UpperCAmelCase )
return hidden_state
class __lowerCAmelCase ( tf.keras.layers.Layer ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = 1 , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = in_channels != out_channels or stride != 1
__lowerCamelCase = max(1 , out_channels // config.groups_width )
__lowerCamelCase = (
TFRegNetShortCut(__UpperCAmelCase , stride=__UpperCAmelCase , name='''shortcut''' )
if should_apply_shortcut
else tf.keras.layers.Activation('''linear''' , name='''shortcut''' )
)
__lowerCamelCase = [
TFRegNetConvLayer(__UpperCAmelCase , kernel_size=1 , activation=config.hidden_act , name='''layer.0''' ),
TFRegNetConvLayer(
__UpperCAmelCase , stride=__UpperCAmelCase , groups=__UpperCAmelCase , activation=config.hidden_act , name='''layer.1''' ),
TFRegNetSELayer(__UpperCAmelCase , reduced_channels=int(round(in_channels / 4 ) ) , name='''layer.2''' ),
TFRegNetConvLayer(__UpperCAmelCase , kernel_size=1 , activation=__UpperCAmelCase , name='''layer.3''' ),
]
__lowerCamelCase = ACTaFN[config.hidden_act]
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = hidden_state
for layer_module in self.layers:
__lowerCamelCase = layer_module(__UpperCAmelCase )
__lowerCamelCase = self.shortcut(__UpperCAmelCase )
hidden_state += residual
__lowerCamelCase = self.activation(__UpperCAmelCase )
return hidden_state
class __lowerCAmelCase ( tf.keras.layers.Layer ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = 2 , __UpperCAmelCase = 2 , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = TFRegNetXLayer if config.layer_type == '''x''' else TFRegNetYLayer
__lowerCamelCase = [
# downsampling is done in the first layer with stride of 2
layer(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , stride=__UpperCAmelCase , name='''layers.0''' ),
*[layer(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , name=F"""layers.{i+1}""" ) for i in range(depth - 1 )],
]
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
for layer_module in self.layers:
__lowerCamelCase = layer_module(__UpperCAmelCase )
return hidden_state
class __lowerCAmelCase ( tf.keras.layers.Layer ):
def __init__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = []
# based on `downsample_in_first_stage`, the first layer of the first stage may or may not downsample the input
self.stages.append(
TFRegNetStage(
__UpperCAmelCase , config.embedding_size , config.hidden_sizes[0] , stride=2 if config.downsample_in_first_stage else 1 , depth=config.depths[0] , name='''stages.0''' , ) )
__lowerCamelCase = zip(config.hidden_sizes , config.hidden_sizes[1:] )
for i, ((in_channels, out_channels), depth) in enumerate(zip(__UpperCAmelCase , config.depths[1:] ) ):
self.stages.append(TFRegNetStage(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , depth=__UpperCAmelCase , name=F"""stages.{i+1}""" ) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False , __UpperCAmelCase = True ):
'''simple docstring'''
__lowerCamelCase = () if output_hidden_states else None
for stage_module in self.stages:
if output_hidden_states:
__lowerCamelCase = hidden_states + (hidden_state,)
__lowerCamelCase = stage_module(__UpperCAmelCase )
if output_hidden_states:
__lowerCamelCase = hidden_states + (hidden_state,)
if not return_dict:
return tuple(v for v in [hidden_state, hidden_states] if v is not None )
return TFBaseModelOutputWithNoAttention(last_hidden_state=__UpperCAmelCase , hidden_states=__UpperCAmelCase )
@keras_serializable
class __lowerCAmelCase ( tf.keras.layers.Layer ):
lowerCAmelCase__ = RegNetConfig
def __init__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = config
__lowerCamelCase = TFRegNetEmbeddings(__UpperCAmelCase , name='''embedder''' )
__lowerCamelCase = TFRegNetEncoder(__UpperCAmelCase , name='''encoder''' )
__lowerCamelCase = tf.keras.layers.GlobalAveragePoolingaD(keepdims=__UpperCAmelCase , name='''pooler''' )
@unpack_inputs
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = False , ):
'''simple docstring'''
__lowerCamelCase = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__lowerCamelCase = return_dict if return_dict is not None else self.config.use_return_dict
__lowerCamelCase = self.embedder(__UpperCAmelCase , training=__UpperCAmelCase )
__lowerCamelCase = self.encoder(
__UpperCAmelCase , output_hidden_states=__UpperCAmelCase , return_dict=__UpperCAmelCase , training=__UpperCAmelCase )
__lowerCamelCase = encoder_outputs[0]
__lowerCamelCase = self.pooler(__UpperCAmelCase )
# Change to NCHW output format have uniformity in the modules
__lowerCamelCase = tf.transpose(__UpperCAmelCase , perm=(0, 3, 1, 2) )
__lowerCamelCase = tf.transpose(__UpperCAmelCase , perm=(0, 3, 1, 2) )
# Change the other hidden state outputs to NCHW as well
if output_hidden_states:
__lowerCamelCase = tuple([tf.transpose(__UpperCAmelCase , perm=(0, 3, 1, 2) ) for h in encoder_outputs[1]] )
if not return_dict:
return (last_hidden_state, pooled_output) + encoder_outputs[1:]
return TFBaseModelOutputWithPoolingAndNoAttention(
last_hidden_state=__UpperCAmelCase , pooler_output=__UpperCAmelCase , hidden_states=hidden_states if output_hidden_states else encoder_outputs.hidden_states , )
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = RegNetConfig
lowerCAmelCase__ = """regnet"""
lowerCAmelCase__ = """pixel_values"""
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return {"pixel_values": tf.TensorSpec(shape=(None, self.config.num_channels, 224, 224) , dtype=tf.floataa )}
a_ = R"""
Parameters:
This model is a Tensorflow
[tf.keras.layers.Layer](https://www.tensorflow.org/api_docs/python/tf/keras/layers/Layer) sub-class. Use it as a
regular Tensorflow Module and refer to the Tensorflow documentation for all matter related to general usage and
behavior.
config ([`RegNetConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the [`~TFPreTrainedModel.from_pretrained`] method to load the model weights.
"""
a_ = R"""
Args:
pixel_values (`tf.Tensor` of shape `(batch_size, num_channels, height, width)`):
Pixel values. Pixel values can be obtained using [`AutoImageProcessor`]. See
[`ConveNextImageProcessor.__call__`] for details.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
"""
@add_start_docstrings(
"""The bare RegNet model outputting raw features without any specific head on top.""" , lowerCAmelCase__ , )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(__UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase )
__lowerCamelCase = TFRegNetMainLayer(__UpperCAmelCase , name='''regnet''' )
@unpack_inputs
@add_start_docstrings_to_model_forward(__UpperCAmelCase )
@add_code_sample_docstrings(
checkpoint=_CHECKPOINT_FOR_DOC , output_type=__UpperCAmelCase , config_class=_CONFIG_FOR_DOC , modality='''vision''' , expected_output=_EXPECTED_OUTPUT_SHAPE , )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=False , ):
'''simple docstring'''
__lowerCamelCase = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__lowerCamelCase = return_dict if return_dict is not None else self.config.use_return_dict
__lowerCamelCase = self.regnet(
pixel_values=__UpperCAmelCase , output_hidden_states=__UpperCAmelCase , return_dict=__UpperCAmelCase , training=__UpperCAmelCase , )
if not return_dict:
return (outputs[0],) + outputs[1:]
return TFBaseModelOutputWithPoolingAndNoAttention(
last_hidden_state=outputs.last_hidden_state , pooler_output=outputs.pooler_output , hidden_states=outputs.hidden_states , )
@add_start_docstrings(
"""
RegNet Model with an image classification head on top (a linear layer on top of the pooled features), e.g. for
ImageNet.
""" , lowerCAmelCase__ , )
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(__UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase )
__lowerCamelCase = config.num_labels
__lowerCamelCase = TFRegNetMainLayer(__UpperCAmelCase , name='''regnet''' )
# classification head
__lowerCamelCase = [
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(config.num_labels , name='''classifier.1''' ) if config.num_labels > 0 else tf.identity,
]
@unpack_inputs
@add_start_docstrings_to_model_forward(__UpperCAmelCase )
@add_code_sample_docstrings(
checkpoint=_IMAGE_CLASS_CHECKPOINT , output_type=__UpperCAmelCase , config_class=_CONFIG_FOR_DOC , expected_output=_IMAGE_CLASS_EXPECTED_OUTPUT , )
def lowerCamelCase ( self , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=False , ):
'''simple docstring'''
__lowerCamelCase = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__lowerCamelCase = return_dict if return_dict is not None else self.config.use_return_dict
__lowerCamelCase = self.regnet(
__UpperCAmelCase , output_hidden_states=__UpperCAmelCase , return_dict=__UpperCAmelCase , training=__UpperCAmelCase )
__lowerCamelCase = outputs.pooler_output if return_dict else outputs[1]
__lowerCamelCase = self.classifier[0](__UpperCAmelCase )
__lowerCamelCase = self.classifier[1](__UpperCAmelCase )
__lowerCamelCase = None if labels is None else self.hf_compute_loss(labels=__UpperCAmelCase , logits=__UpperCAmelCase )
if not return_dict:
__lowerCamelCase = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return TFSequenceClassifierOutput(loss=__UpperCAmelCase , logits=__UpperCAmelCase , hidden_states=outputs.hidden_states )
| 330 |
from transformers import BertTokenizer, EncoderDecoderModel, SeqaSeqTrainer, SeqaSeqTrainingArguments
from transformers.testing_utils import TestCasePlus, require_torch, slow
from transformers.utils import is_datasets_available
if is_datasets_available():
import datasets
class __lowerCAmelCase ( lowerCAmelCase__ ):
@slow
@require_torch
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = EncoderDecoderModel.from_encoder_decoder_pretrained('''prajjwal1/bert-tiny''' , '''prajjwal1/bert-tiny''' )
__lowerCamelCase = BertTokenizer.from_pretrained('''bert-base-uncased''' )
__lowerCamelCase = bertabert.config.encoder.vocab_size
__lowerCamelCase = tokenizer.sep_token_id
__lowerCamelCase = tokenizer.cls_token_id
__lowerCamelCase = 128
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''train[:1%]''' )
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''validation[:1%]''' )
__lowerCamelCase = train_dataset.select(range(32 ) )
__lowerCamelCase = val_dataset.select(range(16 ) )
__lowerCamelCase = 4
def _map_to_encoder_decoder_inputs(__UpperCAmelCase ):
# Tokenizer will automatically set [BOS] <text> [EOS]
__lowerCamelCase = tokenizer(batch['''article'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=512 )
__lowerCamelCase = tokenizer(batch['''highlights'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=128 )
__lowerCamelCase = inputs.input_ids
__lowerCamelCase = inputs.attention_mask
__lowerCamelCase = outputs.input_ids
__lowerCamelCase = outputs.input_ids.copy()
__lowerCamelCase = [
[-100 if token == tokenizer.pad_token_id else token for token in labels] for labels in batch['''labels''']
]
__lowerCamelCase = outputs.attention_mask
assert all(len(__UpperCAmelCase ) == 512 for x in inputs.input_ids )
assert all(len(__UpperCAmelCase ) == 128 for x in outputs.input_ids )
return batch
def _compute_metrics(__UpperCAmelCase ):
__lowerCamelCase = pred.label_ids
__lowerCamelCase = pred.predictions
# all unnecessary tokens are removed
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = sum([int(pred_str[i] == label_str[i] ) for i in range(len(__UpperCAmelCase ) )] ) / len(__UpperCAmelCase )
return {"accuracy": accuracy}
# map train dataset
__lowerCamelCase = train_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
train_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
# same for validation dataset
__lowerCamelCase = val_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
val_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
__lowerCamelCase = self.get_auto_remove_tmp_dir()
__lowerCamelCase = SeqaSeqTrainingArguments(
output_dir=__UpperCAmelCase , per_device_train_batch_size=__UpperCAmelCase , per_device_eval_batch_size=__UpperCAmelCase , predict_with_generate=__UpperCAmelCase , evaluation_strategy='''steps''' , do_train=__UpperCAmelCase , do_eval=__UpperCAmelCase , warmup_steps=0 , eval_steps=2 , logging_steps=2 , )
# instantiate trainer
__lowerCamelCase = SeqaSeqTrainer(
model=__UpperCAmelCase , args=__UpperCAmelCase , compute_metrics=_compute_metrics , train_dataset=__UpperCAmelCase , eval_dataset=__UpperCAmelCase , tokenizer=__UpperCAmelCase , )
# start training
trainer.train()
| 330 | 1 |
import uuid
from typing import Any, Dict, List, Optional, Union
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
if is_torch_available():
import torch
a_ = logging.get_logger(__name__)
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=None , __UpperCAmelCase=None ):
'''simple docstring'''
if not conversation_id:
__lowerCamelCase = uuid.uuida()
if past_user_inputs is None:
__lowerCamelCase = []
if generated_responses is None:
__lowerCamelCase = []
__lowerCamelCase = conversation_id
__lowerCamelCase = past_user_inputs
__lowerCamelCase = generated_responses
__lowerCamelCase = text
def __eq__( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
return False
if self.uuid == other.uuid:
return True
return (
self.new_user_input == other.new_user_input
and self.past_user_inputs == other.past_user_inputs
and self.generated_responses == other.generated_responses
)
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False ):
'''simple docstring'''
if self.new_user_input:
if overwrite:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" was overwritten """
F"""with: \"{text}\".""" )
__lowerCamelCase = text
else:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" new input """
F"""ignored: \"{text}\". Set `overwrite` to True to overwrite unprocessed user input""" )
else:
__lowerCamelCase = text
def lowerCamelCase ( self ):
'''simple docstring'''
if self.new_user_input:
self.past_user_inputs.append(self.new_user_input )
__lowerCamelCase = None
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
self.generated_responses.append(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
for user_input, generated_response in zip(self.past_user_inputs , self.generated_responses ):
yield True, user_input
yield False, generated_response
if self.new_user_input:
yield True, self.new_user_input
def __repr__( self ):
'''simple docstring'''
__lowerCamelCase = F"""Conversation id: {self.uuid} \n"""
for is_user, text in self.iter_texts():
__lowerCamelCase = '''user''' if is_user else '''bot'''
output += F"""{name} >> {text} \n"""
return output
@add_end_docstrings(
lowerCAmelCase__ , r"""
min_length_for_response (`int`, *optional*, defaults to 32):
The minimum length (in number of tokens) for a response.
minimum_tokens (`int`, *optional*, defaults to 10):
The minimum length of tokens to leave for a response.
""" , )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
if self.tokenizer.pad_token_id is None:
__lowerCamelCase = self.tokenizer.eos_token
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = {}
__lowerCamelCase = {}
__lowerCamelCase = {}
if min_length_for_response is not None:
__lowerCamelCase = min_length_for_response
if minimum_tokens is not None:
__lowerCamelCase = minimum_tokens
if "max_length" in generate_kwargs:
__lowerCamelCase = generate_kwargs['''max_length''']
# self.max_length = generate_kwargs.get("max_length", self.model.config.max_length)
if clean_up_tokenization_spaces is not None:
__lowerCamelCase = clean_up_tokenization_spaces
if generate_kwargs:
forward_params.update(__UpperCAmelCase )
return preprocess_params, forward_params, postprocess_params
def __call__( self , __UpperCAmelCase , __UpperCAmelCase=0 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = super().__call__(__UpperCAmelCase , num_workers=__UpperCAmelCase , **__UpperCAmelCase )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) == 1:
return outputs[0]
return outputs
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=32 ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError('''ConversationalPipeline, expects Conversation as inputs''' )
if conversation.new_user_input is None:
raise ValueError(
F"""Conversation with UUID {type(conversation.uuid )} does not contain new user input to process. """
'''Add user inputs with the conversation\'s `add_user_input` method''' )
if hasattr(self.tokenizer , '''_build_conversation_input_ids''' ):
__lowerCamelCase = self.tokenizer._build_conversation_input_ids(__UpperCAmelCase )
else:
# If the tokenizer cannot handle conversations, we default to only the old version
__lowerCamelCase = self._legacy_parse_and_tokenize(__UpperCAmelCase )
if self.framework == "pt":
__lowerCamelCase = torch.LongTensor([input_ids] )
elif self.framework == "tf":
__lowerCamelCase = tf.constant([input_ids] )
return {"input_ids": input_ids, "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=10 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = generate_kwargs.get('''max_length''' , self.model.config.max_length )
__lowerCamelCase = model_inputs['''input_ids'''].shape[1]
if max_length - minimum_tokens < n:
logger.warning(F"""Conversation input is to long ({n}), trimming it to ({max_length} - {minimum_tokens})""" )
__lowerCamelCase = max_length - minimum_tokens
__lowerCamelCase = model_inputs['''input_ids'''][:, -trim:]
if "attention_mask" in model_inputs:
__lowerCamelCase = model_inputs['''attention_mask'''][:, -trim:]
__lowerCamelCase = model_inputs.pop('''conversation''' )
__lowerCamelCase = max_length
__lowerCamelCase = self.model.generate(**__UpperCAmelCase , **__UpperCAmelCase )
if self.model.config.is_encoder_decoder:
__lowerCamelCase = 1
else:
__lowerCamelCase = n
return {"output_ids": output_ids[:, start_position:], "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = model_outputs['''output_ids''']
__lowerCamelCase = self.tokenizer.decode(
output_ids[0] , skip_special_tokens=__UpperCAmelCase , clean_up_tokenization_spaces=__UpperCAmelCase , )
__lowerCamelCase = model_outputs['''conversation''']
conversation.mark_processed()
conversation.append_response(__UpperCAmelCase )
return conversation
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer.eos_token_id
__lowerCamelCase = []
for is_user, text in conversation.iter_texts():
if eos_token_id is not None:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) + [eos_token_id] )
else:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) )
if len(__UpperCAmelCase ) > self.tokenizer.model_max_length:
__lowerCamelCase = input_ids[-self.tokenizer.model_max_length :]
return input_ids
| 330 |
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_timm_backbone""": ["""TimmBackboneConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TimmBackbone"""]
if TYPE_CHECKING:
from .configuration_timm_backbone import TimmBackboneConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_timm_backbone import TimmBackbone
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
import colorsys
from PIL import Image # type: ignore
def a__ ( _UpperCamelCase : float ,_UpperCamelCase : float ,_UpperCamelCase : int ):
__lowerCamelCase = x
__lowerCamelCase = y
for step in range(_UpperCamelCase ): # noqa: B007
__lowerCamelCase = a * a - b * b + x
__lowerCamelCase = 2 * a * b + y
__lowerCamelCase = a_new
# divergence happens for all complex number with an absolute value
# greater than 4
if a * a + b * b > 4:
break
return step / (max_step - 1)
def a__ ( _UpperCamelCase : float ):
if distance == 1:
return (0, 0, 0)
else:
return (2_55, 2_55, 2_55)
def a__ ( _UpperCamelCase : float ):
if distance == 1:
return (0, 0, 0)
else:
return tuple(round(i * 2_55 ) for i in colorsys.hsv_to_rgb(_UpperCamelCase ,1 ,1 ) )
def a__ ( _UpperCamelCase : int = 8_00 ,_UpperCamelCase : int = 6_00 ,_UpperCamelCase : float = -0.6 ,_UpperCamelCase : float = 0 ,_UpperCamelCase : float = 3.2 ,_UpperCamelCase : int = 50 ,_UpperCamelCase : bool = True ,):
__lowerCamelCase = Image.new('''RGB''' ,(image_width, image_height) )
__lowerCamelCase = img.load()
# loop through the image-coordinates
for image_x in range(_UpperCamelCase ):
for image_y in range(_UpperCamelCase ):
# determine the figure-coordinates based on the image-coordinates
__lowerCamelCase = figure_width / image_width * image_height
__lowerCamelCase = figure_center_x + (image_x / image_width - 0.5) * figure_width
__lowerCamelCase = figure_center_y + (image_y / image_height - 0.5) * figure_height
__lowerCamelCase = get_distance(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# color the corresponding pixel based on the selected coloring-function
if use_distance_color_coding:
__lowerCamelCase = get_color_coded_rgb(_UpperCamelCase )
else:
__lowerCamelCase = get_black_and_white_rgb(_UpperCamelCase )
return img
if __name__ == "__main__":
import doctest
doctest.testmod()
# colored version, full figure
a_ = get_image()
# uncomment for colored version, different section, zoomed in
# img = get_image(figure_center_x = -0.6, figure_center_y = -0.4,
# figure_width = 0.8)
# uncomment for black and white version, full figure
# img = get_image(use_distance_color_coding = False)
# uncomment to save the image
# img.save("mandelbrot.png")
img.show()
| 330 |
import random
import unittest
from torch.utils.data import BatchSampler, DataLoader, IterableDataset
from accelerate import Accelerator
from accelerate.data_loader import (
BatchSamplerShard,
DataLoaderDispatcher,
DataLoaderShard,
IterableDatasetShard,
SkipBatchSampler,
SkipDataLoader,
skip_first_batches,
)
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase=0.01 , __UpperCAmelCase=1000 ):
'''simple docstring'''
__lowerCamelCase = p_stop
__lowerCamelCase = max_length
def __iter__( self ):
'''simple docstring'''
__lowerCamelCase = 0
__lowerCamelCase = False
while not stop and count < self.max_length:
yield count
count += 1
__lowerCamelCase = random.random() < self.p_stop
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = [
BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
for i in range(2 )
]
__lowerCamelCase = [list(__UpperCAmelCase ) for batch_sampler_shard in batch_sampler_shards]
if not split_batches:
self.assertListEqual([len(__UpperCAmelCase ) for shard in batch_sampler_shards] , [len(__UpperCAmelCase ) for e in expected] )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [0, 1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 0]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [1, 2, 3]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1, 0]], [[1, 0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 0]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], [[0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = [[0, 1, 2], [3, 4], [5, 6, 7, 8], [9, 10, 11], [12, 13]]
__lowerCamelCase = [BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , even_batches=__UpperCAmelCase ) for i in range(2 )]
self.assertEqual(len(batch_sampler_shards[0] ) , 3 )
self.assertEqual(len(batch_sampler_shards[1] ) , 2 )
self.assertListEqual(list(batch_sampler_shards[0] ) , [[0, 1, 2], [5, 6, 7, 8], [12, 13]] )
self.assertListEqual(list(batch_sampler_shards[1] ) , [[3, 4], [9, 10, 11]] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=2 , __UpperCAmelCase=False ):
'''simple docstring'''
random.seed(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = [
IterableDatasetShard(
__UpperCAmelCase , batch_size=__UpperCAmelCase , drop_last=__UpperCAmelCase , num_processes=__UpperCAmelCase , process_index=__UpperCAmelCase , split_batches=__UpperCAmelCase , )
for i in range(__UpperCAmelCase )
]
__lowerCamelCase = []
for iterable_dataset_shard in iterable_dataset_shards:
# Since our random iterable dataset will be... random... we need to use a seed to get reproducible results.
random.seed(__UpperCAmelCase )
iterable_dataset_lists.append(list(__UpperCAmelCase ) )
__lowerCamelCase = batch_size // num_processes if split_batches else batch_size
# All iterable dataset shard should have the same length, a round multiple of shard_batch_size
__lowerCamelCase = iterable_dataset_lists[0]
for l in iterable_dataset_lists[1:]:
self.assertEqual(len(__UpperCAmelCase ) , len(__UpperCAmelCase ) )
self.assertTrue(len(__UpperCAmelCase ) % shard_batch_size == 0 )
__lowerCamelCase = []
for idx in range(0 , len(__UpperCAmelCase ) , __UpperCAmelCase ):
for l in iterable_dataset_lists:
observed += l[idx : idx + shard_batch_size]
if not drop_last:
while len(__UpperCAmelCase ) < len(__UpperCAmelCase ):
reference += reference
self.assertListEqual(__UpperCAmelCase , reference[: len(__UpperCAmelCase )] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = 42
__lowerCamelCase = RandomIterableDataset()
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
# Edge case with a very small dataset
__lowerCamelCase = RandomIterableDataset(max_length=2 )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BatchSampler(range(16 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = SkipBatchSampler(__UpperCAmelCase , 2 )
self.assertListEqual(list(__UpperCAmelCase ) , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = SkipDataLoader(list(range(16 ) ) , batch_size=4 , skip_batches=2 )
self.assertListEqual([t.tolist() for t in dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoader(list(range(16 ) ) , batch_size=4 )
__lowerCamelCase = skip_first_batches(__UpperCAmelCase , num_batches=2 )
self.assertListEqual([t.tolist() for t in new_dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoaderShard(list(range(16 ) ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
def lowerCamelCase ( self ):
'''simple docstring'''
Accelerator()
__lowerCamelCase = DataLoaderDispatcher(range(16 ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
| 330 | 1 |
from __future__ import annotations
from collections.abc import Iterator
from typing import Generic, TypeVar
a_ = TypeVar("""T""")
class __lowerCAmelCase ( Generic[T] ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = data
__lowerCamelCase = None
def __str__( self ):
'''simple docstring'''
return F"""{self.data}"""
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
__lowerCamelCase = None
def __iter__( self ):
'''simple docstring'''
__lowerCamelCase = self.top
while node:
yield node.data
__lowerCamelCase = node.next
def __str__( self ):
'''simple docstring'''
return "->".join([str(__UpperCAmelCase ) for item in self] )
def __len__( self ):
'''simple docstring'''
return len(tuple(iter(self ) ) )
def lowerCamelCase ( self ):
'''simple docstring'''
return self.top is None
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = Node(__UpperCAmelCase )
if not self.is_empty():
__lowerCamelCase = self.top
__lowerCamelCase = node
def lowerCamelCase ( self ):
'''simple docstring'''
if self.is_empty():
raise IndexError('''pop from empty stack''' )
assert isinstance(self.top , __UpperCAmelCase )
__lowerCamelCase = self.top
__lowerCamelCase = self.top.next
return pop_node.data
def lowerCamelCase ( self ):
'''simple docstring'''
if self.is_empty():
raise IndexError('''peek from empty stack''' )
assert self.top is not None
return self.top.data
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = None
if __name__ == "__main__":
from doctest import testmod
testmod()
| 330 |
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = n ** (1 / 3)
return (val * val * val) == n
if __name__ == "__main__":
print(perfect_cube(27))
print(perfect_cube(4))
| 330 | 1 |
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {
"""configuration_git""": ["""GIT_PRETRAINED_CONFIG_ARCHIVE_MAP""", """GitConfig""", """GitVisionConfig"""],
"""processing_git""": ["""GitProcessor"""],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = [
"""GIT_PRETRAINED_MODEL_ARCHIVE_LIST""",
"""GitForCausalLM""",
"""GitModel""",
"""GitPreTrainedModel""",
"""GitVisionModel""",
]
if TYPE_CHECKING:
from .configuration_git import GIT_PRETRAINED_CONFIG_ARCHIVE_MAP, GitConfig, GitVisionConfig
from .processing_git import GitProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_git import (
GIT_PRETRAINED_MODEL_ARCHIVE_LIST,
GitForCausalLM,
GitModel,
GitPreTrainedModel,
GitVisionModel,
)
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 |
import argparse
import os
# New Code #
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
from accelerate.utils import find_executable_batch_size
########################################################################
# This is a fully working simple example to use Accelerate,
# specifically showcasing how to ensure out-of-memory errors never
# interrupt training, and builds off the `nlp_example.py` script.
#
# This example trains a Bert base model on GLUE MRPC
# in any of the following settings (with the same script):
# - single CPU or single GPU
# - multi GPUS (using PyTorch distributed mode)
# - (multi) TPUs
# - fp16 (mixed-precision) or fp32 (normal precision)
#
# New additions from the base script can be found quickly by
# looking for the # New Code # tags
#
# To run it in each of these various modes, follow the instructions
# in the readme for examples:
# https://github.com/huggingface/accelerate/tree/main/examples
#
########################################################################
a_ = 16
a_ = 32
def a__ ( _UpperCamelCase : Accelerator ,_UpperCamelCase : int = 16 ):
__lowerCamelCase = AutoTokenizer.from_pretrained('''bert-base-cased''' )
__lowerCamelCase = load_dataset('''glue''' ,'''mrpc''' )
def tokenize_function(_UpperCamelCase : Optional[Any] ):
# max_length=None => use the model max length (it's actually the default)
__lowerCamelCase = tokenizer(examples['''sentence1'''] ,examples['''sentence2'''] ,truncation=_UpperCamelCase ,max_length=_UpperCamelCase )
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(
_UpperCamelCase ,batched=_UpperCamelCase ,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(_UpperCamelCase : Any ):
# On TPU it's best to pad everything to the same length or training will be very slow.
__lowerCamelCase = 1_28 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(
_UpperCamelCase ,padding='''longest''' ,max_length=_UpperCamelCase ,pad_to_multiple_of=_UpperCamelCase ,return_tensors='''pt''' ,)
# Instantiate dataloaders.
__lowerCamelCase = DataLoader(
tokenized_datasets['''train'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
__lowerCamelCase = DataLoader(
tokenized_datasets['''validation'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
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
a_ = mocked_dataloaders # noqa: F811
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# For testing only
if os.environ.get('''TESTING_MOCKED_DATALOADERS''' ,_UpperCamelCase ) == "1":
__lowerCamelCase = 2
# Initialize accelerator
__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'''] )
__lowerCamelCase = evaluate.load('''glue''' ,'''mrpc''' )
# New Code #
# We now can define an inner training loop function. It should take a batch size as the only parameter,
# and build the dataloaders in there.
# It also gets our decorator
@find_executable_batch_size(starting_batch_size=_UpperCamelCase )
def inner_training_loop(_UpperCamelCase : Union[str, Any] ):
# And now just move everything below under this function
# We need to bring in the Accelerator object from earlier
nonlocal accelerator
# And reset all of its attributes that could hold onto any memory:
accelerator.free_memory()
# Then we can declare the model, optimizer, and everything else:
set_seed(_UpperCamelCase )
# 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=_UpperCamelCase )
# 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=_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = get_dataloaders(_UpperCamelCase ,_UpperCamelCase )
# Instantiate scheduler
__lowerCamelCase = get_linear_schedule_with_warmup(
optimizer=_UpperCamelCase ,num_warmup_steps=1_00 ,num_training_steps=(len(_UpperCamelCase ) * num_epochs) ,)
# Prepare everything
# There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the
# prepare method.
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Now we train the model
for epoch in range(_UpperCamelCase ):
model.train()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.loss
accelerator.backward(_UpperCamelCase )
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
model.eval()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
with torch.no_grad():
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.logits.argmax(dim=-1 )
__lowerCamelCase ,__lowerCamelCase = accelerator.gather_for_metrics((predictions, batch['''labels''']) )
metric.add_batch(
predictions=_UpperCamelCase ,references=_UpperCamelCase ,)
__lowerCamelCase = metric.compute()
# Use accelerator.print to print only on the main process.
accelerator.print(F"""epoch {epoch}:""" ,_UpperCamelCase )
# New Code #
# And call it at the end with no arguments
# Note: You could also refactor this outside of your training loop function
inner_training_loop()
def a__ ( ):
__lowerCamelCase = argparse.ArgumentParser(description='''Simple example of training script.''' )
parser.add_argument(
'''--mixed_precision''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,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.''' )
__lowerCamelCase = parser.parse_args()
__lowerCamelCase = {'''lr''': 2e-5, '''num_epochs''': 3, '''seed''': 42, '''batch_size''': 16}
training_function(_UpperCamelCase ,_UpperCamelCase )
if __name__ == "__main__":
main()
| 330 | 1 |
import random
import unittest
import torch
from diffusers import IFImgaImgSuperResolutionPipeline
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_VARIATION_BATCH_PARAMS, TEXT_GUIDED_IMAGE_VARIATION_PARAMS
from ..test_pipelines_common import PipelineTesterMixin
from . import IFPipelineTesterMixin
@skip_mps
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = IFImgaImgSuperResolutionPipeline
lowerCAmelCase__ = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {"""width""", """height"""}
lowerCAmelCase__ = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS.union({"""original_image"""} )
lowerCAmelCase__ = PipelineTesterMixin.required_optional_params - {"""latents"""}
def lowerCamelCase ( self ):
'''simple docstring'''
return self._get_superresolution_dummy_components()
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=0 ):
'''simple docstring'''
if str(__UpperCAmelCase ).startswith('''mps''' ):
__lowerCamelCase = torch.manual_seed(__UpperCAmelCase )
else:
__lowerCamelCase = torch.Generator(device=__UpperCAmelCase ).manual_seed(__UpperCAmelCase )
__lowerCamelCase = floats_tensor((1, 3, 32, 32) , rng=random.Random(__UpperCAmelCase ) ).to(__UpperCAmelCase )
__lowerCamelCase = floats_tensor((1, 3, 16, 16) , rng=random.Random(__UpperCAmelCase ) ).to(__UpperCAmelCase )
__lowerCamelCase = {
'''prompt''': '''A painting of a squirrel eating a burger''',
'''image''': image,
'''original_image''': original_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 lowerCamelCase ( self ):
'''simple docstring'''
self._test_xformers_attention_forwardGenerator_pass(expected_max_diff=1E-3 )
def lowerCamelCase ( self ):
'''simple docstring'''
self._test_save_load_optional_components()
@unittest.skipIf(torch_device != '''cuda''' , reason='''float16 requires CUDA''' )
def lowerCamelCase ( self ):
'''simple docstring'''
# 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 lowerCamelCase ( self ):
'''simple docstring'''
self._test_attention_slicing_forward_pass(expected_max_diff=1E-2 )
def lowerCamelCase ( self ):
'''simple docstring'''
self._test_save_load_local()
def lowerCamelCase ( self ):
'''simple docstring'''
self._test_inference_batch_single_identical(
expected_max_diff=1E-2 , )
| 330 |
import logging
import os
import threading
import time
try:
import warnings
except ImportError:
a_ = None
try:
import msvcrt
except ImportError:
a_ = None
try:
import fcntl
except ImportError:
a_ = None
# Backward compatibility
# ------------------------------------------------
try:
TimeoutError
except NameError:
a_ = OSError
# Data
# ------------------------------------------------
a_ = [
"""Timeout""",
"""BaseFileLock""",
"""WindowsFileLock""",
"""UnixFileLock""",
"""SoftFileLock""",
"""FileLock""",
]
a_ = """3.0.12"""
a_ = None
def a__ ( ):
global _logger
__lowerCamelCase = _logger or logging.getLogger(__name__ )
return _logger
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = lock_file
return None
def __str__( self ):
'''simple docstring'''
__lowerCamelCase = F"""The file lock '{self.lock_file}' could not be acquired."""
return temp
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = lock
return None
def __enter__( self ):
'''simple docstring'''
return self.lock
def __exit__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
self.lock.release()
return None
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
__lowerCamelCase = max_filename_length if max_filename_length is not None else 255
# Hash the filename if it's too long
__lowerCamelCase = self.hash_filename_if_too_long(__UpperCAmelCase , __UpperCAmelCase )
# The path to the lock file.
__lowerCamelCase = lock_file
# The file descriptor for the *_lock_file* as it is returned by the
# os.open() function.
# This file lock is only NOT None, if the object currently holds the
# lock.
__lowerCamelCase = None
# The default timeout value.
__lowerCamelCase = timeout
# We use this lock primarily for the lock counter.
__lowerCamelCase = threading.Lock()
# The lock counter is used for implementing the nested locking
# mechanism. Whenever the lock is acquired, the counter is increased and
# the lock is only released, when this value is 0 again.
__lowerCamelCase = 0
return None
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._lock_file
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._timeout
@timeout.setter
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = float(__UpperCAmelCase )
return None
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError()
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError()
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._lock_file_fd is not None
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=0.05 ):
'''simple docstring'''
# Use the default timeout, if no timeout is provided.
if timeout is None:
__lowerCamelCase = self.timeout
# Increment the number right at the beginning.
# We can still undo it, if something fails.
with self._thread_lock:
self._lock_counter += 1
__lowerCamelCase = id(self )
__lowerCamelCase = self._lock_file
__lowerCamelCase = time.time()
try:
while True:
with self._thread_lock:
if not self.is_locked:
logger().debug(F"""Attempting to acquire lock {lock_id} on {lock_filename}""" )
self._acquire()
if self.is_locked:
logger().debug(F"""Lock {lock_id} acquired on {lock_filename}""" )
break
elif timeout >= 0 and time.time() - start_time > timeout:
logger().debug(F"""Timeout on acquiring lock {lock_id} on {lock_filename}""" )
raise Timeout(self._lock_file )
else:
logger().debug(
F"""Lock {lock_id} not acquired on {lock_filename}, waiting {poll_intervall} seconds ...""" )
time.sleep(__UpperCAmelCase )
except: # noqa
# Something did go wrong, so decrement the counter.
with self._thread_lock:
__lowerCamelCase = max(0 , self._lock_counter - 1 )
raise
return _Acquire_ReturnProxy(lock=self )
def lowerCamelCase ( self , __UpperCAmelCase=False ):
'''simple docstring'''
with self._thread_lock:
if self.is_locked:
self._lock_counter -= 1
if self._lock_counter == 0 or force:
__lowerCamelCase = id(self )
__lowerCamelCase = self._lock_file
logger().debug(F"""Attempting to release lock {lock_id} on {lock_filename}""" )
self._release()
__lowerCamelCase = 0
logger().debug(F"""Lock {lock_id} released on {lock_filename}""" )
return None
def __enter__( self ):
'''simple docstring'''
self.acquire()
return self
def __exit__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
self.release()
return None
def __del__( self ):
'''simple docstring'''
self.release(force=__UpperCAmelCase )
return None
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = os.path.basename(__UpperCAmelCase )
if len(__UpperCAmelCase ) > max_length and max_length > 0:
__lowerCamelCase = os.path.dirname(__UpperCAmelCase )
__lowerCamelCase = str(hash(__UpperCAmelCase ) )
__lowerCamelCase = filename[: max_length - len(__UpperCAmelCase ) - 8] + '''...''' + hashed_filename + '''.lock'''
return os.path.join(__UpperCAmelCase , __UpperCAmelCase )
else:
return path
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
from .file_utils import relative_to_absolute_path
super().__init__(__UpperCAmelCase , timeout=__UpperCAmelCase , max_filename_length=__UpperCAmelCase )
__lowerCamelCase = '''\\\\?\\''' + relative_to_absolute_path(self.lock_file )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_RDWR | os.O_CREAT | os.O_TRUNC
try:
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
except OSError:
pass
else:
try:
msvcrt.locking(__UpperCAmelCase , msvcrt.LK_NBLCK , 1 )
except OSError:
os.close(__UpperCAmelCase )
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self._lock_file_fd
__lowerCamelCase = None
msvcrt.locking(__UpperCAmelCase , msvcrt.LK_UNLCK , 1 )
os.close(__UpperCAmelCase )
try:
os.remove(self._lock_file )
# Probably another instance of the application
# that acquired the file lock.
except OSError:
pass
return None
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
__lowerCamelCase = os.statvfs(os.path.dirname(__UpperCAmelCase ) ).f_namemax
super().__init__(__UpperCAmelCase , timeout=__UpperCAmelCase , max_filename_length=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_RDWR | os.O_CREAT | os.O_TRUNC
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
try:
fcntl.flock(__UpperCAmelCase , fcntl.LOCK_EX | fcntl.LOCK_NB )
except OSError:
os.close(__UpperCAmelCase )
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
# Do not remove the lockfile:
#
# https://github.com/benediktschmitt/py-filelock/issues/31
# https://stackoverflow.com/questions/17708885/flock-removing-locked-file-without-race-condition
__lowerCamelCase = self._lock_file_fd
__lowerCamelCase = None
fcntl.flock(__UpperCAmelCase , fcntl.LOCK_UN )
os.close(__UpperCAmelCase )
return None
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_WRONLY | os.O_CREAT | os.O_EXCL | os.O_TRUNC
try:
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
except OSError:
pass
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
os.close(self._lock_file_fd )
__lowerCamelCase = None
try:
os.remove(self._lock_file )
# The file is already deleted and that's what we want.
except OSError:
pass
return None
a_ = None
if msvcrt:
a_ = WindowsFileLock
elif fcntl:
a_ = UnixFileLock
else:
a_ = SoftFileLock
if warnings is not None:
warnings.warn("""only soft file lock is available""")
| 330 | 1 |
from typing import List, Optional, Union
import numpy as np
from ....audio_utils import mel_filter_bank, optimal_fft_length, spectrogram, window_function
from ....feature_extraction_sequence_utils import SequenceFeatureExtractor
from ....feature_extraction_utils import BatchFeature
from ....file_utils import PaddingStrategy, TensorType
from ....utils import logging
a_ = logging.get_logger(__name__)
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""input_features""", """attention_mask"""]
def __init__( self , __UpperCAmelCase=80 , __UpperCAmelCase=16000 , __UpperCAmelCase=0.0 , __UpperCAmelCase=10 , __UpperCAmelCase=25 , __UpperCAmelCase="hamming_window" , __UpperCAmelCase=32_768.0 , __UpperCAmelCase=0.97 , __UpperCAmelCase=1.0 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=False , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(feature_size=__UpperCAmelCase , sampling_rate=__UpperCAmelCase , padding_value=__UpperCAmelCase , **__UpperCAmelCase )
__lowerCamelCase = feature_size
__lowerCamelCase = sampling_rate
__lowerCamelCase = padding_value
__lowerCamelCase = hop_length
__lowerCamelCase = win_length
__lowerCamelCase = frame_signal_scale
__lowerCamelCase = preemphasis_coeff
__lowerCamelCase = mel_floor
__lowerCamelCase = normalize_means
__lowerCamelCase = normalize_vars
__lowerCamelCase = win_function
__lowerCamelCase = return_attention_mask
__lowerCamelCase = win_length * sampling_rate // 1000
__lowerCamelCase = hop_length * sampling_rate // 1000
__lowerCamelCase = optimal_fft_length(self.sample_size )
__lowerCamelCase = (self.n_fft // 2) + 1
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if self.win_function == "hamming_window":
__lowerCamelCase = window_function(window_length=self.sample_size , name=self.win_function , periodic=__UpperCAmelCase )
else:
__lowerCamelCase = window_function(window_length=self.sample_size , name=self.win_function )
__lowerCamelCase = mel_filter_bank(
num_frequency_bins=self.n_freqs , num_mel_filters=self.feature_size , min_frequency=0.0 , max_frequency=self.sampling_rate / 2.0 , sampling_rate=self.sampling_rate , )
__lowerCamelCase = spectrogram(
one_waveform * self.frame_signal_scale , window=__UpperCAmelCase , frame_length=self.sample_size , hop_length=self.sample_stride , fft_length=self.n_fft , center=__UpperCAmelCase , preemphasis=self.preemphasis_coeff , mel_filters=__UpperCAmelCase , mel_floor=self.mel_floor , log_mel='''log''' , )
return msfc_features.T
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# make sure we normalize float32 arrays
if self.normalize_means:
__lowerCamelCase = x[:input_length].mean(axis=0 )
__lowerCamelCase = np.subtract(__UpperCAmelCase , __UpperCAmelCase )
if self.normalize_vars:
__lowerCamelCase = x[:input_length].std(axis=0 )
__lowerCamelCase = np.divide(__UpperCAmelCase , __UpperCAmelCase )
if input_length < x.shape[0]:
__lowerCamelCase = padding_value
# make sure array is in float32
__lowerCamelCase = x.astype(np.floataa )
return x
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None ):
'''simple docstring'''
__lowerCamelCase = attention_mask.sum(-1 ) if attention_mask is not None else [x.shape[0] for x in input_features]
return [self._normalize_one(__UpperCAmelCase , __UpperCAmelCase , self.padding_value ) for x, n in zip(__UpperCAmelCase , __UpperCAmelCase )]
def __call__( self , __UpperCAmelCase , __UpperCAmelCase = False , __UpperCAmelCase = None , __UpperCAmelCase = False , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
if sampling_rate is not None:
if sampling_rate != self.sampling_rate:
raise ValueError(
F"""The model corresponding to this feature extractor: {self} was trained using a sampling rate of"""
F""" {self.sampling_rate}. Please make sure that the provided `raw_speech` input was sampled with"""
F""" {self.sampling_rate} and not {sampling_rate}.""" )
else:
logger.warning(
'''It is strongly recommended to pass the ``sampling_rate`` argument to this function. '''
'''Failing to do so can result in silent errors that might be hard to debug.''' )
__lowerCamelCase = isinstance(__UpperCAmelCase , np.ndarray ) and len(raw_speech.shape ) > 1
if is_batched_numpy and len(raw_speech.shape ) > 2:
raise ValueError(F"""Only mono-channel audio is supported for input to {self}""" )
__lowerCamelCase = is_batched_numpy or (
isinstance(__UpperCAmelCase , (list, tuple) ) and (isinstance(raw_speech[0] , (np.ndarray, tuple, list) ))
)
if is_batched:
__lowerCamelCase = [np.asarray(__UpperCAmelCase , dtype=np.floataa ) for speech in raw_speech]
elif not is_batched and not isinstance(__UpperCAmelCase , np.ndarray ):
__lowerCamelCase = np.asarray(__UpperCAmelCase , dtype=np.floataa )
elif isinstance(__UpperCAmelCase , np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ):
__lowerCamelCase = raw_speech.astype(np.floataa )
# always return batch
if not is_batched:
__lowerCamelCase = [raw_speech]
# extract fbank features
__lowerCamelCase = [self._extract_mfsc_features(__UpperCAmelCase ) for one_waveform in raw_speech]
# convert into correct format for padding
__lowerCamelCase = BatchFeature({'''input_features''': features} )
__lowerCamelCase = self.pad(
__UpperCAmelCase , padding=__UpperCAmelCase , max_length=__UpperCAmelCase , truncation=__UpperCAmelCase , pad_to_multiple_of=__UpperCAmelCase , return_attention_mask=__UpperCAmelCase , **__UpperCAmelCase , )
# make sure list is in array format
__lowerCamelCase = padded_inputs.get('''input_features''' )
if isinstance(input_features[0] , __UpperCAmelCase ):
__lowerCamelCase = [np.asarray(__UpperCAmelCase , dtype=np.floataa ) for feature in input_features]
__lowerCamelCase = padded_inputs.get('''attention_mask''' )
if attention_mask is not None:
__lowerCamelCase = [np.asarray(__UpperCAmelCase , dtype=np.intaa ) for array in attention_mask]
if self.normalize_means or self.normalize_vars:
__lowerCamelCase = (
np.array(__UpperCAmelCase , dtype=np.intaa )
if self._get_padding_strategies(__UpperCAmelCase , max_length=__UpperCAmelCase ) is not PaddingStrategy.DO_NOT_PAD
and padding
else None
)
__lowerCamelCase = self.normalize(
padded_inputs['''input_features'''] , attention_mask=__UpperCAmelCase )
if return_tensors is not None:
__lowerCamelCase = padded_inputs.convert_to_tensors(__UpperCAmelCase )
return padded_inputs
| 330 |
import copy
import inspect
import unittest
import numpy as np
from huggingface_hub import hf_hub_download
from transformers import TimesformerConfig
from transformers.models.auto import get_values
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 (
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING,
TimesformerForVideoClassification,
TimesformerModel,
)
from transformers.models.timesformer.modeling_timesformer import TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from transformers import VideoMAEImageProcessor
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=13 , __UpperCAmelCase=10 , __UpperCAmelCase=3 , __UpperCAmelCase=2 , __UpperCAmelCase=2 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=32 , __UpperCAmelCase=5 , __UpperCAmelCase=4 , __UpperCAmelCase=37 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=10 , __UpperCAmelCase=0.02 , __UpperCAmelCase="divided_space_time" , __UpperCAmelCase=None , ):
'''simple docstring'''
__lowerCamelCase = parent
__lowerCamelCase = batch_size
__lowerCamelCase = image_size
__lowerCamelCase = num_channels
__lowerCamelCase = patch_size
__lowerCamelCase = num_frames
__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 = attention_type
__lowerCamelCase = initializer_range
__lowerCamelCase = scope
__lowerCamelCase = num_labels
# in TimeSformer, the number of spatial tokens equals num_frames * num_patches per frame + 1 CLS token
__lowerCamelCase = (image_size // patch_size) ** 2
__lowerCamelCase = (num_frames) * self.num_patches_per_frame + 1
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = floats_tensor(
[self.batch_size, self.num_frames, 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 lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , num_frames=self.num_frames , 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 , initializer_range=self.initializer_range , attention_type=self.attention_type , )
__lowerCamelCase = self.num_labels
return config
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
# verify the logits shape
__lowerCamelCase = torch.Size((self.batch_size, self.num_labels) )
self.parent.assertEqual(result.logits.shape , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = config_and_inputs
__lowerCamelCase = {'''pixel_values''': pixel_values}
return config, inputs_dict
@require_torch
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (TimesformerModel, TimesformerForVideoClassification) if is_torch_available() else ()
lowerCAmelCase__ = (
{"""feature-extraction""": TimesformerModel, """video-classification""": TimesformerForVideoClassification}
if is_torch_available()
else {}
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerModelTester(self )
__lowerCamelCase = ConfigTester(
self , config_class=__UpperCAmelCase , has_text_modality=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = copy.deepcopy(__UpperCAmelCase )
if return_labels:
if model_class in get_values(__UpperCAmelCase ):
__lowerCamelCase = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=__UpperCAmelCase )
return inputs_dict
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
@unittest.skip(reason='''TimeSformer does not use inputs_embeds''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
__lowerCamelCase = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(__UpperCAmelCase , nn.Linear ) )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
__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] , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_video_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = TimesformerModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.has_attentions:
pass
else:
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCamelCase = True
for model_class in self.all_model_classes:
__lowerCamelCase = self.model_tester.seq_length
__lowerCamelCase = self.model_tester.num_frames
__lowerCamelCase = True
__lowerCamelCase = False
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
__lowerCamelCase = len(__UpperCAmelCase )
# Check attention is always last and order is fine
__lowerCamelCase = True
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
self.assertEqual(out_len + 1 , len(__UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
def lowerCamelCase ( self ):
'''simple docstring'''
def check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.hidden_states
__lowerCamelCase = self.model_tester.num_hidden_layers + 1
self.assertEqual(len(__UpperCAmelCase ) , __UpperCAmelCase )
__lowerCamelCase = self.model_tester.seq_length
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [seq_length, self.model_tester.hidden_size] , )
__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(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCamelCase = True
check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def a__ ( ):
__lowerCamelCase = hf_hub_download(
repo_id='''hf-internal-testing/spaghetti-video''' ,filename='''eating_spaghetti.npy''' ,repo_type='''dataset''' )
__lowerCamelCase = np.load(_UpperCamelCase )
return list(_UpperCamelCase )
@require_torch
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
@cached_property
def lowerCamelCase ( self ):
'''simple docstring'''
# logits were tested with a different mean and std, so we use the same here
return (
VideoMAEImageProcessor(image_mean=[0.5, 0.5, 0.5] , image_std=[0.5, 0.5, 0.5] )
if is_vision_available()
else None
)
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification.from_pretrained('''facebook/timesformer-base-finetuned-k400''' ).to(
__UpperCAmelCase )
__lowerCamelCase = self.default_image_processor
__lowerCamelCase = prepare_video()
__lowerCamelCase = image_processor(video[:8] , return_tensors='''pt''' ).to(__UpperCAmelCase )
# forward pass
with torch.no_grad():
__lowerCamelCase = model(**__UpperCAmelCase )
# verify the logits
__lowerCamelCase = torch.Size((1, 400) )
self.assertEqual(outputs.logits.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor([-0.3_016, -0.7_713, -0.4_205] ).to(__UpperCAmelCase )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
import argparse
import dataclasses
import json
import logging
import os
import shutil
from typing import List, Optional
import datasets
from accelerate import Accelerator
from datasets import load_dataset
from finetuning import finetune
from tqdm.auto import tqdm
import transformers
from transformers import AutoConfig, set_seed
from transformers.trainer_utils import IntervalStrategy
a_ = logging.getLogger(__name__)
a_ = """pytorch_model.bin"""
@dataclasses.dataclass
class __lowerCAmelCase :
lowerCAmelCase__ = dataclasses.field(
metadata={"""help""": """Path to pretrained model or model identifier from huggingface.co/models."""} )
lowerCAmelCase__ = dataclasses.field(
default=lowerCAmelCase__ , metadata={"""help""": """Where do you want to store the pretrained models downloaded from huggingface.co."""} , )
@dataclasses.dataclass
class __lowerCAmelCase :
lowerCAmelCase__ = dataclasses.field(metadata={"""help""": """A csv or a json file containing the training data."""} )
lowerCAmelCase__ = dataclasses.field(metadata={"""help""": """A csv or a json file containing the data to predict on."""} )
lowerCAmelCase__ = dataclasses.field(
default=lowerCAmelCase__ , metadata={"""help""": """A csv or a json file containing the validation data."""} )
lowerCAmelCase__ = dataclasses.field(
default=lowerCAmelCase__ , metadata={"""help""": """The name of the task to train on."""} , )
lowerCAmelCase__ = dataclasses.field(
default=lowerCAmelCase__ , metadata={"""help""": """The list of labels for the task."""} )
@dataclasses.dataclass
class __lowerCAmelCase :
lowerCAmelCase__ = dataclasses.field(
metadata={"""help""": """The output directory where the model predictions and checkpoints will be written."""} )
lowerCAmelCase__ = dataclasses.field(
default="""accuracy""" , metadata={"""help""": """The evaluation metric used for the task."""} )
lowerCAmelCase__ = dataclasses.field(
default="""no""" , metadata={
"""help""": """The evaluation strategy to adopt during training. Possible values are: [\"no\", \"step\", \"epoch]"""
} , )
lowerCAmelCase__ = dataclasses.field(
default=1_0 , metadata={"""help""": """Number of evaluation calls with no improvement after which training will be stopped."""} , )
lowerCAmelCase__ = dataclasses.field(
default=0.0 , metadata={
"""help""": """How much the specified evaluation metric must improve to satisfy early stopping conditions."""
} , )
lowerCAmelCase__ = dataclasses.field(
default=lowerCAmelCase__ , metadata={"""help""": """Whether to filter the pseudo-labeled data based on the confidence score."""} , )
lowerCAmelCase__ = dataclasses.field(
default=lowerCAmelCase__ , metadata={"""help""": """Whether to filter the pseudo-labeled data based on the validation performance."""} , )
lowerCAmelCase__ = dataclasses.field(
default=lowerCAmelCase__ , metadata={"""help""": """Whether to fine-tune on labeled data after pseudo training."""} , )
lowerCAmelCase__ = dataclasses.field(
default=0.0 , metadata={"""help""": """Confidence threshold for pseudo-labeled data filtering."""} , )
lowerCAmelCase__ = dataclasses.field(
default=1_0_0 , metadata={"""help""": """Number of evaluation calls with no improvement after which training will be stopped."""} , )
lowerCAmelCase__ = dataclasses.field(
default=lowerCAmelCase__ , metadata={"""help""": """Random seed for initialization."""} , )
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : str ,_UpperCamelCase : List[str] ,_UpperCamelCase : List[Any] ,_UpperCamelCase : int ,_UpperCamelCase : str ):
__lowerCamelCase = datasets.concatenate_datasets([infer_input, infer_output] ,axis=1 )
if args.do_filter_by_confidence:
__lowerCamelCase = dataset.filter(lambda _UpperCamelCase : example["probability"] > args.confidence_threshold )
if args.do_filter_by_val_performance:
assert eval_result >= 0.0 and eval_result <= 1.0
__lowerCamelCase = int(eval_result * len(_UpperCamelCase ) )
print(_UpperCamelCase )
__lowerCamelCase = dataset.sort('''probability''' ,reverse=_UpperCamelCase )
__lowerCamelCase = dataset.select(range(_UpperCamelCase ) )
__lowerCamelCase = dataset.remove_columns(['''label''', '''probability'''] )
__lowerCamelCase = dataset.rename_column('''prediction''' ,'''label''' )
__lowerCamelCase = dataset.map(lambda _UpperCamelCase : {"label": idalabel[example["label"]]} )
__lowerCamelCase = dataset.shuffle(seed=args.seed )
__lowerCamelCase = os.path.join(_UpperCamelCase ,F"""train_pseudo.{args.data_file_extension}""" )
if args.data_file_extension == "csv":
dataset.to_csv(_UpperCamelCase ,index=_UpperCamelCase )
else:
dataset.to_json(_UpperCamelCase )
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : List[Any] ,_UpperCamelCase : Optional[int] ,_UpperCamelCase : Dict ,**_UpperCamelCase : Optional[int] ):
__lowerCamelCase = Accelerator()
# Make one log on every process with the configuration for debugging.
logging.basicConfig(
format='''%(asctime)s - %(levelname)s - %(name)s - %(message)s''' ,datefmt='''%m/%d/%Y %H:%M:%S''' ,level=logging.INFO ,)
logger.info(accelerator.state )
# Setup logging, we only want one process per machine to log things on the
# screen. accelerator.is_local_main_process is only True for one process per
# machine.
logger.setLevel(logging.INFO if accelerator.is_local_main_process else logging.ERROR )
if accelerator.is_local_main_process:
datasets.utils.logging.set_verbosity_warning()
transformers.utils.logging.set_verbosity_info()
else:
datasets.utils.logging.set_verbosity_error()
transformers.utils.logging.set_verbosity_error()
__lowerCamelCase = STModelArguments(model_name_or_path=_UpperCamelCase )
__lowerCamelCase = STDataArguments(train_file=_UpperCamelCase ,infer_file=_UpperCamelCase )
__lowerCamelCase = STTrainingArguments(output_dir=_UpperCamelCase )
__lowerCamelCase = argparse.Namespace()
for arg_class in (model_args, data_args, training_args):
for key, value in vars(_UpperCamelCase ).items():
setattr(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
for key, value in kwargs.items():
if hasattr(_UpperCamelCase ,_UpperCamelCase ):
setattr(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Sanity checks
__lowerCamelCase = {}
__lowerCamelCase = None
# You need to provide the training data and the data to predict on
assert args.train_file is not None
assert args.infer_file is not None
__lowerCamelCase = args.train_file
__lowerCamelCase = args.infer_file
if args.evaluation_strategy != IntervalStrategy.NO.value:
assert args.eval_file is not None
__lowerCamelCase = args.eval_file
for key in data_files:
__lowerCamelCase = data_files[key].split('''.''' )[-1]
assert extension in ["csv", "json"], F"""`{key}_file` should be a csv or a json file."""
if args.data_file_extension is None:
__lowerCamelCase = extension
else:
assert extension == args.data_file_extension, F"""`{key}_file` should be a {args.data_file_extension} file`."""
assert (
args.eval_metric in datasets.list_metrics()
), F"""{args.eval_metric} not in the list of supported metrics {datasets.list_metrics()}."""
# If passed along, set the training seed now.
if args.seed is not None:
set_seed(args.seed )
logger.info('''Creating the initial data directory for self-training...''' )
__lowerCamelCase = F"""{args.output_dir}/self-train_iter-{{}}""".format
__lowerCamelCase = data_dir_format(0 )
if accelerator.is_main_process:
if args.output_dir is not None:
os.makedirs(args.output_dir ,exist_ok=_UpperCamelCase )
os.makedirs(_UpperCamelCase ,exist_ok=_UpperCamelCase )
accelerator.wait_for_everyone()
__lowerCamelCase = None
__lowerCamelCase = None
__lowerCamelCase = 0
__lowerCamelCase = False
# Show the progress bar
__lowerCamelCase = tqdm(range(args.max_selftrain_iterations ) ,disable=not accelerator.is_local_main_process )
# Self-train
for iteration in range(0 ,int(args.max_selftrain_iterations ) ):
__lowerCamelCase = data_dir_format(_UpperCamelCase )
assert os.path.exists(_UpperCamelCase )
# Stage 1: initial fine-tuning for iteration = 0 or pseudo-training for
# iteration > 0
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''stage-1''' )
__lowerCamelCase = {
'''accelerator''': accelerator,
'''model_name_or_path''': args.model_name_or_path,
'''cache_dir''': args.cache_dir,
'''do_train''': True,
'''train_file''': data_files['''train'''] if iteration == 0 else data_files['''train_pseudo'''],
'''do_eval''': True if args.eval_file is not None else False,
'''eval_file''': data_files['''eval'''],
'''do_predict''': True,
'''infer_file''': data_files['''infer'''],
'''task_name''': args.task_name,
'''label_list''': args.label_list,
'''output_dir''': current_output_dir,
'''eval_metric''': args.eval_metric,
'''evaluation_strategy''': args.evaluation_strategy,
'''early_stopping_patience''': args.early_stopping_patience,
'''early_stopping_threshold''': args.early_stopping_threshold,
'''seed''': args.seed,
}
# Add additional training arguments
for key, value in kwargs.items():
if key not in arguments_dict and not hasattr(_UpperCamelCase ,_UpperCamelCase ):
arguments_dict.update({key: value} )
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''best-checkpoint''' ,_UpperCamelCase )
if os.path.exists(_UpperCamelCase ):
logger.info(
'''Found existing model checkpoint at %s. Skipping self-training: iteration: %d, stage: 1.''' ,_UpperCamelCase ,_UpperCamelCase ,)
else:
logger.info('''***** Running self-training: iteration: %d, stage: 1 *****''' ,_UpperCamelCase )
finetune(**_UpperCamelCase )
accelerator.wait_for_everyone()
assert os.path.exists(_UpperCamelCase )
logger.info('''Self-training job completed: iteration: %d, stage: 1.''' ,_UpperCamelCase )
if iteration > 0 and args.finetune_on_labeled_data:
# Stage 2 (optional): fine-tuning on the original labeled data
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''best-checkpoint''' )
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''stage-2''' )
# Update arguments_dict
__lowerCamelCase = model_path
__lowerCamelCase = data_files['''train''']
__lowerCamelCase = current_output_dir
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''best-checkpoint''' ,_UpperCamelCase )
if os.path.exists(_UpperCamelCase ):
logger.info(
'''Found existing model checkpoint at %s. Skipping self-training: iteration: %d, stage: 2.''' ,_UpperCamelCase ,_UpperCamelCase ,)
else:
logger.info('''***** Running self-training: iteration: %d, stage: 2 *****''' ,_UpperCamelCase )
finetune(**_UpperCamelCase )
accelerator.wait_for_everyone()
assert os.path.exists(_UpperCamelCase )
logger.info('''Self-training job completed: iteration: %d, stage: 2.''' ,_UpperCamelCase )
__lowerCamelCase = iteration
__lowerCamelCase = data_dir_format(iteration + 1 )
__lowerCamelCase = AutoConfig.from_pretrained(os.path.join(_UpperCamelCase ,'''best-checkpoint''' ) )
__lowerCamelCase = config.idalabel
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''eval_results_best-checkpoint.json''' )
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''test_results_best-checkpoint.json''' )
assert os.path.exists(_UpperCamelCase )
with open(_UpperCamelCase ,'''r''' ) as f:
__lowerCamelCase = float(json.load(_UpperCamelCase )[args.eval_metric] )
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''infer_output_best-checkpoint.csv''' )
assert os.path.exists(_UpperCamelCase )
# Loading the dataset from local csv or json files.
__lowerCamelCase = load_dataset(args.data_file_extension ,data_files={'''data''': data_files['''infer''']} )['''data''']
__lowerCamelCase = load_dataset('''csv''' ,data_files={'''data''': infer_output_file} )['''data''']
if accelerator.is_main_process:
os.makedirs(_UpperCamelCase ,exist_ok=_UpperCamelCase )
shutil.copy(_UpperCamelCase ,os.path.join(_UpperCamelCase ,F"""eval_results_iter-{iteration}.json""" ) )
if os.path.exists(_UpperCamelCase ):
shutil.copy(_UpperCamelCase ,os.path.join(_UpperCamelCase ,F"""test_results_iter-{iteration}.json""" ) )
create_pseudo_labeled_data(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
accelerator.wait_for_everyone()
__lowerCamelCase = os.path.join(_UpperCamelCase ,F"""train_pseudo.{args.data_file_extension}""" )
if args.evaluation_strategy != IntervalStrategy.NO.value:
__lowerCamelCase = eval_result
if best_iteration is None:
__lowerCamelCase = new_iteration
__lowerCamelCase = new_eval_result
else:
if new_eval_result - best_eval_result > args.early_stopping_threshold:
__lowerCamelCase = new_iteration
__lowerCamelCase = new_eval_result
__lowerCamelCase = 0
else:
if new_eval_result == best_eval_result:
__lowerCamelCase = new_iteration
__lowerCamelCase = new_eval_result
early_stopping_patience_counter += 1
if early_stopping_patience_counter >= args.early_stopping_patience:
__lowerCamelCase = True
progress_bar.update(1 )
if should_training_stop:
break
if best_iteration is not None:
# Save the best iteration
logger.info('''Best iteration: %d''' ,_UpperCamelCase )
logger.info('''Best evaluation result: %s = %f''' ,args.eval_metric ,_UpperCamelCase )
accelerator.wait_for_everyone()
if accelerator.is_main_process:
shutil.copy(
os.path.join(_UpperCamelCase ,F"""eval_results_iter-{iteration}.json""" ) ,os.path.join(_UpperCamelCase ,'''eval_results_best-iteration.json''' ) ,)
else:
# Assume that the last iteration is the best
logger.info('''Best iteration: %d''' ,args.max_selftrain_iterations - 1 )
logger.info('''Best evaluation result: %s = %f''' ,args.eval_metric ,_UpperCamelCase )
accelerator.wait_for_everyone()
if accelerator.is_main_process:
shutil.copy(
os.path.join(_UpperCamelCase ,F"""eval_results_iter-{args.max_selftrain_iterations - 1}.json""" ) ,os.path.join(_UpperCamelCase ,'''eval_results_best-iteration.json''' ) ,)
| 330 |
def a__ ( _UpperCamelCase : int ):
if not isinstance(_UpperCamelCase ,_UpperCamelCase ):
__lowerCamelCase = F"""Input value of [number={number}] must be an integer"""
raise TypeError(_UpperCamelCase )
if number < 0:
return False
__lowerCamelCase = number * number
while number > 0:
if number % 10 != number_square % 10:
return False
number //= 10
number_square //= 10
return True
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 | 1 |
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : Any ):
__lowerCamelCase = ''''''
for i in table:
res += inp[i - 1]
return res
def a__ ( _UpperCamelCase : Optional[int] ):
return data[1:] + data[0]
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : Dict ):
__lowerCamelCase = ''''''
for i in range(len(_UpperCamelCase ) ):
if a[i] == b[i]:
res += "0"
else:
res += "1"
return res
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : List[Any] ):
__lowerCamelCase = int('''0b''' + data[0] + data[-1] ,2 )
__lowerCamelCase = int('''0b''' + data[1:3] ,2 )
return bin(s[row][col] )[2:]
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : Dict ,_UpperCamelCase : str ,_UpperCamelCase : int ,_UpperCamelCase : List[str] ):
__lowerCamelCase = message[:4]
__lowerCamelCase = message[4:]
__lowerCamelCase = apply_table(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = xor(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = apply_sbox(_UpperCamelCase ,temp[:4] ) # noqa: E741
__lowerCamelCase = apply_sbox(_UpperCamelCase ,temp[4:] )
__lowerCamelCase = '''0''' * (2 - len(_UpperCamelCase )) + l # noqa: E741
__lowerCamelCase = '''0''' * (2 - len(_UpperCamelCase )) + r
__lowerCamelCase = apply_table(l + r ,_UpperCamelCase )
__lowerCamelCase = xor(_UpperCamelCase ,_UpperCamelCase )
return temp + right
if __name__ == "__main__":
a_ = input("""Enter 10 bit key: """)
a_ = input("""Enter 8 bit message: """)
a_ = [6, 3, 7, 4, 8, 5, 10, 9]
a_ = [3, 5, 2, 7, 4, 10, 1, 9, 8, 6]
a_ = [2, 4, 3, 1]
a_ = [2, 6, 3, 1, 4, 8, 5, 7]
a_ = [4, 1, 3, 5, 7, 2, 8, 6]
a_ = [4, 1, 2, 3, 2, 3, 4, 1]
a_ = [[1, 0, 3, 2], [3, 2, 1, 0], [0, 2, 1, 3], [3, 1, 3, 2]]
a_ = [[0, 1, 2, 3], [2, 0, 1, 3], [3, 0, 1, 0], [2, 1, 0, 3]]
# key generation
a_ = apply_table(key, paa_table)
a_ = temp[:5]
a_ = temp[5:]
a_ = left_shift(left)
a_ = left_shift(right)
a_ = apply_table(left + right, pa_table)
a_ = left_shift(left)
a_ = left_shift(right)
a_ = left_shift(left)
a_ = left_shift(right)
a_ = apply_table(left + right, pa_table)
# encryption
a_ = apply_table(message, IP)
a_ = function(expansion, sa, sa, keya, temp)
a_ = temp[4:] + temp[:4]
a_ = function(expansion, sa, sa, keya, temp)
a_ = apply_table(temp, IP_inv)
print("""Cipher text is:""", CT)
# decryption
a_ = apply_table(CT, IP)
a_ = function(expansion, sa, sa, keya, temp)
a_ = temp[4:] + temp[:4]
a_ = function(expansion, sa, sa, keya, temp)
a_ = apply_table(temp, IP_inv)
print("""Plain text after decypting is:""", PT)
| 330 |
import gc
import unittest
from parameterized import parameterized
from diffusers import FlaxUNetaDConditionModel
from diffusers.utils import is_flax_available
from diffusers.utils.testing_utils import load_hf_numpy, require_flax, slow
if is_flax_available():
import jax
import jax.numpy as jnp
@slow
@require_flax
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
return F"""gaussian_noise_s={seed}_shape={"_".join([str(__UpperCAmelCase ) for s in shape] )}.npy"""
def lowerCamelCase ( self ):
'''simple docstring'''
# clean up the VRAM after each test
super().tearDown()
gc.collect()
def lowerCamelCase ( self , __UpperCAmelCase=0 , __UpperCAmelCase=(4, 4, 64, 64) , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = jnp.array(load_hf_numpy(self.get_file_format(__UpperCAmelCase , __UpperCAmelCase ) ) , dtype=__UpperCAmelCase )
return image
def lowerCamelCase ( self , __UpperCAmelCase=False , __UpperCAmelCase="CompVis/stable-diffusion-v1-4" ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = '''bf16''' if fpaa else None
__lowerCamelCase ,__lowerCamelCase = FlaxUNetaDConditionModel.from_pretrained(
__UpperCAmelCase , subfolder='''unet''' , dtype=__UpperCAmelCase , revision=__UpperCAmelCase )
return model, params
def lowerCamelCase ( self , __UpperCAmelCase=0 , __UpperCAmelCase=(4, 77, 768) , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = jnp.array(load_hf_numpy(self.get_file_format(__UpperCAmelCase , __UpperCAmelCase ) ) , dtype=__UpperCAmelCase )
return hidden_states
@parameterized.expand(
[
# fmt: off
[83, 4, [-0.2_323, -0.1_304, 0.0_813, -0.3_093, -0.0_919, -0.1_571, -0.1_125, -0.5_806]],
[17, 0.55, [-0.0_831, -0.2_443, 0.0_901, -0.0_919, 0.3_396, 0.0_103, -0.3_743, 0.0_701]],
[8, 0.89, [-0.4_863, 0.0_859, 0.0_875, -0.1_658, 0.9_199, -0.0_114, 0.4_839, 0.4_639]],
[3, 1000, [-0.5_649, 0.2_402, -0.5_518, 0.1_248, 1.1_328, -0.2_443, -0.0_325, -1.0_078]],
# fmt: on
] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_unet_model(model_id='''CompVis/stable-diffusion-v1-4''' , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_latents(__UpperCAmelCase , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_encoder_hidden_states(__UpperCAmelCase , fpaa=__UpperCAmelCase )
__lowerCamelCase = model.apply(
{'''params''': params} , __UpperCAmelCase , jnp.array(__UpperCAmelCase , dtype=jnp.intaa ) , encoder_hidden_states=__UpperCAmelCase , ).sample
assert sample.shape == latents.shape
__lowerCamelCase = jnp.asarray(jax.device_get((sample[-1, -2:, -2:, :2].flatten()) ) , dtype=jnp.floataa )
__lowerCamelCase = jnp.array(__UpperCAmelCase , dtype=jnp.floataa )
# Found torch (float16) and flax (bfloat16) outputs to be within this tolerance, in the same hardware
assert jnp.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-2 )
@parameterized.expand(
[
# fmt: off
[83, 4, [0.1_514, 0.0_807, 0.1_624, 0.1_016, -0.1_896, 0.0_263, 0.0_677, 0.2_310]],
[17, 0.55, [0.1_164, -0.0_216, 0.0_170, 0.1_589, -0.3_120, 0.1_005, -0.0_581, -0.1_458]],
[8, 0.89, [-0.1_758, -0.0_169, 0.1_004, -0.1_411, 0.1_312, 0.1_103, -0.1_996, 0.2_139]],
[3, 1000, [0.1_214, 0.0_352, -0.0_731, -0.1_562, -0.0_994, -0.0_906, -0.2_340, -0.0_539]],
# fmt: on
] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_unet_model(model_id='''stabilityai/stable-diffusion-2''' , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_latents(__UpperCAmelCase , shape=(4, 4, 96, 96) , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_encoder_hidden_states(__UpperCAmelCase , shape=(4, 77, 1024) , fpaa=__UpperCAmelCase )
__lowerCamelCase = model.apply(
{'''params''': params} , __UpperCAmelCase , jnp.array(__UpperCAmelCase , dtype=jnp.intaa ) , encoder_hidden_states=__UpperCAmelCase , ).sample
assert sample.shape == latents.shape
__lowerCamelCase = jnp.asarray(jax.device_get((sample[-1, -2:, -2:, :2].flatten()) ) , dtype=jnp.floataa )
__lowerCamelCase = jnp.array(__UpperCAmelCase , dtype=jnp.floataa )
# Found torch (float16) and flax (bfloat16) outputs to be within this tolerance, on the same hardware
assert jnp.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-2 )
| 330 | 1 |
def a__ ( _UpperCamelCase : int ):
stooge(_UpperCamelCase ,0 ,len(_UpperCamelCase ) - 1 )
return arr
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : List[str] ,_UpperCamelCase : Optional[int] ):
if i >= h:
return
# If first element is smaller than the last then swap them
if arr[i] > arr[h]:
__lowerCamelCase ,__lowerCamelCase = arr[h], arr[i]
# If there are more than 2 elements in the array
if h - i + 1 > 2:
__lowerCamelCase = (int)((h - i + 1) / 3 )
# Recursively sort first 2/3 elements
stooge(_UpperCamelCase ,_UpperCamelCase ,(h - t) )
# Recursively sort last 2/3 elements
stooge(_UpperCamelCase ,i + t ,(_UpperCamelCase) )
# Recursively sort first 2/3 elements
stooge(_UpperCamelCase ,_UpperCamelCase ,(h - t) )
if __name__ == "__main__":
a_ = input("""Enter numbers separated by a comma:\n""").strip()
a_ = [int(item) for item in user_input.split(""",""")]
print(stooge_sort(unsorted))
| 330 |
from typing import TYPE_CHECKING
from ....utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_mmbt""": ["""MMBTConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""MMBTForClassification""", """MMBTModel""", """ModalEmbeddings"""]
if TYPE_CHECKING:
from .configuration_mmbt import MMBTConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_mmbt import MMBTForClassification, MMBTModel, ModalEmbeddings
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
from ...processing_utils import ProcessorMixin
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""image_processor""", """feature_extractor"""]
lowerCAmelCase__ = """TvltImageProcessor"""
lowerCAmelCase__ = """TvltFeatureExtractor"""
def __init__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
super().__init__(image_processor=__UpperCAmelCase , feature_extractor=__UpperCAmelCase )
__lowerCamelCase = image_processor
__lowerCamelCase = feature_extractor
def __call__( self , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase=False , __UpperCAmelCase=False , *__UpperCAmelCase , **__UpperCAmelCase , ):
'''simple docstring'''
if images is None and audio is None:
raise ValueError('''You need to specify either an `images` or `audio` input to process.''' )
__lowerCamelCase = None
if images is not None:
__lowerCamelCase = self.image_processor(__UpperCAmelCase , mask_pixel=__UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase )
if images_mixed is not None:
__lowerCamelCase = self.image_processor(__UpperCAmelCase , is_mixed=__UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase )
if audio is not None:
__lowerCamelCase = self.feature_extractor(
__UpperCAmelCase , *__UpperCAmelCase , sampling_rate=__UpperCAmelCase , mask_audio=__UpperCAmelCase , **__UpperCAmelCase )
__lowerCamelCase = {}
if audio is not None:
output_dict.update(__UpperCAmelCase )
if images is not None:
output_dict.update(__UpperCAmelCase )
if images_mixed_dict is not None:
output_dict.update(__UpperCAmelCase )
return output_dict
@property
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.image_processor.model_input_names
__lowerCamelCase = self.feature_extractor.model_input_names
return list(dict.fromkeys(image_processor_input_names + feature_extractor_input_names ) )
| 330 |
import gc
import importlib.metadata
import tempfile
import unittest
from packaging import version
from transformers import (
AutoModel,
AutoModelForCausalLM,
AutoModelForSeqaSeqLM,
AutoModelForSequenceClassification,
AutoTokenizer,
BitsAndBytesConfig,
pipeline,
)
from transformers.testing_utils import (
is_torch_available,
require_accelerate,
require_bitsandbytes,
require_torch,
require_torch_gpu,
require_torch_multi_gpu,
slow,
)
def a__ ( _UpperCamelCase : Optional[int] ):
if model.config.model_type == "gpt2":
return model.transformer.h[0].mlp.c_fc
return model.transformer.h[0].mlp.dense_ah_to_h
if is_torch_available():
import torch
import torch.nn as nn
class __lowerCAmelCase ( nn.Module ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
super().__init__()
__lowerCamelCase = module
__lowerCamelCase = nn.Sequential(
nn.Linear(module.in_features , __UpperCAmelCase , bias=__UpperCAmelCase ) , nn.Linear(__UpperCAmelCase , module.out_features , bias=__UpperCAmelCase ) , )
__lowerCamelCase = (2.0 / (5 * min(module.in_features , module.out_features ))) ** 0.5
nn.init.normal_(self.adapter[0].weight , std=__UpperCAmelCase )
nn.init.zeros_(self.adapter[1].weight )
self.adapter.to(module.weight.device )
def lowerCamelCase ( self , __UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return self.module(__UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ) + self.adapter(__UpperCAmelCase )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class __lowerCAmelCase ( unittest.TestCase ):
# We keep the constants inside the init function and model loading inside setUp function
# We need to test on relatively large models (aka >1b parameters otherwise the quantiztion may not work as expected)
# Therefore here we use only bloom-1b3 to test our module
lowerCAmelCase__ = """bigscience/bloom-1b7"""
# Constant values
lowerCAmelCase__ = 2.1_09_65_95_52_69_25_74
lowerCAmelCase__ = """Hello my name is"""
lowerCAmelCase__ = set()
EXPECTED_OUTPUTS.add("""Hello my name is John and I am a professional photographer. I""" )
EXPECTED_OUTPUTS.add("""Hello my name is John.\nI am a friend of your father.\n""" )
EXPECTED_OUTPUTS.add("""Hello my name is John Doe, I am a student at the University""" )
lowerCAmelCase__ = 1_0
def lowerCamelCase ( self ):
'''simple docstring'''
# Models and tokenizer
__lowerCamelCase = AutoTokenizer.from_pretrained(self.model_name )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# Models and tokenizer
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , torch_dtype=torch.floataa , device_map='''auto''' )
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
def lowerCamelCase ( self ):
'''simple docstring'''
del self.model_fpaa
del self.model_abit
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_abit.config
self.assertTrue(hasattr(__UpperCAmelCase , '''quantization_config''' ) )
__lowerCamelCase = config.to_dict()
__lowerCamelCase = config.to_diff_dict()
__lowerCamelCase = config.to_json_string()
def lowerCamelCase ( self ):
'''simple docstring'''
from bitsandbytes.nn import Paramsabit
__lowerCamelCase = self.model_fpaa.get_memory_footprint()
__lowerCamelCase = self.model_abit.get_memory_footprint()
self.assertAlmostEqual(mem_fpaa / mem_abit , self.EXPECTED_RELATIVE_DIFFERENCE )
__lowerCamelCase = get_some_linear_layer(self.model_abit )
self.assertTrue(linear.weight.__class__ == Paramsabit )
def lowerCamelCase ( self ):
'''simple docstring'''
from transformers import TaPreTrainedModel
self.model_fpaa.get_memory_footprint()
self.model_abit.get_memory_footprint()
for name, module in self.model_abit.named_modules():
if isinstance(__UpperCAmelCase , torch.nn.Linear ):
if name not in ["lm_head"] + TaPreTrainedModel._keep_in_fpaa_modules:
# 4-bit parameters are packed in uint8 variables
self.assertTrue(module.weight.dtype == torch.uinta )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = self.model_abit.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_sequences[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitsAndBytesConfig()
__lowerCamelCase = True
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = model_abit_from_config.generate(
input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_sequences[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
def lowerCamelCase ( self ):
'''simple docstring'''
with self.assertRaises(__UpperCAmelCase ), tempfile.TemporaryDirectory() as tmpdirname:
self.model_abit.save_pretrained(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitsAndBytesConfig()
with self.assertRaises(__UpperCAmelCase ):
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=__UpperCAmelCase , load_in_abit=__UpperCAmelCase , device_map='''auto''' , bnb_abit_quant_type='''nf4''' , )
def lowerCamelCase ( self ):
'''simple docstring'''
with self.assertRaises(__UpperCAmelCase ):
# Tries with `str`
self.model_abit.to('''cpu''' )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `dtype``
self.model_abit.to(torch.floataa )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.to(torch.device('''cuda:0''' ) )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.float()
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.half()
# Test if we did not break anything
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = self.model_fpaa.to(torch.floataa )
__lowerCamelCase = self.model_fpaa.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.to('''cpu''' )
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.half()
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.float()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = AutoModelForSeqaSeqLM.from_pretrained('''t5-small''' , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
self.assertTrue(model.decoder.block[0].layer[2].DenseReluDense.wo.weight.dtype == torch.floataa )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class __lowerCAmelCase ( unittest.TestCase ):
@classmethod
def lowerCamelCase ( cls ):
'''simple docstring'''
__lowerCamelCase = '''t5-small'''
__lowerCamelCase = '''google/flan-t5-small''' # flan-t5 uses dense-act instead of dense-relu-dense
__lowerCamelCase = AutoTokenizer.from_pretrained(cls.model_name )
__lowerCamelCase = '''Translate in German: Hello, my dog is cute'''
def lowerCamelCase ( self ):
'''simple docstring'''
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
from transformers import TaForConditionalGeneration
__lowerCamelCase = TaForConditionalGeneration._keep_in_fpaa_modules
__lowerCamelCase = None
# test with `t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
# test with `flan-t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
__lowerCamelCase = modules
def lowerCamelCase ( self ):
'''simple docstring'''
import bitsandbytes as bnb
from transformers import TaForConditionalGeneration
# test with `t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# there was a bug with decoders - this test checks that it is fixed
self.assertTrue(isinstance(model.decoder.block[0].layer[0].SelfAttention.q , bnb.nn.Linearabit ) )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
# test with `flan-t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# model_name
__lowerCamelCase = '''bigscience/bloom-560m'''
__lowerCamelCase = '''t5-small'''
# Different types of model
__lowerCamelCase = AutoModel.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# Sequence classification model
__lowerCamelCase = AutoModelForSequenceClassification.from_pretrained(
self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# CausalLM model
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# Seq2seq model
__lowerCamelCase = AutoModelForSeqaSeqLM.from_pretrained(
self.seq_to_seq_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
def lowerCamelCase ( self ):
'''simple docstring'''
del self.base_model
del self.sequence_model
del self.model_abit
del self.seq_to_seq_model
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
from bitsandbytes.nn import Paramsabit
self.assertTrue(self.base_model.h[-1].mlp.dense_ah_to_h.weight.__class__ == Paramsabit )
# Other heads should be nn.Parameter
self.assertTrue(self.model_abit.lm_head.weight.__class__ == torch.nn.Parameter )
self.assertTrue(self.sequence_model.score.weight.__class__ == torch.nn.Parameter )
self.assertTrue(self.seq_to_seq_model.lm_head.weight.__class__ == torch.nn.Parameter )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
del self.pipe
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = pipeline(
'''text-generation''' , model=self.model_name , model_kwargs={'''device_map''': '''auto''', '''load_in_4bit''': True, '''torch_dtype''': torch.floataa} , max_new_tokens=self.MAX_NEW_TOKENS , )
# Real second forward pass
__lowerCamelCase = self.pipe(self.input_text )
self.assertIn(pipeline_output[0]['''generated_text'''] , self.EXPECTED_OUTPUTS )
@require_torch_multi_gpu
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , load_in_abit=__UpperCAmelCase , device_map='''balanced''' )
# Check correct device map
self.assertEqual(set(model_parallel.hf_device_map.values() ) , {0, 1} )
# Check that inference pass works on the model
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
# Second real batch
__lowerCamelCase = model_parallel.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_parallel[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''facebook/opt-350m'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
if version.parse(importlib.metadata.version('''bitsandbytes''' ) ) < version.parse('''0.37.0''' ):
return
# Step 1: freeze all parameters
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase )
self.assertEqual(set(model.hf_device_map.values() ) , {torch.cuda.current_device()} )
for param in model.parameters():
__lowerCamelCase = False # freeze the model - train adapters later
if param.ndim == 1:
# cast the small parameters (e.g. layernorm) to fp32 for stability
__lowerCamelCase = param.data.to(torch.floataa )
# Step 2: add adapters
for _, module in model.named_modules():
if "OPTAttention" in repr(type(__UpperCAmelCase ) ):
__lowerCamelCase = LoRALayer(module.q_proj , rank=16 )
__lowerCamelCase = LoRALayer(module.k_proj , rank=16 )
__lowerCamelCase = LoRALayer(module.v_proj , rank=16 )
# Step 3: dummy batch
__lowerCamelCase = self.tokenizer('''Test batch ''' , return_tensors='''pt''' ).to(0 )
# Step 4: Check if the gradient is not None
with torch.cuda.amp.autocast():
__lowerCamelCase = model.forward(**__UpperCAmelCase )
out.logits.norm().backward()
for module in model.modules():
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
self.assertTrue(module.adapter[1].weight.grad is not None )
self.assertTrue(module.adapter[1].weight.grad.norm().item() > 0 )
elif isinstance(__UpperCAmelCase , nn.Embedding ):
self.assertTrue(module.weight.grad is None )
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """gpt2-xl"""
lowerCAmelCase__ = 3.31_91_85_48_54_15_21_87
| 330 | 1 |
import itertools
import json
import os
import unittest
from transformers import AddedToken, RobertaTokenizer, RobertaTokenizerFast
from transformers.models.roberta.tokenization_roberta import VOCAB_FILES_NAMES
from transformers.testing_utils import require_tokenizers, slow
from ...test_tokenization_common import TokenizerTesterMixin
@require_tokenizers
class __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = RobertaTokenizer
lowerCAmelCase__ = RobertaTokenizerFast
lowerCAmelCase__ = True
lowerCAmelCase__ = {"""cls_token""": """<s>"""}
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
__lowerCamelCase = [
'''l''',
'''o''',
'''w''',
'''e''',
'''r''',
'''s''',
'''t''',
'''i''',
'''d''',
'''n''',
'''\u0120''',
'''\u0120l''',
'''\u0120n''',
'''\u0120lo''',
'''\u0120low''',
'''er''',
'''\u0120lowest''',
'''\u0120newer''',
'''\u0120wider''',
'''<unk>''',
]
__lowerCamelCase = dict(zip(__UpperCAmelCase , range(len(__UpperCAmelCase ) ) ) )
__lowerCamelCase = ['''#version: 0.2''', '''\u0120 l''', '''\u0120l o''', '''\u0120lo w''', '''e r''', '''''']
__lowerCamelCase = {'''unk_token''': '''<unk>'''}
__lowerCamelCase = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['''vocab_file'''] )
__lowerCamelCase = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['''merges_file'''] )
with open(self.vocab_file , '''w''' , encoding='''utf-8''' ) as fp:
fp.write(json.dumps(__UpperCAmelCase ) + '''\n''' )
with open(self.merges_file , '''w''' , encoding='''utf-8''' ) as fp:
fp.write('''\n'''.join(__UpperCAmelCase ) )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
kwargs.update(self.special_tokens_map )
return self.tokenizer_class.from_pretrained(self.tmpdirname , **__UpperCAmelCase )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
kwargs.update(self.special_tokens_map )
return RobertaTokenizerFast.from_pretrained(self.tmpdirname , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = '''lower newer'''
__lowerCamelCase = '''lower newer'''
return input_text, output_text
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer_class(self.vocab_file , self.merges_file , **self.special_tokens_map )
__lowerCamelCase = '''lower newer'''
__lowerCamelCase = ['''l''', '''o''', '''w''', '''er''', '''\u0120''', '''n''', '''e''', '''w''', '''er''']
__lowerCamelCase = tokenizer.tokenize(__UpperCAmelCase ) # , add_prefix_space=True)
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = tokens + [tokenizer.unk_token]
__lowerCamelCase = [0, 1, 2, 15, 10, 9, 3, 2, 15, 19]
self.assertListEqual(tokenizer.convert_tokens_to_ids(__UpperCAmelCase ) , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_tokenizer()
self.assertListEqual(tokenizer.encode('''Hello world!''' , add_special_tokens=__UpperCAmelCase ) , [0, 31414, 232, 328, 2] )
self.assertListEqual(
tokenizer.encode('''Hello world! cécé herlolip 418''' , add_special_tokens=__UpperCAmelCase ) , [0, 31414, 232, 328, 740, 1140, 12695, 69, 46078, 1588, 2] , )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer_class.from_pretrained('''roberta-base''' )
__lowerCamelCase = tokenizer.encode('''sequence builders''' , add_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.encode('''multi-sequence build''' , add_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.encode(
'''sequence builders''' , add_special_tokens=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase )
__lowerCamelCase = tokenizer.encode(
'''sequence builders''' , '''multi-sequence build''' , add_special_tokens=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase )
__lowerCamelCase = tokenizer.build_inputs_with_special_tokens(__UpperCAmelCase )
__lowerCamelCase = tokenizer.build_inputs_with_special_tokens(__UpperCAmelCase , __UpperCAmelCase )
assert encoded_sentence == encoded_text_from_decode
assert encoded_pair == encoded_pair_from_decode
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_tokenizer()
__lowerCamelCase = '''Encode this sequence.'''
__lowerCamelCase = tokenizer.byte_encoder[''' '''.encode('''utf-8''' )[0]]
# Testing encoder arguments
__lowerCamelCase = tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase )
__lowerCamelCase = tokenizer.convert_ids_to_tokens(encoded[0] )[0]
self.assertNotEqual(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase )
__lowerCamelCase = tokenizer.convert_ids_to_tokens(encoded[0] )[0]
self.assertEqual(__UpperCAmelCase , __UpperCAmelCase )
tokenizer.add_special_tokens({'''bos_token''': '''<s>'''} )
__lowerCamelCase = tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.convert_ids_to_tokens(encoded[1] )[0]
self.assertNotEqual(__UpperCAmelCase , __UpperCAmelCase )
# Testing spaces after special tokens
__lowerCamelCase = '''<mask>'''
tokenizer.add_special_tokens(
{'''mask_token''': AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase )} ) # mask token has a left space
__lowerCamelCase = tokenizer.convert_tokens_to_ids(__UpperCAmelCase )
__lowerCamelCase = '''Encode <mask> sequence'''
__lowerCamelCase = '''Encode <mask>sequence'''
__lowerCamelCase = tokenizer.encode(__UpperCAmelCase )
__lowerCamelCase = encoded.index(__UpperCAmelCase )
__lowerCamelCase = tokenizer.convert_ids_to_tokens(encoded[mask_loc + 1] )[0]
self.assertEqual(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = tokenizer.encode(__UpperCAmelCase )
__lowerCamelCase = encoded.index(__UpperCAmelCase )
__lowerCamelCase = tokenizer.convert_ids_to_tokens(encoded[mask_loc + 1] )[0]
self.assertNotEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
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(__UpperCAmelCase , **__UpperCAmelCase )
__lowerCamelCase = self.tokenizer_class.from_pretrained(__UpperCAmelCase , **__UpperCAmelCase )
__lowerCamelCase = '''A, <mask> AllenNLP sentence.'''
__lowerCamelCase = tokenizer_r.encode_plus(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase , return_token_type_ids=__UpperCAmelCase )
__lowerCamelCase = tokenizer_p.encode_plus(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase , return_token_type_ids=__UpperCAmelCase )
# token_type_ids should put 0 everywhere
self.assertEqual(sum(tokens_r['''token_type_ids'''] ) , sum(tokens_p['''token_type_ids'''] ) )
# attention_mask should put 1 everywhere, so sum over length should be 1
self.assertEqual(
sum(tokens_r['''attention_mask'''] ) / len(tokens_r['''attention_mask'''] ) , sum(tokens_p['''attention_mask'''] ) / len(tokens_p['''attention_mask'''] ) , )
__lowerCamelCase = tokenizer_r.convert_ids_to_tokens(tokens_r['''input_ids'''] )
__lowerCamelCase = tokenizer_p.convert_ids_to_tokens(tokens_p['''input_ids'''] )
# Rust correctly handles the space before the mask while python doesnt
self.assertSequenceEqual(tokens_p['''input_ids'''] , [0, 250, 6, 50264, 3823, 487, 21992, 3645, 4, 2] )
self.assertSequenceEqual(tokens_r['''input_ids'''] , [0, 250, 6, 50264, 3823, 487, 21992, 3645, 4, 2] )
self.assertSequenceEqual(
__UpperCAmelCase , ['''<s>''', '''A''', ''',''', '''<mask>''', '''ĠAllen''', '''N''', '''LP''', '''Ġsentence''', '''.''', '''</s>'''] )
self.assertSequenceEqual(
__UpperCAmelCase , ['''<s>''', '''A''', ''',''', '''<mask>''', '''ĠAllen''', '''N''', '''LP''', '''Ġsentence''', '''.''', '''</s>'''] )
def lowerCamelCase ( self ):
'''simple docstring'''
for trim_offsets, add_prefix_space in itertools.product([True, False] , repeat=2 ):
__lowerCamelCase = self.rust_tokenizer_class.from_pretrained(
self.tmpdirname , use_fast=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , trim_offsets=__UpperCAmelCase )
__lowerCamelCase = json.loads(tokenizer_r.backend_tokenizer.pre_tokenizer.__getstate__() )
__lowerCamelCase = json.loads(tokenizer_r.backend_tokenizer.post_processor.__getstate__() )
self.assertEqual(pre_tokenizer_state['''add_prefix_space'''] , __UpperCAmelCase )
self.assertEqual(post_processor_state['''add_prefix_space'''] , __UpperCAmelCase )
self.assertEqual(post_processor_state['''trim_offsets'''] , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Test which aims to verify that the offsets are well adapted to the argument `add_prefix_space` and
# `trim_offsets`
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(F"""{tokenizer.__class__.__name__} ({pretrained_name})""" ):
__lowerCamelCase = '''hello''' # `hello` is a token in the vocabulary of `pretrained_name`
__lowerCamelCase = F"""{text_of_1_token} {text_of_1_token}"""
__lowerCamelCase = self.rust_tokenizer_class.from_pretrained(
__UpperCAmelCase , use_fast=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , trim_offsets=__UpperCAmelCase )
__lowerCamelCase = tokenizer_r(__UpperCAmelCase , return_offsets_mapping=__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
self.assertEqual(encoding.offset_mapping[0] , (0, len(__UpperCAmelCase )) )
self.assertEqual(
encoding.offset_mapping[1] , (len(__UpperCAmelCase ) + 1, len(__UpperCAmelCase ) + 1 + len(__UpperCAmelCase )) , )
__lowerCamelCase = self.rust_tokenizer_class.from_pretrained(
__UpperCAmelCase , use_fast=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , trim_offsets=__UpperCAmelCase )
__lowerCamelCase = tokenizer_r(__UpperCAmelCase , return_offsets_mapping=__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
self.assertEqual(encoding.offset_mapping[0] , (0, len(__UpperCAmelCase )) )
self.assertEqual(
encoding.offset_mapping[1] , (len(__UpperCAmelCase ) + 1, len(__UpperCAmelCase ) + 1 + len(__UpperCAmelCase )) , )
__lowerCamelCase = self.rust_tokenizer_class.from_pretrained(
__UpperCAmelCase , use_fast=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , trim_offsets=__UpperCAmelCase )
__lowerCamelCase = tokenizer_r(__UpperCAmelCase , return_offsets_mapping=__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
self.assertEqual(encoding.offset_mapping[0] , (0, len(__UpperCAmelCase )) )
self.assertEqual(
encoding.offset_mapping[1] , (len(__UpperCAmelCase ), len(__UpperCAmelCase ) + 1 + len(__UpperCAmelCase )) , )
__lowerCamelCase = self.rust_tokenizer_class.from_pretrained(
__UpperCAmelCase , use_fast=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , trim_offsets=__UpperCAmelCase )
__lowerCamelCase = tokenizer_r(__UpperCAmelCase , return_offsets_mapping=__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
self.assertEqual(encoding.offset_mapping[0] , (0, len(__UpperCAmelCase )) )
self.assertEqual(
encoding.offset_mapping[1] , (len(__UpperCAmelCase ), len(__UpperCAmelCase ) + 1 + len(__UpperCAmelCase )) , )
__lowerCamelCase = F""" {text}"""
# tokenizer_r = self.rust_tokenizer_class.from_pretrained(
# pretrained_name, use_fast=True, add_prefix_space=True, trim_offsets=True
# )
# encoding = tokenizer_r(text, return_offsets_mapping=True, add_special_tokens=False)
# self.assertEqual(encoding.offset_mapping[0], (1, 1 + len(text_of_1_token)))
# self.assertEqual(
# encoding.offset_mapping[1],
# (1 + len(text_of_1_token) + 1, 1 + len(text_of_1_token) + 1 + len(text_of_1_token)),
# )
__lowerCamelCase = self.rust_tokenizer_class.from_pretrained(
__UpperCAmelCase , use_fast=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , trim_offsets=__UpperCAmelCase )
__lowerCamelCase = tokenizer_r(__UpperCAmelCase , return_offsets_mapping=__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
self.assertEqual(encoding.offset_mapping[0] , (1, 1 + len(__UpperCAmelCase )) )
self.assertEqual(
encoding.offset_mapping[1] , (1 + len(__UpperCAmelCase ) + 1, 1 + len(__UpperCAmelCase ) + 1 + len(__UpperCAmelCase )) , )
__lowerCamelCase = self.rust_tokenizer_class.from_pretrained(
__UpperCAmelCase , use_fast=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , trim_offsets=__UpperCAmelCase )
__lowerCamelCase = tokenizer_r(__UpperCAmelCase , return_offsets_mapping=__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
self.assertEqual(encoding.offset_mapping[0] , (0, 1 + len(__UpperCAmelCase )) )
self.assertEqual(
encoding.offset_mapping[1] , (1 + len(__UpperCAmelCase ), 1 + len(__UpperCAmelCase ) + 1 + len(__UpperCAmelCase )) , )
__lowerCamelCase = self.rust_tokenizer_class.from_pretrained(
__UpperCAmelCase , use_fast=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , trim_offsets=__UpperCAmelCase )
__lowerCamelCase = tokenizer_r(__UpperCAmelCase , return_offsets_mapping=__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
self.assertEqual(encoding.offset_mapping[0] , (0, 1 + len(__UpperCAmelCase )) )
self.assertEqual(
encoding.offset_mapping[1] , (1 + len(__UpperCAmelCase ), 1 + len(__UpperCAmelCase ) + 1 + len(__UpperCAmelCase )) , )
| 330 |
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 __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 42
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ ):
lowerCAmelCase__ = True
@register_to_config
def __init__( self , __UpperCAmelCase = 3 , __UpperCAmelCase = 3 , __UpperCAmelCase = ("DownEncoderBlock2D",) , __UpperCAmelCase = ("UpDecoderBlock2D",) , __UpperCAmelCase = (64,) , __UpperCAmelCase = 1 , __UpperCAmelCase = "silu" , __UpperCAmelCase = 4 , __UpperCAmelCase = 32 , __UpperCAmelCase = 32 , __UpperCAmelCase = 0.18_215 , ):
'''simple docstring'''
super().__init__()
# pass init params to Encoder
__lowerCamelCase = Encoder(
in_channels=__UpperCAmelCase , out_channels=__UpperCAmelCase , down_block_types=__UpperCAmelCase , block_out_channels=__UpperCAmelCase , layers_per_block=__UpperCAmelCase , act_fn=__UpperCAmelCase , norm_num_groups=__UpperCAmelCase , double_z=__UpperCAmelCase , )
# pass init params to Decoder
__lowerCamelCase = Decoder(
in_channels=__UpperCAmelCase , out_channels=__UpperCAmelCase , up_block_types=__UpperCAmelCase , block_out_channels=__UpperCAmelCase , layers_per_block=__UpperCAmelCase , norm_num_groups=__UpperCAmelCase , act_fn=__UpperCAmelCase , )
__lowerCamelCase = nn.Convad(2 * latent_channels , 2 * latent_channels , 1 )
__lowerCamelCase = nn.Convad(__UpperCAmelCase , __UpperCAmelCase , 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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
if isinstance(__UpperCAmelCase , (Encoder, Decoder) ):
__lowerCamelCase = value
def lowerCamelCase ( self , __UpperCAmelCase = True ):
'''simple docstring'''
__lowerCamelCase = use_tiling
def lowerCamelCase ( self ):
'''simple docstring'''
self.enable_tiling(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = True
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = False
@property
# Copied from diffusers.models.unet_2d_condition.UNet2DConditionModel.attn_processors
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = {}
def fn_recursive_add_processors(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
if hasattr(__UpperCAmelCase , '''set_processor''' ):
__lowerCamelCase = module.processor
for sub_name, child in module.named_children():
fn_recursive_add_processors(F"""{name}.{sub_name}""" , __UpperCAmelCase , __UpperCAmelCase )
return processors
for name, module in self.named_children():
fn_recursive_add_processors(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
return processors
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = len(self.attn_processors.keys() )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) != count:
raise ValueError(
F"""A dict of processors was passed, but the number of processors {len(__UpperCAmelCase )} does not match the"""
F""" number of attention layers: {count}. Please make sure to pass {count} processor classes.""" )
def fn_recursive_attn_processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
if hasattr(__UpperCAmelCase , '''set_processor''' ):
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
module.set_processor(__UpperCAmelCase )
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}""" , __UpperCAmelCase , __UpperCAmelCase )
for name, module in self.named_children():
fn_recursive_attn_processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
self.set_attn_processor(AttnProcessor() )
@apply_forward_hook
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
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(__UpperCAmelCase , return_dict=__UpperCAmelCase )
if self.use_slicing and x.shape[0] > 1:
__lowerCamelCase = [self.encoder(__UpperCAmelCase ) for x_slice in x.split(1 )]
__lowerCamelCase = torch.cat(__UpperCAmelCase )
else:
__lowerCamelCase = self.encoder(__UpperCAmelCase )
__lowerCamelCase = self.quant_conv(__UpperCAmelCase )
__lowerCamelCase = DiagonalGaussianDistribution(__UpperCAmelCase )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
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(__UpperCAmelCase , return_dict=__UpperCAmelCase )
__lowerCamelCase = self.post_quant_conv(__UpperCAmelCase )
__lowerCamelCase = self.decoder(__UpperCAmelCase )
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
@apply_forward_hook
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
if self.use_slicing and z.shape[0] > 1:
__lowerCamelCase = [self._decode(__UpperCAmelCase ).sample for z_slice in z.split(1 )]
__lowerCamelCase = torch.cat(__UpperCAmelCase )
else:
__lowerCamelCase = self._decode(__UpperCAmelCase ).sample
if not return_dict:
return (decoded,)
return DecoderOutput(sample=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(a.shape[2] , b.shape[2] , __UpperCAmelCase )
for y in range(__UpperCAmelCase ):
__lowerCamelCase = a[:, :, -blend_extent + y, :] * (1 - y / blend_extent) + b[:, :, y, :] * (y / blend_extent)
return b
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(a.shape[3] , b.shape[3] , __UpperCAmelCase )
for x in range(__UpperCAmelCase ):
__lowerCamelCase = a[:, :, :, -blend_extent + x] * (1 - x / blend_extent) + b[:, :, :, x] * (x / blend_extent)
return b
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
__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] , __UpperCAmelCase ):
__lowerCamelCase = []
for j in range(0 , x.shape[3] , __UpperCAmelCase ):
__lowerCamelCase = x[:, :, i : i + self.tile_sample_min_size, j : j + self.tile_sample_min_size]
__lowerCamelCase = self.encoder(__UpperCAmelCase )
__lowerCamelCase = self.quant_conv(__UpperCAmelCase )
row.append(__UpperCAmelCase )
rows.append(__UpperCAmelCase )
__lowerCamelCase = []
for i, row in enumerate(__UpperCAmelCase ):
__lowerCamelCase = []
for j, tile in enumerate(__UpperCAmelCase ):
# 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] , __UpperCAmelCase , __UpperCAmelCase )
if j > 0:
__lowerCamelCase = self.blend_h(row[j - 1] , __UpperCAmelCase , __UpperCAmelCase )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(__UpperCAmelCase , dim=3 ) )
__lowerCamelCase = torch.cat(__UpperCAmelCase , dim=2 )
__lowerCamelCase = DiagonalGaussianDistribution(__UpperCAmelCase )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
__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] , __UpperCAmelCase ):
__lowerCamelCase = []
for j in range(0 , z.shape[3] , __UpperCAmelCase ):
__lowerCamelCase = z[:, :, i : i + self.tile_latent_min_size, j : j + self.tile_latent_min_size]
__lowerCamelCase = self.post_quant_conv(__UpperCAmelCase )
__lowerCamelCase = self.decoder(__UpperCAmelCase )
row.append(__UpperCAmelCase )
rows.append(__UpperCAmelCase )
__lowerCamelCase = []
for i, row in enumerate(__UpperCAmelCase ):
__lowerCamelCase = []
for j, tile in enumerate(__UpperCAmelCase ):
# 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] , __UpperCAmelCase , __UpperCAmelCase )
if j > 0:
__lowerCamelCase = self.blend_h(row[j - 1] , __UpperCAmelCase , __UpperCAmelCase )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(__UpperCAmelCase , dim=3 ) )
__lowerCamelCase = torch.cat(__UpperCAmelCase , dim=2 )
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False , __UpperCAmelCase = True , __UpperCAmelCase = None , ):
'''simple docstring'''
__lowerCamelCase = sample
__lowerCamelCase = self.encode(__UpperCAmelCase ).latent_dist
if sample_posterior:
__lowerCamelCase = posterior.sample(generator=__UpperCAmelCase )
else:
__lowerCamelCase = posterior.mode()
__lowerCamelCase = self.decode(__UpperCAmelCase ).sample
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
| 330 | 1 |
import inspect
import unittest
from transformers import BitConfig
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_backbone_common import BackboneTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import BitBackbone, BitForImageClassification, BitImageProcessor, BitModel
from transformers.models.bit.modeling_bit import BIT_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=3 , __UpperCAmelCase=32 , __UpperCAmelCase=3 , __UpperCAmelCase=10 , __UpperCAmelCase=[8, 16, 32, 64] , __UpperCAmelCase=[1, 1, 2, 1] , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase="relu" , __UpperCAmelCase=3 , __UpperCAmelCase=None , __UpperCAmelCase=["stage2", "stage3", "stage4"] , __UpperCAmelCase=[2, 3, 4] , __UpperCAmelCase=1 , ):
'''simple docstring'''
__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(__UpperCAmelCase )
__lowerCamelCase = out_features
__lowerCamelCase = out_indices
__lowerCamelCase = num_groups
def lowerCamelCase ( self ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
return BitConfig(
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 , out_features=self.out_features , out_indices=self.out_indices , num_groups=self.num_groups , )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = BitModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(
result.last_hidden_state.shape , (self.batch_size, self.hidden_sizes[-1], self.image_size // 32, self.image_size // 32) , )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = BitForImageClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = BitBackbone(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) , len(config.out_features ) )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) , [self.batch_size, self.hidden_sizes[1], 4, 4] )
# verify channels
self.parent.assertEqual(len(model.channels ) , len(config.out_features ) )
self.parent.assertListEqual(model.channels , config.hidden_sizes[1:] )
# verify backbone works with out_features=None
__lowerCamelCase = None
__lowerCamelCase = BitBackbone(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
# verify feature maps
self.parent.assertEqual(len(result.feature_maps ) , 1 )
self.parent.assertListEqual(list(result.feature_maps[0].shape ) , [self.batch_size, self.hidden_sizes[-1], 1, 1] )
# verify channels
self.parent.assertEqual(len(model.channels ) , 1 )
self.parent.assertListEqual(model.channels , [config.hidden_sizes[-1]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = config_and_inputs
__lowerCamelCase = {'''pixel_values''': pixel_values}
return config, inputs_dict
@require_torch
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (BitModel, BitForImageClassification, BitBackbone) if is_torch_available() else ()
lowerCAmelCase__ = (
{"""feature-extraction""": BitModel, """image-classification""": BitForImageClassification}
if is_torch_available()
else {}
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitModelTester(self )
__lowerCamelCase = ConfigTester(self , config_class=__UpperCAmelCase , has_text_modality=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
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 lowerCamelCase ( self ):
'''simple docstring'''
return
@unittest.skip(reason='''Bit does not output attentions''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
@unittest.skip(reason='''Bit does not use inputs_embeds''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
@unittest.skip(reason='''Bit does not support input and output embeddings''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
__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] , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_backbone(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(config=__UpperCAmelCase )
for name, module in model.named_modules():
if isinstance(__UpperCAmelCase , (nn.BatchNormad, nn.GroupNorm) ):
self.assertTrue(
torch.all(module.weight == 1 ) , msg=F"""Parameter {name} of model {model_class} seems not properly initialized""" , )
self.assertTrue(
torch.all(module.bias == 0 ) , msg=F"""Parameter {name} of model {model_class} seems not properly initialized""" , )
def lowerCamelCase ( self ):
'''simple docstring'''
def check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.encoder_hidden_states if config.is_encoder_decoder else outputs.hidden_states
__lowerCamelCase = self.model_tester.num_stages
self.assertEqual(len(__UpperCAmelCase ) , expected_num_stages + 1 )
# Bit'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()
__lowerCamelCase = ['''preactivation''', '''bottleneck''']
for model_class in self.all_model_classes:
for layer_type in layers_type:
__lowerCamelCase = layer_type
__lowerCamelCase = True
check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCamelCase = True
check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
@unittest.skip(reason='''Bit does not use feedforward chunking''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in BIT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = BitModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
def a__ ( ):
__lowerCamelCase = Image.open('''./tests/fixtures/tests_samples/COCO/000000039769.png''' )
return image
@require_torch
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
@cached_property
def lowerCamelCase ( self ):
'''simple docstring'''
return (
BitImageProcessor.from_pretrained(BIT_PRETRAINED_MODEL_ARCHIVE_LIST[0] ) if is_vision_available() else None
)
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitForImageClassification.from_pretrained(BIT_PRETRAINED_MODEL_ARCHIVE_LIST[0] ).to(__UpperCAmelCase )
__lowerCamelCase = self.default_image_processor
__lowerCamelCase = prepare_img()
__lowerCamelCase = image_processor(images=__UpperCAmelCase , return_tensors='''pt''' ).to(__UpperCAmelCase )
# forward pass
with torch.no_grad():
__lowerCamelCase = model(**__UpperCAmelCase )
# verify the logits
__lowerCamelCase = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor([[-0.6_526, -0.5_263, -1.4_398]] ).to(__UpperCAmelCase )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , __UpperCAmelCase , atol=1E-4 ) )
@require_torch
class __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (BitBackbone,) if is_torch_available() else ()
lowerCAmelCase__ = BitConfig
lowerCAmelCase__ = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitModelTester(self )
| 330 |
import argparse
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import BigBirdPegasusConfig, BigBirdPegasusForConditionalGeneration
a_ = [
# tf -> hf
("""/""", """."""),
("""layer_""", """layers."""),
("""kernel""", """weight"""),
("""beta""", """bias"""),
("""gamma""", """weight"""),
("""pegasus""", """model"""),
]
a_ = [
(""".output.dense""", """.fc2"""),
("""intermediate.LayerNorm""", """final_layer_norm"""),
("""intermediate.dense""", """fc1"""),
]
a_ = (
INIT_COMMON
+ [
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.out_proj"""),
("""attention.self""", """self_attn"""),
("""attention.encdec.LayerNorm""", """encoder_attn_layer_norm"""),
("""attention.encdec_output.dense""", """encoder_attn.out_proj"""),
("""attention.encdec""", """encoder_attn"""),
("""key""", """k_proj"""),
("""value""", """v_proj"""),
("""query""", """q_proj"""),
("""decoder.LayerNorm""", """decoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = (
INIT_COMMON
+ [
("""embeddings.word_embeddings""", """shared.weight"""),
("""embeddings.position_embeddings""", """embed_positions.weight"""),
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.output"""),
("""attention.self""", """self_attn.self"""),
("""encoder.LayerNorm""", """encoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = [
"""encdec/key/bias""",
"""encdec/query/bias""",
"""encdec/value/bias""",
"""self/key/bias""",
"""self/query/bias""",
"""self/value/bias""",
"""encdec_output/dense/bias""",
"""attention/output/dense/bias""",
]
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : Optional[Any] ):
for tf_name, hf_name in patterns:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = BigBirdPegasusConfig(**_UpperCamelCase )
__lowerCamelCase = BigBirdPegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
__lowerCamelCase = {}
# separating decoder weights
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if k.startswith('''pegasus/decoder''' )}
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if not k.startswith('''pegasus/decoder''' )}
for k, v in tqdm(decoder_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = DECODER_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
for k, v in tqdm(remaining_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = REMAINING_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict and k != "pegasus/embeddings/position_embeddings":
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
if k != "pegasus/embeddings/position_embeddings":
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
__lowerCamelCase = mapping['''model.embed_positions.weight''']
__lowerCamelCase = mapping.pop('''model.embed_positions.weight''' )
__lowerCamelCase ,__lowerCamelCase = torch_model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k
for k in missing
if k
not in [
'''final_logits_bias''',
'''model.encoder.embed_tokens.weight''',
'''model.decoder.embed_tokens.weight''',
'''lm_head.weight''',
]
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ,_UpperCamelCase : dict ):
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = convert_bigbird_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""--save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
a_ = {}
convert_bigbird_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir, config_update=config_update)
| 330 | 1 |
from math import ceil
from typing import List, Optional, Union
import numpy as np
from ...audio_utils import mel_filter_bank, spectrogram, window_function
from ...feature_extraction_sequence_utils import BatchFeature, SequenceFeatureExtractor
from ...utils import TensorType, logging
a_ = logging.get_logger(__name__)
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""audio_values""", """audio_mask"""]
def __init__( self , __UpperCAmelCase=2048 , __UpperCAmelCase=1 , __UpperCAmelCase=[16, 16] , __UpperCAmelCase=128 , __UpperCAmelCase=44100 , __UpperCAmelCase=86 , __UpperCAmelCase=2048 , __UpperCAmelCase=0.0 , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(
feature_size=__UpperCAmelCase , sampling_rate=__UpperCAmelCase , padding_value=__UpperCAmelCase , **__UpperCAmelCase , )
__lowerCamelCase = spectrogram_length
__lowerCamelCase = num_channels
__lowerCamelCase = patch_size
__lowerCamelCase = feature_size // self.patch_size[1]
__lowerCamelCase = n_fft
__lowerCamelCase = sampling_rate // hop_length_to_sampling_rate
__lowerCamelCase = sampling_rate
__lowerCamelCase = padding_value
__lowerCamelCase = mel_filter_bank(
num_frequency_bins=1 + n_fft // 2 , num_mel_filters=__UpperCAmelCase , min_frequency=0.0 , max_frequency=22_050.0 , sampling_rate=__UpperCAmelCase , norm='''slaney''' , mel_scale='''slaney''' , ).T
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = spectrogram(
__UpperCAmelCase , window_function(self.n_fft , '''hann''' ) , frame_length=self.n_fft , hop_length=self.hop_length , power=2.0 , mel_filters=self.mel_filters.T , log_mel='''dB''' , db_range=80.0 , )
__lowerCamelCase = log_spec[:, :-1]
__lowerCamelCase = log_spec - 20.0
__lowerCamelCase = np.clip(log_spec / 40.0 , -2.0 , 0.0 ) + 1.0
return log_spec
def __call__( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = False , __UpperCAmelCase = False , **__UpperCAmelCase , ):
'''simple docstring'''
if sampling_rate is not None:
if sampling_rate != self.sampling_rate:
raise ValueError(
'''This feature extractor is set to support sampling rate'''
F""" of {self.sampling_rate}. Please make sure that the provided `raw_speech` input was sampled"""
F""" with {self.sampling_rate} and not {sampling_rate}.""" )
else:
logger.warning(
'''It is strongly recommended to pass the `sampling_rate` argument to this function. '''
'''Failing to do so can result in silent errors that might be hard to debug.''' )
__lowerCamelCase = isinstance(__UpperCAmelCase , np.ndarray ) and len(raw_speech.shape ) > 1
if is_batched_numpy and len(raw_speech.shape ) > 2:
raise ValueError(F"""Only mono-channel audio is supported for input to {self}""" )
__lowerCamelCase = is_batched_numpy or (
isinstance(__UpperCAmelCase , (list, tuple) ) and (isinstance(raw_speech[0] , (np.ndarray, tuple, list) ))
)
if is_batched:
__lowerCamelCase = [np.asarray([speech] , dtype=np.floataa ).T for speech in raw_speech]
elif not is_batched and not isinstance(__UpperCAmelCase , np.ndarray ):
__lowerCamelCase = np.asarray(__UpperCAmelCase , dtype=np.floataa )
elif isinstance(__UpperCAmelCase , np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ):
__lowerCamelCase = raw_speech.astype(np.floataa )
# always return batch
if not is_batched:
__lowerCamelCase = [np.asarray([raw_speech] ).T]
# Convert audio signals to log mel spectrograms, truncate by time axis
__lowerCamelCase = [
self._np_extract_fbank_features(waveform.squeeze() ).T[: self.spectrogram_length] for waveform in raw_speech
]
if isinstance(audio_features[0] , __UpperCAmelCase ):
__lowerCamelCase = [np.asarray(__UpperCAmelCase , dtype=np.floataa ) for feature in audio_features]
# Create audio attention mask
__lowerCamelCase = max(
[ceil(feature.shape[0] / self.patch_size[0] ) * self.freq_len for feature in audio_features] ) # The maximum number of audio patches in a batch
if return_attention_mask:
__lowerCamelCase = [
(ceil(feature.shape[0] / self.patch_size[0] ) * self.freq_len) * [1]
+ (max_patch_len - ceil(feature.shape[0] / self.patch_size[0] ) * self.freq_len) * [0]
for feature in audio_features
]
__lowerCamelCase = np.array(__UpperCAmelCase ).astype(np.floataa )
# convert into correct format for padding
__lowerCamelCase = max_patch_len // self.freq_len * self.patch_size[0] # The maximum audio size in a batch
__lowerCamelCase = np.ones([len(__UpperCAmelCase ), 1, max_time_len, self.feature_size] ).astype(np.floataa )
__lowerCamelCase = padded_audio_features * self.padding_value
for i in range(len(__UpperCAmelCase ) ):
__lowerCamelCase = audio_features[i]
__lowerCamelCase = feature
# return as BatchFeature
if return_attention_mask:
__lowerCamelCase = {'''audio_values''': padded_audio_features, '''audio_mask''': audio_mask}
else:
__lowerCamelCase = {'''audio_values''': padded_audio_features}
__lowerCamelCase = BatchFeature(data=__UpperCAmelCase , tensor_type=__UpperCAmelCase )
return encoded_inputs
| 330 |
import uuid
from typing import Any, Dict, List, Optional, Union
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
if is_torch_available():
import torch
a_ = logging.get_logger(__name__)
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=None , __UpperCAmelCase=None ):
'''simple docstring'''
if not conversation_id:
__lowerCamelCase = uuid.uuida()
if past_user_inputs is None:
__lowerCamelCase = []
if generated_responses is None:
__lowerCamelCase = []
__lowerCamelCase = conversation_id
__lowerCamelCase = past_user_inputs
__lowerCamelCase = generated_responses
__lowerCamelCase = text
def __eq__( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
return False
if self.uuid == other.uuid:
return True
return (
self.new_user_input == other.new_user_input
and self.past_user_inputs == other.past_user_inputs
and self.generated_responses == other.generated_responses
)
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False ):
'''simple docstring'''
if self.new_user_input:
if overwrite:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" was overwritten """
F"""with: \"{text}\".""" )
__lowerCamelCase = text
else:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" new input """
F"""ignored: \"{text}\". Set `overwrite` to True to overwrite unprocessed user input""" )
else:
__lowerCamelCase = text
def lowerCamelCase ( self ):
'''simple docstring'''
if self.new_user_input:
self.past_user_inputs.append(self.new_user_input )
__lowerCamelCase = None
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
self.generated_responses.append(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
for user_input, generated_response in zip(self.past_user_inputs , self.generated_responses ):
yield True, user_input
yield False, generated_response
if self.new_user_input:
yield True, self.new_user_input
def __repr__( self ):
'''simple docstring'''
__lowerCamelCase = F"""Conversation id: {self.uuid} \n"""
for is_user, text in self.iter_texts():
__lowerCamelCase = '''user''' if is_user else '''bot'''
output += F"""{name} >> {text} \n"""
return output
@add_end_docstrings(
lowerCAmelCase__ , r"""
min_length_for_response (`int`, *optional*, defaults to 32):
The minimum length (in number of tokens) for a response.
minimum_tokens (`int`, *optional*, defaults to 10):
The minimum length of tokens to leave for a response.
""" , )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
if self.tokenizer.pad_token_id is None:
__lowerCamelCase = self.tokenizer.eos_token
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = {}
__lowerCamelCase = {}
__lowerCamelCase = {}
if min_length_for_response is not None:
__lowerCamelCase = min_length_for_response
if minimum_tokens is not None:
__lowerCamelCase = minimum_tokens
if "max_length" in generate_kwargs:
__lowerCamelCase = generate_kwargs['''max_length''']
# self.max_length = generate_kwargs.get("max_length", self.model.config.max_length)
if clean_up_tokenization_spaces is not None:
__lowerCamelCase = clean_up_tokenization_spaces
if generate_kwargs:
forward_params.update(__UpperCAmelCase )
return preprocess_params, forward_params, postprocess_params
def __call__( self , __UpperCAmelCase , __UpperCAmelCase=0 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = super().__call__(__UpperCAmelCase , num_workers=__UpperCAmelCase , **__UpperCAmelCase )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) == 1:
return outputs[0]
return outputs
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=32 ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError('''ConversationalPipeline, expects Conversation as inputs''' )
if conversation.new_user_input is None:
raise ValueError(
F"""Conversation with UUID {type(conversation.uuid )} does not contain new user input to process. """
'''Add user inputs with the conversation\'s `add_user_input` method''' )
if hasattr(self.tokenizer , '''_build_conversation_input_ids''' ):
__lowerCamelCase = self.tokenizer._build_conversation_input_ids(__UpperCAmelCase )
else:
# If the tokenizer cannot handle conversations, we default to only the old version
__lowerCamelCase = self._legacy_parse_and_tokenize(__UpperCAmelCase )
if self.framework == "pt":
__lowerCamelCase = torch.LongTensor([input_ids] )
elif self.framework == "tf":
__lowerCamelCase = tf.constant([input_ids] )
return {"input_ids": input_ids, "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=10 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = generate_kwargs.get('''max_length''' , self.model.config.max_length )
__lowerCamelCase = model_inputs['''input_ids'''].shape[1]
if max_length - minimum_tokens < n:
logger.warning(F"""Conversation input is to long ({n}), trimming it to ({max_length} - {minimum_tokens})""" )
__lowerCamelCase = max_length - minimum_tokens
__lowerCamelCase = model_inputs['''input_ids'''][:, -trim:]
if "attention_mask" in model_inputs:
__lowerCamelCase = model_inputs['''attention_mask'''][:, -trim:]
__lowerCamelCase = model_inputs.pop('''conversation''' )
__lowerCamelCase = max_length
__lowerCamelCase = self.model.generate(**__UpperCAmelCase , **__UpperCAmelCase )
if self.model.config.is_encoder_decoder:
__lowerCamelCase = 1
else:
__lowerCamelCase = n
return {"output_ids": output_ids[:, start_position:], "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = model_outputs['''output_ids''']
__lowerCamelCase = self.tokenizer.decode(
output_ids[0] , skip_special_tokens=__UpperCAmelCase , clean_up_tokenization_spaces=__UpperCAmelCase , )
__lowerCamelCase = model_outputs['''conversation''']
conversation.mark_processed()
conversation.append_response(__UpperCAmelCase )
return conversation
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer.eos_token_id
__lowerCamelCase = []
for is_user, text in conversation.iter_texts():
if eos_token_id is not None:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) + [eos_token_id] )
else:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) )
if len(__UpperCAmelCase ) > self.tokenizer.model_max_length:
__lowerCamelCase = input_ids[-self.tokenizer.model_max_length :]
return input_ids
| 330 | 1 |
import argparse
import random
import joblib
import numpy as np
import torch
from igf.igf import (
SecondaryLearner,
collect_objective_set,
compute_perplexity,
generate_datasets,
load_gpta,
recopy_gpta,
set_seed,
train_secondary_learner,
)
from torch.utils.data import DataLoader, RandomSampler
from transformers import GPTaLMHeadModel
def a__ ( _UpperCamelCase : str=32 ,_UpperCamelCase : Optional[int]=10 ,_UpperCamelCase : Optional[Any]=1_00 ,_UpperCamelCase : List[Any]=10_26 ,_UpperCamelCase : Optional[int]=True ,_UpperCamelCase : str="data/tokenized_stories_train_wikitext103.jbl" ,_UpperCamelCase : Optional[int]="igf_context_pairs.jbl" ,):
set_seed(3 )
# generate train_data and objective_set
__lowerCamelCase ,__lowerCamelCase = generate_datasets(
_UpperCamelCase ,_UpperCamelCase ,number=_UpperCamelCase ,min_len=10_26 ,trim=_UpperCamelCase )
# keeps model same across runs
set_seed(4 )
# model, lm_optimizer, lm_scheduler = recopy_gpt2(model, device, max_steps) # store original model weights
# can we train on GPU?
__lowerCamelCase = torch.device('''cuda:0''' if torch.cuda.is_available() else '''cpu''' )
# load pretrained model
__lowerCamelCase = load_gpta('''gpt2''' ).to(_UpperCamelCase )
print('''computing perplexity on objective set''' )
__lowerCamelCase = compute_perplexity(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ).item()
print('''perplexity on objective set:''' ,_UpperCamelCase )
# collect igf pairs and save to file demo.jbl
collect_objective_set(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# clean up, delete model and data we don't need anymore
del model, train_data, objective_set
torch.cuda.empty_cache()
def a__ ( _UpperCamelCase : Any ,_UpperCamelCase : str=15 ,_UpperCamelCase : Dict=1_28 ,_UpperCamelCase : Optional[int]=1_00 ,_UpperCamelCase : Union[str, Any]="igf_model.pt" ,):
set_seed(42 )
# Load pre-trained model
__lowerCamelCase = GPTaLMHeadModel.from_pretrained('''gpt2''' )
# Initialize secondary learner to use embedding weights of model
__lowerCamelCase = SecondaryLearner(_UpperCamelCase )
# Train secondary learner
__lowerCamelCase = train_secondary_learner(
_UpperCamelCase ,_UpperCamelCase ,max_epochs=_UpperCamelCase ,batch_size=_UpperCamelCase ,eval_freq=1_00 ,igf_model_path=_UpperCamelCase ,)
del model, secondary_learner_train_data
torch.cuda.empty_cache()
return secondary_learner
def a__ ( _UpperCamelCase : Tuple ,_UpperCamelCase : Dict ,_UpperCamelCase : List[str] ,_UpperCamelCase : List[Any]=32 ,_UpperCamelCase : str=10_00 ,_UpperCamelCase : Any=16 ,_UpperCamelCase : Any=1.0 ,_UpperCamelCase : Any=recopy_gpta ,_UpperCamelCase : Dict=None ,_UpperCamelCase : Union[str, Any]=10 ,_UpperCamelCase : Optional[int]="gpt2_finetuned.pt" ,):
__lowerCamelCase = torch.device('''cuda:0''' if torch.cuda.is_available() else '''cpu''' )
__lowerCamelCase = RandomSampler(_UpperCamelCase )
__lowerCamelCase = DataLoader(_UpperCamelCase ,sampler=_UpperCamelCase )
__lowerCamelCase = max_steps // (len(_UpperCamelCase )) + 1
__lowerCamelCase = 0
__lowerCamelCase = torch.zeros((1, context_len) ,dtype=torch.long ,device=_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = recopy_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
model.train()
if secondary_learner is not None:
secondary_learner.to(_UpperCamelCase )
secondary_learner.eval()
__lowerCamelCase = []
__lowerCamelCase = 0
__lowerCamelCase = []
__lowerCamelCase = []
# Compute the performance of the transformer model at the beginning
__lowerCamelCase = compute_perplexity(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
test_perps.append(_UpperCamelCase )
print('''Test perplexity, step''' ,_UpperCamelCase ,''':''' ,_UpperCamelCase )
for epoch in range(int(_UpperCamelCase ) ):
for step, example in enumerate(_UpperCamelCase ):
torch.cuda.empty_cache()
__lowerCamelCase = random.randint(0 ,example.size(2 ) - context_len - 1 )
__lowerCamelCase = example[0, 0, start : start + context_len]
lm_optimizer.zero_grad()
__lowerCamelCase = model(_UpperCamelCase ,labels=_UpperCamelCase )
__lowerCamelCase = True
if secondary_learner is not None:
__lowerCamelCase = secondary_learner.forward(
torch.tensor(_UpperCamelCase ,dtype=torch.long ,device=_UpperCamelCase ).unsqueeze(0 ) )[0].item()
observed_qs.append(float(_UpperCamelCase ) )
# Here we implement the simple non-constant threshold for the predicted IG(X) value
# We will decay the selectivity of our secondary learner filter from
# 1 standard deviation above average to 1 below average after 10 batches.
if global_step == 10:
__lowerCamelCase = -1
if predicted_q < threshold:
__lowerCamelCase = False
# If we passed the filter, add the context to the batch!
if do_backprop:
contexts.append(np.array(context.cpu() ) )
__lowerCamelCase = outputs[0]
lm_loss.backward()
examples += 1
del outputs
# Once the batch is filled with enough contexts, backprop on the batch.
if examples == batch_size:
torch.cuda.empty_cache()
__lowerCamelCase = 0
# Do LM backprop
torch.nn.utils.clip_grad_norm_(model.parameters() ,3.0 )
lm_optimizer.step()
lm_scheduler.step() # Update learning rate schedule
global_step += 1
# Compute the performance of the transformer model at this batch
if global_step % eval_interval == 0:
__lowerCamelCase = compute_perplexity(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
test_perps.append(_UpperCamelCase )
print('''Test perplexity, step''' ,_UpperCamelCase ,''':''' ,_UpperCamelCase )
# Break out of the loop after 60 batches
if max_steps > 0 and global_step > 60:
break
if max_steps > 0 and global_step > 60:
break
# save finetuned transformer model
torch.save(model.state_dict() ,_UpperCamelCase )
torch.cuda.empty_cache()
# Do some cleaning up so we can reinitialize for the next run of this function
del lm_optimizer
del lm_scheduler
return model
def a__ ( ):
__lowerCamelCase = argparse.ArgumentParser(description='''Fine-tune a transformer model with IGF on a language modeling task''' )
# Required parameters
parser.add_argument(
'''--data_dir''' ,default=_UpperCamelCase ,type=_UpperCamelCase ,required=_UpperCamelCase ,help='''The input data dir. Should contain data files for WikiText.''' ,)
parser.add_argument(
'''--model_name_or_path''' ,default=_UpperCamelCase ,type=_UpperCamelCase ,required=_UpperCamelCase ,help='''Path to pretrained model or model identifier from huggingface.co/models''' ,)
parser.add_argument(
'''--data_file''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,help=(
'''A jbl file containing tokenized data which can be split as objective dataset, '''
'''train_dataset and test_dataset.'''
) ,)
parser.add_argument(
'''--igf_data_file''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,help='''A jbl file containing the context and information gain pairs to train secondary learner.''' ,)
parser.add_argument(
'''--output_dir''' ,default=_UpperCamelCase ,type=_UpperCamelCase ,required=_UpperCamelCase ,help='''The output directory where the final fine-tuned model is stored.''' ,)
parser.add_argument(
'''--tokenizer_name''' ,default=_UpperCamelCase ,type=_UpperCamelCase ,help='''Pretrained tokenizer name or path if not the same as model_name''' ,)
parser.add_argument('''--seed''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,help='''A seed for reproducible training.''' )
parser.add_argument(
'''--context_len''' ,default=32 ,type=_UpperCamelCase ,help=(
'''The maximum total input sequence length after tokenization. Sequences longer '''
'''than this will be truncated, sequences shorter will be padded.'''
) ,)
parser.add_argument(
'''--size_objective_set''' ,default=1_00 ,type=_UpperCamelCase ,help='''number of articles that are long enough to be used as our objective set''' ,)
parser.add_argument(
'''--eval_freq''' ,default=1_00 ,type=_UpperCamelCase ,help='''secondary model evaluation is triggered at eval_freq''' )
parser.add_argument('''--max_steps''' ,default=10_00 ,type=_UpperCamelCase ,help='''To calculate training epochs''' )
parser.add_argument(
'''--secondary_learner_batch_size''' ,default=1_28 ,type=_UpperCamelCase ,help='''batch size of training data for secondary learner''' ,)
parser.add_argument(
'''--batch_size''' ,default=16 ,type=_UpperCamelCase ,help='''batch size of training data of language model(gpt2) ''' )
parser.add_argument(
'''--eval_interval''' ,default=10 ,type=_UpperCamelCase ,help=(
'''decay the selectivity of our secondary learner filter from'''
'''1 standard deviation above average to 1 below average after 10 batches'''
) ,)
parser.add_argument(
'''--number''' ,default=1_00 ,type=_UpperCamelCase ,help='''The number of examples split to be used as objective_set/test_data''' )
parser.add_argument(
'''--min_len''' ,default=10_26 ,type=_UpperCamelCase ,help='''The minimum length of the article to be used as objective set''' )
parser.add_argument(
'''--secondary_learner_max_epochs''' ,default=15 ,type=_UpperCamelCase ,help='''number of epochs to train secondary learner''' )
parser.add_argument('''--trim''' ,default=_UpperCamelCase ,type=_UpperCamelCase ,help='''truncate the example if it exceeds context length''' )
parser.add_argument(
'''--threshold''' ,default=1.0 ,type=_UpperCamelCase ,help=(
'''The threshold value used by secondary learner to filter the train_data and allow only'''
''' informative data as input to the model'''
) ,)
parser.add_argument('''--finetuned_model_name''' ,default='''gpt2_finetuned.pt''' ,type=_UpperCamelCase ,help='''finetuned_model_name''' )
parser.add_argument(
'''--recopy_model''' ,default=_UpperCamelCase ,type=_UpperCamelCase ,help='''Reset the model to the original pretrained GPT-2 weights after each iteration''' ,)
# function calls
# Collecting *n* pairs of context and information gain(X, IG(X)) for training the secondary learner
generate_n_pairs(
context_len=32 ,max_steps=10 ,size_objective_set=1_00 ,min_len=10_26 ,trim=_UpperCamelCase ,data_file='''data/tokenized_stories_train_wikitext103.jbl''' ,igf_data_file='''igf_context_pairs.jbl''' ,)
# Load train data for secondary learner
__lowerCamelCase = joblib.load('''data/IGF_values.jbl''' )
# Train secondary learner
__lowerCamelCase = training_secondary_learner(
_UpperCamelCase ,secondary_learner_max_epochs=15 ,secondary_learner_batch_size=1_28 ,eval_freq=1_00 ,igf_model_path='''igf_model.pt''' ,)
# load pretrained gpt2 model
__lowerCamelCase = GPTaLMHeadModel.from_pretrained('''gpt2''' )
set_seed(42 )
# Generate train and test data to train and evaluate gpt2 model
__lowerCamelCase ,__lowerCamelCase = generate_datasets(
context_len=32 ,file='''data/tokenized_stories_train_wikitext103.jbl''' ,number=1_00 ,min_len=10_26 ,trim=_UpperCamelCase )
# fine-tuning of the gpt2 model using igf (Information Gain Filtration)
finetune(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,context_len=32 ,max_steps=10_00 ,batch_size=16 ,threshold=1.0 ,recopy_model=_UpperCamelCase ,secondary_learner=_UpperCamelCase ,eval_interval=10 ,finetuned_model_name='''gpt2_finetuned.pt''' ,)
if __name__ == "__main__":
main()
| 330 |
import argparse
import os
from pathlib import Path
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import PegasusConfig, PegasusForConditionalGeneration, PegasusTokenizer
from transformers.models.pegasus.configuration_pegasus import DEFAULTS, task_specific_params
a_ = [
# replace left string with right string to get the relevant state_dict key (identical state dict to bart)
["""memory_attention""", """encoder_attn"""],
["""attention""", """attn"""],
["""/""", """."""],
[""".LayerNorm.gamma""", """_layer_norm.weight"""],
[""".LayerNorm.beta""", """_layer_norm.bias"""],
["""r.layer_""", """r.layers."""],
["""output_proj""", """out_proj"""],
["""ffn.dense_1.""", """fc2."""],
["""ffn.dense.""", """fc1."""],
["""ffn_layer_norm""", """final_layer_norm"""],
["""kernel""", """weight"""],
["""encoder_layer_norm.""", """encoder.layer_norm."""],
["""decoder_layer_norm.""", """decoder.layer_norm."""],
["""embeddings.weights""", """shared.weight"""],
]
def a__ ( _UpperCamelCase : int ):
for pegasus_name, hf_name in PATTERNS:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = DEFAULTS.copy()
cfg_kwargs.update(_UpperCamelCase )
__lowerCamelCase = PegasusConfig(**_UpperCamelCase )
__lowerCamelCase = PegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.model.state_dict()
__lowerCamelCase = {}
for k, v in tf_weights.items():
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase )
if new_k not in sd:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if "dense" in k or "proj" in new_k:
__lowerCamelCase = v.T
__lowerCamelCase = torch.tensor(_UpperCamelCase ,dtype=sd[new_k].dtype )
assert v.shape == sd[new_k].shape, F"""{new_k}, {k}, {v.shape}, {sd[new_k].shape}"""
# make sure embedding.padding_idx is respected
__lowerCamelCase = torch.zeros_like(mapping['''shared.weight'''][cfg.pad_token_id + 1] )
__lowerCamelCase = mapping['''shared.weight''']
__lowerCamelCase = mapping['''shared.weight''']
__lowerCamelCase = {k: torch.zeros_like(_UpperCamelCase ) for k, v in sd.items() if k.endswith('''bias''' ) and k not in mapping}
mapping.update(**_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = torch_model.model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k for k in missing if k not in ['''encoder.embed_positions.weight''', '''decoder.embed_positions.weight''']
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : str="./ckpt/aeslc/model.ckpt-32000" ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''Adafactor''', '''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# save tokenizer first
__lowerCamelCase = Path(_UpperCamelCase ).parent.name
__lowerCamelCase = task_specific_params[F"""summarization_{dataset}"""]['''max_position_embeddings''']
__lowerCamelCase = PegasusTokenizer.from_pretrained('''sshleifer/pegasus''' ,model_max_length=_UpperCamelCase )
assert tok.model_max_length == desired_max_model_length
tok.save_pretrained(_UpperCamelCase )
# convert model
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = task_specific_params[F"""summarization_{dataset}"""]
if dataset == "large":
__lowerCamelCase = task_specific_params
__lowerCamelCase = convert_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
sd.pop('''model.decoder.embed_positions.weight''' )
sd.pop('''model.encoder.embed_positions.weight''' )
torch.save(_UpperCamelCase ,Path(_UpperCamelCase ) / '''pytorch_model.bin''' )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument("""tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
if args.save_dir is None:
a_ = Path(args.tf_ckpt_path).parent.name
a_ = os.path.join("""pegasus""", dataset)
convert_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir)
| 330 | 1 |
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = int(_UpperCamelCase )
if decimal in (0, 1): # Exit cases for the recursion
return str(_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = divmod(_UpperCamelCase ,2 )
return binary_recursive(_UpperCamelCase ) + str(_UpperCamelCase )
def a__ ( _UpperCamelCase : str ):
__lowerCamelCase = str(_UpperCamelCase ).strip()
if not number:
raise ValueError('''No input value was provided''' )
__lowerCamelCase = '''-''' if number.startswith('''-''' ) else ''''''
__lowerCamelCase = number.lstrip('''-''' )
if not number.isnumeric():
raise ValueError('''Input value is not an integer''' )
return F"""{negative}0b{binary_recursive(int(_UpperCamelCase ) )}"""
if __name__ == "__main__":
from doctest import testmod
testmod()
| 330 |
import argparse
import fairseq
import torch
from transformers import UniSpeechSatConfig, UniSpeechSatForCTC, UniSpeechSatForPreTraining, logging
logging.set_verbosity_info()
a_ = logging.get_logger(__name__)
a_ = {
"""post_extract_proj""": """feature_projection.projection""",
"""encoder.pos_conv.0""": """encoder.pos_conv_embed.conv""",
"""self_attn.k_proj""": """encoder.layers.*.attention.k_proj""",
"""self_attn.v_proj""": """encoder.layers.*.attention.v_proj""",
"""self_attn.q_proj""": """encoder.layers.*.attention.q_proj""",
"""self_attn.out_proj""": """encoder.layers.*.attention.out_proj""",
"""self_attn_layer_norm""": """encoder.layers.*.layer_norm""",
"""fc1""": """encoder.layers.*.feed_forward.intermediate_dense""",
"""fc2""": """encoder.layers.*.feed_forward.output_dense""",
"""final_layer_norm""": """encoder.layers.*.final_layer_norm""",
"""encoder.layer_norm""": """encoder.layer_norm""",
"""encoder.layer_norm_for_extract""": """layer_norm_for_extract""",
"""w2v_model.layer_norm""": """feature_projection.layer_norm""",
"""quantizer.weight_proj""": """quantizer.weight_proj""",
"""quantizer.vars""": """quantizer.codevectors""",
"""project_q""": """project_q""",
"""final_proj""": """project_hid""",
"""w2v_encoder.proj""": """lm_head""",
"""label_embs_concat""": """label_embeddings_concat""",
"""mask_emb""": """masked_spec_embed""",
"""spk_proj""": """speaker_proj""",
}
a_ = [
"""lm_head""",
"""quantizer.weight_proj""",
"""quantizer.codevectors""",
"""project_q""",
"""project_hid""",
"""label_embeddings_concat""",
"""speaker_proj""",
"""layer_norm_for_extract""",
]
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : Tuple ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Dict ):
for attribute in key.split('''.''' ):
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if weight_type is not None:
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase ).shape
else:
__lowerCamelCase = hf_pointer.shape
if hf_shape != value.shape:
raise ValueError(
F"""Shape of hf {key + "." + weight_type if weight_type is not None else ""} is {hf_shape}, but should be"""
F""" {value.shape} for {full_name}""" )
if weight_type == "weight":
__lowerCamelCase = value
elif weight_type == "weight_g":
__lowerCamelCase = value
elif weight_type == "weight_v":
__lowerCamelCase = value
elif weight_type == "bias":
__lowerCamelCase = value
else:
__lowerCamelCase = value
logger.info(F"""{key + "." + weight_type if weight_type is not None else ""} was initialized from {full_name}.""" )
def a__ ( _UpperCamelCase : Any ,_UpperCamelCase : Any ):
__lowerCamelCase = []
__lowerCamelCase = fairseq_model.state_dict()
__lowerCamelCase = hf_model.unispeech_sat.feature_extractor
for name, value in fairseq_dict.items():
__lowerCamelCase = False
if "conv_layers" in name:
load_conv_layer(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,hf_model.config.feat_extract_norm == '''group''' ,)
__lowerCamelCase = True
else:
for key, mapped_key in MAPPING.items():
__lowerCamelCase = '''unispeech_sat.''' + mapped_key if mapped_key not in TOP_LEVEL_KEYS else mapped_key
if key in name or key.split('''w2v_model.''' )[-1] == name.split('''.''' )[0]:
if "layer_norm_for_extract" in name and (".".join(name.split('''.''' )[:-1] ) != key):
# special case since naming is very similar
continue
__lowerCamelCase = True
if "*" in mapped_key:
__lowerCamelCase = name.split(_UpperCamelCase )[0].split('''.''' )[-2]
__lowerCamelCase = mapped_key.replace('''*''' ,_UpperCamelCase )
if "weight_g" in name:
__lowerCamelCase = '''weight_g'''
elif "weight_v" in name:
__lowerCamelCase = '''weight_v'''
elif "bias" in name:
__lowerCamelCase = '''bias'''
elif "weight" in name:
# TODO: don't match quantizer.weight_proj
__lowerCamelCase = '''weight'''
else:
__lowerCamelCase = None
set_recursively(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
continue
if not is_used:
unused_weights.append(_UpperCamelCase )
logger.warning(F"""Unused weights: {unused_weights}""" )
def a__ ( _UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Dict ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = full_name.split('''conv_layers.''' )[-1]
__lowerCamelCase = name.split('''.''' )
__lowerCamelCase = int(items[0] )
__lowerCamelCase = int(items[1] )
if type_id == 0:
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif (type_id == 2 and not use_group_norm) or (type_id == 2 and layer_id == 0 and use_group_norm):
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(_UpperCamelCase )
@torch.no_grad()
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : List[str] ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : List[Any]=True ):
if config_path is not None:
__lowerCamelCase = UniSpeechSatConfig.from_pretrained(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatConfig()
__lowerCamelCase = ''''''
if is_finetuned:
__lowerCamelCase = UniSpeechSatForCTC(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatForPreTraining(_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
__lowerCamelCase = model[0].eval()
recursively_load_weights(_UpperCamelCase ,_UpperCamelCase )
hf_wavavec.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument("""--checkpoint_path""", default=None, type=str, help="""Path to fairseq checkpoint""")
parser.add_argument("""--dict_path""", default=None, type=str, help="""Path to dict of fine-tuned model""")
parser.add_argument("""--config_path""", default=None, type=str, help="""Path to hf config.json of model to convert""")
parser.add_argument(
"""--not_finetuned""", action="""store_true""", help="""Whether the model to convert is a fine-tuned model or not"""
)
a_ = parser.parse_args()
convert_unispeech_sat_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 330 | 1 |
def a__ ( _UpperCamelCase : int ):
if not isinstance(_UpperCamelCase ,_UpperCamelCase ):
__lowerCamelCase = F"""Input value of [number={number}] must be an integer"""
raise TypeError(_UpperCamelCase )
if number < 0:
return False
__lowerCamelCase = number * number
while number > 0:
if number % 10 != number_square % 10:
return False
number //= 10
number_square //= 10
return True
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 |
from typing import List, Union
import numpy as np
from ..utils import add_end_docstrings, is_torch_available, is_vision_available, logging, requires_backends
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_torch_available():
import torch
from ..models.auto.modeling_auto import MODEL_FOR_DEPTH_ESTIMATION_MAPPING
a_ = logging.get_logger(__name__)
@add_end_docstrings(lowerCAmelCase__ )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
requires_backends(self , '''vision''' )
self.check_model_type(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return super().__call__(__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
return {}, {}, {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = load_image(__UpperCAmelCase )
__lowerCamelCase = image.size
__lowerCamelCase = self.image_processor(images=__UpperCAmelCase , return_tensors=self.framework )
return model_inputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.model(**__UpperCAmelCase )
return model_outputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = model_outputs.predicted_depth
__lowerCamelCase = torch.nn.functional.interpolate(
predicted_depth.unsqueeze(1 ) , size=self.image_size[::-1] , mode='''bicubic''' , align_corners=__UpperCAmelCase )
__lowerCamelCase = prediction.squeeze().cpu().numpy()
__lowerCamelCase = (output * 255 / np.max(__UpperCAmelCase )).astype('''uint8''' )
__lowerCamelCase = Image.fromarray(__UpperCAmelCase )
__lowerCamelCase = {}
__lowerCamelCase = predicted_depth
__lowerCamelCase = depth
return output_dict
| 330 | 1 |
import numpy as np
a_ = [
["""a""", """b""", """c""", """d""", """e"""],
["""f""", """g""", """h""", """i""", """k"""],
["""l""", """m""", """n""", """o""", """p"""],
["""q""", """r""", """s""", """t""", """u"""],
["""v""", """w""", """x""", """y""", """z"""],
]
class __lowerCAmelCase :
def __init__( self ):
'''simple docstring'''
__lowerCamelCase = np.array(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = np.where(letter == self.SQUARE )
__lowerCamelCase = np.concatenate([indexa + 1, indexa + 1] )
return indexes
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.SQUARE[indexa - 1, indexa - 1]
return letter
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = message.lower()
__lowerCamelCase = message.replace(''' ''' , '''''' )
__lowerCamelCase = message.replace('''j''' , '''i''' )
__lowerCamelCase = np.empty((2, len(__UpperCAmelCase )) )
for letter_index in range(len(__UpperCAmelCase ) ):
__lowerCamelCase = self.letter_to_numbers(message[letter_index] )
__lowerCamelCase = numbers[0]
__lowerCamelCase = numbers[1]
__lowerCamelCase = first_step.reshape(2 * len(__UpperCAmelCase ) )
__lowerCamelCase = ''''''
for numbers_index in range(len(__UpperCAmelCase ) ):
__lowerCamelCase = int(second_step[numbers_index * 2] )
__lowerCamelCase = int(second_step[(numbers_index * 2) + 1] )
__lowerCamelCase = self.numbers_to_letter(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = encoded_message + letter
return encoded_message
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = message.lower()
message.replace(''' ''' , '''''' )
__lowerCamelCase = np.empty(2 * len(__UpperCAmelCase ) )
for letter_index in range(len(__UpperCAmelCase ) ):
__lowerCamelCase = self.letter_to_numbers(message[letter_index] )
__lowerCamelCase = numbers[0]
__lowerCamelCase = numbers[1]
__lowerCamelCase = first_step.reshape((2, len(__UpperCAmelCase )) )
__lowerCamelCase = ''''''
for numbers_index in range(len(__UpperCAmelCase ) ):
__lowerCamelCase = int(second_step[0, numbers_index] )
__lowerCamelCase = int(second_step[1, numbers_index] )
__lowerCamelCase = self.numbers_to_letter(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = decoded_message + letter
return decoded_message
| 330 |
from typing import Dict, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
convert_to_rgb,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
OPENAI_CLIP_MEAN,
OPENAI_CLIP_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_vision_available, logging
a_ = logging.get_logger(__name__)
if is_vision_available():
import PIL
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""pixel_values"""]
def __init__( self , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = True , __UpperCAmelCase = 1 / 255 , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = True , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = size if size is not None else {'''shortest_edge''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
__lowerCamelCase = crop_size if crop_size is not None else {'''height''': 224, '''width''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase , 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 OPENAI_CLIP_MEAN
__lowerCamelCase = image_std if image_std is not None else OPENAI_CLIP_STD
__lowerCamelCase = do_convert_rgb
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
if "shortest_edge" not in size:
raise ValueError(F"""The `size` parameter must contain the key `shortest_edge`. Got {size.keys()}""" )
__lowerCamelCase = get_resize_output_image_size(__UpperCAmelCase , size=size['''shortest_edge'''] , default_to_square=__UpperCAmelCase )
return resize(__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase )
if "height" not in size or "width" not in size:
raise ValueError(F"""The `size` parameter must contain the keys (height, width). Got {size.keys()}""" )
return center_crop(__UpperCAmelCase , size=(size['''height'''], size['''width''']) , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return rescale(__UpperCAmelCase , scale=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return normalize(__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = ChannelDimension.FIRST , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = do_resize if do_resize is not None else self.do_resize
__lowerCamelCase = size if size is not None else self.size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''size''' , default_to_square=__UpperCAmelCase )
__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 = crop_size if crop_size is not None else self.crop_size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''crop_size''' , default_to_square=__UpperCAmelCase )
__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 = do_convert_rgb if do_convert_rgb is not None else self.do_convert_rgb
__lowerCamelCase = make_list_of_images(__UpperCAmelCase )
if not valid_images(__UpperCAmelCase ):
raise ValueError(
'''Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '''
'''torch.Tensor, tf.Tensor or jax.ndarray.''' )
if do_resize and size is None:
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.''' )
# PIL RGBA images are converted to RGB
if do_convert_rgb:
__lowerCamelCase = [convert_to_rgb(__UpperCAmelCase ) for image in images]
# All transformations expect numpy arrays.
__lowerCamelCase = [to_numpy_array(__UpperCAmelCase ) for image in images]
if do_resize:
__lowerCamelCase = [self.resize(image=__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase ) for image in images]
if do_center_crop:
__lowerCamelCase = [self.center_crop(image=__UpperCAmelCase , size=__UpperCAmelCase ) for image in images]
if do_rescale:
__lowerCamelCase = [self.rescale(image=__UpperCAmelCase , scale=__UpperCAmelCase ) for image in images]
if do_normalize:
__lowerCamelCase = [self.normalize(image=__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase ) for image in images]
__lowerCamelCase = [to_channel_dimension_format(__UpperCAmelCase , __UpperCAmelCase ) for image in images]
__lowerCamelCase = {'''pixel_values''': images}
return BatchFeature(data=__UpperCAmelCase , tensor_type=__UpperCAmelCase )
| 330 | 1 |
from ...utils import (
OptionalDependencyNotAvailable,
is_torch_available,
is_transformers_available,
is_transformers_version,
)
try:
if not (is_transformers_available() and is_torch_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from ...utils.dummy_torch_and_transformers_objects import (
ImageTextPipelineOutput,
UniDiffuserPipeline,
)
else:
from .modeling_text_decoder import UniDiffuserTextDecoder
from .modeling_uvit import UniDiffuserModel, UTransformeraDModel
from .pipeline_unidiffuser import ImageTextPipelineOutput, UniDiffuserPipeline
| 330 |
from __future__ import annotations
from typing import Generic, TypeVar
a_ = TypeVar("""T""")
class __lowerCAmelCase ( Generic[T] ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = data
__lowerCamelCase = self
__lowerCamelCase = 0
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# map from node name to the node object
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# create a new set with x as its member
__lowerCamelCase = DisjointSetTreeNode(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# find the set x belongs to (with path-compression)
__lowerCamelCase = self.map[data]
if elem_ref != elem_ref.parent:
__lowerCamelCase = self.find_set(elem_ref.parent.data )
return elem_ref.parent
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# helper function for union operation
if nodea.rank > nodea.rank:
__lowerCamelCase = nodea
else:
__lowerCamelCase = nodea
if nodea.rank == nodea.rank:
nodea.rank += 1
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# merge 2 disjoint sets
self.link(self.find_set(__UpperCAmelCase ) , self.find_set(__UpperCAmelCase ) )
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# connections: map from the node to the neighbouring nodes (with weights)
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# add a node ONLY if its not present in the graph
if node not in self.connections:
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# add an edge with the given weight
self.add_node(__UpperCAmelCase )
self.add_node(__UpperCAmelCase )
__lowerCamelCase = weight
__lowerCamelCase = weight
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = []
__lowerCamelCase = set()
for start in self.connections:
for end in self.connections[start]:
if (start, end) not in seen:
seen.add((end, start) )
edges.append((start, end, self.connections[start][end]) )
edges.sort(key=lambda __UpperCAmelCase : x[2] )
# creating the disjoint set
__lowerCamelCase = DisjointSetTree[T]()
for node in self.connections:
disjoint_set.make_set(__UpperCAmelCase )
# MST generation
__lowerCamelCase = 0
__lowerCamelCase = 0
__lowerCamelCase = GraphUndirectedWeighted[T]()
while num_edges < len(self.connections ) - 1:
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = edges[index]
index += 1
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
if parent_u != parent_v:
num_edges += 1
graph.add_edge(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
disjoint_set.union(__UpperCAmelCase , __UpperCAmelCase )
return graph
| 330 | 1 |
from typing import List, Optional, Union
import numpy as np
from ...feature_extraction_sequence_utils import SequenceFeatureExtractor
from ...feature_extraction_utils import BatchFeature
from ...utils import PaddingStrategy, TensorType, logging
a_ = logging.get_logger(__name__)
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""input_values""", """padding_mask"""]
def __init__( self , __UpperCAmelCase = 1 , __UpperCAmelCase = 24000 , __UpperCAmelCase = 0.0 , __UpperCAmelCase = None , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(feature_size=__UpperCAmelCase , sampling_rate=__UpperCAmelCase , padding_value=__UpperCAmelCase , **__UpperCAmelCase )
__lowerCamelCase = chunk_length_s
__lowerCamelCase = overlap
@property
def lowerCamelCase ( self ):
'''simple docstring'''
if self.chunk_length_s is None:
return None
else:
return int(self.chunk_length_s * self.sampling_rate )
@property
def lowerCamelCase ( self ):
'''simple docstring'''
if self.chunk_length_s is None or self.overlap is None:
return None
else:
return max(1 , int((1.0 - self.overlap) * self.chunk_length ) )
def __call__( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = False , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , ):
'''simple docstring'''
if sampling_rate is not None:
if sampling_rate != self.sampling_rate:
raise ValueError(
F"""The model corresponding to this feature extractor: {self} was trained using a sampling rate of"""
F""" {self.sampling_rate}. Please make sure that the provided audio input was sampled with"""
F""" {self.sampling_rate} and not {sampling_rate}.""" )
else:
logger.warning(
'''It is strongly recommended to pass the `sampling_rate` argument to this function. '''
'''Failing to do so can result in silent errors that might be hard to debug.''' )
if padding and truncation:
raise ValueError('''Both padding and truncation were set. Make sure you only set one.''' )
elif padding is None:
# by default let's pad the inputs
__lowerCamelCase = True
__lowerCamelCase = bool(
isinstance(__UpperCAmelCase , (list, tuple) ) and (isinstance(raw_audio[0] , (np.ndarray, tuple, list) )) )
if is_batched:
__lowerCamelCase = [np.asarray(__UpperCAmelCase , dtype=np.floataa ).T for audio in raw_audio]
elif not is_batched and not isinstance(__UpperCAmelCase , np.ndarray ):
__lowerCamelCase = np.asarray(__UpperCAmelCase , dtype=np.floataa )
elif isinstance(__UpperCAmelCase , np.ndarray ) and raw_audio.dtype is np.dtype(np.floataa ):
__lowerCamelCase = raw_audio.astype(np.floataa )
# always return batch
if not is_batched:
__lowerCamelCase = [np.asarray(__UpperCAmelCase ).T]
# verify inputs are valid
for idx, example in enumerate(__UpperCAmelCase ):
if example.ndim > 2:
raise ValueError(F"""Expected input shape (channels, length) but got shape {example.shape}""" )
if self.feature_size == 1 and example.ndim != 1:
raise ValueError(F"""Expected mono audio but example has {example.shape[-1]} channels""" )
if self.feature_size == 2 and example.shape[-1] != 2:
raise ValueError(F"""Expected stereo audio but example has {example.shape[-1]} channels""" )
__lowerCamelCase = None
__lowerCamelCase = BatchFeature({'''input_values''': raw_audio} )
if self.chunk_stride is not None and self.chunk_length is not None and max_length is None:
if truncation:
__lowerCamelCase = min(array.shape[0] for array in raw_audio )
__lowerCamelCase = int(np.floor(max_length / self.chunk_stride ) )
__lowerCamelCase = (nb_step - 1) * self.chunk_stride + self.chunk_length
elif padding:
__lowerCamelCase = max(array.shape[0] for array in raw_audio )
__lowerCamelCase = int(np.ceil(max_length / self.chunk_stride ) )
__lowerCamelCase = (nb_step - 1) * self.chunk_stride + self.chunk_length
__lowerCamelCase = '''max_length'''
else:
__lowerCamelCase = input_values
# normal padding on batch
if padded_inputs is None:
__lowerCamelCase = self.pad(
__UpperCAmelCase , max_length=__UpperCAmelCase , truncation=__UpperCAmelCase , padding=__UpperCAmelCase , return_attention_mask=__UpperCAmelCase , )
if padding:
__lowerCamelCase = padded_inputs.pop('''attention_mask''' )
__lowerCamelCase = []
for example in padded_inputs.pop('''input_values''' ):
if self.feature_size == 1:
__lowerCamelCase = example[..., None]
input_values.append(example.T )
__lowerCamelCase = input_values
if return_tensors is not None:
__lowerCamelCase = padded_inputs.convert_to_tensors(__UpperCAmelCase )
return padded_inputs
| 330 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=2 , __UpperCAmelCase=8 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=99 , __UpperCAmelCase=16 , __UpperCAmelCase=5 , __UpperCAmelCase=2 , __UpperCAmelCase=36 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.0 , __UpperCAmelCase=512 , __UpperCAmelCase=16 , __UpperCAmelCase=2 , __UpperCAmelCase=0.02 , __UpperCAmelCase=3 , __UpperCAmelCase=4 , __UpperCAmelCase=None , ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
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=__UpperCAmelCase , initializer_range=self.initializer_range , )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_config()
__lowerCamelCase = 300
return config
def lowerCamelCase ( self ):
'''simple docstring'''
(
(
__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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = True
__lowerCamelCase = MraModel(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , )
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , )
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForQuestionAnswering(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , start_positions=__UpperCAmelCase , end_positions=__UpperCAmelCase , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForSequenceClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForTokenClassification(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_choices
__lowerCamelCase = MraForMultipleChoice(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
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(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowerCamelCase ( self ):
'''simple docstring'''
__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 __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = ()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModelTester(self )
__lowerCamelCase = ConfigTester(self , config_class=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__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(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = MraModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
@unittest.skip(reason='''MRA does not output attentions''' )
def lowerCamelCase ( self ):
'''simple docstring'''
return
@require_torch
class __lowerCAmelCase ( unittest.TestCase ):
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = torch.Size((1, 256, 768) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[-0.0_140, 0.0_830, -0.0_381], [0.1_546, 0.1_402, 0.0_220], [0.1_162, 0.0_851, 0.0_165]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 256, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[9.2_595, -3.6_038, 11.8_819], [9.3_869, -3.2_693, 11.0_956], [11.8_524, -3.4_938, 13.1_210]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' )
__lowerCamelCase = torch.arange(4096 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 4096, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[5.4_789, -2.3_564, 7.5_064], [7.9_067, -1.3_369, 9.9_668], [9.0_712, -1.8_106, 7.0_380]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
import argparse
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import BigBirdPegasusConfig, BigBirdPegasusForConditionalGeneration
a_ = [
# tf -> hf
("""/""", """."""),
("""layer_""", """layers."""),
("""kernel""", """weight"""),
("""beta""", """bias"""),
("""gamma""", """weight"""),
("""pegasus""", """model"""),
]
a_ = [
(""".output.dense""", """.fc2"""),
("""intermediate.LayerNorm""", """final_layer_norm"""),
("""intermediate.dense""", """fc1"""),
]
a_ = (
INIT_COMMON
+ [
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.out_proj"""),
("""attention.self""", """self_attn"""),
("""attention.encdec.LayerNorm""", """encoder_attn_layer_norm"""),
("""attention.encdec_output.dense""", """encoder_attn.out_proj"""),
("""attention.encdec""", """encoder_attn"""),
("""key""", """k_proj"""),
("""value""", """v_proj"""),
("""query""", """q_proj"""),
("""decoder.LayerNorm""", """decoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = (
INIT_COMMON
+ [
("""embeddings.word_embeddings""", """shared.weight"""),
("""embeddings.position_embeddings""", """embed_positions.weight"""),
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.output"""),
("""attention.self""", """self_attn.self"""),
("""encoder.LayerNorm""", """encoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = [
"""encdec/key/bias""",
"""encdec/query/bias""",
"""encdec/value/bias""",
"""self/key/bias""",
"""self/query/bias""",
"""self/value/bias""",
"""encdec_output/dense/bias""",
"""attention/output/dense/bias""",
]
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : Optional[Any] ):
for tf_name, hf_name in patterns:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = BigBirdPegasusConfig(**_UpperCamelCase )
__lowerCamelCase = BigBirdPegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
__lowerCamelCase = {}
# separating decoder weights
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if k.startswith('''pegasus/decoder''' )}
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if not k.startswith('''pegasus/decoder''' )}
for k, v in tqdm(decoder_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = DECODER_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
for k, v in tqdm(remaining_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = REMAINING_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict and k != "pegasus/embeddings/position_embeddings":
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
if k != "pegasus/embeddings/position_embeddings":
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
__lowerCamelCase = mapping['''model.embed_positions.weight''']
__lowerCamelCase = mapping.pop('''model.embed_positions.weight''' )
__lowerCamelCase ,__lowerCamelCase = torch_model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k
for k in missing
if k
not in [
'''final_logits_bias''',
'''model.encoder.embed_tokens.weight''',
'''model.decoder.embed_tokens.weight''',
'''lm_head.weight''',
]
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ,_UpperCamelCase : dict ):
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = convert_bigbird_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""--save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
a_ = {}
convert_bigbird_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir, config_update=config_update)
| 330 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_torch_available,
)
a_ = {"""configuration_encoder_decoder""": ["""EncoderDecoderConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""EncoderDecoderModel"""]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TFEncoderDecoderModel"""]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""FlaxEncoderDecoderModel"""]
if TYPE_CHECKING:
from .configuration_encoder_decoder import EncoderDecoderConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_encoder_decoder import EncoderDecoderModel
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_encoder_decoder import TFEncoderDecoderModel
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_encoder_decoder import FlaxEncoderDecoderModel
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
from PIL import Image
def a__ ( _UpperCamelCase : Image ,_UpperCamelCase : int ):
__lowerCamelCase = (2_59 * (level + 2_55)) / (2_55 * (2_59 - level))
def contrast(_UpperCamelCase : int ) -> int:
return int(1_28 + factor * (c - 1_28) )
return img.point(_UpperCamelCase )
if __name__ == "__main__":
# Load image
with Image.open("""image_data/lena.jpg""") as img:
# Change contrast to 170
a_ = change_contrast(img, 170)
cont_img.save("""image_data/lena_high_contrast.png""", format="""png""")
| 330 |
from string import ascii_lowercase, ascii_uppercase
def a__ ( _UpperCamelCase : str ):
if not sentence:
return ""
__lowerCamelCase = dict(zip(_UpperCamelCase ,_UpperCamelCase ) )
return lower_to_upper.get(sentence[0] ,sentence[0] ) + sentence[1:]
if __name__ == "__main__":
from doctest import testmod
testmod()
| 330 | 1 |
import argparse
import os
import numpy as np
import tensorflow as tf
import torch
from transformers import BertModel
def a__ ( _UpperCamelCase : BertModel ,_UpperCamelCase : str ,_UpperCamelCase : str ):
__lowerCamelCase = ('''dense.weight''', '''attention.self.query''', '''attention.self.key''', '''attention.self.value''')
__lowerCamelCase = (
('''layer.''', '''layer_'''),
('''word_embeddings.weight''', '''word_embeddings'''),
('''position_embeddings.weight''', '''position_embeddings'''),
('''token_type_embeddings.weight''', '''token_type_embeddings'''),
('''.''', '''/'''),
('''LayerNorm/weight''', '''LayerNorm/gamma'''),
('''LayerNorm/bias''', '''LayerNorm/beta'''),
('''weight''', '''kernel'''),
)
if not os.path.isdir(_UpperCamelCase ):
os.makedirs(_UpperCamelCase )
__lowerCamelCase = model.state_dict()
def to_tf_var_name(_UpperCamelCase : str ):
for patt, repl in iter(_UpperCamelCase ):
__lowerCamelCase = name.replace(_UpperCamelCase ,_UpperCamelCase )
return F"""bert/{name}"""
def create_tf_var(_UpperCamelCase : np.ndarray ,_UpperCamelCase : str ,_UpperCamelCase : tf.Session ):
__lowerCamelCase = tf.dtypes.as_dtype(tensor.dtype )
__lowerCamelCase = tf.get_variable(dtype=_UpperCamelCase ,shape=tensor.shape ,name=_UpperCamelCase ,initializer=tf.zeros_initializer() )
session.run(tf.variables_initializer([tf_var] ) )
session.run(_UpperCamelCase )
return tf_var
tf.reset_default_graph()
with tf.Session() as session:
for var_name in state_dict:
__lowerCamelCase = to_tf_var_name(_UpperCamelCase )
__lowerCamelCase = state_dict[var_name].numpy()
if any(x in var_name for x in tensors_to_transpose ):
__lowerCamelCase = torch_tensor.T
__lowerCamelCase = create_tf_var(tensor=_UpperCamelCase ,name=_UpperCamelCase ,session=_UpperCamelCase )
tf.keras.backend.set_value(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = session.run(_UpperCamelCase )
print(F"""Successfully created {tf_name}: {np.allclose(_UpperCamelCase ,_UpperCamelCase )}""" )
__lowerCamelCase = tf.train.Saver(tf.trainable_variables() )
saver.save(_UpperCamelCase ,os.path.join(_UpperCamelCase ,model_name.replace('''-''' ,'''_''' ) + '''.ckpt''' ) )
def a__ ( _UpperCamelCase : Union[str, Any]=None ):
__lowerCamelCase = argparse.ArgumentParser()
parser.add_argument('''--model_name''' ,type=_UpperCamelCase ,required=_UpperCamelCase ,help='''model name e.g. bert-base-uncased''' )
parser.add_argument(
'''--cache_dir''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,required=_UpperCamelCase ,help='''Directory containing pytorch model''' )
parser.add_argument('''--pytorch_model_path''' ,type=_UpperCamelCase ,required=_UpperCamelCase ,help='''/path/to/<pytorch-model-name>.bin''' )
parser.add_argument('''--tf_cache_dir''' ,type=_UpperCamelCase ,required=_UpperCamelCase ,help='''Directory in which to save tensorflow model''' )
__lowerCamelCase = parser.parse_args(_UpperCamelCase )
__lowerCamelCase = BertModel.from_pretrained(
pretrained_model_name_or_path=args.model_name ,state_dict=torch.load(args.pytorch_model_path ) ,cache_dir=args.cache_dir ,)
convert_pytorch_checkpoint_to_tf(model=_UpperCamelCase ,ckpt_dir=args.tf_cache_dir ,model_name=args.model_name )
if __name__ == "__main__":
main()
| 330 |
from transformers import BertTokenizer, EncoderDecoderModel, SeqaSeqTrainer, SeqaSeqTrainingArguments
from transformers.testing_utils import TestCasePlus, require_torch, slow
from transformers.utils import is_datasets_available
if is_datasets_available():
import datasets
class __lowerCAmelCase ( lowerCAmelCase__ ):
@slow
@require_torch
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = EncoderDecoderModel.from_encoder_decoder_pretrained('''prajjwal1/bert-tiny''' , '''prajjwal1/bert-tiny''' )
__lowerCamelCase = BertTokenizer.from_pretrained('''bert-base-uncased''' )
__lowerCamelCase = bertabert.config.encoder.vocab_size
__lowerCamelCase = tokenizer.sep_token_id
__lowerCamelCase = tokenizer.cls_token_id
__lowerCamelCase = 128
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''train[:1%]''' )
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''validation[:1%]''' )
__lowerCamelCase = train_dataset.select(range(32 ) )
__lowerCamelCase = val_dataset.select(range(16 ) )
__lowerCamelCase = 4
def _map_to_encoder_decoder_inputs(__UpperCAmelCase ):
# Tokenizer will automatically set [BOS] <text> [EOS]
__lowerCamelCase = tokenizer(batch['''article'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=512 )
__lowerCamelCase = tokenizer(batch['''highlights'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=128 )
__lowerCamelCase = inputs.input_ids
__lowerCamelCase = inputs.attention_mask
__lowerCamelCase = outputs.input_ids
__lowerCamelCase = outputs.input_ids.copy()
__lowerCamelCase = [
[-100 if token == tokenizer.pad_token_id else token for token in labels] for labels in batch['''labels''']
]
__lowerCamelCase = outputs.attention_mask
assert all(len(__UpperCAmelCase ) == 512 for x in inputs.input_ids )
assert all(len(__UpperCAmelCase ) == 128 for x in outputs.input_ids )
return batch
def _compute_metrics(__UpperCAmelCase ):
__lowerCamelCase = pred.label_ids
__lowerCamelCase = pred.predictions
# all unnecessary tokens are removed
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = sum([int(pred_str[i] == label_str[i] ) for i in range(len(__UpperCAmelCase ) )] ) / len(__UpperCAmelCase )
return {"accuracy": accuracy}
# map train dataset
__lowerCamelCase = train_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
train_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
# same for validation dataset
__lowerCamelCase = val_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
val_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
__lowerCamelCase = self.get_auto_remove_tmp_dir()
__lowerCamelCase = SeqaSeqTrainingArguments(
output_dir=__UpperCAmelCase , per_device_train_batch_size=__UpperCAmelCase , per_device_eval_batch_size=__UpperCAmelCase , predict_with_generate=__UpperCAmelCase , evaluation_strategy='''steps''' , do_train=__UpperCAmelCase , do_eval=__UpperCAmelCase , warmup_steps=0 , eval_steps=2 , logging_steps=2 , )
# instantiate trainer
__lowerCamelCase = SeqaSeqTrainer(
model=__UpperCAmelCase , args=__UpperCAmelCase , compute_metrics=_compute_metrics , train_dataset=__UpperCAmelCase , eval_dataset=__UpperCAmelCase , tokenizer=__UpperCAmelCase , )
# start training
trainer.train()
| 330 | 1 |
import importlib
import math
import os
from dataclasses import dataclass
from enum import Enum
from typing import Any, Dict, Optional, Tuple, Union
import flax
import jax.numpy as jnp
from ..utils import BaseOutput
a_ = """scheduler_config.json"""
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 1
lowerCAmelCase__ = 2
lowerCAmelCase__ = 3
lowerCAmelCase__ = 4
lowerCAmelCase__ = 5
@dataclass
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 42
class __lowerCAmelCase :
lowerCAmelCase__ = SCHEDULER_CONFIG_NAME
lowerCAmelCase__ = ["""dtype"""]
lowerCAmelCase__ = []
lowerCAmelCase__ = True
@classmethod
def lowerCamelCase ( cls , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=False , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = cls.load_config(
pretrained_model_name_or_path=__UpperCAmelCase , subfolder=__UpperCAmelCase , return_unused_kwargs=__UpperCAmelCase , **__UpperCAmelCase , )
__lowerCamelCase ,__lowerCamelCase = cls.from_config(__UpperCAmelCase , return_unused_kwargs=__UpperCAmelCase , **__UpperCAmelCase )
if hasattr(__UpperCAmelCase , '''create_state''' ) and getattr(__UpperCAmelCase , '''has_state''' , __UpperCAmelCase ):
__lowerCamelCase = scheduler.create_state()
if return_unused_kwargs:
return scheduler, state, unused_kwargs
return scheduler, state
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False , **__UpperCAmelCase ):
'''simple docstring'''
self.save_config(save_directory=__UpperCAmelCase , push_to_hub=__UpperCAmelCase , **__UpperCAmelCase )
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._get_compatibles()
@classmethod
def lowerCamelCase ( cls ):
'''simple docstring'''
__lowerCamelCase = list(set([cls.__name__] + cls._compatibles ) )
__lowerCamelCase = importlib.import_module(__name__.split('''.''' )[0] )
__lowerCamelCase = [
getattr(__UpperCAmelCase , __UpperCAmelCase ) for c in compatible_classes_str if hasattr(__UpperCAmelCase , __UpperCAmelCase )
]
return compatible_classes
def a__ ( _UpperCamelCase : jnp.ndarray ,_UpperCamelCase : Tuple[int] ):
assert len(_UpperCamelCase ) >= x.ndim
return jnp.broadcast_to(x.reshape(x.shape + (1,) * (len(_UpperCamelCase ) - x.ndim) ) ,_UpperCamelCase )
def a__ ( _UpperCamelCase : int ,_UpperCamelCase : str=0.999 ,_UpperCamelCase : Optional[int]=jnp.floataa ):
def alpha_bar(_UpperCamelCase : List[str] ):
return math.cos((time_step + 0.008) / 1.008 * math.pi / 2 ) ** 2
__lowerCamelCase = []
for i in range(_UpperCamelCase ):
__lowerCamelCase = i / num_diffusion_timesteps
__lowerCamelCase = (i + 1) / num_diffusion_timesteps
betas.append(min(1 - alpha_bar(_UpperCamelCase ) / alpha_bar(_UpperCamelCase ) ,_UpperCamelCase ) )
return jnp.array(_UpperCamelCase ,dtype=_UpperCamelCase )
@flax.struct.dataclass
class __lowerCAmelCase :
lowerCAmelCase__ = 42
lowerCAmelCase__ = 42
lowerCAmelCase__ = 42
@classmethod
def lowerCamelCase ( cls , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = scheduler.config
if config.trained_betas is not None:
__lowerCamelCase = jnp.asarray(config.trained_betas , dtype=scheduler.dtype )
elif config.beta_schedule == "linear":
__lowerCamelCase = jnp.linspace(config.beta_start , config.beta_end , config.num_train_timesteps , dtype=scheduler.dtype )
elif config.beta_schedule == "scaled_linear":
# this schedule is very specific to the latent diffusion model.
__lowerCamelCase = (
jnp.linspace(
config.beta_start**0.5 , config.beta_end**0.5 , config.num_train_timesteps , dtype=scheduler.dtype )
** 2
)
elif config.beta_schedule == "squaredcos_cap_v2":
# Glide cosine schedule
__lowerCamelCase = betas_for_alpha_bar(config.num_train_timesteps , dtype=scheduler.dtype )
else:
raise NotImplementedError(
F"""beta_schedule {config.beta_schedule} is not implemented for scheduler {scheduler.__class__.__name__}""" )
__lowerCamelCase = 1.0 - betas
__lowerCamelCase = jnp.cumprod(__UpperCAmelCase , axis=0 )
return cls(
alphas=__UpperCAmelCase , betas=__UpperCAmelCase , alphas_cumprod=__UpperCAmelCase , )
def a__ ( _UpperCamelCase : CommonSchedulerState ,_UpperCamelCase : jnp.ndarray ,_UpperCamelCase : jnp.ndarray ,_UpperCamelCase : jnp.ndarray ):
__lowerCamelCase = state.alphas_cumprod
__lowerCamelCase = alphas_cumprod[timesteps] ** 0.5
__lowerCamelCase = sqrt_alpha_prod.flatten()
__lowerCamelCase = broadcast_to_shape_from_left(_UpperCamelCase ,original_samples.shape )
__lowerCamelCase = (1 - alphas_cumprod[timesteps]) ** 0.5
__lowerCamelCase = sqrt_one_minus_alpha_prod.flatten()
__lowerCamelCase = broadcast_to_shape_from_left(_UpperCamelCase ,original_samples.shape )
return sqrt_alpha_prod, sqrt_one_minus_alpha_prod
def a__ ( _UpperCamelCase : CommonSchedulerState ,_UpperCamelCase : jnp.ndarray ,_UpperCamelCase : jnp.ndarray ,_UpperCamelCase : jnp.ndarray ):
__lowerCamelCase ,__lowerCamelCase = get_sqrt_alpha_prod(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = sqrt_alpha_prod * original_samples + sqrt_one_minus_alpha_prod * noise
return noisy_samples
def a__ ( _UpperCamelCase : CommonSchedulerState ,_UpperCamelCase : jnp.ndarray ,_UpperCamelCase : jnp.ndarray ,_UpperCamelCase : jnp.ndarray ):
__lowerCamelCase ,__lowerCamelCase = get_sqrt_alpha_prod(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = sqrt_alpha_prod * noise - sqrt_one_minus_alpha_prod * sample
return velocity
| 330 |
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_timm_backbone""": ["""TimmBackboneConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TimmBackbone"""]
if TYPE_CHECKING:
from .configuration_timm_backbone import TimmBackboneConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_timm_backbone import TimmBackbone
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
from __future__ import annotations
from functools import lru_cache
from math import ceil
a_ = 100
a_ = set(range(3, NUM_PRIMES, 2))
primes.add(2)
a_ = 42
for prime in range(3, ceil(NUM_PRIMES**0.5), 2):
if prime not in primes:
continue
primes.difference_update(set(range(prime * prime, NUM_PRIMES, prime)))
@lru_cache(maxsize=1_00 )
def a__ ( _UpperCamelCase : int ):
if number_to_partition < 0:
return set()
elif number_to_partition == 0:
return {1}
__lowerCamelCase = set()
__lowerCamelCase = 42
__lowerCamelCase = 42
for prime in primes:
if prime > number_to_partition:
continue
for sub in partition(number_to_partition - prime ):
ret.add(sub * prime )
return ret
def a__ ( _UpperCamelCase : int = 50_00 ):
for number_to_partition in range(1 ,_UpperCamelCase ):
if len(partition(_UpperCamelCase ) ) > number_unique_partitions:
return number_to_partition
return None
if __name__ == "__main__":
print(f"{solution() = }")
| 330 |
import random
import unittest
from torch.utils.data import BatchSampler, DataLoader, IterableDataset
from accelerate import Accelerator
from accelerate.data_loader import (
BatchSamplerShard,
DataLoaderDispatcher,
DataLoaderShard,
IterableDatasetShard,
SkipBatchSampler,
SkipDataLoader,
skip_first_batches,
)
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase=0.01 , __UpperCAmelCase=1000 ):
'''simple docstring'''
__lowerCamelCase = p_stop
__lowerCamelCase = max_length
def __iter__( self ):
'''simple docstring'''
__lowerCamelCase = 0
__lowerCamelCase = False
while not stop and count < self.max_length:
yield count
count += 1
__lowerCamelCase = random.random() < self.p_stop
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = [
BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
for i in range(2 )
]
__lowerCamelCase = [list(__UpperCAmelCase ) for batch_sampler_shard in batch_sampler_shards]
if not split_batches:
self.assertListEqual([len(__UpperCAmelCase ) for shard in batch_sampler_shards] , [len(__UpperCAmelCase ) for e in expected] )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [0, 1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 0]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [1, 2, 3]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1, 0]], [[1, 0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 0]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], [[0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = [[0, 1, 2], [3, 4], [5, 6, 7, 8], [9, 10, 11], [12, 13]]
__lowerCamelCase = [BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , even_batches=__UpperCAmelCase ) for i in range(2 )]
self.assertEqual(len(batch_sampler_shards[0] ) , 3 )
self.assertEqual(len(batch_sampler_shards[1] ) , 2 )
self.assertListEqual(list(batch_sampler_shards[0] ) , [[0, 1, 2], [5, 6, 7, 8], [12, 13]] )
self.assertListEqual(list(batch_sampler_shards[1] ) , [[3, 4], [9, 10, 11]] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=2 , __UpperCAmelCase=False ):
'''simple docstring'''
random.seed(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = [
IterableDatasetShard(
__UpperCAmelCase , batch_size=__UpperCAmelCase , drop_last=__UpperCAmelCase , num_processes=__UpperCAmelCase , process_index=__UpperCAmelCase , split_batches=__UpperCAmelCase , )
for i in range(__UpperCAmelCase )
]
__lowerCamelCase = []
for iterable_dataset_shard in iterable_dataset_shards:
# Since our random iterable dataset will be... random... we need to use a seed to get reproducible results.
random.seed(__UpperCAmelCase )
iterable_dataset_lists.append(list(__UpperCAmelCase ) )
__lowerCamelCase = batch_size // num_processes if split_batches else batch_size
# All iterable dataset shard should have the same length, a round multiple of shard_batch_size
__lowerCamelCase = iterable_dataset_lists[0]
for l in iterable_dataset_lists[1:]:
self.assertEqual(len(__UpperCAmelCase ) , len(__UpperCAmelCase ) )
self.assertTrue(len(__UpperCAmelCase ) % shard_batch_size == 0 )
__lowerCamelCase = []
for idx in range(0 , len(__UpperCAmelCase ) , __UpperCAmelCase ):
for l in iterable_dataset_lists:
observed += l[idx : idx + shard_batch_size]
if not drop_last:
while len(__UpperCAmelCase ) < len(__UpperCAmelCase ):
reference += reference
self.assertListEqual(__UpperCAmelCase , reference[: len(__UpperCAmelCase )] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = 42
__lowerCamelCase = RandomIterableDataset()
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
# Edge case with a very small dataset
__lowerCamelCase = RandomIterableDataset(max_length=2 )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BatchSampler(range(16 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = SkipBatchSampler(__UpperCAmelCase , 2 )
self.assertListEqual(list(__UpperCAmelCase ) , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = SkipDataLoader(list(range(16 ) ) , batch_size=4 , skip_batches=2 )
self.assertListEqual([t.tolist() for t in dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoader(list(range(16 ) ) , batch_size=4 )
__lowerCamelCase = skip_first_batches(__UpperCAmelCase , num_batches=2 )
self.assertListEqual([t.tolist() for t in new_dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoaderShard(list(range(16 ) ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
def lowerCamelCase ( self ):
'''simple docstring'''
Accelerator()
__lowerCamelCase = DataLoaderDispatcher(range(16 ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
| 330 | 1 |
from __future__ import annotations
a_ = [-10, -5, 0, 5, 5.1, 11, 13, 21, 3, 4, -21, -10, -5, -1, 0]
a_ = [-5, 0, 5, 5.1, 11, 13, 21, -1, 4, -1, -10, -5, -1, 0, -1]
def a__ ( _UpperCamelCase : list[float] ):
__lowerCamelCase = []
__lowerCamelCase = len(_UpperCamelCase )
for i in range(_UpperCamelCase ):
__lowerCamelCase = -1
for j in range(i + 1 ,_UpperCamelCase ):
if arr[i] < arr[j]:
__lowerCamelCase = arr[j]
break
result.append(_UpperCamelCase )
return result
def a__ ( _UpperCamelCase : list[float] ):
__lowerCamelCase = []
for i, outer in enumerate(_UpperCamelCase ):
__lowerCamelCase = -1
for inner in arr[i + 1 :]:
if outer < inner:
__lowerCamelCase = inner
break
result.append(_UpperCamelCase )
return result
def a__ ( _UpperCamelCase : list[float] ):
__lowerCamelCase = len(_UpperCamelCase )
__lowerCamelCase = []
__lowerCamelCase = [-1] * arr_size
for index in reversed(range(_UpperCamelCase ) ):
if stack:
while stack[-1] <= arr[index]:
stack.pop()
if not stack:
break
if stack:
__lowerCamelCase = stack[-1]
stack.append(arr[index] )
return result
if __name__ == "__main__":
from doctest import testmod
from timeit import timeit
testmod()
print(next_greatest_element_slow(arr))
print(next_greatest_element_fast(arr))
print(next_greatest_element(arr))
a_ = (
"""from __main__ import arr, next_greatest_element_slow, """
"""next_greatest_element_fast, next_greatest_element"""
)
print(
"""next_greatest_element_slow():""",
timeit("""next_greatest_element_slow(arr)""", setup=setup),
)
print(
"""next_greatest_element_fast():""",
timeit("""next_greatest_element_fast(arr)""", setup=setup),
)
print(
""" next_greatest_element():""",
timeit("""next_greatest_element(arr)""", setup=setup),
)
| 330 |
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = n ** (1 / 3)
return (val * val * val) == n
if __name__ == "__main__":
print(perfect_cube(27))
print(perfect_cube(4))
| 330 | 1 |
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Features, Value
from .base import TaskTemplate
@dataclass(frozen=lowerCAmelCase__ )
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = field(default="""language-modeling""" , metadata={"""include_in_asdict_even_if_is_default""": True} )
lowerCAmelCase__ = Features({"""text""": Value("""string""" )} )
lowerCAmelCase__ = Features({} )
lowerCAmelCase__ = "text"
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return {self.text_column: "text"}
| 330 |
import argparse
import os
# New Code #
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
from accelerate.utils import find_executable_batch_size
########################################################################
# This is a fully working simple example to use Accelerate,
# specifically showcasing how to ensure out-of-memory errors never
# interrupt training, and builds off the `nlp_example.py` script.
#
# This example trains a Bert base model on GLUE MRPC
# in any of the following settings (with the same script):
# - single CPU or single GPU
# - multi GPUS (using PyTorch distributed mode)
# - (multi) TPUs
# - fp16 (mixed-precision) or fp32 (normal precision)
#
# New additions from the base script can be found quickly by
# looking for the # New Code # tags
#
# To run it in each of these various modes, follow the instructions
# in the readme for examples:
# https://github.com/huggingface/accelerate/tree/main/examples
#
########################################################################
a_ = 16
a_ = 32
def a__ ( _UpperCamelCase : Accelerator ,_UpperCamelCase : int = 16 ):
__lowerCamelCase = AutoTokenizer.from_pretrained('''bert-base-cased''' )
__lowerCamelCase = load_dataset('''glue''' ,'''mrpc''' )
def tokenize_function(_UpperCamelCase : Optional[Any] ):
# max_length=None => use the model max length (it's actually the default)
__lowerCamelCase = tokenizer(examples['''sentence1'''] ,examples['''sentence2'''] ,truncation=_UpperCamelCase ,max_length=_UpperCamelCase )
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(
_UpperCamelCase ,batched=_UpperCamelCase ,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(_UpperCamelCase : Any ):
# On TPU it's best to pad everything to the same length or training will be very slow.
__lowerCamelCase = 1_28 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(
_UpperCamelCase ,padding='''longest''' ,max_length=_UpperCamelCase ,pad_to_multiple_of=_UpperCamelCase ,return_tensors='''pt''' ,)
# Instantiate dataloaders.
__lowerCamelCase = DataLoader(
tokenized_datasets['''train'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
__lowerCamelCase = DataLoader(
tokenized_datasets['''validation'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
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
a_ = mocked_dataloaders # noqa: F811
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# For testing only
if os.environ.get('''TESTING_MOCKED_DATALOADERS''' ,_UpperCamelCase ) == "1":
__lowerCamelCase = 2
# Initialize accelerator
__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'''] )
__lowerCamelCase = evaluate.load('''glue''' ,'''mrpc''' )
# New Code #
# We now can define an inner training loop function. It should take a batch size as the only parameter,
# and build the dataloaders in there.
# It also gets our decorator
@find_executable_batch_size(starting_batch_size=_UpperCamelCase )
def inner_training_loop(_UpperCamelCase : Union[str, Any] ):
# And now just move everything below under this function
# We need to bring in the Accelerator object from earlier
nonlocal accelerator
# And reset all of its attributes that could hold onto any memory:
accelerator.free_memory()
# Then we can declare the model, optimizer, and everything else:
set_seed(_UpperCamelCase )
# 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=_UpperCamelCase )
# 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=_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = get_dataloaders(_UpperCamelCase ,_UpperCamelCase )
# Instantiate scheduler
__lowerCamelCase = get_linear_schedule_with_warmup(
optimizer=_UpperCamelCase ,num_warmup_steps=1_00 ,num_training_steps=(len(_UpperCamelCase ) * num_epochs) ,)
# Prepare everything
# There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the
# prepare method.
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Now we train the model
for epoch in range(_UpperCamelCase ):
model.train()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.loss
accelerator.backward(_UpperCamelCase )
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
model.eval()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
with torch.no_grad():
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.logits.argmax(dim=-1 )
__lowerCamelCase ,__lowerCamelCase = accelerator.gather_for_metrics((predictions, batch['''labels''']) )
metric.add_batch(
predictions=_UpperCamelCase ,references=_UpperCamelCase ,)
__lowerCamelCase = metric.compute()
# Use accelerator.print to print only on the main process.
accelerator.print(F"""epoch {epoch}:""" ,_UpperCamelCase )
# New Code #
# And call it at the end with no arguments
# Note: You could also refactor this outside of your training loop function
inner_training_loop()
def a__ ( ):
__lowerCamelCase = argparse.ArgumentParser(description='''Simple example of training script.''' )
parser.add_argument(
'''--mixed_precision''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,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.''' )
__lowerCamelCase = parser.parse_args()
__lowerCamelCase = {'''lr''': 2e-5, '''num_epochs''': 3, '''seed''': 42, '''batch_size''': 16}
training_function(_UpperCamelCase ,_UpperCamelCase )
if __name__ == "__main__":
main()
| 330 | 1 |
import inspect
import jax
import jax.lax as lax
import jax.numpy as jnp
from ..utils import add_start_docstrings
from ..utils.logging import get_logger
a_ = get_logger(__name__)
a_ = R"""
Args:
input_ids (`jnp.ndarray` of shape `(batch_size, sequence_length)`):
Indices of input sequence tokens in the vocabulary.
Indices can be obtained using [`PreTrainedTokenizer`]. See [`PreTrainedTokenizer.encode`] and
[`PreTrainedTokenizer.__call__`] for details.
[What are input IDs?](../glossary#input-ids)
scores (`jnp.ndarray` of shape `(batch_size, config.vocab_size)`):
Prediction scores of a language modeling head. These can be logits for each vocabulary when not using beam
search or log softmax for each vocabulary token when using beam search
kwargs (`Dict[str, Any]`, *optional*):
Additional logits processor specific kwargs.
Return:
`jnp.ndarray` of shape `(batch_size, config.vocab_size)`: The processed prediction scores.
"""
class __lowerCAmelCase :
@add_start_docstrings(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
raise NotImplementedError(
F"""{self.__class__} is an abstract class. Only classes inheriting this class can be called.""" )
class __lowerCAmelCase :
@add_start_docstrings(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
raise NotImplementedError(
F"""{self.__class__} is an abstract class. Only classes inheriting this class can be called.""" )
class __lowerCAmelCase ( lowerCAmelCase__ ):
@add_start_docstrings(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
for processor in self:
__lowerCamelCase = inspect.signature(processor.__call__ ).parameters
if len(__UpperCAmelCase ) > 3:
if not all(arg in kwargs for arg in list(function_args.keys() )[2:] ):
raise ValueError(
F"""Make sure that all the required parameters: {list(function_args.keys() )} for """
F"""{processor.__class__} are passed to the logits processor.""" )
__lowerCamelCase = processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , **__UpperCAmelCase )
else:
__lowerCamelCase = processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
return scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or not (temperature > 0):
raise ValueError(F"""`temperature` has to be a strictly positive float, but is {temperature}""" )
__lowerCamelCase = temperature
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = scores / self.temperature
return scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase = -float('''Inf''' ) , __UpperCAmelCase = 1 ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or (top_p < 0 or top_p > 1.0):
raise ValueError(F"""`top_p` has to be a float > 0 and < 1, but is {top_p}""" )
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or (min_tokens_to_keep < 1):
raise ValueError(F"""`min_tokens_to_keep` has to be a positive integer, but is {min_tokens_to_keep}""" )
__lowerCamelCase = top_p
__lowerCamelCase = filter_value
__lowerCamelCase = min_tokens_to_keep
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = lax.top_k(__UpperCAmelCase , scores.shape[-1] )
__lowerCamelCase = jnp.full_like(__UpperCAmelCase , self.filter_value )
__lowerCamelCase = jax.nn.softmax(__UpperCAmelCase , axis=-1 ).cumsum(axis=-1 )
__lowerCamelCase = cumulative_probs < self.top_p
# include the token that is higher than top_p as well
__lowerCamelCase = jnp.roll(__UpperCAmelCase , 1 )
score_mask |= score_mask.at[:, 0].set(__UpperCAmelCase )
# min tokens to keep
__lowerCamelCase = score_mask.at[:, : self.min_tokens_to_keep].set(__UpperCAmelCase )
__lowerCamelCase = jnp.where(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = jax.lax.sort_key_val(__UpperCAmelCase , __UpperCAmelCase )[-1]
return next_scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase = -float('''Inf''' ) , __UpperCAmelCase = 1 ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or top_k <= 0:
raise ValueError(F"""`top_k` has to be a strictly positive integer, but is {top_k}""" )
__lowerCamelCase = max(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = filter_value
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = scores.shape
__lowerCamelCase = jnp.full(batch_size * vocab_size , self.filter_value )
__lowerCamelCase = min(self.top_k , scores.shape[-1] ) # Safety check
__lowerCamelCase ,__lowerCamelCase = lax.top_k(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = jnp.broadcast_to((jnp.arange(__UpperCAmelCase ) * vocab_size)[:, None] , (batch_size, topk) ).flatten()
__lowerCamelCase = topk_scores.flatten()
__lowerCamelCase = topk_indices.flatten() + shift
__lowerCamelCase = next_scores_flat.at[topk_indices_flat].set(__UpperCAmelCase )
__lowerCamelCase = next_scores_flat.reshape(__UpperCAmelCase , __UpperCAmelCase )
return next_scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = bos_token_id
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = jnp.full(scores.shape , -float('''inf''' ) )
__lowerCamelCase = 1 - jnp.bool_(cur_len - 1 )
__lowerCamelCase = jnp.where(__UpperCAmelCase , new_scores.at[:, self.bos_token_id].set(0 ) , __UpperCAmelCase )
return scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = max_length
__lowerCamelCase = eos_token_id
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = jnp.full(scores.shape , -float('''inf''' ) )
__lowerCamelCase = 1 - jnp.bool_(cur_len - self.max_length + 1 )
__lowerCamelCase = jnp.where(__UpperCAmelCase , new_scores.at[:, self.eos_token_id].set(0 ) , __UpperCAmelCase )
return scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or min_length < 0:
raise ValueError(F"""`min_length` has to be a positive integer, but is {min_length}""" )
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or eos_token_id < 0:
raise ValueError(F"""`eos_token_id` has to be a positive integer, but is {eos_token_id}""" )
__lowerCamelCase = min_length
__lowerCamelCase = eos_token_id
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# create boolean flag to decide if min length penalty should be applied
__lowerCamelCase = 1 - jnp.clip(cur_len - self.min_length , 0 , 1 )
__lowerCamelCase = jnp.where(__UpperCAmelCase , scores.at[:, self.eos_token_id].set(-float('''inf''' ) ) , __UpperCAmelCase )
return scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = begin_index
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = 1 - jnp.bool_(cur_len - self.begin_index )
__lowerCamelCase = jnp.where(__UpperCAmelCase , scores.at[:, self.begin_suppress_tokens].set(-float('''inf''' ) ) , __UpperCAmelCase )
return scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = list(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = scores.at[..., self.suppress_tokens].set(-float('''inf''' ) )
return scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = dict(__UpperCAmelCase )
# Converts the dictionary of format {index: token} containing the tokens to be forced to an array, where the
# index of the array corresponds to the index of the token to be forced, for XLA compatibility.
# Indexes without forced tokens will have a negative value.
__lowerCamelCase = jnp.ones((max(force_token_map.keys() ) + 1) , dtype=jnp.intaa ) * -1
for index, token in force_token_map.items():
if token is not None:
__lowerCamelCase = force_token_array.at[index].set(__UpperCAmelCase )
__lowerCamelCase = jnp.intaa(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
def _force_token(__UpperCAmelCase ):
__lowerCamelCase = scores.shape[0]
__lowerCamelCase = self.force_token_array[generation_idx]
__lowerCamelCase = jnp.ones_like(__UpperCAmelCase , dtype=scores.dtype ) * -float('''inf''' )
__lowerCamelCase = jnp.zeros((batch_size, 1) , dtype=scores.dtype )
__lowerCamelCase = lax.dynamic_update_slice(__UpperCAmelCase , __UpperCAmelCase , (0, current_token) )
return new_scores
__lowerCamelCase = lax.cond(
cur_len >= self.force_token_array.shape[0] , lambda: scores , lambda: lax.cond(
self.force_token_array[cur_len] >= 0 , lambda: _force_token(__UpperCAmelCase ) , lambda: scores , ) , )
return scores
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = generate_config.eos_token_id
__lowerCamelCase = generate_config.no_timestamps_token_id
__lowerCamelCase = generate_config.no_timestamps_token_id + 1
__lowerCamelCase = decoder_input_length + 1
if generate_config.is_multilingual:
# room for language token and task token
self.begin_index += 2
if hasattr(__UpperCAmelCase , '''max_initial_timestamp_index''' ):
__lowerCamelCase = generate_config.max_initial_timestamp_index
else:
__lowerCamelCase = model_config.vocab_size
if self.max_initial_timestamp_index is None:
__lowerCamelCase = model_config.vocab_size
def __call__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# suppress <|notimestamps|> which is handled by without_timestamps
__lowerCamelCase = scores.at[:, self.no_timestamps_token_id].set(-float('''inf''' ) )
def handle_pairs(__UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = jnp.where((cur_len - self.begin_index) >= 1 , __UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = jnp.where(
input_ids_k[cur_len - 1] >= self.timestamp_begin , True and last_was_timestamp , __UpperCAmelCase , )
__lowerCamelCase = jnp.where((cur_len - self.begin_index) < 2 , __UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = jnp.where(
input_ids_k[cur_len - 2] >= self.timestamp_begin , __UpperCAmelCase , __UpperCAmelCase , )
return jnp.where(
__UpperCAmelCase , jnp.where(
penultimate_was_timestamp > 0 , scores_k.at[self.timestamp_begin :].set(-float('''inf''' ) ) , scores_k.at[: self.eos_token_id].set(-float('''inf''' ) ) , ) , __UpperCAmelCase , )
__lowerCamelCase = jax.vmap(__UpperCAmelCase )(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = jnp.where(cur_len == self.begin_index , __UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = jnp.where(
self.max_initial_timestamp_index is not None , True and apply_max_initial_timestamp , __UpperCAmelCase , )
__lowerCamelCase = self.timestamp_begin + self.max_initial_timestamp_index
__lowerCamelCase = jnp.where(
__UpperCAmelCase , scores.at[:, last_allowed + 1 :].set(-float('''inf''' ) ) , __UpperCAmelCase , )
# if sum of probability over timestamps is above any other token, sample timestamp
__lowerCamelCase = jax.nn.log_softmax(__UpperCAmelCase , axis=-1 )
def handle_cumulative_probs(__UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = jax.nn.logsumexp(logprobs_k[self.timestamp_begin :] , axis=-1 )
__lowerCamelCase = jnp.max(logprobs_k[: self.timestamp_begin] )
return jnp.where(
timestamp_logprob > max_text_token_logprob , scores_k.at[: self.timestamp_begin].set(-float('''inf''' ) ) , __UpperCAmelCase , )
__lowerCamelCase = jax.vmap(__UpperCAmelCase )(__UpperCAmelCase , __UpperCAmelCase )
return scores
| 330 |
import logging
import os
import threading
import time
try:
import warnings
except ImportError:
a_ = None
try:
import msvcrt
except ImportError:
a_ = None
try:
import fcntl
except ImportError:
a_ = None
# Backward compatibility
# ------------------------------------------------
try:
TimeoutError
except NameError:
a_ = OSError
# Data
# ------------------------------------------------
a_ = [
"""Timeout""",
"""BaseFileLock""",
"""WindowsFileLock""",
"""UnixFileLock""",
"""SoftFileLock""",
"""FileLock""",
]
a_ = """3.0.12"""
a_ = None
def a__ ( ):
global _logger
__lowerCamelCase = _logger or logging.getLogger(__name__ )
return _logger
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = lock_file
return None
def __str__( self ):
'''simple docstring'''
__lowerCamelCase = F"""The file lock '{self.lock_file}' could not be acquired."""
return temp
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = lock
return None
def __enter__( self ):
'''simple docstring'''
return self.lock
def __exit__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
self.lock.release()
return None
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
__lowerCamelCase = max_filename_length if max_filename_length is not None else 255
# Hash the filename if it's too long
__lowerCamelCase = self.hash_filename_if_too_long(__UpperCAmelCase , __UpperCAmelCase )
# The path to the lock file.
__lowerCamelCase = lock_file
# The file descriptor for the *_lock_file* as it is returned by the
# os.open() function.
# This file lock is only NOT None, if the object currently holds the
# lock.
__lowerCamelCase = None
# The default timeout value.
__lowerCamelCase = timeout
# We use this lock primarily for the lock counter.
__lowerCamelCase = threading.Lock()
# The lock counter is used for implementing the nested locking
# mechanism. Whenever the lock is acquired, the counter is increased and
# the lock is only released, when this value is 0 again.
__lowerCamelCase = 0
return None
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._lock_file
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._timeout
@timeout.setter
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = float(__UpperCAmelCase )
return None
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError()
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError()
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return self._lock_file_fd is not None
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=0.05 ):
'''simple docstring'''
# Use the default timeout, if no timeout is provided.
if timeout is None:
__lowerCamelCase = self.timeout
# Increment the number right at the beginning.
# We can still undo it, if something fails.
with self._thread_lock:
self._lock_counter += 1
__lowerCamelCase = id(self )
__lowerCamelCase = self._lock_file
__lowerCamelCase = time.time()
try:
while True:
with self._thread_lock:
if not self.is_locked:
logger().debug(F"""Attempting to acquire lock {lock_id} on {lock_filename}""" )
self._acquire()
if self.is_locked:
logger().debug(F"""Lock {lock_id} acquired on {lock_filename}""" )
break
elif timeout >= 0 and time.time() - start_time > timeout:
logger().debug(F"""Timeout on acquiring lock {lock_id} on {lock_filename}""" )
raise Timeout(self._lock_file )
else:
logger().debug(
F"""Lock {lock_id} not acquired on {lock_filename}, waiting {poll_intervall} seconds ...""" )
time.sleep(__UpperCAmelCase )
except: # noqa
# Something did go wrong, so decrement the counter.
with self._thread_lock:
__lowerCamelCase = max(0 , self._lock_counter - 1 )
raise
return _Acquire_ReturnProxy(lock=self )
def lowerCamelCase ( self , __UpperCAmelCase=False ):
'''simple docstring'''
with self._thread_lock:
if self.is_locked:
self._lock_counter -= 1
if self._lock_counter == 0 or force:
__lowerCamelCase = id(self )
__lowerCamelCase = self._lock_file
logger().debug(F"""Attempting to release lock {lock_id} on {lock_filename}""" )
self._release()
__lowerCamelCase = 0
logger().debug(F"""Lock {lock_id} released on {lock_filename}""" )
return None
def __enter__( self ):
'''simple docstring'''
self.acquire()
return self
def __exit__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
self.release()
return None
def __del__( self ):
'''simple docstring'''
self.release(force=__UpperCAmelCase )
return None
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = os.path.basename(__UpperCAmelCase )
if len(__UpperCAmelCase ) > max_length and max_length > 0:
__lowerCamelCase = os.path.dirname(__UpperCAmelCase )
__lowerCamelCase = str(hash(__UpperCAmelCase ) )
__lowerCamelCase = filename[: max_length - len(__UpperCAmelCase ) - 8] + '''...''' + hashed_filename + '''.lock'''
return os.path.join(__UpperCAmelCase , __UpperCAmelCase )
else:
return path
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
from .file_utils import relative_to_absolute_path
super().__init__(__UpperCAmelCase , timeout=__UpperCAmelCase , max_filename_length=__UpperCAmelCase )
__lowerCamelCase = '''\\\\?\\''' + relative_to_absolute_path(self.lock_file )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_RDWR | os.O_CREAT | os.O_TRUNC
try:
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
except OSError:
pass
else:
try:
msvcrt.locking(__UpperCAmelCase , msvcrt.LK_NBLCK , 1 )
except OSError:
os.close(__UpperCAmelCase )
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self._lock_file_fd
__lowerCamelCase = None
msvcrt.locking(__UpperCAmelCase , msvcrt.LK_UNLCK , 1 )
os.close(__UpperCAmelCase )
try:
os.remove(self._lock_file )
# Probably another instance of the application
# that acquired the file lock.
except OSError:
pass
return None
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=-1 , __UpperCAmelCase=None ):
'''simple docstring'''
__lowerCamelCase = os.statvfs(os.path.dirname(__UpperCAmelCase ) ).f_namemax
super().__init__(__UpperCAmelCase , timeout=__UpperCAmelCase , max_filename_length=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_RDWR | os.O_CREAT | os.O_TRUNC
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
try:
fcntl.flock(__UpperCAmelCase , fcntl.LOCK_EX | fcntl.LOCK_NB )
except OSError:
os.close(__UpperCAmelCase )
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
# Do not remove the lockfile:
#
# https://github.com/benediktschmitt/py-filelock/issues/31
# https://stackoverflow.com/questions/17708885/flock-removing-locked-file-without-race-condition
__lowerCamelCase = self._lock_file_fd
__lowerCamelCase = None
fcntl.flock(__UpperCAmelCase , fcntl.LOCK_UN )
os.close(__UpperCAmelCase )
return None
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = os.O_WRONLY | os.O_CREAT | os.O_EXCL | os.O_TRUNC
try:
__lowerCamelCase = os.open(self._lock_file , __UpperCAmelCase )
except OSError:
pass
else:
__lowerCamelCase = fd
return None
def lowerCamelCase ( self ):
'''simple docstring'''
os.close(self._lock_file_fd )
__lowerCamelCase = None
try:
os.remove(self._lock_file )
# The file is already deleted and that's what we want.
except OSError:
pass
return None
a_ = None
if msvcrt:
a_ = WindowsFileLock
elif fcntl:
a_ = UnixFileLock
else:
a_ = SoftFileLock
if warnings is not None:
warnings.warn("""only soft file lock is available""")
| 330 | 1 |
import qiskit
def a__ ( _UpperCamelCase : int = 2 ):
__lowerCamelCase = qubits
# Using Aer's simulator
__lowerCamelCase = qiskit.Aer.get_backend('''aer_simulator''' )
# Creating a Quantum Circuit acting on the q register
__lowerCamelCase = qiskit.QuantumCircuit(_UpperCamelCase ,_UpperCamelCase )
# Adding a H gate on qubit 0 (now q0 in superposition)
circuit.h(0 )
for i in range(1 ,_UpperCamelCase ):
# Adding CX (CNOT) gate
circuit.cx(i - 1 ,_UpperCamelCase )
# Mapping the quantum measurement to the classical bits
circuit.measure(list(range(_UpperCamelCase ) ) ,list(range(_UpperCamelCase ) ) )
# Now measuring any one qubit would affect other qubits to collapse
# their super position and have same state as the measured one.
# Executing the circuit on the simulator
__lowerCamelCase = qiskit.execute(_UpperCamelCase ,_UpperCamelCase ,shots=10_00 )
return job.result().get_counts(_UpperCamelCase )
if __name__ == "__main__":
print(f"Total count for various states are: {quantum_entanglement(3)}")
| 330 |
import copy
import inspect
import unittest
import numpy as np
from huggingface_hub import hf_hub_download
from transformers import TimesformerConfig
from transformers.models.auto import get_values
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 (
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING,
TimesformerForVideoClassification,
TimesformerModel,
)
from transformers.models.timesformer.modeling_timesformer import TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from transformers import VideoMAEImageProcessor
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=13 , __UpperCAmelCase=10 , __UpperCAmelCase=3 , __UpperCAmelCase=2 , __UpperCAmelCase=2 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=32 , __UpperCAmelCase=5 , __UpperCAmelCase=4 , __UpperCAmelCase=37 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=10 , __UpperCAmelCase=0.02 , __UpperCAmelCase="divided_space_time" , __UpperCAmelCase=None , ):
'''simple docstring'''
__lowerCamelCase = parent
__lowerCamelCase = batch_size
__lowerCamelCase = image_size
__lowerCamelCase = num_channels
__lowerCamelCase = patch_size
__lowerCamelCase = num_frames
__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 = attention_type
__lowerCamelCase = initializer_range
__lowerCamelCase = scope
__lowerCamelCase = num_labels
# in TimeSformer, the number of spatial tokens equals num_frames * num_patches per frame + 1 CLS token
__lowerCamelCase = (image_size // patch_size) ** 2
__lowerCamelCase = (num_frames) * self.num_patches_per_frame + 1
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = floats_tensor(
[self.batch_size, self.num_frames, 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 lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , num_frames=self.num_frames , 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 , initializer_range=self.initializer_range , attention_type=self.attention_type , )
__lowerCamelCase = self.num_labels
return config
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
# verify the logits shape
__lowerCamelCase = torch.Size((self.batch_size, self.num_labels) )
self.parent.assertEqual(result.logits.shape , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = config_and_inputs
__lowerCamelCase = {'''pixel_values''': pixel_values}
return config, inputs_dict
@require_torch
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (TimesformerModel, TimesformerForVideoClassification) if is_torch_available() else ()
lowerCAmelCase__ = (
{"""feature-extraction""": TimesformerModel, """video-classification""": TimesformerForVideoClassification}
if is_torch_available()
else {}
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerModelTester(self )
__lowerCamelCase = ConfigTester(
self , config_class=__UpperCAmelCase , has_text_modality=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = copy.deepcopy(__UpperCAmelCase )
if return_labels:
if model_class in get_values(__UpperCAmelCase ):
__lowerCamelCase = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=__UpperCAmelCase )
return inputs_dict
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
@unittest.skip(reason='''TimeSformer does not use inputs_embeds''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
__lowerCamelCase = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(__UpperCAmelCase , nn.Linear ) )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
__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] , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_video_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = TimesformerModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.has_attentions:
pass
else:
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCamelCase = True
for model_class in self.all_model_classes:
__lowerCamelCase = self.model_tester.seq_length
__lowerCamelCase = self.model_tester.num_frames
__lowerCamelCase = True
__lowerCamelCase = False
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
__lowerCamelCase = len(__UpperCAmelCase )
# Check attention is always last and order is fine
__lowerCamelCase = True
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
self.assertEqual(out_len + 1 , len(__UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
def lowerCamelCase ( self ):
'''simple docstring'''
def check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.hidden_states
__lowerCamelCase = self.model_tester.num_hidden_layers + 1
self.assertEqual(len(__UpperCAmelCase ) , __UpperCAmelCase )
__lowerCamelCase = self.model_tester.seq_length
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [seq_length, self.model_tester.hidden_size] , )
__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(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCamelCase = True
check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def a__ ( ):
__lowerCamelCase = hf_hub_download(
repo_id='''hf-internal-testing/spaghetti-video''' ,filename='''eating_spaghetti.npy''' ,repo_type='''dataset''' )
__lowerCamelCase = np.load(_UpperCamelCase )
return list(_UpperCamelCase )
@require_torch
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
@cached_property
def lowerCamelCase ( self ):
'''simple docstring'''
# logits were tested with a different mean and std, so we use the same here
return (
VideoMAEImageProcessor(image_mean=[0.5, 0.5, 0.5] , image_std=[0.5, 0.5, 0.5] )
if is_vision_available()
else None
)
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification.from_pretrained('''facebook/timesformer-base-finetuned-k400''' ).to(
__UpperCAmelCase )
__lowerCamelCase = self.default_image_processor
__lowerCamelCase = prepare_video()
__lowerCamelCase = image_processor(video[:8] , return_tensors='''pt''' ).to(__UpperCAmelCase )
# forward pass
with torch.no_grad():
__lowerCamelCase = model(**__UpperCAmelCase )
# verify the logits
__lowerCamelCase = torch.Size((1, 400) )
self.assertEqual(outputs.logits.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor([-0.3_016, -0.7_713, -0.4_205] ).to(__UpperCAmelCase )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
import argparse
import re
import requests
import torch
# git clone https://github.com/salesforce/BLIP.git
from models.blip import blip_decoder
from models.blip_itm import blip_itm
from models.blip_vqa import blip_vqa
from PIL import Image
from torchvision import transforms
from torchvision.transforms.functional import InterpolationMode
from transformers import (
BertTokenizer,
BlipConfig,
BlipForConditionalGeneration,
BlipForImageTextRetrieval,
BlipForQuestionAnswering,
)
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : List[Any] ):
__lowerCamelCase = '''https://storage.googleapis.com/sfr-vision-language-research/BLIP/demo.jpg'''
__lowerCamelCase = Image.open(requests.get(_UpperCamelCase ,stream=_UpperCamelCase ).raw ).convert('''RGB''' )
__lowerCamelCase = transforms.Compose(
[
transforms.Resize((image_size, image_size) ,interpolation=InterpolationMode.BICUBIC ),
transforms.ToTensor(),
transforms.Normalize((0.48_145_466, 0.4_578_275, 0.40_821_073) ,(0.26_862_954, 0.26_130_258, 0.27_577_711) ),
] )
__lowerCamelCase = transform(_UpperCamelCase ).unsqueeze(0 ).to(_UpperCamelCase )
return image
def a__ ( _UpperCamelCase : List[Any] ):
if "visual_encoder" in key:
__lowerCamelCase = re.sub('''visual_encoder*''' ,'''vision_model.encoder''' ,_UpperCamelCase )
if "blocks" in key:
__lowerCamelCase = re.sub(R'''blocks''' ,'''layers''' ,_UpperCamelCase )
if "attn" in key:
__lowerCamelCase = re.sub(R'''attn''' ,'''self_attn''' ,_UpperCamelCase )
if "norm1" in key:
__lowerCamelCase = re.sub(R'''norm1''' ,'''layer_norm1''' ,_UpperCamelCase )
if "norm2" in key:
__lowerCamelCase = re.sub(R'''norm2''' ,'''layer_norm2''' ,_UpperCamelCase )
if "encoder.norm" in key:
__lowerCamelCase = re.sub(R'''encoder.norm''' ,'''post_layernorm''' ,_UpperCamelCase )
if "encoder.patch_embed.proj" in key:
__lowerCamelCase = re.sub(R'''encoder.patch_embed.proj''' ,'''embeddings.patch_embedding''' ,_UpperCamelCase )
if "encoder.pos_embed" in key:
__lowerCamelCase = re.sub(R'''encoder.pos_embed''' ,'''embeddings.position_embedding''' ,_UpperCamelCase )
if "encoder.cls_token" in key:
__lowerCamelCase = re.sub(R'''encoder.cls_token''' ,'''embeddings.class_embedding''' ,_UpperCamelCase )
if "self_attn" in key:
__lowerCamelCase = re.sub(R'''self_attn.proj''' ,'''self_attn.projection''' ,_UpperCamelCase )
return key
@torch.no_grad()
def a__ ( _UpperCamelCase : int ,_UpperCamelCase : Union[str, Any]=None ):
if config_path is not None:
__lowerCamelCase = BlipConfig.from_pretrained(_UpperCamelCase )
else:
__lowerCamelCase = BlipConfig(projection_dim=5_12 ,text_config={} ,vision_config={} )
__lowerCamelCase = BlipForConditionalGeneration(_UpperCamelCase ).eval()
__lowerCamelCase = '''https://storage.googleapis.com/sfr-vision-language-research/BLIP/models/model_base_capfilt_large.pth'''
__lowerCamelCase = blip_decoder(pretrained=_UpperCamelCase ,image_size=3_84 ,vit='''base''' )
__lowerCamelCase = pt_model.eval()
__lowerCamelCase = pt_model.state_dict()
for key in modified_state_dict.copy():
__lowerCamelCase = modified_state_dict.pop(_UpperCamelCase )
__lowerCamelCase = rename_key(_UpperCamelCase )
__lowerCamelCase = value
hf_model.load_state_dict(_UpperCamelCase )
__lowerCamelCase = 3_84
__lowerCamelCase = load_demo_image(image_size=_UpperCamelCase ,device='''cpu''' )
__lowerCamelCase = BertTokenizer.from_pretrained('''bert-base-uncased''' )
__lowerCamelCase = tokenizer(['''a picture of'''] ).input_ids
__lowerCamelCase = hf_model.generate(_UpperCamelCase ,_UpperCamelCase )
assert out[0].tolist() == [3_05_22, 10_37, 38_61, 19_97, 10_37, 24_50, 35_64, 20_06, 19_96, 35_09, 20_07, 20_14, 38_99, 1_02]
__lowerCamelCase = hf_model.generate(_UpperCamelCase )
assert out[0].tolist() == [3_05_22, 10_37, 24_50, 35_64, 20_06, 19_96, 35_09, 20_07, 20_14, 38_99, 1_02]
if pytorch_dump_folder_path is not None:
hf_model.save_pretrained(_UpperCamelCase )
# model_url = 'https://storage.googleapis.com/sfr-vision-language-research/BLIP/models/model_vqa.pth'
__lowerCamelCase = (
'''https://storage.googleapis.com/sfr-vision-language-research/BLIP/models/model_base_vqa_capfilt_large.pth'''
)
__lowerCamelCase = blip_vqa(pretrained=_UpperCamelCase ,image_size=_UpperCamelCase ,vit='''base''' )
vqa_model.eval()
__lowerCamelCase = vqa_model.state_dict()
for key in modified_state_dict.copy():
__lowerCamelCase = modified_state_dict.pop(_UpperCamelCase )
__lowerCamelCase = rename_key(_UpperCamelCase )
__lowerCamelCase = value
__lowerCamelCase = BlipForQuestionAnswering(_UpperCamelCase )
hf_vqa_model.load_state_dict(_UpperCamelCase )
__lowerCamelCase = ['''How many dogs are in this image?''']
__lowerCamelCase = tokenizer(_UpperCamelCase ,return_tensors='''pt''' ).input_ids
__lowerCamelCase = hf_vqa_model.generate(_UpperCamelCase ,_UpperCamelCase )
print(tokenizer.decode(answer[0] ) )
assert tokenizer.decode(answer[0] ) == "[UNK] 1 [SEP]"
if pytorch_dump_folder_path is not None:
hf_vqa_model.save_pretrained(pytorch_dump_folder_path + '''_vqa''' )
__lowerCamelCase = '''https://storage.googleapis.com/sfr-vision-language-research/BLIP/models/model_base_retrieval_coco.pth'''
__lowerCamelCase = blip_itm(pretrained=_UpperCamelCase ,image_size=_UpperCamelCase ,vit='''base''' )
itm_model.eval()
__lowerCamelCase = itm_model.state_dict()
for key in modified_state_dict.copy():
__lowerCamelCase = modified_state_dict.pop(_UpperCamelCase )
__lowerCamelCase = rename_key(_UpperCamelCase )
__lowerCamelCase = value
__lowerCamelCase = BlipForImageTextRetrieval(_UpperCamelCase )
__lowerCamelCase = ['''A picture of a woman with a dog sitting in a beach''']
__lowerCamelCase = tokenizer(
_UpperCamelCase ,return_tensors='''pt''' ,padding='''max_length''' ,truncation=_UpperCamelCase ,max_length=35 ,).input_ids
hf_itm_model.load_state_dict(_UpperCamelCase )
hf_itm_model.eval()
__lowerCamelCase = hf_itm_model(_UpperCamelCase ,_UpperCamelCase ,use_itm_head=_UpperCamelCase )
__lowerCamelCase = hf_itm_model(_UpperCamelCase ,_UpperCamelCase ,use_itm_head=_UpperCamelCase )
assert out[0].item() == 0.2_110_687_494_277_954
assert torch.nn.functional.softmax(out_itm[0] ,dim=1 )[:, 1].item() == 0.45_698_845_386_505_127
if pytorch_dump_folder_path is not None:
hf_itm_model.save_pretrained(pytorch_dump_folder_path + '''_itm''' )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument("""--config_path""", default=None, type=str, help="""Path to hf config.json of model to convert""")
a_ = parser.parse_args()
convert_blip_checkpoint(args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path)
| 330 |
def a__ ( _UpperCamelCase : int ):
if not isinstance(_UpperCamelCase ,_UpperCamelCase ):
__lowerCamelCase = F"""Input value of [number={number}] must be an integer"""
raise TypeError(_UpperCamelCase )
if number < 0:
return False
__lowerCamelCase = number * number
while number > 0:
if number % 10 != number_square % 10:
return False
number //= 10
number_square //= 10
return True
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 | 1 |
from copy import deepcopy
import torch
import torch.nn.functional as F
from torch.optim import AdamW
from torch.optim.lr_scheduler import LambdaLR
from torch.utils.data import DataLoader
from accelerate.accelerator import Accelerator
from accelerate.state import GradientState
from accelerate.test_utils import RegressionDataset, RegressionModel
from accelerate.utils import DistributedType, is_torch_version, set_seed
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : Optional[int] ,_UpperCamelCase : Optional[int] ,_UpperCamelCase : Union[str, Any] ):
for param, grad_param in zip(model_a.parameters() ,model_b.parameters() ):
if not param.requires_grad:
continue
if not did_step:
# Grads should not be in sync
assert (
torch.allclose(param.grad ,grad_param.grad ) is False
), F"""Gradients in sync when they should not be at iteration {iteration}:\nmodel_a grad ({param.grad}) == model_b grad ({grad_param.grad})"""
else:
# Grads should be in sync
assert (
torch.allclose(param.grad ,grad_param.grad ) is True
), F"""Gradients not in sync when they should be at iteration {iteration}:\nmodel_a grad ({param.grad}) != model_b grad ({grad_param.grad})"""
def a__ ( _UpperCamelCase : Tuple ,_UpperCamelCase : List[str] ,_UpperCamelCase : int ,_UpperCamelCase : List[Any] ,_UpperCamelCase : Dict=True ):
model.train()
__lowerCamelCase = model(_UpperCamelCase )
__lowerCamelCase = F.mse_loss(_UpperCamelCase ,target.to(output.device ) )
if not do_backward:
loss /= accelerator.gradient_accumulation_steps
loss.backward()
else:
accelerator.backward(_UpperCamelCase )
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : List[Any]=False ):
set_seed(42 )
__lowerCamelCase = RegressionModel()
__lowerCamelCase = deepcopy(_UpperCamelCase )
__lowerCamelCase = RegressionDataset(length=80 )
__lowerCamelCase = DataLoader(_UpperCamelCase ,batch_size=16 )
model.to(accelerator.device )
if sched:
__lowerCamelCase = AdamW(params=model.parameters() ,lr=1e-3 )
__lowerCamelCase = AdamW(params=ddp_model.parameters() ,lr=1e-3 )
__lowerCamelCase = LambdaLR(_UpperCamelCase ,lr_lambda=lambda _UpperCamelCase : epoch**0.65 )
__lowerCamelCase = LambdaLR(_UpperCamelCase ,lr_lambda=lambda _UpperCamelCase : epoch**0.65 )
# Make a copy of `model`
if sched:
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
else:
__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(_UpperCamelCase ,_UpperCamelCase )
if sched:
return (model, opt, sched, dataloader, ddp_model, ddp_opt, ddp_sched)
return model, ddp_model, dataloader
def a__ ( _UpperCamelCase : Union[str, Any] ):
# Test when on a single CPU or GPU that the context manager does nothing
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = get_training_setup(_UpperCamelCase )
# Use a single batch
__lowerCamelCase ,__lowerCamelCase = next(iter(_UpperCamelCase ) ).values()
for iteration in range(3 ):
# Gather the distributed inputs and targs for the base model
__lowerCamelCase ,__lowerCamelCase = accelerator.gather((ddp_input, ddp_target) )
__lowerCamelCase ,__lowerCamelCase = input.to(accelerator.device ), target.to(accelerator.device )
# Perform our initial ground truth step in non "DDP"
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Do "gradient accumulation" (noop)
if iteration % 2 == 0:
# Accumulate grads locally
with accelerator.no_sync(_UpperCamelCase ):
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
else:
# Sync grads
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Since `no_sync` is a noop, `ddp_model` and `model` grads should always be in sync
check_model_parameters(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
for param, ddp_param in zip(model.parameters() ,ddp_model.parameters() ):
if not param.requires_grad:
continue
assert torch.allclose(
param.grad ,ddp_param.grad ), F"""Gradients not in sync when they should be:\nModel grad ({param.grad}) != DDP grad ({ddp_param.grad})"""
# Shuffle ddp_input on each iteration
torch.manual_seed(13_37 + iteration )
__lowerCamelCase = ddp_input[torch.randperm(len(_UpperCamelCase ) )]
def a__ ( _UpperCamelCase : Any ):
# Test on distributed setup that context manager behaves properly
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = get_training_setup(_UpperCamelCase )
# Use a single batch
__lowerCamelCase ,__lowerCamelCase = next(iter(_UpperCamelCase ) ).values()
for iteration in range(3 ):
# Gather the distributed inputs and targs for the base model
__lowerCamelCase ,__lowerCamelCase = accelerator.gather((ddp_input, ddp_target) )
__lowerCamelCase ,__lowerCamelCase = input.to(accelerator.device ), target.to(accelerator.device )
# Perform our initial ground truth step in non "DDP"
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Do "gradient accumulation" (noop)
if iteration % 2 == 0:
# Accumulate grads locally
with accelerator.no_sync(_UpperCamelCase ):
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
else:
# Sync grads
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# DDP model and model should only be in sync when not (iteration % 2 == 0)
for param, ddp_param in zip(model.parameters() ,ddp_model.parameters() ):
if not param.requires_grad:
continue
if iteration % 2 == 0:
# Grads should not be in sync
assert (
torch.allclose(param.grad ,ddp_param.grad ) is False
), F"""Gradients in sync when they should not be:\nModel grad ({param.grad}) == DDP grad ({ddp_param.grad})"""
else:
# Grads should be in sync
assert (
torch.allclose(param.grad ,ddp_param.grad ) is True
), F"""Gradients not in sync when they should be:\nModel grad ({param.grad}) != DDP grad ({ddp_param.grad})"""
# Shuffle ddp_input on each iteration
torch.manual_seed(13_37 + iteration )
__lowerCamelCase = ddp_input[torch.randperm(len(_UpperCamelCase ) )]
def a__ ( _UpperCamelCase : int=False ,_UpperCamelCase : str=False ):
__lowerCamelCase = Accelerator(
split_batches=_UpperCamelCase ,dispatch_batches=_UpperCamelCase ,gradient_accumulation_steps=2 )
# Test that context manager behaves properly
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = get_training_setup(_UpperCamelCase )
for iteration, batch in enumerate(_UpperCamelCase ):
__lowerCamelCase ,__lowerCamelCase = batch.values()
# Gather the distributed inputs and targs for the base model
__lowerCamelCase ,__lowerCamelCase = accelerator.gather((ddp_input, ddp_target) )
__lowerCamelCase ,__lowerCamelCase = input.to(accelerator.device ), target.to(accelerator.device )
# Perform our initial ground truth step in non "DDP"
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Do "gradient accumulation" (noop)
with accelerator.accumulate(_UpperCamelCase ):
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# DDP model and model should only be in sync when not (iteration % 2 == 0)
for param, ddp_param in zip(model.parameters() ,ddp_model.parameters() ):
if not param.requires_grad:
continue
if ((iteration + 1) % 2 == 0) or (iteration == len(_UpperCamelCase ) - 1):
# Grads should be in sync
assert (
torch.allclose(param.grad ,ddp_param.grad ) is True
), F"""Gradients not in sync when they should be at iteration {iteration}:\nModel grad ({param.grad}) != DDP grad ({ddp_param.grad})"""
else:
# Grads should not be in sync
assert (
torch.allclose(param.grad ,ddp_param.grad ) is False
), F"""Gradients in sync when they should not be at iteration {iteration}:\nModel grad ({param.grad}) == DDP grad ({ddp_param.grad})"""
# Shuffle ddp_input on each iteration
torch.manual_seed(13_37 + iteration )
__lowerCamelCase = ddp_input[torch.randperm(len(_UpperCamelCase ) )]
GradientState._reset_state()
def a__ ( _UpperCamelCase : Any=False ,_UpperCamelCase : Optional[int]=False ):
__lowerCamelCase = Accelerator(
split_batches=_UpperCamelCase ,dispatch_batches=_UpperCamelCase ,gradient_accumulation_steps=2 )
# Test that context manager behaves properly
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = get_training_setup(_UpperCamelCase ,_UpperCamelCase )
for iteration, batch in enumerate(_UpperCamelCase ):
__lowerCamelCase ,__lowerCamelCase = batch.values()
# Gather the distributed inputs and targs for the base model
__lowerCamelCase ,__lowerCamelCase = accelerator.gather((ddp_input, ddp_target) )
__lowerCamelCase ,__lowerCamelCase = input.to(accelerator.device ), target.to(accelerator.device )
# Perform our initial ground truth step in non "DDP"
model.train()
ddp_model.train()
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
opt.step()
if ((iteration + 1) % 2 == 0) or ((iteration + 1) == len(_UpperCamelCase )):
if split_batches:
sched.step()
else:
for _ in range(accelerator.num_processes ):
sched.step()
opt.zero_grad()
# Perform gradient accumulation under wrapper
with accelerator.accumulate(_UpperCamelCase ):
step_model(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
ddp_opt.step()
ddp_sched.step()
ddp_opt.zero_grad()
# Learning rates should be the same
assert (
opt.param_groups[0]["lr"] == ddp_opt.param_groups[0]["lr"]
), F"""Learning rates found in each optimizer did not align\nopt: {opt.param_groups[0]["lr"]}\nDDP opt: {ddp_opt.param_groups[0]["lr"]}\n"""
__lowerCamelCase = (((iteration + 1) % 2) == 0) or ((iteration + 1) == len(_UpperCamelCase ))
if accelerator.num_processes > 1:
check_model_parameters(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Shuffle ddp_input on each iteration
torch.manual_seed(13_37 + iteration )
GradientState._reset_state()
def a__ ( ):
__lowerCamelCase = Accelerator()
__lowerCamelCase = RegressionDataset(length=80 )
__lowerCamelCase = DataLoader(_UpperCamelCase ,batch_size=16 )
__lowerCamelCase = RegressionDataset(length=96 )
__lowerCamelCase = DataLoader(_UpperCamelCase ,batch_size=16 )
__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(_UpperCamelCase ,_UpperCamelCase )
assert accelerator.gradient_state.active_dataloader is None
for iteration, _ in enumerate(_UpperCamelCase ):
assert id(accelerator.gradient_state.active_dataloader ) == id(_UpperCamelCase )
if iteration < len(_UpperCamelCase ) - 1:
assert not accelerator.gradient_state.end_of_dataloader
if iteration == 1:
for batch_num, _ in enumerate(_UpperCamelCase ):
assert id(accelerator.gradient_state.active_dataloader ) == id(_UpperCamelCase )
if batch_num < len(_UpperCamelCase ) - 1:
assert not accelerator.gradient_state.end_of_dataloader
else:
assert accelerator.gradient_state.end_of_dataloader
else:
assert accelerator.gradient_state.end_of_dataloader
assert accelerator.gradient_state.active_dataloader is None
def a__ ( ):
__lowerCamelCase = Accelerator()
__lowerCamelCase = accelerator.state
if state.local_process_index == 0:
print('''**Test `accumulate` gradient accumulation with dataloader break**''' )
test_dataloader_break()
if state.distributed_type == DistributedType.NO:
if state.local_process_index == 0:
print('''**Test NOOP `no_sync` context manager**''' )
test_noop_sync(_UpperCamelCase )
if state.distributed_type in (DistributedType.MULTI_GPU, DistributedType.MULTI_CPU):
if state.local_process_index == 0:
print('''**Test Distributed `no_sync` context manager**''' )
test_distributed_sync(_UpperCamelCase )
if state.distributed_type == DistributedType.MULTI_GPU:
for split_batch in [True, False]:
for dispatch_batches in [True, False]:
if state.local_process_index == 0:
print(
'''**Test `accumulate` gradient accumulation, ''' ,F"""`split_batches={split_batch}` and `dispatch_batches={dispatch_batches}`**""" ,)
test_gradient_accumulation(_UpperCamelCase ,_UpperCamelCase )
# Currently will break on torch 2.0 +, need to investigate why
if is_torch_version('''<''' ,'''2.0''' ) or state.distributed_type == DistributedType.NO:
if state.local_process_index == 0:
print(
'''**Test `accumulate` gradient accumulation with optimizer and scheduler, ''' ,'''`split_batches=False`, `dispatch_batches=False`**''' ,)
test_gradient_accumulation_with_opt_and_scheduler()
if state.distributed_type == DistributedType.MULTI_GPU:
for split_batch in [True, False]:
for dispatch_batches in [True, False]:
if not split_batch and not dispatch_batches:
continue
if state.local_process_index == 0:
print(
'''**Test `accumulate` gradient accumulation with optimizer and scheduler, ''' ,F"""`split_batches={split_batch}` and `dispatch_batches={dispatch_batches}`**""" ,)
test_gradient_accumulation_with_opt_and_scheduler(_UpperCamelCase ,_UpperCamelCase )
def a__ ( _UpperCamelCase : Optional[Any] ):
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()
| 330 |
import gc
import unittest
from parameterized import parameterized
from diffusers import FlaxUNetaDConditionModel
from diffusers.utils import is_flax_available
from diffusers.utils.testing_utils import load_hf_numpy, require_flax, slow
if is_flax_available():
import jax
import jax.numpy as jnp
@slow
@require_flax
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
return F"""gaussian_noise_s={seed}_shape={"_".join([str(__UpperCAmelCase ) for s in shape] )}.npy"""
def lowerCamelCase ( self ):
'''simple docstring'''
# clean up the VRAM after each test
super().tearDown()
gc.collect()
def lowerCamelCase ( self , __UpperCAmelCase=0 , __UpperCAmelCase=(4, 4, 64, 64) , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = jnp.array(load_hf_numpy(self.get_file_format(__UpperCAmelCase , __UpperCAmelCase ) ) , dtype=__UpperCAmelCase )
return image
def lowerCamelCase ( self , __UpperCAmelCase=False , __UpperCAmelCase="CompVis/stable-diffusion-v1-4" ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = '''bf16''' if fpaa else None
__lowerCamelCase ,__lowerCamelCase = FlaxUNetaDConditionModel.from_pretrained(
__UpperCAmelCase , subfolder='''unet''' , dtype=__UpperCAmelCase , revision=__UpperCAmelCase )
return model, params
def lowerCamelCase ( self , __UpperCAmelCase=0 , __UpperCAmelCase=(4, 77, 768) , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = jnp.bfloataa if fpaa else jnp.floataa
__lowerCamelCase = jnp.array(load_hf_numpy(self.get_file_format(__UpperCAmelCase , __UpperCAmelCase ) ) , dtype=__UpperCAmelCase )
return hidden_states
@parameterized.expand(
[
# fmt: off
[83, 4, [-0.2_323, -0.1_304, 0.0_813, -0.3_093, -0.0_919, -0.1_571, -0.1_125, -0.5_806]],
[17, 0.55, [-0.0_831, -0.2_443, 0.0_901, -0.0_919, 0.3_396, 0.0_103, -0.3_743, 0.0_701]],
[8, 0.89, [-0.4_863, 0.0_859, 0.0_875, -0.1_658, 0.9_199, -0.0_114, 0.4_839, 0.4_639]],
[3, 1000, [-0.5_649, 0.2_402, -0.5_518, 0.1_248, 1.1_328, -0.2_443, -0.0_325, -1.0_078]],
# fmt: on
] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_unet_model(model_id='''CompVis/stable-diffusion-v1-4''' , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_latents(__UpperCAmelCase , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_encoder_hidden_states(__UpperCAmelCase , fpaa=__UpperCAmelCase )
__lowerCamelCase = model.apply(
{'''params''': params} , __UpperCAmelCase , jnp.array(__UpperCAmelCase , dtype=jnp.intaa ) , encoder_hidden_states=__UpperCAmelCase , ).sample
assert sample.shape == latents.shape
__lowerCamelCase = jnp.asarray(jax.device_get((sample[-1, -2:, -2:, :2].flatten()) ) , dtype=jnp.floataa )
__lowerCamelCase = jnp.array(__UpperCAmelCase , dtype=jnp.floataa )
# Found torch (float16) and flax (bfloat16) outputs to be within this tolerance, in the same hardware
assert jnp.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-2 )
@parameterized.expand(
[
# fmt: off
[83, 4, [0.1_514, 0.0_807, 0.1_624, 0.1_016, -0.1_896, 0.0_263, 0.0_677, 0.2_310]],
[17, 0.55, [0.1_164, -0.0_216, 0.0_170, 0.1_589, -0.3_120, 0.1_005, -0.0_581, -0.1_458]],
[8, 0.89, [-0.1_758, -0.0_169, 0.1_004, -0.1_411, 0.1_312, 0.1_103, -0.1_996, 0.2_139]],
[3, 1000, [0.1_214, 0.0_352, -0.0_731, -0.1_562, -0.0_994, -0.0_906, -0.2_340, -0.0_539]],
# fmt: on
] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.get_unet_model(model_id='''stabilityai/stable-diffusion-2''' , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_latents(__UpperCAmelCase , shape=(4, 4, 96, 96) , fpaa=__UpperCAmelCase )
__lowerCamelCase = self.get_encoder_hidden_states(__UpperCAmelCase , shape=(4, 77, 1024) , fpaa=__UpperCAmelCase )
__lowerCamelCase = model.apply(
{'''params''': params} , __UpperCAmelCase , jnp.array(__UpperCAmelCase , dtype=jnp.intaa ) , encoder_hidden_states=__UpperCAmelCase , ).sample
assert sample.shape == latents.shape
__lowerCamelCase = jnp.asarray(jax.device_get((sample[-1, -2:, -2:, :2].flatten()) ) , dtype=jnp.floataa )
__lowerCamelCase = jnp.array(__UpperCAmelCase , dtype=jnp.floataa )
# Found torch (float16) and flax (bfloat16) outputs to be within this tolerance, on the same hardware
assert jnp.allclose(__UpperCAmelCase , __UpperCAmelCase , atol=1E-2 )
| 330 | 1 |
from scipy.stats import pearsonr, spearmanr
from sklearn.metrics import fa_score, matthews_corrcoef
import datasets
a_ = """\
@inproceedings{wang2019glue,
title={{GLUE}: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding},
author={Wang, Alex and Singh, Amanpreet and Michael, Julian and Hill, Felix and Levy, Omer and Bowman, Samuel R.},
note={In the Proceedings of ICLR.},
year={2019}
}
"""
a_ = """\
GLUE, the General Language Understanding Evaluation benchmark
(https://gluebenchmark.com/) is a collection of resources for training,
evaluating, and analyzing natural language understanding systems.
"""
a_ = """
Compute GLUE evaluation metric associated to each GLUE dataset.
Args:
predictions: list of predictions to score.
Each translation should be tokenized into a list of tokens.
references: list of lists of references for each translation.
Each reference should be tokenized into a list of tokens.
Returns: depending on the GLUE subset, one or several of:
\"accuracy\": Accuracy
\"f1\": F1 score
\"pearson\": Pearson Correlation
\"spearmanr\": Spearman Correlation
\"matthews_correlation\": Matthew Correlation
Examples:
>>> glue_metric = datasets.load_metric('glue', 'sst2') # 'sst2' or any of [\"mnli\", \"mnli_mismatched\", \"mnli_matched\", \"qnli\", \"rte\", \"wnli\", \"hans\"]
>>> references = [0, 1]
>>> predictions = [0, 1]
>>> results = glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{'accuracy': 1.0}
>>> glue_metric = datasets.load_metric('glue', 'mrpc') # 'mrpc' or 'qqp'
>>> references = [0, 1]
>>> predictions = [0, 1]
>>> results = glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{'accuracy': 1.0, 'f1': 1.0}
>>> glue_metric = datasets.load_metric('glue', 'stsb')
>>> references = [0., 1., 2., 3., 4., 5.]
>>> predictions = [0., 1., 2., 3., 4., 5.]
>>> results = glue_metric.compute(predictions=predictions, references=references)
>>> print({\"pearson\": round(results[\"pearson\"], 2), \"spearmanr\": round(results[\"spearmanr\"], 2)})
{'pearson': 1.0, 'spearmanr': 1.0}
>>> glue_metric = datasets.load_metric('glue', 'cola')
>>> references = [0, 1]
>>> predictions = [0, 1]
>>> results = glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{'matthews_correlation': 1.0}
"""
def a__ ( _UpperCamelCase : List[str] ,_UpperCamelCase : Any ):
return float((preds == labels).mean() )
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : List[Any] ):
__lowerCamelCase = simple_accuracy(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = float(fa_score(y_true=_UpperCamelCase ,y_pred=_UpperCamelCase ) )
return {
"accuracy": acc,
"f1": fa,
}
def a__ ( _UpperCamelCase : Tuple ,_UpperCamelCase : Dict ):
__lowerCamelCase = float(pearsonr(_UpperCamelCase ,_UpperCamelCase )[0] )
__lowerCamelCase = float(spearmanr(_UpperCamelCase ,_UpperCamelCase )[0] )
return {
"pearson": pearson_corr,
"spearmanr": spearman_corr,
}
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class __lowerCAmelCase ( datasets.Metric ):
def lowerCamelCase ( self ):
'''simple docstring'''
if self.config_name not in [
"sst2",
"mnli",
"mnli_mismatched",
"mnli_matched",
"cola",
"stsb",
"mrpc",
"qqp",
"qnli",
"rte",
"wnli",
"hans",
]:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["sst2", "mnli", "mnli_mismatched", "mnli_matched", '''
'''"cola", "stsb", "mrpc", "qqp", "qnli", "rte", "wnli", "hans"]''' )
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(
{
'''predictions''': datasets.Value('''int64''' if self.config_name != '''stsb''' else '''float32''' ),
'''references''': datasets.Value('''int64''' if self.config_name != '''stsb''' else '''float32''' ),
} ) , codebase_urls=[] , reference_urls=[] , format='''numpy''' , )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
if self.config_name == "cola":
return {"matthews_correlation": matthews_corrcoef(__UpperCAmelCase , __UpperCAmelCase )}
elif self.config_name == "stsb":
return pearson_and_spearman(__UpperCAmelCase , __UpperCAmelCase )
elif self.config_name in ["mrpc", "qqp"]:
return acc_and_fa(__UpperCAmelCase , __UpperCAmelCase )
elif self.config_name in ["sst2", "mnli", "mnli_mismatched", "mnli_matched", "qnli", "rte", "wnli", "hans"]:
return {"accuracy": simple_accuracy(__UpperCAmelCase , __UpperCAmelCase )}
else:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["sst2", "mnli", "mnli_mismatched", "mnli_matched", '''
'''"cola", "stsb", "mrpc", "qqp", "qnli", "rte", "wnli", "hans"]''' )
| 330 |
from typing import TYPE_CHECKING
from ....utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_mmbt""": ["""MMBTConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""MMBTForClassification""", """MMBTModel""", """ModalEmbeddings"""]
if TYPE_CHECKING:
from .configuration_mmbt import MMBTConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_mmbt import MMBTForClassification, MMBTModel, ModalEmbeddings
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
from math import factorial
def a__ ( _UpperCamelCase : 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(_UpperCamelCase ) / (factorial(_UpperCamelCase ) * factorial(n - k )) )
if __name__ == "__main__":
import sys
if len(sys.argv) == 1:
print(solution(20))
else:
try:
a_ = int(sys.argv[1])
print(solution(n))
except ValueError:
print("""Invalid entry - please enter a number.""")
| 330 |
import gc
import importlib.metadata
import tempfile
import unittest
from packaging import version
from transformers import (
AutoModel,
AutoModelForCausalLM,
AutoModelForSeqaSeqLM,
AutoModelForSequenceClassification,
AutoTokenizer,
BitsAndBytesConfig,
pipeline,
)
from transformers.testing_utils import (
is_torch_available,
require_accelerate,
require_bitsandbytes,
require_torch,
require_torch_gpu,
require_torch_multi_gpu,
slow,
)
def a__ ( _UpperCamelCase : Optional[int] ):
if model.config.model_type == "gpt2":
return model.transformer.h[0].mlp.c_fc
return model.transformer.h[0].mlp.dense_ah_to_h
if is_torch_available():
import torch
import torch.nn as nn
class __lowerCAmelCase ( nn.Module ):
def __init__( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
super().__init__()
__lowerCamelCase = module
__lowerCamelCase = nn.Sequential(
nn.Linear(module.in_features , __UpperCAmelCase , bias=__UpperCAmelCase ) , nn.Linear(__UpperCAmelCase , module.out_features , bias=__UpperCAmelCase ) , )
__lowerCamelCase = (2.0 / (5 * min(module.in_features , module.out_features ))) ** 0.5
nn.init.normal_(self.adapter[0].weight , std=__UpperCAmelCase )
nn.init.zeros_(self.adapter[1].weight )
self.adapter.to(module.weight.device )
def lowerCamelCase ( self , __UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return self.module(__UpperCAmelCase , *__UpperCAmelCase , **__UpperCAmelCase ) + self.adapter(__UpperCAmelCase )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class __lowerCAmelCase ( unittest.TestCase ):
# We keep the constants inside the init function and model loading inside setUp function
# We need to test on relatively large models (aka >1b parameters otherwise the quantiztion may not work as expected)
# Therefore here we use only bloom-1b3 to test our module
lowerCAmelCase__ = """bigscience/bloom-1b7"""
# Constant values
lowerCAmelCase__ = 2.1_09_65_95_52_69_25_74
lowerCAmelCase__ = """Hello my name is"""
lowerCAmelCase__ = set()
EXPECTED_OUTPUTS.add("""Hello my name is John and I am a professional photographer. I""" )
EXPECTED_OUTPUTS.add("""Hello my name is John.\nI am a friend of your father.\n""" )
EXPECTED_OUTPUTS.add("""Hello my name is John Doe, I am a student at the University""" )
lowerCAmelCase__ = 1_0
def lowerCamelCase ( self ):
'''simple docstring'''
# Models and tokenizer
__lowerCamelCase = AutoTokenizer.from_pretrained(self.model_name )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# Models and tokenizer
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , torch_dtype=torch.floataa , device_map='''auto''' )
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
def lowerCamelCase ( self ):
'''simple docstring'''
del self.model_fpaa
del self.model_abit
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_abit.config
self.assertTrue(hasattr(__UpperCAmelCase , '''quantization_config''' ) )
__lowerCamelCase = config.to_dict()
__lowerCamelCase = config.to_diff_dict()
__lowerCamelCase = config.to_json_string()
def lowerCamelCase ( self ):
'''simple docstring'''
from bitsandbytes.nn import Paramsabit
__lowerCamelCase = self.model_fpaa.get_memory_footprint()
__lowerCamelCase = self.model_abit.get_memory_footprint()
self.assertAlmostEqual(mem_fpaa / mem_abit , self.EXPECTED_RELATIVE_DIFFERENCE )
__lowerCamelCase = get_some_linear_layer(self.model_abit )
self.assertTrue(linear.weight.__class__ == Paramsabit )
def lowerCamelCase ( self ):
'''simple docstring'''
from transformers import TaPreTrainedModel
self.model_fpaa.get_memory_footprint()
self.model_abit.get_memory_footprint()
for name, module in self.model_abit.named_modules():
if isinstance(__UpperCAmelCase , torch.nn.Linear ):
if name not in ["lm_head"] + TaPreTrainedModel._keep_in_fpaa_modules:
# 4-bit parameters are packed in uint8 variables
self.assertTrue(module.weight.dtype == torch.uinta )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = self.model_abit.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_sequences[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitsAndBytesConfig()
__lowerCamelCase = True
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = model_abit_from_config.generate(
input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_sequences[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
def lowerCamelCase ( self ):
'''simple docstring'''
with self.assertRaises(__UpperCAmelCase ), tempfile.TemporaryDirectory() as tmpdirname:
self.model_abit.save_pretrained(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BitsAndBytesConfig()
with self.assertRaises(__UpperCAmelCase ):
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=__UpperCAmelCase , load_in_abit=__UpperCAmelCase , device_map='''auto''' , bnb_abit_quant_type='''nf4''' , )
def lowerCamelCase ( self ):
'''simple docstring'''
with self.assertRaises(__UpperCAmelCase ):
# Tries with `str`
self.model_abit.to('''cpu''' )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `dtype``
self.model_abit.to(torch.floataa )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.to(torch.device('''cuda:0''' ) )
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.float()
with self.assertRaises(__UpperCAmelCase ):
# Tries with a `device`
self.model_abit.half()
# Test if we did not break anything
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
__lowerCamelCase = self.model_fpaa.to(torch.floataa )
__lowerCamelCase = self.model_fpaa.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.to('''cpu''' )
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.half()
# Check this does not throw an error
__lowerCamelCase = self.model_fpaa.float()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = AutoModelForSeqaSeqLM.from_pretrained('''t5-small''' , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
self.assertTrue(model.decoder.block[0].layer[2].DenseReluDense.wo.weight.dtype == torch.floataa )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class __lowerCAmelCase ( unittest.TestCase ):
@classmethod
def lowerCamelCase ( cls ):
'''simple docstring'''
__lowerCamelCase = '''t5-small'''
__lowerCamelCase = '''google/flan-t5-small''' # flan-t5 uses dense-act instead of dense-relu-dense
__lowerCamelCase = AutoTokenizer.from_pretrained(cls.model_name )
__lowerCamelCase = '''Translate in German: Hello, my dog is cute'''
def lowerCamelCase ( self ):
'''simple docstring'''
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
from transformers import TaForConditionalGeneration
__lowerCamelCase = TaForConditionalGeneration._keep_in_fpaa_modules
__lowerCamelCase = None
# test with `t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
# test with `flan-t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
__lowerCamelCase = modules
def lowerCamelCase ( self ):
'''simple docstring'''
import bitsandbytes as bnb
from transformers import TaForConditionalGeneration
# test with `t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# there was a bug with decoders - this test checks that it is fixed
self.assertTrue(isinstance(model.decoder.block[0].layer[0].SelfAttention.q , bnb.nn.Linearabit ) )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
# test with `flan-t5-small`
__lowerCamelCase = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' ).to(0 )
__lowerCamelCase = model.generate(**__UpperCAmelCase )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
# model_name
__lowerCamelCase = '''bigscience/bloom-560m'''
__lowerCamelCase = '''t5-small'''
# Different types of model
__lowerCamelCase = AutoModel.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# Sequence classification model
__lowerCamelCase = AutoModelForSequenceClassification.from_pretrained(
self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# CausalLM model
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
# Seq2seq model
__lowerCamelCase = AutoModelForSeqaSeqLM.from_pretrained(
self.seq_to_seq_name , load_in_abit=__UpperCAmelCase , device_map='''auto''' )
def lowerCamelCase ( self ):
'''simple docstring'''
del self.base_model
del self.sequence_model
del self.model_abit
del self.seq_to_seq_model
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
from bitsandbytes.nn import Paramsabit
self.assertTrue(self.base_model.h[-1].mlp.dense_ah_to_h.weight.__class__ == Paramsabit )
# Other heads should be nn.Parameter
self.assertTrue(self.model_abit.lm_head.weight.__class__ == torch.nn.Parameter )
self.assertTrue(self.sequence_model.score.weight.__class__ == torch.nn.Parameter )
self.assertTrue(self.seq_to_seq_model.lm_head.weight.__class__ == torch.nn.Parameter )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
del self.pipe
gc.collect()
torch.cuda.empty_cache()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = pipeline(
'''text-generation''' , model=self.model_name , model_kwargs={'''device_map''': '''auto''', '''load_in_4bit''': True, '''torch_dtype''': torch.floataa} , max_new_tokens=self.MAX_NEW_TOKENS , )
# Real second forward pass
__lowerCamelCase = self.pipe(self.input_text )
self.assertIn(pipeline_output[0]['''generated_text'''] , self.EXPECTED_OUTPUTS )
@require_torch_multi_gpu
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(
self.model_name , load_in_abit=__UpperCAmelCase , device_map='''balanced''' )
# Check correct device map
self.assertEqual(set(model_parallel.hf_device_map.values() ) , {0, 1} )
# Check that inference pass works on the model
__lowerCamelCase = self.tokenizer(self.input_text , return_tensors='''pt''' )
# Second real batch
__lowerCamelCase = model_parallel.generate(input_ids=encoded_input['''input_ids'''].to(0 ) , max_new_tokens=10 )
self.assertIn(self.tokenizer.decode(output_parallel[0] , skip_special_tokens=__UpperCAmelCase ) , self.EXPECTED_OUTPUTS )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''facebook/opt-350m'''
super().setUp()
def lowerCamelCase ( self ):
'''simple docstring'''
if version.parse(importlib.metadata.version('''bitsandbytes''' ) ) < version.parse('''0.37.0''' ):
return
# Step 1: freeze all parameters
__lowerCamelCase = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=__UpperCAmelCase )
self.assertEqual(set(model.hf_device_map.values() ) , {torch.cuda.current_device()} )
for param in model.parameters():
__lowerCamelCase = False # freeze the model - train adapters later
if param.ndim == 1:
# cast the small parameters (e.g. layernorm) to fp32 for stability
__lowerCamelCase = param.data.to(torch.floataa )
# Step 2: add adapters
for _, module in model.named_modules():
if "OPTAttention" in repr(type(__UpperCAmelCase ) ):
__lowerCamelCase = LoRALayer(module.q_proj , rank=16 )
__lowerCamelCase = LoRALayer(module.k_proj , rank=16 )
__lowerCamelCase = LoRALayer(module.v_proj , rank=16 )
# Step 3: dummy batch
__lowerCamelCase = self.tokenizer('''Test batch ''' , return_tensors='''pt''' ).to(0 )
# Step 4: Check if the gradient is not None
with torch.cuda.amp.autocast():
__lowerCamelCase = model.forward(**__UpperCAmelCase )
out.logits.norm().backward()
for module in model.modules():
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
self.assertTrue(module.adapter[1].weight.grad is not None )
self.assertTrue(module.adapter[1].weight.grad.norm().item() > 0 )
elif isinstance(__UpperCAmelCase , nn.Embedding ):
self.assertTrue(module.weight.grad is None )
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = """gpt2-xl"""
lowerCAmelCase__ = 3.31_91_85_48_54_15_21_87
| 330 | 1 |
from __future__ import annotations
from math import pow, sqrt
def a__ ( _UpperCamelCase : float ,_UpperCamelCase : float ,_UpperCamelCase : float ):
if (resistance, reactance, impedance).count(0 ) != 1:
raise ValueError('''One and only one argument must be 0''' )
if resistance == 0:
return {"resistance": sqrt(pow(_UpperCamelCase ,2 ) - pow(_UpperCamelCase ,2 ) )}
elif reactance == 0:
return {"reactance": sqrt(pow(_UpperCamelCase ,2 ) - pow(_UpperCamelCase ,2 ) )}
elif impedance == 0:
return {"impedance": sqrt(pow(_UpperCamelCase ,2 ) + pow(_UpperCamelCase ,2 ) )}
else:
raise ValueError('''Exactly one argument must be 0''' )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 330 |
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 __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 42
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ ):
lowerCAmelCase__ = True
@register_to_config
def __init__( self , __UpperCAmelCase = 3 , __UpperCAmelCase = 3 , __UpperCAmelCase = ("DownEncoderBlock2D",) , __UpperCAmelCase = ("UpDecoderBlock2D",) , __UpperCAmelCase = (64,) , __UpperCAmelCase = 1 , __UpperCAmelCase = "silu" , __UpperCAmelCase = 4 , __UpperCAmelCase = 32 , __UpperCAmelCase = 32 , __UpperCAmelCase = 0.18_215 , ):
'''simple docstring'''
super().__init__()
# pass init params to Encoder
__lowerCamelCase = Encoder(
in_channels=__UpperCAmelCase , out_channels=__UpperCAmelCase , down_block_types=__UpperCAmelCase , block_out_channels=__UpperCAmelCase , layers_per_block=__UpperCAmelCase , act_fn=__UpperCAmelCase , norm_num_groups=__UpperCAmelCase , double_z=__UpperCAmelCase , )
# pass init params to Decoder
__lowerCamelCase = Decoder(
in_channels=__UpperCAmelCase , out_channels=__UpperCAmelCase , up_block_types=__UpperCAmelCase , block_out_channels=__UpperCAmelCase , layers_per_block=__UpperCAmelCase , norm_num_groups=__UpperCAmelCase , act_fn=__UpperCAmelCase , )
__lowerCamelCase = nn.Convad(2 * latent_channels , 2 * latent_channels , 1 )
__lowerCamelCase = nn.Convad(__UpperCAmelCase , __UpperCAmelCase , 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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
if isinstance(__UpperCAmelCase , (Encoder, Decoder) ):
__lowerCamelCase = value
def lowerCamelCase ( self , __UpperCAmelCase = True ):
'''simple docstring'''
__lowerCamelCase = use_tiling
def lowerCamelCase ( self ):
'''simple docstring'''
self.enable_tiling(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = True
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = False
@property
# Copied from diffusers.models.unet_2d_condition.UNet2DConditionModel.attn_processors
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = {}
def fn_recursive_add_processors(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
if hasattr(__UpperCAmelCase , '''set_processor''' ):
__lowerCamelCase = module.processor
for sub_name, child in module.named_children():
fn_recursive_add_processors(F"""{name}.{sub_name}""" , __UpperCAmelCase , __UpperCAmelCase )
return processors
for name, module in self.named_children():
fn_recursive_add_processors(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
return processors
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = len(self.attn_processors.keys() )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) != count:
raise ValueError(
F"""A dict of processors was passed, but the number of processors {len(__UpperCAmelCase )} does not match the"""
F""" number of attention layers: {count}. Please make sure to pass {count} processor classes.""" )
def fn_recursive_attn_processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
if hasattr(__UpperCAmelCase , '''set_processor''' ):
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
module.set_processor(__UpperCAmelCase )
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}""" , __UpperCAmelCase , __UpperCAmelCase )
for name, module in self.named_children():
fn_recursive_attn_processor(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
self.set_attn_processor(AttnProcessor() )
@apply_forward_hook
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
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(__UpperCAmelCase , return_dict=__UpperCAmelCase )
if self.use_slicing and x.shape[0] > 1:
__lowerCamelCase = [self.encoder(__UpperCAmelCase ) for x_slice in x.split(1 )]
__lowerCamelCase = torch.cat(__UpperCAmelCase )
else:
__lowerCamelCase = self.encoder(__UpperCAmelCase )
__lowerCamelCase = self.quant_conv(__UpperCAmelCase )
__lowerCamelCase = DiagonalGaussianDistribution(__UpperCAmelCase )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
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(__UpperCAmelCase , return_dict=__UpperCAmelCase )
__lowerCamelCase = self.post_quant_conv(__UpperCAmelCase )
__lowerCamelCase = self.decoder(__UpperCAmelCase )
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
@apply_forward_hook
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
if self.use_slicing and z.shape[0] > 1:
__lowerCamelCase = [self._decode(__UpperCAmelCase ).sample for z_slice in z.split(1 )]
__lowerCamelCase = torch.cat(__UpperCAmelCase )
else:
__lowerCamelCase = self._decode(__UpperCAmelCase ).sample
if not return_dict:
return (decoded,)
return DecoderOutput(sample=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(a.shape[2] , b.shape[2] , __UpperCAmelCase )
for y in range(__UpperCAmelCase ):
__lowerCamelCase = a[:, :, -blend_extent + y, :] * (1 - y / blend_extent) + b[:, :, y, :] * (y / blend_extent)
return b
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = min(a.shape[3] , b.shape[3] , __UpperCAmelCase )
for x in range(__UpperCAmelCase ):
__lowerCamelCase = a[:, :, :, -blend_extent + x] * (1 - x / blend_extent) + b[:, :, :, x] * (x / blend_extent)
return b
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
__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] , __UpperCAmelCase ):
__lowerCamelCase = []
for j in range(0 , x.shape[3] , __UpperCAmelCase ):
__lowerCamelCase = x[:, :, i : i + self.tile_sample_min_size, j : j + self.tile_sample_min_size]
__lowerCamelCase = self.encoder(__UpperCAmelCase )
__lowerCamelCase = self.quant_conv(__UpperCAmelCase )
row.append(__UpperCAmelCase )
rows.append(__UpperCAmelCase )
__lowerCamelCase = []
for i, row in enumerate(__UpperCAmelCase ):
__lowerCamelCase = []
for j, tile in enumerate(__UpperCAmelCase ):
# 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] , __UpperCAmelCase , __UpperCAmelCase )
if j > 0:
__lowerCamelCase = self.blend_h(row[j - 1] , __UpperCAmelCase , __UpperCAmelCase )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(__UpperCAmelCase , dim=3 ) )
__lowerCamelCase = torch.cat(__UpperCAmelCase , dim=2 )
__lowerCamelCase = DiagonalGaussianDistribution(__UpperCAmelCase )
if not return_dict:
return (posterior,)
return AutoencoderKLOutput(latent_dist=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = True ):
'''simple docstring'''
__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] , __UpperCAmelCase ):
__lowerCamelCase = []
for j in range(0 , z.shape[3] , __UpperCAmelCase ):
__lowerCamelCase = z[:, :, i : i + self.tile_latent_min_size, j : j + self.tile_latent_min_size]
__lowerCamelCase = self.post_quant_conv(__UpperCAmelCase )
__lowerCamelCase = self.decoder(__UpperCAmelCase )
row.append(__UpperCAmelCase )
rows.append(__UpperCAmelCase )
__lowerCamelCase = []
for i, row in enumerate(__UpperCAmelCase ):
__lowerCamelCase = []
for j, tile in enumerate(__UpperCAmelCase ):
# 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] , __UpperCAmelCase , __UpperCAmelCase )
if j > 0:
__lowerCamelCase = self.blend_h(row[j - 1] , __UpperCAmelCase , __UpperCAmelCase )
result_row.append(tile[:, :, :row_limit, :row_limit] )
result_rows.append(torch.cat(__UpperCAmelCase , dim=3 ) )
__lowerCamelCase = torch.cat(__UpperCAmelCase , dim=2 )
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False , __UpperCAmelCase = True , __UpperCAmelCase = None , ):
'''simple docstring'''
__lowerCamelCase = sample
__lowerCamelCase = self.encode(__UpperCAmelCase ).latent_dist
if sample_posterior:
__lowerCamelCase = posterior.sample(generator=__UpperCAmelCase )
else:
__lowerCamelCase = posterior.mode()
__lowerCamelCase = self.decode(__UpperCAmelCase ).sample
if not return_dict:
return (dec,)
return DecoderOutput(sample=__UpperCAmelCase )
| 330 | 1 |
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {
"""configuration_blip_2""": [
"""BLIP_2_PRETRAINED_CONFIG_ARCHIVE_MAP""",
"""Blip2Config""",
"""Blip2QFormerConfig""",
"""Blip2VisionConfig""",
],
"""processing_blip_2""": ["""Blip2Processor"""],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = [
"""BLIP_2_PRETRAINED_MODEL_ARCHIVE_LIST""",
"""Blip2Model""",
"""Blip2QFormerModel""",
"""Blip2PreTrainedModel""",
"""Blip2ForConditionalGeneration""",
"""Blip2VisionModel""",
]
if TYPE_CHECKING:
from .configuration_blip_a import (
BLIP_2_PRETRAINED_CONFIG_ARCHIVE_MAP,
BlipaConfig,
BlipaQFormerConfig,
BlipaVisionConfig,
)
from .processing_blip_a import BlipaProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_blip_a import (
BLIP_2_PRETRAINED_MODEL_ARCHIVE_LIST,
BlipaForConditionalGeneration,
BlipaModel,
BlipaPreTrainedModel,
BlipaQFormerModel,
BlipaVisionModel,
)
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 |
import argparse
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import BigBirdPegasusConfig, BigBirdPegasusForConditionalGeneration
a_ = [
# tf -> hf
("""/""", """."""),
("""layer_""", """layers."""),
("""kernel""", """weight"""),
("""beta""", """bias"""),
("""gamma""", """weight"""),
("""pegasus""", """model"""),
]
a_ = [
(""".output.dense""", """.fc2"""),
("""intermediate.LayerNorm""", """final_layer_norm"""),
("""intermediate.dense""", """fc1"""),
]
a_ = (
INIT_COMMON
+ [
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.out_proj"""),
("""attention.self""", """self_attn"""),
("""attention.encdec.LayerNorm""", """encoder_attn_layer_norm"""),
("""attention.encdec_output.dense""", """encoder_attn.out_proj"""),
("""attention.encdec""", """encoder_attn"""),
("""key""", """k_proj"""),
("""value""", """v_proj"""),
("""query""", """q_proj"""),
("""decoder.LayerNorm""", """decoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = (
INIT_COMMON
+ [
("""embeddings.word_embeddings""", """shared.weight"""),
("""embeddings.position_embeddings""", """embed_positions.weight"""),
("""attention.self.LayerNorm""", """self_attn_layer_norm"""),
("""attention.output.dense""", """self_attn.output"""),
("""attention.self""", """self_attn.self"""),
("""encoder.LayerNorm""", """encoder.layernorm_embedding"""),
]
+ END_COMMON
)
a_ = [
"""encdec/key/bias""",
"""encdec/query/bias""",
"""encdec/value/bias""",
"""self/key/bias""",
"""self/query/bias""",
"""self/value/bias""",
"""encdec_output/dense/bias""",
"""attention/output/dense/bias""",
]
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : Optional[Any] ):
for tf_name, hf_name in patterns:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = BigBirdPegasusConfig(**_UpperCamelCase )
__lowerCamelCase = BigBirdPegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
__lowerCamelCase = {}
# separating decoder weights
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if k.startswith('''pegasus/decoder''' )}
__lowerCamelCase = {k: tf_weights[k] for k in tf_weights if not k.startswith('''pegasus/decoder''' )}
for k, v in tqdm(decoder_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = DECODER_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
for k, v in tqdm(remaining_weights.items() ,'''tf -> hf conversion''' ):
__lowerCamelCase = [k.endswith(_UpperCamelCase ) for ending in KEYS_TO_IGNORE]
if any(_UpperCamelCase ):
continue
__lowerCamelCase = REMAINING_PATTERNS
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase ,_UpperCamelCase )
if new_k not in state_dict and k != "pegasus/embeddings/position_embeddings":
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if any(True if i in k else False for i in ['''dense''', '''query''', '''key''', '''value'''] ):
__lowerCamelCase = v.T
__lowerCamelCase = torch.from_numpy(_UpperCamelCase )
if k != "pegasus/embeddings/position_embeddings":
assert v.shape == state_dict[new_k].shape, F"""{new_k}, {k}, {v.shape}, {state_dict[new_k].shape}"""
__lowerCamelCase = mapping['''model.embed_positions.weight''']
__lowerCamelCase = mapping.pop('''model.embed_positions.weight''' )
__lowerCamelCase ,__lowerCamelCase = torch_model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k
for k in missing
if k
not in [
'''final_logits_bias''',
'''model.encoder.embed_tokens.weight''',
'''model.decoder.embed_tokens.weight''',
'''lm_head.weight''',
]
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ,_UpperCamelCase : dict ):
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = convert_bigbird_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""--save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
a_ = {}
convert_bigbird_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir, config_update=config_update)
| 330 | 1 |
from __future__ import annotations
import bisect
def a__ ( _UpperCamelCase : list[int] ,_UpperCamelCase : int ,_UpperCamelCase : int = 0 ,_UpperCamelCase : int = -1 ):
if hi < 0:
__lowerCamelCase = len(_UpperCamelCase )
while lo < hi:
__lowerCamelCase = lo + (hi - lo) // 2
if sorted_collection[mid] < item:
__lowerCamelCase = mid + 1
else:
__lowerCamelCase = mid
return lo
def a__ ( _UpperCamelCase : list[int] ,_UpperCamelCase : int ,_UpperCamelCase : int = 0 ,_UpperCamelCase : int = -1 ):
if hi < 0:
__lowerCamelCase = len(_UpperCamelCase )
while lo < hi:
__lowerCamelCase = lo + (hi - lo) // 2
if sorted_collection[mid] <= item:
__lowerCamelCase = mid + 1
else:
__lowerCamelCase = mid
return lo
def a__ ( _UpperCamelCase : list[int] ,_UpperCamelCase : int ,_UpperCamelCase : int = 0 ,_UpperCamelCase : int = -1 ):
sorted_collection.insert(bisect_left(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ) ,_UpperCamelCase )
def a__ ( _UpperCamelCase : list[int] ,_UpperCamelCase : int ,_UpperCamelCase : int = 0 ,_UpperCamelCase : int = -1 ):
sorted_collection.insert(bisect_right(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ) ,_UpperCamelCase )
def a__ ( _UpperCamelCase : list[int] ,_UpperCamelCase : int ):
__lowerCamelCase = 0
__lowerCamelCase = len(_UpperCamelCase ) - 1
while left <= right:
__lowerCamelCase = left + (right - left) // 2
__lowerCamelCase = sorted_collection[midpoint]
if current_item == item:
return midpoint
elif item < current_item:
__lowerCamelCase = midpoint - 1
else:
__lowerCamelCase = midpoint + 1
return None
def a__ ( _UpperCamelCase : list[int] ,_UpperCamelCase : int ):
__lowerCamelCase = bisect.bisect_left(_UpperCamelCase ,_UpperCamelCase )
if index != len(_UpperCamelCase ) and sorted_collection[index] == item:
return index
return None
def a__ ( _UpperCamelCase : list[int] ,_UpperCamelCase : int ,_UpperCamelCase : int ,_UpperCamelCase : int ):
if right < left:
return None
__lowerCamelCase = left + (right - left) // 2
if sorted_collection[midpoint] == item:
return midpoint
elif sorted_collection[midpoint] > item:
return binary_search_by_recursion(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,midpoint - 1 )
else:
return binary_search_by_recursion(_UpperCamelCase ,_UpperCamelCase ,midpoint + 1 ,_UpperCamelCase )
if __name__ == "__main__":
a_ = input("""Enter numbers separated by comma:\n""").strip()
a_ = sorted(int(item) for item in user_input.split(""","""))
a_ = int(input("""Enter a single number to be found in the list:\n"""))
a_ = binary_search(collection, target)
if result is None:
print(f"{target} was not found in {collection}.")
else:
print(f"{target} was found at position {result} in {collection}.")
| 330 |
import uuid
from typing import Any, Dict, List, Optional, Union
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
if is_torch_available():
import torch
a_ = logging.get_logger(__name__)
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase=None , __UpperCAmelCase=None ):
'''simple docstring'''
if not conversation_id:
__lowerCamelCase = uuid.uuida()
if past_user_inputs is None:
__lowerCamelCase = []
if generated_responses is None:
__lowerCamelCase = []
__lowerCamelCase = conversation_id
__lowerCamelCase = past_user_inputs
__lowerCamelCase = generated_responses
__lowerCamelCase = text
def __eq__( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
return False
if self.uuid == other.uuid:
return True
return (
self.new_user_input == other.new_user_input
and self.past_user_inputs == other.past_user_inputs
and self.generated_responses == other.generated_responses
)
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = False ):
'''simple docstring'''
if self.new_user_input:
if overwrite:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" was overwritten """
F"""with: \"{text}\".""" )
__lowerCamelCase = text
else:
logger.warning(
F"""User input added while unprocessed input was existing: \"{self.new_user_input}\" new input """
F"""ignored: \"{text}\". Set `overwrite` to True to overwrite unprocessed user input""" )
else:
__lowerCamelCase = text
def lowerCamelCase ( self ):
'''simple docstring'''
if self.new_user_input:
self.past_user_inputs.append(self.new_user_input )
__lowerCamelCase = None
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
self.generated_responses.append(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
for user_input, generated_response in zip(self.past_user_inputs , self.generated_responses ):
yield True, user_input
yield False, generated_response
if self.new_user_input:
yield True, self.new_user_input
def __repr__( self ):
'''simple docstring'''
__lowerCamelCase = F"""Conversation id: {self.uuid} \n"""
for is_user, text in self.iter_texts():
__lowerCamelCase = '''user''' if is_user else '''bot'''
output += F"""{name} >> {text} \n"""
return output
@add_end_docstrings(
lowerCAmelCase__ , r"""
min_length_for_response (`int`, *optional*, defaults to 32):
The minimum length (in number of tokens) for a response.
minimum_tokens (`int`, *optional*, defaults to 10):
The minimum length of tokens to leave for a response.
""" , )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
if self.tokenizer.pad_token_id is None:
__lowerCamelCase = self.tokenizer.eos_token
def lowerCamelCase ( self , __UpperCAmelCase=None , __UpperCAmelCase=None , __UpperCAmelCase=None , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = {}
__lowerCamelCase = {}
__lowerCamelCase = {}
if min_length_for_response is not None:
__lowerCamelCase = min_length_for_response
if minimum_tokens is not None:
__lowerCamelCase = minimum_tokens
if "max_length" in generate_kwargs:
__lowerCamelCase = generate_kwargs['''max_length''']
# self.max_length = generate_kwargs.get("max_length", self.model.config.max_length)
if clean_up_tokenization_spaces is not None:
__lowerCamelCase = clean_up_tokenization_spaces
if generate_kwargs:
forward_params.update(__UpperCAmelCase )
return preprocess_params, forward_params, postprocess_params
def __call__( self , __UpperCAmelCase , __UpperCAmelCase=0 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = super().__call__(__UpperCAmelCase , num_workers=__UpperCAmelCase , **__UpperCAmelCase )
if isinstance(__UpperCAmelCase , __UpperCAmelCase ) and len(__UpperCAmelCase ) == 1:
return outputs[0]
return outputs
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=32 ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError('''ConversationalPipeline, expects Conversation as inputs''' )
if conversation.new_user_input is None:
raise ValueError(
F"""Conversation with UUID {type(conversation.uuid )} does not contain new user input to process. """
'''Add user inputs with the conversation\'s `add_user_input` method''' )
if hasattr(self.tokenizer , '''_build_conversation_input_ids''' ):
__lowerCamelCase = self.tokenizer._build_conversation_input_ids(__UpperCAmelCase )
else:
# If the tokenizer cannot handle conversations, we default to only the old version
__lowerCamelCase = self._legacy_parse_and_tokenize(__UpperCAmelCase )
if self.framework == "pt":
__lowerCamelCase = torch.LongTensor([input_ids] )
elif self.framework == "tf":
__lowerCamelCase = tf.constant([input_ids] )
return {"input_ids": input_ids, "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=10 , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = generate_kwargs.get('''max_length''' , self.model.config.max_length )
__lowerCamelCase = model_inputs['''input_ids'''].shape[1]
if max_length - minimum_tokens < n:
logger.warning(F"""Conversation input is to long ({n}), trimming it to ({max_length} - {minimum_tokens})""" )
__lowerCamelCase = max_length - minimum_tokens
__lowerCamelCase = model_inputs['''input_ids'''][:, -trim:]
if "attention_mask" in model_inputs:
__lowerCamelCase = model_inputs['''attention_mask'''][:, -trim:]
__lowerCamelCase = model_inputs.pop('''conversation''' )
__lowerCamelCase = max_length
__lowerCamelCase = self.model.generate(**__UpperCAmelCase , **__UpperCAmelCase )
if self.model.config.is_encoder_decoder:
__lowerCamelCase = 1
else:
__lowerCamelCase = n
return {"output_ids": output_ids[:, start_position:], "conversation": conversation}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = model_outputs['''output_ids''']
__lowerCamelCase = self.tokenizer.decode(
output_ids[0] , skip_special_tokens=__UpperCAmelCase , clean_up_tokenization_spaces=__UpperCAmelCase , )
__lowerCamelCase = model_outputs['''conversation''']
conversation.mark_processed()
conversation.append_response(__UpperCAmelCase )
return conversation
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.tokenizer.eos_token_id
__lowerCamelCase = []
for is_user, text in conversation.iter_texts():
if eos_token_id is not None:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) + [eos_token_id] )
else:
input_ids.extend(self.tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase ) )
if len(__UpperCAmelCase ) > self.tokenizer.model_max_length:
__lowerCamelCase = input_ids[-self.tokenizer.model_max_length :]
return input_ids
| 330 | 1 |
import argparse
import re
from typing import Dict
import torch
from datasets import Audio, Dataset, load_dataset, load_metric
from transformers import AutoFeatureExtractor, pipeline
def a__ ( _UpperCamelCase : Dataset ,_UpperCamelCase : Dict[str, str] ):
__lowerCamelCase = args.log_outputs
__lowerCamelCase = '''_'''.join(args.dataset.split('''/''' ) + [args.config, args.split] )
# load metric
__lowerCamelCase = load_metric('''wer''' )
__lowerCamelCase = load_metric('''cer''' )
# compute metrics
__lowerCamelCase = wer.compute(references=result['''target'''] ,predictions=result['''prediction'''] )
__lowerCamelCase = cer.compute(references=result['''target'''] ,predictions=result['''prediction'''] )
# print & log results
__lowerCamelCase = F"""WER: {wer_result}\nCER: {cer_result}"""
print(_UpperCamelCase )
with open(F"""{dataset_id}_eval_results.txt""" ,'''w''' ) as f:
f.write(_UpperCamelCase )
# log all results in text file. Possibly interesting for analysis
if log_outputs is not None:
__lowerCamelCase = F"""log_{dataset_id}_predictions.txt"""
__lowerCamelCase = F"""log_{dataset_id}_targets.txt"""
with open(_UpperCamelCase ,'''w''' ) as p, open(_UpperCamelCase ,'''w''' ) as t:
# mapping function to write output
def write_to_file(_UpperCamelCase : Optional[Any] ,_UpperCamelCase : int ):
p.write(F"""{i}""" + '''\n''' )
p.write(batch['''prediction'''] + '''\n''' )
t.write(F"""{i}""" + '''\n''' )
t.write(batch['''target'''] + '''\n''' )
result.map(_UpperCamelCase ,with_indices=_UpperCamelCase )
def a__ ( _UpperCamelCase : str ):
__lowerCamelCase = '''[,?.!\-\;\:"“%‘”�—’…–]''' # noqa: W605 IMPORTANT: this should correspond to the chars that were ignored during training
__lowerCamelCase = re.sub(_UpperCamelCase ,'''''' ,text.lower() )
# In addition, we can normalize the target text, e.g. removing new lines characters etc...
# note that order is important here!
__lowerCamelCase = ['''\n\n''', '''\n''', ''' ''', ''' ''']
for t in token_sequences_to_ignore:
__lowerCamelCase = ''' '''.join(text.split(_UpperCamelCase ) )
return text
def a__ ( _UpperCamelCase : Optional[Any] ):
# load dataset
__lowerCamelCase = load_dataset(args.dataset ,args.config ,split=args.split ,use_auth_token=_UpperCamelCase )
# for testing: only process the first two examples as a test
# dataset = dataset.select(range(10))
# load processor
__lowerCamelCase = AutoFeatureExtractor.from_pretrained(args.model_id )
__lowerCamelCase = feature_extractor.sampling_rate
# resample audio
__lowerCamelCase = dataset.cast_column('''audio''' ,Audio(sampling_rate=_UpperCamelCase ) )
# load eval pipeline
if args.device is None:
__lowerCamelCase = 0 if torch.cuda.is_available() else -1
__lowerCamelCase = pipeline('''automatic-speech-recognition''' ,model=args.model_id ,device=args.device )
# map function to decode audio
def map_to_pred(_UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = asr(
batch['''audio''']['''array'''] ,chunk_length_s=args.chunk_length_s ,stride_length_s=args.stride_length_s )
__lowerCamelCase = prediction['''text''']
__lowerCamelCase = normalize_text(batch['''sentence'''] )
return batch
# run inference on all examples
__lowerCamelCase = dataset.map(_UpperCamelCase ,remove_columns=dataset.column_names )
# compute and log_results
# do not change function below
log_results(_UpperCamelCase ,_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument(
"""--model_id""", type=str, required=True, help="""Model identifier. Should be loadable with 🤗 Transformers"""
)
parser.add_argument(
"""--dataset""",
type=str,
required=True,
help="""Dataset name to evaluate the `model_id`. Should be loadable with 🤗 Datasets""",
)
parser.add_argument(
"""--config""", type=str, required=True, help="""Config of the dataset. *E.g.* `'en'` for Common Voice"""
)
parser.add_argument("""--split""", type=str, required=True, help="""Split of the dataset. *E.g.* `'test'`""")
parser.add_argument(
"""--chunk_length_s""", type=float, default=None, help="""Chunk length in seconds. Defaults to 5 seconds."""
)
parser.add_argument(
"""--stride_length_s""", type=float, default=None, help="""Stride of the audio chunks. Defaults to 1 second."""
)
parser.add_argument(
"""--log_outputs""", action="""store_true""", help="""If defined, write outputs to log file for analysis."""
)
parser.add_argument(
"""--device""",
type=int,
default=None,
help="""The device to run the pipeline on. -1 for CPU (default), 0 for the first GPU and so on.""",
)
a_ = parser.parse_args()
main(args)
| 330 |
import argparse
import os
from pathlib import Path
from typing import Dict
import tensorflow as tf
import torch
from tqdm import tqdm
from transformers import PegasusConfig, PegasusForConditionalGeneration, PegasusTokenizer
from transformers.models.pegasus.configuration_pegasus import DEFAULTS, task_specific_params
a_ = [
# replace left string with right string to get the relevant state_dict key (identical state dict to bart)
["""memory_attention""", """encoder_attn"""],
["""attention""", """attn"""],
["""/""", """."""],
[""".LayerNorm.gamma""", """_layer_norm.weight"""],
[""".LayerNorm.beta""", """_layer_norm.bias"""],
["""r.layer_""", """r.layers."""],
["""output_proj""", """out_proj"""],
["""ffn.dense_1.""", """fc2."""],
["""ffn.dense.""", """fc1."""],
["""ffn_layer_norm""", """final_layer_norm"""],
["""kernel""", """weight"""],
["""encoder_layer_norm.""", """encoder.layer_norm."""],
["""decoder_layer_norm.""", """decoder.layer_norm."""],
["""embeddings.weights""", """shared.weight"""],
]
def a__ ( _UpperCamelCase : int ):
for pegasus_name, hf_name in PATTERNS:
__lowerCamelCase = k.replace(_UpperCamelCase ,_UpperCamelCase )
return k
def a__ ( _UpperCamelCase : dict ,_UpperCamelCase : dict ):
__lowerCamelCase = DEFAULTS.copy()
cfg_kwargs.update(_UpperCamelCase )
__lowerCamelCase = PegasusConfig(**_UpperCamelCase )
__lowerCamelCase = PegasusForConditionalGeneration(_UpperCamelCase )
__lowerCamelCase = torch_model.model.state_dict()
__lowerCamelCase = {}
for k, v in tf_weights.items():
__lowerCamelCase = rename_state_dict_key(_UpperCamelCase )
if new_k not in sd:
raise ValueError(F"""could not find new key {new_k} in state dict. (converted from {k})""" )
if "dense" in k or "proj" in new_k:
__lowerCamelCase = v.T
__lowerCamelCase = torch.tensor(_UpperCamelCase ,dtype=sd[new_k].dtype )
assert v.shape == sd[new_k].shape, F"""{new_k}, {k}, {v.shape}, {sd[new_k].shape}"""
# make sure embedding.padding_idx is respected
__lowerCamelCase = torch.zeros_like(mapping['''shared.weight'''][cfg.pad_token_id + 1] )
__lowerCamelCase = mapping['''shared.weight''']
__lowerCamelCase = mapping['''shared.weight''']
__lowerCamelCase = {k: torch.zeros_like(_UpperCamelCase ) for k, v in sd.items() if k.endswith('''bias''' ) and k not in mapping}
mapping.update(**_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = torch_model.model.load_state_dict(_UpperCamelCase ,strict=_UpperCamelCase )
__lowerCamelCase = [
k for k in missing if k not in ['''encoder.embed_positions.weight''', '''decoder.embed_positions.weight''']
]
assert unexpected_missing == [], F"""no matches found for the following torch keys {unexpected_missing}"""
assert extra == [], F"""no matches found for the following tf keys {extra}"""
return torch_model
def a__ ( _UpperCamelCase : str="./ckpt/aeslc/model.ckpt-32000" ):
__lowerCamelCase = tf.train.list_variables(_UpperCamelCase )
__lowerCamelCase = {}
__lowerCamelCase = ['''Adafactor''', '''global_step''']
for name, shape in tqdm(_UpperCamelCase ,desc='''converting tf checkpoint to dict''' ):
__lowerCamelCase = any(pat in name for pat in ignore_name )
if skip_key:
continue
__lowerCamelCase = tf.train.load_variable(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = array
return tf_weights
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# save tokenizer first
__lowerCamelCase = Path(_UpperCamelCase ).parent.name
__lowerCamelCase = task_specific_params[F"""summarization_{dataset}"""]['''max_position_embeddings''']
__lowerCamelCase = PegasusTokenizer.from_pretrained('''sshleifer/pegasus''' ,model_max_length=_UpperCamelCase )
assert tok.model_max_length == desired_max_model_length
tok.save_pretrained(_UpperCamelCase )
# convert model
__lowerCamelCase = get_tf_weights_as_numpy(_UpperCamelCase )
__lowerCamelCase = task_specific_params[F"""summarization_{dataset}"""]
if dataset == "large":
__lowerCamelCase = task_specific_params
__lowerCamelCase = convert_pegasus(_UpperCamelCase ,_UpperCamelCase )
torch_model.save_pretrained(_UpperCamelCase )
__lowerCamelCase = torch_model.state_dict()
sd.pop('''model.decoder.embed_positions.weight''' )
sd.pop('''model.encoder.embed_positions.weight''' )
torch.save(_UpperCamelCase ,Path(_UpperCamelCase ) / '''pytorch_model.bin''' )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument("""tf_ckpt_path""", type=str, help="""passed to tf.train.list_variables""")
parser.add_argument("""save_dir""", default=None, type=str, help="""Path to the output PyTorch model.""")
a_ = parser.parse_args()
if args.save_dir is None:
a_ = Path(args.tf_ckpt_path).parent.name
a_ = os.path.join("""pegasus""", dataset)
convert_pegasus_ckpt_to_pytorch(args.tf_ckpt_path, args.save_dir)
| 330 | 1 |
import unittest
from transformers import BarthezTokenizer, BarthezTokenizerFast, BatchEncoding
from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow
from ...test_tokenization_common import TokenizerTesterMixin
@require_tokenizers
@require_sentencepiece
@slow # see https://github.com/huggingface/transformers/issues/11457
class __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = BarthezTokenizer
lowerCAmelCase__ = BarthezTokenizerFast
lowerCAmelCase__ = True
lowerCAmelCase__ = True
def lowerCamelCase ( self ):
'''simple docstring'''
super().setUp()
__lowerCamelCase = BarthezTokenizerFast.from_pretrained('''moussaKam/mbarthez''' )
tokenizer.save_pretrained(self.tmpdirname )
tokenizer.save_pretrained(self.tmpdirname , legacy_format=__UpperCAmelCase )
__lowerCamelCase = tokenizer
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = '''<pad>'''
__lowerCamelCase = 1
self.assertEqual(self.get_tokenizer()._convert_token_to_id(__UpperCAmelCase ) , __UpperCAmelCase )
self.assertEqual(self.get_tokenizer()._convert_id_to_token(__UpperCAmelCase ) , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = list(self.get_tokenizer().get_vocab().keys() )
self.assertEqual(vocab_keys[0] , '''<s>''' )
self.assertEqual(vocab_keys[1] , '''<pad>''' )
self.assertEqual(vocab_keys[-1] , '''<mask>''' )
self.assertEqual(len(__UpperCAmelCase ) , 101122 )
def lowerCamelCase ( self ):
'''simple docstring'''
self.assertEqual(self.get_tokenizer().vocab_size , 101122 )
@require_torch
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = ['''A long paragraph for summarization.''', '''Another paragraph for summarization.''']
__lowerCamelCase = [0, 57, 3018, 70307, 91, 2]
__lowerCamelCase = self.tokenizer(
__UpperCAmelCase , max_length=len(__UpperCAmelCase ) , padding=__UpperCAmelCase , truncation=__UpperCAmelCase , return_tensors='''pt''' )
self.assertIsInstance(__UpperCAmelCase , __UpperCAmelCase )
self.assertEqual((2, 6) , batch.input_ids.shape )
self.assertEqual((2, 6) , batch.attention_mask.shape )
__lowerCamelCase = batch.input_ids.tolist()[0]
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.test_rust_tokenizer:
return
__lowerCamelCase = self.get_tokenizer()
__lowerCamelCase = self.get_rust_tokenizer()
__lowerCamelCase = '''I was born in 92000, and this is falsé.'''
__lowerCamelCase = tokenizer.tokenize(__UpperCAmelCase )
__lowerCamelCase = rust_tokenizer.tokenize(__UpperCAmelCase )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
__lowerCamelCase = rust_tokenizer.encode(__UpperCAmelCase , add_special_tokens=__UpperCAmelCase )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = self.get_rust_tokenizer()
__lowerCamelCase = tokenizer.encode(__UpperCAmelCase )
__lowerCamelCase = rust_tokenizer.encode(__UpperCAmelCase )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
# fmt: off
__lowerCamelCase = {'''input_ids''': [[0, 490, 14328, 4507, 354, 47, 43669, 95, 25, 78117, 20215, 19779, 190, 22, 400, 4, 35343, 80310, 603, 86, 24937, 105, 33438, 94762, 196, 39642, 7, 15, 15933, 173, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [0, 10534, 87, 25, 66, 3358, 196, 55289, 8, 82961, 81, 2204, 75203, 7, 15, 763, 12956, 216, 178, 14328, 9595, 1377, 69693, 7, 448, 71021, 196, 18106, 1437, 13974, 108, 9083, 4, 49315, 7, 39, 86, 1326, 2793, 46333, 4, 448, 196, 74588, 7, 49315, 7, 39, 21, 822, 38470, 74, 21, 66723, 62480, 8, 22050, 5, 2]], '''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, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]]} # noqa: E501
# fmt: on
# moussaKam/mbarthez is a french model. So we also use french texts.
__lowerCamelCase = [
'''Le transformeur est un modèle d\'apprentissage profond introduit en 2017, '''
'''utilisé principalement dans le domaine du traitement automatique des langues (TAL).''',
'''À l\'instar des réseaux de neurones récurrents (RNN), les transformeurs sont conçus '''
'''pour gérer des données séquentielles, telles que le langage naturel, pour des tâches '''
'''telles que la traduction et la synthèse de texte.''',
]
self.tokenizer_integration_test_util(
expected_encoding=__UpperCAmelCase , model_name='''moussaKam/mbarthez''' , revision='''c2e4ecbca5e3cd2c37fe1ac285ca4fbdf1366fb6''' , sequences=__UpperCAmelCase , )
| 330 |
import argparse
import fairseq
import torch
from transformers import UniSpeechSatConfig, UniSpeechSatForCTC, UniSpeechSatForPreTraining, logging
logging.set_verbosity_info()
a_ = logging.get_logger(__name__)
a_ = {
"""post_extract_proj""": """feature_projection.projection""",
"""encoder.pos_conv.0""": """encoder.pos_conv_embed.conv""",
"""self_attn.k_proj""": """encoder.layers.*.attention.k_proj""",
"""self_attn.v_proj""": """encoder.layers.*.attention.v_proj""",
"""self_attn.q_proj""": """encoder.layers.*.attention.q_proj""",
"""self_attn.out_proj""": """encoder.layers.*.attention.out_proj""",
"""self_attn_layer_norm""": """encoder.layers.*.layer_norm""",
"""fc1""": """encoder.layers.*.feed_forward.intermediate_dense""",
"""fc2""": """encoder.layers.*.feed_forward.output_dense""",
"""final_layer_norm""": """encoder.layers.*.final_layer_norm""",
"""encoder.layer_norm""": """encoder.layer_norm""",
"""encoder.layer_norm_for_extract""": """layer_norm_for_extract""",
"""w2v_model.layer_norm""": """feature_projection.layer_norm""",
"""quantizer.weight_proj""": """quantizer.weight_proj""",
"""quantizer.vars""": """quantizer.codevectors""",
"""project_q""": """project_q""",
"""final_proj""": """project_hid""",
"""w2v_encoder.proj""": """lm_head""",
"""label_embs_concat""": """label_embeddings_concat""",
"""mask_emb""": """masked_spec_embed""",
"""spk_proj""": """speaker_proj""",
}
a_ = [
"""lm_head""",
"""quantizer.weight_proj""",
"""quantizer.codevectors""",
"""project_q""",
"""project_hid""",
"""label_embeddings_concat""",
"""speaker_proj""",
"""layer_norm_for_extract""",
]
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : Tuple ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Dict ):
for attribute in key.split('''.''' ):
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if weight_type is not None:
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase ).shape
else:
__lowerCamelCase = hf_pointer.shape
if hf_shape != value.shape:
raise ValueError(
F"""Shape of hf {key + "." + weight_type if weight_type is not None else ""} is {hf_shape}, but should be"""
F""" {value.shape} for {full_name}""" )
if weight_type == "weight":
__lowerCamelCase = value
elif weight_type == "weight_g":
__lowerCamelCase = value
elif weight_type == "weight_v":
__lowerCamelCase = value
elif weight_type == "bias":
__lowerCamelCase = value
else:
__lowerCamelCase = value
logger.info(F"""{key + "." + weight_type if weight_type is not None else ""} was initialized from {full_name}.""" )
def a__ ( _UpperCamelCase : Any ,_UpperCamelCase : Any ):
__lowerCamelCase = []
__lowerCamelCase = fairseq_model.state_dict()
__lowerCamelCase = hf_model.unispeech_sat.feature_extractor
for name, value in fairseq_dict.items():
__lowerCamelCase = False
if "conv_layers" in name:
load_conv_layer(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,hf_model.config.feat_extract_norm == '''group''' ,)
__lowerCamelCase = True
else:
for key, mapped_key in MAPPING.items():
__lowerCamelCase = '''unispeech_sat.''' + mapped_key if mapped_key not in TOP_LEVEL_KEYS else mapped_key
if key in name or key.split('''w2v_model.''' )[-1] == name.split('''.''' )[0]:
if "layer_norm_for_extract" in name and (".".join(name.split('''.''' )[:-1] ) != key):
# special case since naming is very similar
continue
__lowerCamelCase = True
if "*" in mapped_key:
__lowerCamelCase = name.split(_UpperCamelCase )[0].split('''.''' )[-2]
__lowerCamelCase = mapped_key.replace('''*''' ,_UpperCamelCase )
if "weight_g" in name:
__lowerCamelCase = '''weight_g'''
elif "weight_v" in name:
__lowerCamelCase = '''weight_v'''
elif "bias" in name:
__lowerCamelCase = '''bias'''
elif "weight" in name:
# TODO: don't match quantizer.weight_proj
__lowerCamelCase = '''weight'''
else:
__lowerCamelCase = None
set_recursively(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
continue
if not is_used:
unused_weights.append(_UpperCamelCase )
logger.warning(F"""Unused weights: {unused_weights}""" )
def a__ ( _UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Dict ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = full_name.split('''conv_layers.''' )[-1]
__lowerCamelCase = name.split('''.''' )
__lowerCamelCase = int(items[0] )
__lowerCamelCase = int(items[1] )
if type_id == 0:
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif (type_id == 2 and not use_group_norm) or (type_id == 2 and layer_id == 0 and use_group_norm):
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(_UpperCamelCase )
@torch.no_grad()
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : List[str] ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : List[Any]=True ):
if config_path is not None:
__lowerCamelCase = UniSpeechSatConfig.from_pretrained(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatConfig()
__lowerCamelCase = ''''''
if is_finetuned:
__lowerCamelCase = UniSpeechSatForCTC(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatForPreTraining(_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
__lowerCamelCase = model[0].eval()
recursively_load_weights(_UpperCamelCase ,_UpperCamelCase )
hf_wavavec.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument("""--checkpoint_path""", default=None, type=str, help="""Path to fairseq checkpoint""")
parser.add_argument("""--dict_path""", default=None, type=str, help="""Path to dict of fine-tuned model""")
parser.add_argument("""--config_path""", default=None, type=str, help="""Path to hf config.json of model to convert""")
parser.add_argument(
"""--not_finetuned""", action="""store_true""", help="""Whether the model to convert is a fine-tuned model or not"""
)
a_ = parser.parse_args()
convert_unispeech_sat_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 330 | 1 |
import argparse
import re
import numpy as np
import requests
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import (
SamConfig,
SamImageProcessor,
SamModel,
SamProcessor,
SamVisionConfig,
)
a_ = {
"""iou_prediction_head.layers.0""": """iou_prediction_head.proj_in""",
"""iou_prediction_head.layers.1""": """iou_prediction_head.layers.0""",
"""iou_prediction_head.layers.2""": """iou_prediction_head.proj_out""",
"""mask_decoder.output_upscaling.0""": """mask_decoder.upscale_conv1""",
"""mask_decoder.output_upscaling.1""": """mask_decoder.upscale_layer_norm""",
"""mask_decoder.output_upscaling.3""": """mask_decoder.upscale_conv2""",
"""mask_downscaling.0""": """mask_embed.conv1""",
"""mask_downscaling.1""": """mask_embed.layer_norm1""",
"""mask_downscaling.3""": """mask_embed.conv2""",
"""mask_downscaling.4""": """mask_embed.layer_norm2""",
"""mask_downscaling.6""": """mask_embed.conv3""",
"""point_embeddings""": """point_embed""",
"""pe_layer.positional_encoding_gaussian_matrix""": """shared_embedding.positional_embedding""",
"""image_encoder""": """vision_encoder""",
"""neck.0""": """neck.conv1""",
"""neck.1""": """neck.layer_norm1""",
"""neck.2""": """neck.conv2""",
"""neck.3""": """neck.layer_norm2""",
"""patch_embed.proj""": """patch_embed.projection""",
""".norm""": """.layer_norm""",
"""blocks""": """layers""",
}
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = {}
state_dict.pop('''pixel_mean''' ,_UpperCamelCase )
state_dict.pop('''pixel_std''' ,_UpperCamelCase )
__lowerCamelCase = R'''.*.output_hypernetworks_mlps.(\d+).layers.(\d+).*'''
for key, value in state_dict.items():
for key_to_modify, new_key in KEYS_TO_MODIFY_MAPPING.items():
if key_to_modify in key:
__lowerCamelCase = key.replace(_UpperCamelCase ,_UpperCamelCase )
if re.match(_UpperCamelCase ,_UpperCamelCase ):
__lowerCamelCase = int(re.match(_UpperCamelCase ,_UpperCamelCase ).group(2 ) )
if layer_nb == 0:
__lowerCamelCase = key.replace('''layers.0''' ,'''proj_in''' )
elif layer_nb == 1:
__lowerCamelCase = key.replace('''layers.1''' ,'''layers.0''' )
elif layer_nb == 2:
__lowerCamelCase = key.replace('''layers.2''' ,'''proj_out''' )
__lowerCamelCase = value
__lowerCamelCase = model_state_dict[
'''prompt_encoder.shared_embedding.positional_embedding'''
]
return model_state_dict
def a__ ( _UpperCamelCase : Optional[Any] ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Dict ,_UpperCamelCase : List[str]="ybelkada/segment-anything" ):
__lowerCamelCase = hf_hub_download(_UpperCamelCase ,F"""checkpoints/{model_name}.pth""" )
if "sam_vit_b" in model_name:
__lowerCamelCase = SamConfig()
elif "sam_vit_l" in model_name:
__lowerCamelCase = SamVisionConfig(
hidden_size=10_24 ,num_hidden_layers=24 ,num_attention_heads=16 ,global_attn_indexes=[5, 11, 17, 23] ,)
__lowerCamelCase = SamConfig(
vision_config=_UpperCamelCase ,)
elif "sam_vit_h" in model_name:
__lowerCamelCase = SamVisionConfig(
hidden_size=12_80 ,num_hidden_layers=32 ,num_attention_heads=16 ,global_attn_indexes=[7, 15, 23, 31] ,)
__lowerCamelCase = SamConfig(
vision_config=_UpperCamelCase ,)
__lowerCamelCase = torch.load(_UpperCamelCase ,map_location='''cpu''' )
__lowerCamelCase = replace_keys(_UpperCamelCase )
__lowerCamelCase = SamImageProcessor()
__lowerCamelCase = SamProcessor(image_processor=_UpperCamelCase )
__lowerCamelCase = SamModel(_UpperCamelCase )
hf_model.load_state_dict(_UpperCamelCase )
__lowerCamelCase = hf_model.to('''cuda''' )
__lowerCamelCase = '''https://huggingface.co/ybelkada/segment-anything/resolve/main/assets/car.png'''
__lowerCamelCase = Image.open(requests.get(_UpperCamelCase ,stream=_UpperCamelCase ).raw ).convert('''RGB''' )
__lowerCamelCase = [[[4_00, 6_50]]]
__lowerCamelCase = [[1]]
__lowerCamelCase = processor(images=np.array(_UpperCamelCase ) ,return_tensors='''pt''' ).to('''cuda''' )
with torch.no_grad():
__lowerCamelCase = hf_model(**_UpperCamelCase )
__lowerCamelCase = output.iou_scores.squeeze()
if model_name == "sam_vit_h_4b8939":
assert scores[-1].item() == 0.579_890_251_159_668
__lowerCamelCase = processor(
images=np.array(_UpperCamelCase ) ,input_points=_UpperCamelCase ,input_labels=_UpperCamelCase ,return_tensors='''pt''' ).to('''cuda''' )
with torch.no_grad():
__lowerCamelCase = hf_model(**_UpperCamelCase )
__lowerCamelCase = output.iou_scores.squeeze()
assert scores[-1].item() == 0.9_712_603_092_193_604
__lowerCamelCase = ((75, 2_75, 17_25, 8_50),)
__lowerCamelCase = processor(images=np.array(_UpperCamelCase ) ,input_boxes=_UpperCamelCase ,return_tensors='''pt''' ).to('''cuda''' )
with torch.no_grad():
__lowerCamelCase = hf_model(**_UpperCamelCase )
__lowerCamelCase = output.iou_scores.squeeze()
assert scores[-1].item() == 0.8_686_015_605_926_514
# Test with 2 points and 1 image.
__lowerCamelCase = [[[4_00, 6_50], [8_00, 6_50]]]
__lowerCamelCase = [[1, 1]]
__lowerCamelCase = processor(
images=np.array(_UpperCamelCase ) ,input_points=_UpperCamelCase ,input_labels=_UpperCamelCase ,return_tensors='''pt''' ).to('''cuda''' )
with torch.no_grad():
__lowerCamelCase = hf_model(**_UpperCamelCase )
__lowerCamelCase = output.iou_scores.squeeze()
assert scores[-1].item() == 0.9_936_047_792_434_692
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
a_ = ["""sam_vit_b_01ec64""", """sam_vit_h_4b8939""", """sam_vit_l_0b3195"""]
parser.add_argument(
"""--model_name""",
default="""sam_vit_h_4b8939""",
choices=choices,
type=str,
help="""Path to hf config.json of model to convert""",
)
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument(
"""--push_to_hub""",
action="""store_true""",
help="""Whether to push the model and processor to the hub after converting""",
)
parser.add_argument(
"""--model_hub_id""",
default="""ybelkada/segment-anything""",
choices=choices,
type=str,
help="""Path to hf config.json of model to convert""",
)
a_ = parser.parse_args()
convert_sam_checkpoint(args.model_name, args.pytorch_dump_folder_path, args.push_to_hub, args.model_hub_id)
| 330 |
from typing import List, Union
import numpy as np
from ..utils import add_end_docstrings, is_torch_available, is_vision_available, logging, requires_backends
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_torch_available():
import torch
from ..models.auto.modeling_auto import MODEL_FOR_DEPTH_ESTIMATION_MAPPING
a_ = logging.get_logger(__name__)
@add_end_docstrings(lowerCAmelCase__ )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , *__UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
super().__init__(*__UpperCAmelCase , **__UpperCAmelCase )
requires_backends(self , '''vision''' )
self.check_model_type(__UpperCAmelCase )
def __call__( self , __UpperCAmelCase , **__UpperCAmelCase ):
'''simple docstring'''
return super().__call__(__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
return {}, {}, {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = load_image(__UpperCAmelCase )
__lowerCamelCase = image.size
__lowerCamelCase = self.image_processor(images=__UpperCAmelCase , return_tensors=self.framework )
return model_inputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.model(**__UpperCAmelCase )
return model_outputs
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = model_outputs.predicted_depth
__lowerCamelCase = torch.nn.functional.interpolate(
predicted_depth.unsqueeze(1 ) , size=self.image_size[::-1] , mode='''bicubic''' , align_corners=__UpperCAmelCase )
__lowerCamelCase = prediction.squeeze().cpu().numpy()
__lowerCamelCase = (output * 255 / np.max(__UpperCAmelCase )).astype('''uint8''' )
__lowerCamelCase = Image.fromarray(__UpperCAmelCase )
__lowerCamelCase = {}
__lowerCamelCase = predicted_depth
__lowerCamelCase = depth
return output_dict
| 330 | 1 |
import argparse
import fairseq
import torch
from transformers import UniSpeechSatConfig, UniSpeechSatForCTC, UniSpeechSatForPreTraining, logging
logging.set_verbosity_info()
a_ = logging.get_logger(__name__)
a_ = {
"""post_extract_proj""": """feature_projection.projection""",
"""encoder.pos_conv.0""": """encoder.pos_conv_embed.conv""",
"""self_attn.k_proj""": """encoder.layers.*.attention.k_proj""",
"""self_attn.v_proj""": """encoder.layers.*.attention.v_proj""",
"""self_attn.q_proj""": """encoder.layers.*.attention.q_proj""",
"""self_attn.out_proj""": """encoder.layers.*.attention.out_proj""",
"""self_attn_layer_norm""": """encoder.layers.*.layer_norm""",
"""fc1""": """encoder.layers.*.feed_forward.intermediate_dense""",
"""fc2""": """encoder.layers.*.feed_forward.output_dense""",
"""final_layer_norm""": """encoder.layers.*.final_layer_norm""",
"""encoder.layer_norm""": """encoder.layer_norm""",
"""encoder.layer_norm_for_extract""": """layer_norm_for_extract""",
"""w2v_model.layer_norm""": """feature_projection.layer_norm""",
"""quantizer.weight_proj""": """quantizer.weight_proj""",
"""quantizer.vars""": """quantizer.codevectors""",
"""project_q""": """project_q""",
"""final_proj""": """project_hid""",
"""w2v_encoder.proj""": """lm_head""",
"""label_embs_concat""": """label_embeddings_concat""",
"""mask_emb""": """masked_spec_embed""",
"""spk_proj""": """speaker_proj""",
}
a_ = [
"""lm_head""",
"""quantizer.weight_proj""",
"""quantizer.codevectors""",
"""project_q""",
"""project_hid""",
"""label_embeddings_concat""",
"""speaker_proj""",
"""layer_norm_for_extract""",
]
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : Tuple ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Dict ):
for attribute in key.split('''.''' ):
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if weight_type is not None:
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase ).shape
else:
__lowerCamelCase = hf_pointer.shape
if hf_shape != value.shape:
raise ValueError(
F"""Shape of hf {key + "." + weight_type if weight_type is not None else ""} is {hf_shape}, but should be"""
F""" {value.shape} for {full_name}""" )
if weight_type == "weight":
__lowerCamelCase = value
elif weight_type == "weight_g":
__lowerCamelCase = value
elif weight_type == "weight_v":
__lowerCamelCase = value
elif weight_type == "bias":
__lowerCamelCase = value
else:
__lowerCamelCase = value
logger.info(F"""{key + "." + weight_type if weight_type is not None else ""} was initialized from {full_name}.""" )
def a__ ( _UpperCamelCase : Any ,_UpperCamelCase : Any ):
__lowerCamelCase = []
__lowerCamelCase = fairseq_model.state_dict()
__lowerCamelCase = hf_model.unispeech_sat.feature_extractor
for name, value in fairseq_dict.items():
__lowerCamelCase = False
if "conv_layers" in name:
load_conv_layer(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,hf_model.config.feat_extract_norm == '''group''' ,)
__lowerCamelCase = True
else:
for key, mapped_key in MAPPING.items():
__lowerCamelCase = '''unispeech_sat.''' + mapped_key if mapped_key not in TOP_LEVEL_KEYS else mapped_key
if key in name or key.split('''w2v_model.''' )[-1] == name.split('''.''' )[0]:
if "layer_norm_for_extract" in name and (".".join(name.split('''.''' )[:-1] ) != key):
# special case since naming is very similar
continue
__lowerCamelCase = True
if "*" in mapped_key:
__lowerCamelCase = name.split(_UpperCamelCase )[0].split('''.''' )[-2]
__lowerCamelCase = mapped_key.replace('''*''' ,_UpperCamelCase )
if "weight_g" in name:
__lowerCamelCase = '''weight_g'''
elif "weight_v" in name:
__lowerCamelCase = '''weight_v'''
elif "bias" in name:
__lowerCamelCase = '''bias'''
elif "weight" in name:
# TODO: don't match quantizer.weight_proj
__lowerCamelCase = '''weight'''
else:
__lowerCamelCase = None
set_recursively(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
continue
if not is_used:
unused_weights.append(_UpperCamelCase )
logger.warning(F"""Unused weights: {unused_weights}""" )
def a__ ( _UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Dict ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = full_name.split('''conv_layers.''' )[-1]
__lowerCamelCase = name.split('''.''' )
__lowerCamelCase = int(items[0] )
__lowerCamelCase = int(items[1] )
if type_id == 0:
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif (type_id == 2 and not use_group_norm) or (type_id == 2 and layer_id == 0 and use_group_norm):
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor[layer_id].layer_norm.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(_UpperCamelCase )
@torch.no_grad()
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : List[str] ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : List[Any]=True ):
if config_path is not None:
__lowerCamelCase = UniSpeechSatConfig.from_pretrained(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatConfig()
__lowerCamelCase = ''''''
if is_finetuned:
__lowerCamelCase = UniSpeechSatForCTC(_UpperCamelCase )
else:
__lowerCamelCase = UniSpeechSatForPreTraining(_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
__lowerCamelCase = model[0].eval()
recursively_load_weights(_UpperCamelCase ,_UpperCamelCase )
hf_wavavec.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument("""--checkpoint_path""", default=None, type=str, help="""Path to fairseq checkpoint""")
parser.add_argument("""--dict_path""", default=None, type=str, help="""Path to dict of fine-tuned model""")
parser.add_argument("""--config_path""", default=None, type=str, help="""Path to hf config.json of model to convert""")
parser.add_argument(
"""--not_finetuned""", action="""store_true""", help="""Whether the model to convert is a fine-tuned model or not"""
)
a_ = parser.parse_args()
convert_unispeech_sat_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 330 |
from typing import Dict, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
convert_to_rgb,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
OPENAI_CLIP_MEAN,
OPENAI_CLIP_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_vision_available, logging
a_ = logging.get_logger(__name__)
if is_vision_available():
import PIL
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = ["""pixel_values"""]
def __init__( self , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = True , __UpperCAmelCase = 1 / 255 , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = True , **__UpperCAmelCase , ):
'''simple docstring'''
super().__init__(**__UpperCAmelCase )
__lowerCamelCase = size if size is not None else {'''shortest_edge''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
__lowerCamelCase = crop_size if crop_size is not None else {'''height''': 224, '''width''': 224}
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase , 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 OPENAI_CLIP_MEAN
__lowerCamelCase = image_std if image_std is not None else OPENAI_CLIP_STD
__lowerCamelCase = do_convert_rgb
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = PILImageResampling.BICUBIC , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase , default_to_square=__UpperCAmelCase )
if "shortest_edge" not in size:
raise ValueError(F"""The `size` parameter must contain the key `shortest_edge`. Got {size.keys()}""" )
__lowerCamelCase = get_resize_output_image_size(__UpperCAmelCase , size=size['''shortest_edge'''] , default_to_square=__UpperCAmelCase )
return resize(__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = get_size_dict(__UpperCAmelCase )
if "height" not in size or "width" not in size:
raise ValueError(F"""The `size` parameter must contain the keys (height, width). Got {size.keys()}""" )
return center_crop(__UpperCAmelCase , size=(size['''height'''], size['''width''']) , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return rescale(__UpperCAmelCase , scale=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase = None , **__UpperCAmelCase , ):
'''simple docstring'''
return normalize(__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase , data_format=__UpperCAmelCase , **__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = ChannelDimension.FIRST , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = do_resize if do_resize is not None else self.do_resize
__lowerCamelCase = size if size is not None else self.size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''size''' , default_to_square=__UpperCAmelCase )
__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 = crop_size if crop_size is not None else self.crop_size
__lowerCamelCase = get_size_dict(__UpperCAmelCase , param_name='''crop_size''' , default_to_square=__UpperCAmelCase )
__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 = do_convert_rgb if do_convert_rgb is not None else self.do_convert_rgb
__lowerCamelCase = make_list_of_images(__UpperCAmelCase )
if not valid_images(__UpperCAmelCase ):
raise ValueError(
'''Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '''
'''torch.Tensor, tf.Tensor or jax.ndarray.''' )
if do_resize and size is None:
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.''' )
# PIL RGBA images are converted to RGB
if do_convert_rgb:
__lowerCamelCase = [convert_to_rgb(__UpperCAmelCase ) for image in images]
# All transformations expect numpy arrays.
__lowerCamelCase = [to_numpy_array(__UpperCAmelCase ) for image in images]
if do_resize:
__lowerCamelCase = [self.resize(image=__UpperCAmelCase , size=__UpperCAmelCase , resample=__UpperCAmelCase ) for image in images]
if do_center_crop:
__lowerCamelCase = [self.center_crop(image=__UpperCAmelCase , size=__UpperCAmelCase ) for image in images]
if do_rescale:
__lowerCamelCase = [self.rescale(image=__UpperCAmelCase , scale=__UpperCAmelCase ) for image in images]
if do_normalize:
__lowerCamelCase = [self.normalize(image=__UpperCAmelCase , mean=__UpperCAmelCase , std=__UpperCAmelCase ) for image in images]
__lowerCamelCase = [to_channel_dimension_format(__UpperCAmelCase , __UpperCAmelCase ) for image in images]
__lowerCamelCase = {'''pixel_values''': images}
return BatchFeature(data=__UpperCAmelCase , tensor_type=__UpperCAmelCase )
| 330 | 1 |
a_ = {
"""Pillow""": """Pillow<10.0.0""",
"""accelerate""": """accelerate>=0.20.3""",
"""av""": """av==9.2.0""",
"""beautifulsoup4""": """beautifulsoup4""",
"""black""": """black~=23.1""",
"""codecarbon""": """codecarbon==1.2.0""",
"""cookiecutter""": """cookiecutter==1.7.3""",
"""dataclasses""": """dataclasses""",
"""datasets""": """datasets!=2.5.0""",
"""decord""": """decord==0.6.0""",
"""deepspeed""": """deepspeed>=0.9.3""",
"""diffusers""": """diffusers""",
"""dill""": """dill<0.3.5""",
"""evaluate""": """evaluate>=0.2.0""",
"""fairscale""": """fairscale>0.3""",
"""faiss-cpu""": """faiss-cpu""",
"""fastapi""": """fastapi""",
"""filelock""": """filelock""",
"""flax""": """flax>=0.4.1,<=0.7.0""",
"""ftfy""": """ftfy""",
"""fugashi""": """fugashi>=1.0""",
"""GitPython""": """GitPython<3.1.19""",
"""hf-doc-builder""": """hf-doc-builder>=0.3.0""",
"""huggingface-hub""": """huggingface-hub>=0.14.1,<1.0""",
"""importlib_metadata""": """importlib_metadata""",
"""ipadic""": """ipadic>=1.0.0,<2.0""",
"""isort""": """isort>=5.5.4""",
"""jax""": """jax>=0.2.8,!=0.3.2,<=0.4.13""",
"""jaxlib""": """jaxlib>=0.1.65,<=0.4.13""",
"""jieba""": """jieba""",
"""kenlm""": """kenlm""",
"""keras-nlp""": """keras-nlp>=0.3.1""",
"""librosa""": """librosa""",
"""nltk""": """nltk""",
"""natten""": """natten>=0.14.6""",
"""numpy""": """numpy>=1.17""",
"""onnxconverter-common""": """onnxconverter-common""",
"""onnxruntime-tools""": """onnxruntime-tools>=1.4.2""",
"""onnxruntime""": """onnxruntime>=1.4.0""",
"""opencv-python""": """opencv-python""",
"""optuna""": """optuna""",
"""optax""": """optax>=0.0.8,<=0.1.4""",
"""packaging""": """packaging>=20.0""",
"""parameterized""": """parameterized""",
"""phonemizer""": """phonemizer""",
"""protobuf""": """protobuf""",
"""psutil""": """psutil""",
"""pyyaml""": """pyyaml>=5.1""",
"""pydantic""": """pydantic<2""",
"""pytest""": """pytest>=7.2.0""",
"""pytest-timeout""": """pytest-timeout""",
"""pytest-xdist""": """pytest-xdist""",
"""python""": """python>=3.8.0""",
"""ray[tune]""": """ray[tune]""",
"""regex""": """regex!=2019.12.17""",
"""requests""": """requests""",
"""rhoknp""": """rhoknp>=1.1.0,<1.3.1""",
"""rjieba""": """rjieba""",
"""rouge-score""": """rouge-score!=0.0.7,!=0.0.8,!=0.1,!=0.1.1""",
"""ruff""": """ruff>=0.0.241,<=0.0.259""",
"""sacrebleu""": """sacrebleu>=1.4.12,<2.0.0""",
"""sacremoses""": """sacremoses""",
"""safetensors""": """safetensors>=0.3.1""",
"""sagemaker""": """sagemaker>=2.31.0""",
"""scikit-learn""": """scikit-learn""",
"""sentencepiece""": """sentencepiece>=0.1.91,!=0.1.92""",
"""sigopt""": """sigopt""",
"""starlette""": """starlette""",
"""sudachipy""": """sudachipy>=0.6.6""",
"""sudachidict_core""": """sudachidict_core>=20220729""",
"""tensorflow-cpu""": """tensorflow-cpu>=2.6,<2.14""",
"""tensorflow""": """tensorflow>=2.6,<2.14""",
"""tensorflow-text""": """tensorflow-text<2.14""",
"""tf2onnx""": """tf2onnx""",
"""timeout-decorator""": """timeout-decorator""",
"""timm""": """timm""",
"""tokenizers""": """tokenizers>=0.11.1,!=0.11.3,<0.14""",
"""torch""": """torch>=1.9,!=1.12.0""",
"""torchaudio""": """torchaudio""",
"""torchvision""": """torchvision""",
"""pyctcdecode""": """pyctcdecode>=0.4.0""",
"""tqdm""": """tqdm>=4.27""",
"""unidic""": """unidic>=1.0.2""",
"""unidic_lite""": """unidic_lite>=1.0.7""",
"""urllib3""": """urllib3<2.0.0""",
"""uvicorn""": """uvicorn""",
}
| 330 |
from __future__ import annotations
from typing import Generic, TypeVar
a_ = TypeVar("""T""")
class __lowerCAmelCase ( Generic[T] ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = data
__lowerCamelCase = self
__lowerCamelCase = 0
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# map from node name to the node object
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# create a new set with x as its member
__lowerCamelCase = DisjointSetTreeNode(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# find the set x belongs to (with path-compression)
__lowerCamelCase = self.map[data]
if elem_ref != elem_ref.parent:
__lowerCamelCase = self.find_set(elem_ref.parent.data )
return elem_ref.parent
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# helper function for union operation
if nodea.rank > nodea.rank:
__lowerCamelCase = nodea
else:
__lowerCamelCase = nodea
if nodea.rank == nodea.rank:
nodea.rank += 1
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# merge 2 disjoint sets
self.link(self.find_set(__UpperCAmelCase ) , self.find_set(__UpperCAmelCase ) )
class __lowerCAmelCase ( Generic[T] ):
def __init__( self ):
'''simple docstring'''
# connections: map from the node to the neighbouring nodes (with weights)
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
# add a node ONLY if its not present in the graph
if node not in self.connections:
__lowerCamelCase = {}
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
# add an edge with the given weight
self.add_node(__UpperCAmelCase )
self.add_node(__UpperCAmelCase )
__lowerCamelCase = weight
__lowerCamelCase = weight
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = []
__lowerCamelCase = set()
for start in self.connections:
for end in self.connections[start]:
if (start, end) not in seen:
seen.add((end, start) )
edges.append((start, end, self.connections[start][end]) )
edges.sort(key=lambda __UpperCAmelCase : x[2] )
# creating the disjoint set
__lowerCamelCase = DisjointSetTree[T]()
for node in self.connections:
disjoint_set.make_set(__UpperCAmelCase )
# MST generation
__lowerCamelCase = 0
__lowerCamelCase = 0
__lowerCamelCase = GraphUndirectedWeighted[T]()
while num_edges < len(self.connections ) - 1:
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = edges[index]
index += 1
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
__lowerCamelCase = disjoint_set.find_set(__UpperCAmelCase )
if parent_u != parent_v:
num_edges += 1
graph.add_edge(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
disjoint_set.union(__UpperCAmelCase , __UpperCAmelCase )
return graph
| 330 | 1 |
import json
import os
from functools import lru_cache
from typing import List, Optional, Tuple
import regex as re
from ...tokenization_utils import AddedToken, PreTrainedTokenizer
from ...utils import logging
a_ = logging.get_logger(__name__)
a_ = {"""vocab_file""": """vocab.json""", """merges_file""": """merges.txt"""}
a_ = {
"""vocab_file""": {
"""allenai/longformer-base-4096""": """https://huggingface.co/allenai/longformer-base-4096/resolve/main/vocab.json""",
"""allenai/longformer-large-4096""": (
"""https://huggingface.co/allenai/longformer-large-4096/resolve/main/vocab.json"""
),
"""allenai/longformer-large-4096-finetuned-triviaqa""": (
"""https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/vocab.json"""
),
"""allenai/longformer-base-4096-extra.pos.embd.only""": (
"""https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/vocab.json"""
),
"""allenai/longformer-large-4096-extra.pos.embd.only""": (
"""https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/vocab.json"""
),
},
"""merges_file""": {
"""allenai/longformer-base-4096""": """https://huggingface.co/allenai/longformer-base-4096/resolve/main/merges.txt""",
"""allenai/longformer-large-4096""": (
"""https://huggingface.co/allenai/longformer-large-4096/resolve/main/merges.txt"""
),
"""allenai/longformer-large-4096-finetuned-triviaqa""": (
"""https://huggingface.co/allenai/longformer-large-4096-finetuned-triviaqa/resolve/main/merges.txt"""
),
"""allenai/longformer-base-4096-extra.pos.embd.only""": (
"""https://huggingface.co/allenai/longformer-base-4096-extra.pos.embd.only/resolve/main/merges.txt"""
),
"""allenai/longformer-large-4096-extra.pos.embd.only""": (
"""https://huggingface.co/allenai/longformer-large-4096-extra.pos.embd.only/resolve/main/merges.txt"""
),
},
}
a_ = {
"""allenai/longformer-base-4096""": 4_096,
"""allenai/longformer-large-4096""": 4_096,
"""allenai/longformer-large-4096-finetuned-triviaqa""": 4_096,
"""allenai/longformer-base-4096-extra.pos.embd.only""": 4_096,
"""allenai/longformer-large-4096-extra.pos.embd.only""": 4_096,
}
@lru_cache()
# Copied from transformers.models.roberta.tokenization_roberta.bytes_to_unicode
def a__ ( ):
__lowerCamelCase = (
list(range(ord('''!''' ) ,ord('''~''' ) + 1 ) ) + list(range(ord('''¡''' ) ,ord('''¬''' ) + 1 ) ) + list(range(ord('''®''' ) ,ord('''ÿ''' ) + 1 ) )
)
__lowerCamelCase = bs[:]
__lowerCamelCase = 0
for b in range(2**8 ):
if b not in bs:
bs.append(_UpperCamelCase )
cs.append(2**8 + n )
n += 1
__lowerCamelCase = [chr(_UpperCamelCase ) for n in cs]
return dict(zip(_UpperCamelCase ,_UpperCamelCase ) )
def a__ ( _UpperCamelCase : List[str] ):
__lowerCamelCase = set()
__lowerCamelCase = word[0]
for char in word[1:]:
pairs.add((prev_char, char) )
__lowerCamelCase = char
return pairs
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = VOCAB_FILES_NAMES
lowerCAmelCase__ = PRETRAINED_VOCAB_FILES_MAP
lowerCAmelCase__ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
lowerCAmelCase__ = ["""input_ids""", """attention_mask"""]
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase="replace" , __UpperCAmelCase="<s>" , __UpperCAmelCase="</s>" , __UpperCAmelCase="</s>" , __UpperCAmelCase="<s>" , __UpperCAmelCase="<unk>" , __UpperCAmelCase="<pad>" , __UpperCAmelCase="<mask>" , __UpperCAmelCase=False , **__UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else bos_token
__lowerCamelCase = AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else eos_token
__lowerCamelCase = AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else sep_token
__lowerCamelCase = AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else cls_token
__lowerCamelCase = AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else unk_token
__lowerCamelCase = AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else pad_token
# Mask token behave like a normal word, i.e. include the space before it
__lowerCamelCase = AddedToken(__UpperCAmelCase , lstrip=__UpperCAmelCase , rstrip=__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else mask_token
super().__init__(
errors=__UpperCAmelCase , bos_token=__UpperCAmelCase , eos_token=__UpperCAmelCase , unk_token=__UpperCAmelCase , sep_token=__UpperCAmelCase , cls_token=__UpperCAmelCase , pad_token=__UpperCAmelCase , mask_token=__UpperCAmelCase , add_prefix_space=__UpperCAmelCase , **__UpperCAmelCase , )
with open(__UpperCAmelCase , encoding='''utf-8''' ) as vocab_handle:
__lowerCamelCase = json.load(__UpperCAmelCase )
__lowerCamelCase = {v: k for k, v in self.encoder.items()}
__lowerCamelCase = errors # how to handle errors in decoding
__lowerCamelCase = bytes_to_unicode()
__lowerCamelCase = {v: k for k, v in self.byte_encoder.items()}
with open(__UpperCAmelCase , encoding='''utf-8''' ) as merges_handle:
__lowerCamelCase = merges_handle.read().split('''\n''' )[1:-1]
__lowerCamelCase = [tuple(merge.split() ) for merge in bpe_merges]
__lowerCamelCase = dict(zip(__UpperCAmelCase , range(len(__UpperCAmelCase ) ) ) )
__lowerCamelCase = {}
__lowerCamelCase = add_prefix_space
# Should have added re.IGNORECASE so BPE merges can happen for capitalized versions of contractions
__lowerCamelCase = re.compile(r'''\'s|\'t|\'re|\'ve|\'m|\'ll|\'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+''' )
@property
def lowerCamelCase ( self ):
'''simple docstring'''
return len(self.encoder )
def lowerCamelCase ( self ):
'''simple docstring'''
return dict(self.encoder , **self.added_tokens_encoder )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if token in self.cache:
return self.cache[token]
__lowerCamelCase = tuple(__UpperCAmelCase )
__lowerCamelCase = get_pairs(__UpperCAmelCase )
if not pairs:
return token
while True:
__lowerCamelCase = min(__UpperCAmelCase , key=lambda __UpperCAmelCase : self.bpe_ranks.get(__UpperCAmelCase , float('''inf''' ) ) )
if bigram not in self.bpe_ranks:
break
__lowerCamelCase ,__lowerCamelCase = bigram
__lowerCamelCase = []
__lowerCamelCase = 0
while i < len(__UpperCAmelCase ):
try:
__lowerCamelCase = word.index(__UpperCAmelCase , __UpperCAmelCase )
except ValueError:
new_word.extend(word[i:] )
break
else:
new_word.extend(word[i:j] )
__lowerCamelCase = j
if word[i] == first and i < len(__UpperCAmelCase ) - 1 and word[i + 1] == second:
new_word.append(first + second )
i += 2
else:
new_word.append(word[i] )
i += 1
__lowerCamelCase = tuple(__UpperCAmelCase )
__lowerCamelCase = new_word
if len(__UpperCAmelCase ) == 1:
break
else:
__lowerCamelCase = get_pairs(__UpperCAmelCase )
__lowerCamelCase = ''' '''.join(__UpperCAmelCase )
__lowerCamelCase = word
return word
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = []
for token in re.findall(self.pat , __UpperCAmelCase ):
__lowerCamelCase = ''''''.join(
self.byte_encoder[b] for b in token.encode('''utf-8''' ) ) # Maps all our bytes to unicode strings, avoiding control tokens of the BPE (spaces in our case)
bpe_tokens.extend(bpe_token for bpe_token in self.bpe(__UpperCAmelCase ).split(''' ''' ) )
return bpe_tokens
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
return self.encoder.get(__UpperCAmelCase , self.encoder.get(self.unk_token ) )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
return self.decoder.get(__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = ''''''.join(__UpperCAmelCase )
__lowerCamelCase = bytearray([self.byte_decoder[c] for c in text] ).decode('''utf-8''' , errors=self.errors )
return text
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None ):
'''simple docstring'''
if not os.path.isdir(__UpperCAmelCase ):
logger.error(F"""Vocabulary path ({save_directory}) should be a directory""" )
return
__lowerCamelCase = os.path.join(
__UpperCAmelCase , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] )
__lowerCamelCase = os.path.join(
__UpperCAmelCase , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''merges_file'''] )
with open(__UpperCAmelCase , '''w''' , encoding='''utf-8''' ) as f:
f.write(json.dumps(self.encoder , indent=2 , sort_keys=__UpperCAmelCase , ensure_ascii=__UpperCAmelCase ) + '''\n''' )
__lowerCamelCase = 0
with open(__UpperCAmelCase , '''w''' , encoding='''utf-8''' ) as writer:
writer.write('''#version: 0.2\n''' )
for bpe_tokens, token_index in sorted(self.bpe_ranks.items() , key=lambda __UpperCAmelCase : kv[1] ):
if index != token_index:
logger.warning(
F"""Saving vocabulary to {merge_file}: BPE merge indices are not consecutive."""
''' Please check that the tokenizer is not corrupted!''' )
__lowerCamelCase = token_index
writer.write(''' '''.join(__UpperCAmelCase ) + '''\n''' )
index += 1
return vocab_file, merge_file
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None ):
'''simple docstring'''
if token_ids_a is None:
return [self.cls_token_id] + token_ids_a + [self.sep_token_id]
__lowerCamelCase = [self.cls_token_id]
__lowerCamelCase = [self.sep_token_id]
return cls + token_ids_a + sep + sep + token_ids_a + sep
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = False ):
'''simple docstring'''
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=__UpperCAmelCase , token_ids_a=__UpperCAmelCase , already_has_special_tokens=__UpperCAmelCase )
if token_ids_a is None:
return [1] + ([0] * len(__UpperCAmelCase )) + [1]
return [1] + ([0] * len(__UpperCAmelCase )) + [1, 1] + ([0] * len(__UpperCAmelCase )) + [1]
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase = None ):
'''simple docstring'''
__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 + sep + token_ids_a + sep ) * [0]
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase=False , **__UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = kwargs.pop('''add_prefix_space''' , self.add_prefix_space )
if (is_split_into_words or add_prefix_space) and (len(__UpperCAmelCase ) > 0 and not text[0].isspace()):
__lowerCamelCase = ''' ''' + text
return (text, kwargs)
| 330 |
import unittest
from transformers import MraConfig, is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
if is_torch_available():
import torch
from transformers import (
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
MraModel,
)
from transformers.models.mra.modeling_mra import MRA_PRETRAINED_MODEL_ARCHIVE_LIST
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=2 , __UpperCAmelCase=8 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=99 , __UpperCAmelCase=16 , __UpperCAmelCase=5 , __UpperCAmelCase=2 , __UpperCAmelCase=36 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.0 , __UpperCAmelCase=0.0 , __UpperCAmelCase=512 , __UpperCAmelCase=16 , __UpperCAmelCase=2 , __UpperCAmelCase=0.02 , __UpperCAmelCase=3 , __UpperCAmelCase=4 , __UpperCAmelCase=None , ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
__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 lowerCamelCase ( self ):
'''simple docstring'''
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=__UpperCAmelCase , initializer_range=self.initializer_range , )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_config()
__lowerCamelCase = 300
return config
def lowerCamelCase ( self ):
'''simple docstring'''
(
(
__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 lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
__lowerCamelCase = True
__lowerCamelCase = MraModel(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , encoder_attention_mask=__UpperCAmelCase , )
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , )
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = MraForQuestionAnswering(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , start_positions=__UpperCAmelCase , end_positions=__UpperCAmelCase , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForSequenceClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_labels
__lowerCamelCase = MraForTokenClassification(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.num_choices
__lowerCamelCase = MraForMultipleChoice(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
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(
__UpperCAmelCase , attention_mask=__UpperCAmelCase , token_type_ids=__UpperCAmelCase , labels=__UpperCAmelCase , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowerCamelCase ( self ):
'''simple docstring'''
__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 __lowerCAmelCase ( lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (
(
MraModel,
MraForMaskedLM,
MraForMultipleChoice,
MraForQuestionAnswering,
MraForSequenceClassification,
MraForTokenClassification,
)
if is_torch_available()
else ()
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = ()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModelTester(self )
__lowerCamelCase = ConfigTester(self , config_class=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__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(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in MRA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = MraModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
@unittest.skip(reason='''MRA does not output attentions''' )
def lowerCamelCase ( self ):
'''simple docstring'''
return
@require_torch
class __lowerCAmelCase ( unittest.TestCase ):
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraModel.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = torch.Size((1, 256, 768) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[-0.0_140, 0.0_830, -0.0_381], [0.1_546, 0.1_402, 0.0_220], [0.1_162, 0.0_851, 0.0_165]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-512-4''' )
__lowerCamelCase = torch.arange(256 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 256, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[9.2_595, -3.6_038, 11.8_819], [9.3_869, -3.2_693, 11.0_956], [11.8_524, -3.4_938, 13.1_210]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = MraForMaskedLM.from_pretrained('''uw-madison/mra-base-4096-8-d3''' )
__lowerCamelCase = torch.arange(4096 ).unsqueeze(0 )
with torch.no_grad():
__lowerCamelCase = model(__UpperCAmelCase )[0]
__lowerCamelCase = 50265
__lowerCamelCase = torch.Size((1, 4096, vocab_size) )
self.assertEqual(output.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor(
[[[5.4_789, -2.3_564, 7.5_064], [7.9_067, -1.3_369, 9.9_668], [9.0_712, -1.8_106, 7.0_380]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 | 1 |
import argparse
import json
import math
import os
import time
import traceback
import zipfile
from collections import Counter
import requests
def a__ ( _UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Optional[Any]=None ):
__lowerCamelCase = None
if token is not None:
__lowerCamelCase = {'''Accept''': '''application/vnd.github+json''', '''Authorization''': F"""Bearer {token}"""}
__lowerCamelCase = F"""https://api.github.com/repos/huggingface/transformers/actions/runs/{workflow_run_id}/jobs?per_page=100"""
__lowerCamelCase = requests.get(_UpperCamelCase ,headers=_UpperCamelCase ).json()
__lowerCamelCase = {}
try:
job_links.update({job['''name''']: job['''html_url'''] for job in result['''jobs''']} )
__lowerCamelCase = math.ceil((result['''total_count'''] - 1_00) / 1_00 )
for i in range(_UpperCamelCase ):
__lowerCamelCase = requests.get(url + F"""&page={i + 2}""" ,headers=_UpperCamelCase ).json()
job_links.update({job['''name''']: job['''html_url'''] for job in result['''jobs''']} )
return job_links
except Exception:
print(F"""Unknown error, could not fetch links:\n{traceback.format_exc()}""" )
return {}
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : int=None ):
__lowerCamelCase = None
if token is not None:
__lowerCamelCase = {'''Accept''': '''application/vnd.github+json''', '''Authorization''': F"""Bearer {token}"""}
__lowerCamelCase = F"""https://api.github.com/repos/huggingface/transformers/actions/runs/{worflow_run_id}/artifacts?per_page=100"""
__lowerCamelCase = requests.get(_UpperCamelCase ,headers=_UpperCamelCase ).json()
__lowerCamelCase = {}
try:
artifacts.update({artifact['''name''']: artifact['''archive_download_url'''] for artifact in result['''artifacts''']} )
__lowerCamelCase = math.ceil((result['''total_count'''] - 1_00) / 1_00 )
for i in range(_UpperCamelCase ):
__lowerCamelCase = requests.get(url + F"""&page={i + 2}""" ,headers=_UpperCamelCase ).json()
artifacts.update({artifact['''name''']: artifact['''archive_download_url'''] for artifact in result['''artifacts''']} )
return artifacts
except Exception:
print(F"""Unknown error, could not fetch links:\n{traceback.format_exc()}""" )
return {}
def a__ ( _UpperCamelCase : List[str] ,_UpperCamelCase : Any ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[int] ):
__lowerCamelCase = None
if token is not None:
__lowerCamelCase = {'''Accept''': '''application/vnd.github+json''', '''Authorization''': F"""Bearer {token}"""}
__lowerCamelCase = requests.get(_UpperCamelCase ,headers=_UpperCamelCase ,allow_redirects=_UpperCamelCase )
__lowerCamelCase = result.headers['''Location''']
__lowerCamelCase = requests.get(_UpperCamelCase ,allow_redirects=_UpperCamelCase )
__lowerCamelCase = os.path.join(_UpperCamelCase ,F"""{artifact_name}.zip""" )
with open(_UpperCamelCase ,'''wb''' ) as fp:
fp.write(response.content )
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : List[str]=None ):
__lowerCamelCase = []
__lowerCamelCase = []
__lowerCamelCase = None
with zipfile.ZipFile(_UpperCamelCase ) as z:
for filename in z.namelist():
if not os.path.isdir(_UpperCamelCase ):
# read the file
if filename in ["failures_line.txt", "summary_short.txt", "job_name.txt"]:
with z.open(_UpperCamelCase ) as f:
for line in f:
__lowerCamelCase = line.decode('''UTF-8''' ).strip()
if filename == "failures_line.txt":
try:
# `error_line` is the place where `error` occurs
__lowerCamelCase = line[: line.index(''': ''' )]
__lowerCamelCase = line[line.index(''': ''' ) + len(''': ''' ) :]
errors.append([error_line, error] )
except Exception:
# skip un-related lines
pass
elif filename == "summary_short.txt" and line.startswith('''FAILED ''' ):
# `test` is the test method that failed
__lowerCamelCase = line[len('''FAILED ''' ) :]
failed_tests.append(_UpperCamelCase )
elif filename == "job_name.txt":
__lowerCamelCase = line
if len(_UpperCamelCase ) != len(_UpperCamelCase ):
raise ValueError(
F"""`errors` and `failed_tests` should have the same number of elements. Got {len(_UpperCamelCase )} for `errors` """
F"""and {len(_UpperCamelCase )} for `failed_tests` instead. The test reports in {artifact_zip_path} have some"""
''' problem.''' )
__lowerCamelCase = None
if job_name and job_links:
__lowerCamelCase = job_links.get(_UpperCamelCase ,_UpperCamelCase )
# A list with elements of the form (line of error, error, failed test)
__lowerCamelCase = [x + [y] + [job_link] for x, y in zip(_UpperCamelCase ,_UpperCamelCase )]
return result
def a__ ( _UpperCamelCase : Tuple ,_UpperCamelCase : Dict=None ):
__lowerCamelCase = []
__lowerCamelCase = [os.path.join(_UpperCamelCase ,_UpperCamelCase ) for p in os.listdir(_UpperCamelCase ) if p.endswith('''.zip''' )]
for p in paths:
errors.extend(get_errors_from_single_artifact(_UpperCamelCase ,job_links=_UpperCamelCase ) )
return errors
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : Union[str, Any]=None ):
__lowerCamelCase = Counter()
counter.update([x[1] for x in logs] )
__lowerCamelCase = counter.most_common()
__lowerCamelCase = {}
for error, count in counts:
if error_filter is None or error not in error_filter:
__lowerCamelCase = {'''count''': count, '''failed_tests''': [(x[2], x[0]) for x in logs if x[1] == error]}
__lowerCamelCase = dict(sorted(r.items() ,key=lambda _UpperCamelCase : item[1]["count"] ,reverse=_UpperCamelCase ) )
return r
def a__ ( _UpperCamelCase : Optional[int] ):
__lowerCamelCase = test.split('''::''' )[0]
if test.startswith('''tests/models/''' ):
__lowerCamelCase = test.split('''/''' )[2]
else:
__lowerCamelCase = None
return test
def a__ ( _UpperCamelCase : Tuple ,_UpperCamelCase : List[str]=None ):
__lowerCamelCase = [(x[0], x[1], get_model(x[2] )) for x in logs]
__lowerCamelCase = [x for x in logs if x[2] is not None]
__lowerCamelCase = {x[2] for x in logs}
__lowerCamelCase = {}
for test in tests:
__lowerCamelCase = Counter()
# count by errors in `test`
counter.update([x[1] for x in logs if x[2] == test] )
__lowerCamelCase = counter.most_common()
__lowerCamelCase = {error: count for error, count in counts if (error_filter is None or error not in error_filter)}
__lowerCamelCase = sum(error_counts.values() )
if n_errors > 0:
__lowerCamelCase = {'''count''': n_errors, '''errors''': error_counts}
__lowerCamelCase = dict(sorted(r.items() ,key=lambda _UpperCamelCase : item[1]["count"] ,reverse=_UpperCamelCase ) )
return r
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = '''| no. | error | status |'''
__lowerCamelCase = '''|-:|:-|:-|'''
__lowerCamelCase = [header, sep]
for error in reduced_by_error:
__lowerCamelCase = reduced_by_error[error]['''count''']
__lowerCamelCase = F"""| {count} | {error[:1_00]} | |"""
lines.append(_UpperCamelCase )
return "\n".join(_UpperCamelCase )
def a__ ( _UpperCamelCase : Any ):
__lowerCamelCase = '''| model | no. of errors | major error | count |'''
__lowerCamelCase = '''|-:|-:|-:|-:|'''
__lowerCamelCase = [header, sep]
for model in reduced_by_model:
__lowerCamelCase = reduced_by_model[model]['''count''']
__lowerCamelCase ,__lowerCamelCase = list(reduced_by_model[model]['''errors'''].items() )[0]
__lowerCamelCase = F"""| {model} | {count} | {error[:60]} | {_count} |"""
lines.append(_UpperCamelCase )
return "\n".join(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument("""--workflow_run_id""", type=str, required=True, help="""A GitHub Actions workflow run id.""")
parser.add_argument(
"""--output_dir""",
type=str,
required=True,
help="""Where to store the downloaded artifacts and other result files.""",
)
parser.add_argument("""--token""", default=None, type=str, help="""A token that has actions:read permission.""")
a_ = parser.parse_args()
os.makedirs(args.output_dir, exist_ok=True)
a_ = get_job_links(args.workflow_run_id, token=args.token)
a_ = {}
# To deal with `workflow_call` event, where a job name is the combination of the job names in the caller and callee.
# For example, `PyTorch 1.11 / Model tests (models/albert, single-gpu)`.
if _job_links:
for k, v in _job_links.items():
# This is how GitHub actions combine job names.
if " / " in k:
a_ = k.find(""" / """)
a_ = k[index + len(""" / """) :]
a_ = v
with open(os.path.join(args.output_dir, """job_links.json"""), """w""", encoding="""UTF-8""") as fp:
json.dump(job_links, fp, ensure_ascii=False, indent=4)
a_ = get_artifacts_links(args.workflow_run_id, token=args.token)
with open(os.path.join(args.output_dir, """artifacts.json"""), """w""", encoding="""UTF-8""") as fp:
json.dump(artifacts, fp, ensure_ascii=False, indent=4)
for idx, (name, url) in enumerate(artifacts.items()):
download_artifact(name, url, args.output_dir, args.token)
# Be gentle to GitHub
time.sleep(1)
a_ = get_all_errors(args.output_dir, job_links=job_links)
# `e[1]` is the error
a_ = Counter()
counter.update([e[1] for e in errors])
# print the top 30 most common test errors
a_ = counter.most_common(30)
for item in most_common:
print(item)
with open(os.path.join(args.output_dir, """errors.json"""), """w""", encoding="""UTF-8""") as fp:
json.dump(errors, fp, ensure_ascii=False, indent=4)
a_ = reduce_by_error(errors)
a_ = reduce_by_model(errors)
a_ = make_github_table(reduced_by_error)
a_ = make_github_table_per_model(reduced_by_model)
with open(os.path.join(args.output_dir, """reduced_by_error.txt"""), """w""", encoding="""UTF-8""") as fp:
fp.write(sa)
with open(os.path.join(args.output_dir, """reduced_by_model.txt"""), """w""", encoding="""UTF-8""") as fp:
fp.write(sa)
| 330 |
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_torch_available,
)
a_ = {"""configuration_encoder_decoder""": ["""EncoderDecoderConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""EncoderDecoderModel"""]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TFEncoderDecoderModel"""]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""FlaxEncoderDecoderModel"""]
if TYPE_CHECKING:
from .configuration_encoder_decoder import EncoderDecoderConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_encoder_decoder import EncoderDecoderModel
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_encoder_decoder import TFEncoderDecoderModel
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_encoder_decoder import FlaxEncoderDecoderModel
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
import shutil
import tempfile
import unittest
import numpy as np
import pytest
from transformers.testing_utils import require_vision
from transformers.utils import is_vision_available
if is_vision_available():
from PIL import Image
from transformers import AutoProcessor, BlipaProcessor, BlipImageProcessor, GPTaTokenizer, PreTrainedTokenizerFast
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = tempfile.mkdtemp()
__lowerCamelCase = BlipImageProcessor()
__lowerCamelCase = GPTaTokenizer.from_pretrained('''hf-internal-testing/tiny-random-GPT2Model''' )
__lowerCamelCase = BlipaProcessor(__UpperCAmelCase , __UpperCAmelCase )
processor.save_pretrained(self.tmpdirname )
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
return AutoProcessor.from_pretrained(self.tmpdirname , **__UpperCAmelCase ).tokenizer
def lowerCamelCase ( self , **__UpperCAmelCase ):
'''simple docstring'''
return AutoProcessor.from_pretrained(self.tmpdirname , **__UpperCAmelCase ).image_processor
def lowerCamelCase ( self ):
'''simple docstring'''
shutil.rmtree(self.tmpdirname )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = [np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )]
__lowerCamelCase = [Image.fromarray(np.moveaxis(__UpperCAmelCase , 0 , -1 ) ) for x in image_inputs]
return image_inputs
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BlipaProcessor(tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() )
processor.save_pretrained(self.tmpdirname )
__lowerCamelCase = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' )
__lowerCamelCase = self.get_image_processor(do_normalize=__UpperCAmelCase , padding_value=1.0 )
__lowerCamelCase = BlipaProcessor.from_pretrained(
self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=__UpperCAmelCase , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , __UpperCAmelCase )
self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.image_processor , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_image_processor()
__lowerCamelCase = self.get_tokenizer()
__lowerCamelCase = BlipaProcessor(tokenizer=__UpperCAmelCase , image_processor=__UpperCAmelCase )
__lowerCamelCase = self.prepare_image_inputs()
__lowerCamelCase = image_processor(__UpperCAmelCase , return_tensors='''np''' )
__lowerCamelCase = processor(images=__UpperCAmelCase , return_tensors='''np''' )
for key in input_feat_extract.keys():
self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1E-2 )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_image_processor()
__lowerCamelCase = self.get_tokenizer()
__lowerCamelCase = BlipaProcessor(tokenizer=__UpperCAmelCase , image_processor=__UpperCAmelCase )
__lowerCamelCase = '''lower newer'''
__lowerCamelCase = processor(text=__UpperCAmelCase )
__lowerCamelCase = tokenizer(__UpperCAmelCase , return_token_type_ids=__UpperCAmelCase )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_image_processor()
__lowerCamelCase = self.get_tokenizer()
__lowerCamelCase = BlipaProcessor(tokenizer=__UpperCAmelCase , image_processor=__UpperCAmelCase )
__lowerCamelCase = '''lower newer'''
__lowerCamelCase = self.prepare_image_inputs()
__lowerCamelCase = processor(text=__UpperCAmelCase , images=__UpperCAmelCase )
self.assertListEqual(list(inputs.keys() ) , ['''pixel_values''', '''input_ids''', '''attention_mask'''] )
# test if it raises when no input is passed
with pytest.raises(__UpperCAmelCase ):
processor()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_image_processor()
__lowerCamelCase = self.get_tokenizer()
__lowerCamelCase = BlipaProcessor(tokenizer=__UpperCAmelCase , image_processor=__UpperCAmelCase )
__lowerCamelCase = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
__lowerCamelCase = processor.batch_decode(__UpperCAmelCase )
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.get_image_processor()
__lowerCamelCase = self.get_tokenizer()
__lowerCamelCase = BlipaProcessor(tokenizer=__UpperCAmelCase , image_processor=__UpperCAmelCase )
__lowerCamelCase = '''lower newer'''
__lowerCamelCase = self.prepare_image_inputs()
__lowerCamelCase = processor(text=__UpperCAmelCase , images=__UpperCAmelCase )
# For now the processor supports only ['pixel_values', 'input_ids', 'attention_mask']
self.assertListEqual(list(inputs.keys() ) , ['''pixel_values''', '''input_ids''', '''attention_mask'''] )
| 330 |
from string import ascii_lowercase, ascii_uppercase
def a__ ( _UpperCamelCase : str ):
if not sentence:
return ""
__lowerCamelCase = dict(zip(_UpperCamelCase ,_UpperCamelCase ) )
return lower_to_upper.get(sentence[0] ,sentence[0] ) + sentence[1:]
if __name__ == "__main__":
from doctest import testmod
testmod()
| 330 | 1 |
from typing import Callable, List, Optional, Tuple, Union
import torch
from transformers import CLIPTextModel, CLIPTokenizer
from ...configuration_utils import ConfigMixin, register_to_config
from ...models import ModelMixin, TransformeraDModel, VQModel
from ...schedulers import VQDiffusionScheduler
from ...utils import logging
from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
a_ = logging.get_logger(__name__) # pylint: disable=invalid-name
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ ):
@register_to_config
def __init__( self , __UpperCAmelCase , __UpperCAmelCase = None , __UpperCAmelCase = None ):
'''simple docstring'''
super().__init__()
__lowerCamelCase = learnable
if self.learnable:
assert hidden_size is not None, "learnable=True requires `hidden_size` to be set"
assert length is not None, "learnable=True requires `length` to be set"
__lowerCamelCase = torch.zeros(__UpperCAmelCase , __UpperCAmelCase )
else:
__lowerCamelCase = None
__lowerCamelCase = torch.nn.Parameter(__UpperCAmelCase )
class __lowerCAmelCase ( lowerCAmelCase__ ):
lowerCAmelCase__ = 42
lowerCAmelCase__ = 42
lowerCAmelCase__ = 42
lowerCAmelCase__ = 42
lowerCAmelCase__ = 42
lowerCAmelCase__ = 42
def __init__( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , ):
'''simple docstring'''
super().__init__()
self.register_modules(
vqvae=__UpperCAmelCase , transformer=__UpperCAmelCase , text_encoder=__UpperCAmelCase , tokenizer=__UpperCAmelCase , scheduler=__UpperCAmelCase , learned_classifier_free_sampling_embeddings=__UpperCAmelCase , )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = len(__UpperCAmelCase ) if isinstance(__UpperCAmelCase , __UpperCAmelCase ) else 1
# get prompt text embeddings
__lowerCamelCase = self.tokenizer(
__UpperCAmelCase , padding='''max_length''' , max_length=self.tokenizer.model_max_length , return_tensors='''pt''' , )
__lowerCamelCase = text_inputs.input_ids
if text_input_ids.shape[-1] > self.tokenizer.model_max_length:
__lowerCamelCase = self.tokenizer.batch_decode(text_input_ids[:, self.tokenizer.model_max_length :] )
logger.warning(
'''The following part of your input was truncated because CLIP can only handle sequences up to'''
F""" {self.tokenizer.model_max_length} tokens: {removed_text}""" )
__lowerCamelCase = text_input_ids[:, : self.tokenizer.model_max_length]
__lowerCamelCase = self.text_encoder(text_input_ids.to(self.device ) )[0]
# NOTE: This additional step of normalizing the text embeddings is from VQ-Diffusion.
# While CLIP does normalize the pooled output of the text transformer when combining
# the image and text embeddings, CLIP does not directly normalize the last hidden state.
#
# CLIP normalizing the pooled output.
# https://github.com/huggingface/transformers/blob/d92e22d1f28324f513f3080e5c47c071a3916721/src/transformers/models/clip/modeling_clip.py#L1052-L1053
__lowerCamelCase = prompt_embeds / prompt_embeds.norm(dim=-1 , keepdim=__UpperCAmelCase )
# duplicate text embeddings for each generation per prompt
__lowerCamelCase = prompt_embeds.repeat_interleave(__UpperCAmelCase , dim=0 )
if do_classifier_free_guidance:
if self.learned_classifier_free_sampling_embeddings.learnable:
__lowerCamelCase = self.learned_classifier_free_sampling_embeddings.embeddings
__lowerCamelCase = negative_prompt_embeds.unsqueeze(0 ).repeat(__UpperCAmelCase , 1 , 1 )
else:
__lowerCamelCase = [''''''] * batch_size
__lowerCamelCase = text_input_ids.shape[-1]
__lowerCamelCase = self.tokenizer(
__UpperCAmelCase , padding='''max_length''' , max_length=__UpperCAmelCase , truncation=__UpperCAmelCase , return_tensors='''pt''' , )
__lowerCamelCase = self.text_encoder(uncond_input.input_ids.to(self.device ) )[0]
# See comment for normalizing text embeddings
__lowerCamelCase = negative_prompt_embeds / negative_prompt_embeds.norm(dim=-1 , keepdim=__UpperCAmelCase )
# duplicate unconditional embeddings for each generation per prompt, using mps friendly method
__lowerCamelCase = negative_prompt_embeds.shape[1]
__lowerCamelCase = negative_prompt_embeds.repeat(1 , __UpperCAmelCase , 1 )
__lowerCamelCase = negative_prompt_embeds.view(batch_size * num_images_per_prompt , __UpperCAmelCase , -1 )
# For classifier free guidance, we need to do two forward passes.
# Here we concatenate the unconditional and text embeddings into a single batch
# to avoid doing two forward passes
__lowerCamelCase = torch.cat([negative_prompt_embeds, prompt_embeds] )
return prompt_embeds
@torch.no_grad()
def __call__( self , __UpperCAmelCase , __UpperCAmelCase = 100 , __UpperCAmelCase = 5.0 , __UpperCAmelCase = 1.0 , __UpperCAmelCase = 1 , __UpperCAmelCase = None , __UpperCAmelCase = None , __UpperCAmelCase = "pil" , __UpperCAmelCase = True , __UpperCAmelCase = None , __UpperCAmelCase = 1 , ):
'''simple docstring'''
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = 1
elif isinstance(__UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = len(__UpperCAmelCase )
else:
raise ValueError(F"""`prompt` has to be of type `str` or `list` but is {type(__UpperCAmelCase )}""" )
__lowerCamelCase = batch_size * num_images_per_prompt
__lowerCamelCase = guidance_scale > 1.0
__lowerCamelCase = self._encode_prompt(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
if (callback_steps is None) or (
callback_steps is not None and (not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or callback_steps <= 0)
):
raise ValueError(
F"""`callback_steps` has to be a positive integer but is {callback_steps} of type"""
F""" {type(__UpperCAmelCase )}.""" )
# get the initial completely masked latents unless the user supplied it
__lowerCamelCase = (batch_size, self.transformer.num_latent_pixels)
if latents is None:
__lowerCamelCase = self.transformer.num_vector_embeds - 1
__lowerCamelCase = torch.full(__UpperCAmelCase , __UpperCAmelCase ).to(self.device )
else:
if latents.shape != latents_shape:
raise ValueError(F"""Unexpected latents shape, got {latents.shape}, expected {latents_shape}""" )
if (latents < 0).any() or (latents >= self.transformer.num_vector_embeds).any():
raise ValueError(
'''Unexpected latents value(s). All latents be valid embedding indices i.e. in the range 0,'''
F""" {self.transformer.num_vector_embeds - 1} (inclusive).""" )
__lowerCamelCase = latents.to(self.device )
# set timesteps
self.scheduler.set_timesteps(__UpperCAmelCase , device=self.device )
__lowerCamelCase = self.scheduler.timesteps.to(self.device )
__lowerCamelCase = latents
for i, t in enumerate(self.progress_bar(__UpperCAmelCase ) ):
# expand the sample if we are doing classifier free guidance
__lowerCamelCase = torch.cat([sample] * 2 ) if do_classifier_free_guidance else sample
# predict the un-noised image
# model_output == `log_p_x_0`
__lowerCamelCase = self.transformer(__UpperCAmelCase , encoder_hidden_states=__UpperCAmelCase , timestep=__UpperCAmelCase ).sample
if do_classifier_free_guidance:
__lowerCamelCase ,__lowerCamelCase = model_output.chunk(2 )
__lowerCamelCase = model_output_uncond + guidance_scale * (model_output_text - model_output_uncond)
model_output -= torch.logsumexp(__UpperCAmelCase , dim=1 , keepdim=__UpperCAmelCase )
__lowerCamelCase = self.truncate(__UpperCAmelCase , __UpperCAmelCase )
# remove `log(0)`'s (`-inf`s)
__lowerCamelCase = model_output.clamp(-70 )
# compute the previous noisy sample x_t -> x_t-1
__lowerCamelCase = self.scheduler.step(__UpperCAmelCase , timestep=__UpperCAmelCase , sample=__UpperCAmelCase , generator=__UpperCAmelCase ).prev_sample
# call the callback, if provided
if callback is not None and i % callback_steps == 0:
callback(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = self.vqvae.config.vq_embed_dim
__lowerCamelCase = (batch_size, self.transformer.height, self.transformer.width, embedding_channels)
__lowerCamelCase = self.vqvae.quantize.get_codebook_entry(__UpperCAmelCase , shape=__UpperCAmelCase )
__lowerCamelCase = self.vqvae.decode(__UpperCAmelCase , force_not_quantize=__UpperCAmelCase ).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(__UpperCAmelCase )
if not return_dict:
return (image,)
return ImagePipelineOutput(images=__UpperCAmelCase )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = torch.sort(__UpperCAmelCase , 1 , descending=__UpperCAmelCase )
__lowerCamelCase = torch.exp(__UpperCAmelCase )
__lowerCamelCase = sorted_p_x_0.cumsum(dim=1 ) < truncation_rate
# Ensure that at least the largest probability is not zeroed out
__lowerCamelCase = torch.full_like(keep_mask[:, 0:1, :] , __UpperCAmelCase )
__lowerCamelCase = torch.cat((all_true, keep_mask) , dim=1 )
__lowerCamelCase = keep_mask[:, :-1, :]
__lowerCamelCase = keep_mask.gather(1 , indices.argsort(1 ) )
__lowerCamelCase = log_p_x_0.clone()
__lowerCamelCase = -torch.inf # -inf = log(0)
return rv
| 330 |
from transformers import BertTokenizer, EncoderDecoderModel, SeqaSeqTrainer, SeqaSeqTrainingArguments
from transformers.testing_utils import TestCasePlus, require_torch, slow
from transformers.utils import is_datasets_available
if is_datasets_available():
import datasets
class __lowerCAmelCase ( lowerCAmelCase__ ):
@slow
@require_torch
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = EncoderDecoderModel.from_encoder_decoder_pretrained('''prajjwal1/bert-tiny''' , '''prajjwal1/bert-tiny''' )
__lowerCamelCase = BertTokenizer.from_pretrained('''bert-base-uncased''' )
__lowerCamelCase = bertabert.config.encoder.vocab_size
__lowerCamelCase = tokenizer.sep_token_id
__lowerCamelCase = tokenizer.cls_token_id
__lowerCamelCase = 128
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''train[:1%]''' )
__lowerCamelCase = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''validation[:1%]''' )
__lowerCamelCase = train_dataset.select(range(32 ) )
__lowerCamelCase = val_dataset.select(range(16 ) )
__lowerCamelCase = 4
def _map_to_encoder_decoder_inputs(__UpperCAmelCase ):
# Tokenizer will automatically set [BOS] <text> [EOS]
__lowerCamelCase = tokenizer(batch['''article'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=512 )
__lowerCamelCase = tokenizer(batch['''highlights'''] , padding='''max_length''' , truncation=__UpperCAmelCase , max_length=128 )
__lowerCamelCase = inputs.input_ids
__lowerCamelCase = inputs.attention_mask
__lowerCamelCase = outputs.input_ids
__lowerCamelCase = outputs.input_ids.copy()
__lowerCamelCase = [
[-100 if token == tokenizer.pad_token_id else token for token in labels] for labels in batch['''labels''']
]
__lowerCamelCase = outputs.attention_mask
assert all(len(__UpperCAmelCase ) == 512 for x in inputs.input_ids )
assert all(len(__UpperCAmelCase ) == 128 for x in outputs.input_ids )
return batch
def _compute_metrics(__UpperCAmelCase ):
__lowerCamelCase = pred.label_ids
__lowerCamelCase = pred.predictions
# all unnecessary tokens are removed
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = tokenizer.batch_decode(__UpperCAmelCase , skip_special_tokens=__UpperCAmelCase )
__lowerCamelCase = sum([int(pred_str[i] == label_str[i] ) for i in range(len(__UpperCAmelCase ) )] ) / len(__UpperCAmelCase )
return {"accuracy": accuracy}
# map train dataset
__lowerCamelCase = train_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
train_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
# same for validation dataset
__lowerCamelCase = val_dataset.map(
_map_to_encoder_decoder_inputs , batched=__UpperCAmelCase , batch_size=__UpperCAmelCase , remove_columns=['''article''', '''highlights'''] , )
val_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
__lowerCamelCase = self.get_auto_remove_tmp_dir()
__lowerCamelCase = SeqaSeqTrainingArguments(
output_dir=__UpperCAmelCase , per_device_train_batch_size=__UpperCAmelCase , per_device_eval_batch_size=__UpperCAmelCase , predict_with_generate=__UpperCAmelCase , evaluation_strategy='''steps''' , do_train=__UpperCAmelCase , do_eval=__UpperCAmelCase , warmup_steps=0 , eval_steps=2 , logging_steps=2 , )
# instantiate trainer
__lowerCamelCase = SeqaSeqTrainer(
model=__UpperCAmelCase , args=__UpperCAmelCase , compute_metrics=_compute_metrics , train_dataset=__UpperCAmelCase , eval_dataset=__UpperCAmelCase , tokenizer=__UpperCAmelCase , )
# start training
trainer.train()
| 330 | 1 |
import copy
import inspect
import unittest
import numpy as np
from huggingface_hub import hf_hub_download
from transformers import TimesformerConfig
from transformers.models.auto import get_values
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 (
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING,
TimesformerForVideoClassification,
TimesformerModel,
)
from transformers.models.timesformer.modeling_timesformer import TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from transformers import VideoMAEImageProcessor
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=13 , __UpperCAmelCase=10 , __UpperCAmelCase=3 , __UpperCAmelCase=2 , __UpperCAmelCase=2 , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=32 , __UpperCAmelCase=5 , __UpperCAmelCase=4 , __UpperCAmelCase=37 , __UpperCAmelCase="gelu" , __UpperCAmelCase=0.1 , __UpperCAmelCase=0.1 , __UpperCAmelCase=10 , __UpperCAmelCase=0.02 , __UpperCAmelCase="divided_space_time" , __UpperCAmelCase=None , ):
'''simple docstring'''
__lowerCamelCase = parent
__lowerCamelCase = batch_size
__lowerCamelCase = image_size
__lowerCamelCase = num_channels
__lowerCamelCase = patch_size
__lowerCamelCase = num_frames
__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 = attention_type
__lowerCamelCase = initializer_range
__lowerCamelCase = scope
__lowerCamelCase = num_labels
# in TimeSformer, the number of spatial tokens equals num_frames * num_patches per frame + 1 CLS token
__lowerCamelCase = (image_size // patch_size) ** 2
__lowerCamelCase = (num_frames) * self.num_patches_per_frame + 1
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = floats_tensor(
[self.batch_size, self.num_frames, 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 lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , num_frames=self.num_frames , 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 , initializer_range=self.initializer_range , attention_type=self.attention_type , )
__lowerCamelCase = self.num_labels
return config
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerModel(config=__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
__lowerCamelCase = model(__UpperCAmelCase )
# verify the logits shape
__lowerCamelCase = torch.Size((self.batch_size, self.num_labels) )
self.parent.assertEqual(result.logits.shape , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.prepare_config_and_inputs()
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = config_and_inputs
__lowerCamelCase = {'''pixel_values''': pixel_values}
return config, inputs_dict
@require_torch
class __lowerCAmelCase ( lowerCAmelCase__ , lowerCAmelCase__ , unittest.TestCase ):
lowerCAmelCase__ = (TimesformerModel, TimesformerForVideoClassification) if is_torch_available() else ()
lowerCAmelCase__ = (
{"""feature-extraction""": TimesformerModel, """video-classification""": TimesformerForVideoClassification}
if is_torch_available()
else {}
)
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
lowerCAmelCase__ = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerModelTester(self )
__lowerCamelCase = ConfigTester(
self , config_class=__UpperCAmelCase , has_text_modality=__UpperCAmelCase , hidden_size=37 )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = copy.deepcopy(__UpperCAmelCase )
if return_labels:
if model_class in get_values(__UpperCAmelCase ):
__lowerCamelCase = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=__UpperCAmelCase )
return inputs_dict
def lowerCamelCase ( self ):
'''simple docstring'''
self.config_tester.run_common_tests()
@unittest.skip(reason='''TimeSformer does not use inputs_embeds''' )
def lowerCamelCase ( self ):
'''simple docstring'''
pass
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
__lowerCamelCase = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(__UpperCAmelCase , nn.Linear ) )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__lowerCamelCase = model_class(__UpperCAmelCase )
__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] , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_video_classification(*__UpperCAmelCase )
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
for model_name in TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__lowerCamelCase = TimesformerModel.from_pretrained(__UpperCAmelCase )
self.assertIsNotNone(__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
if not self.has_attentions:
pass
else:
__lowerCamelCase ,__lowerCamelCase = self.model_tester.prepare_config_and_inputs_for_common()
__lowerCamelCase = True
for model_class in self.all_model_classes:
__lowerCamelCase = self.model_tester.seq_length
__lowerCamelCase = self.model_tester.num_frames
__lowerCamelCase = True
__lowerCamelCase = False
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
__lowerCamelCase = len(__UpperCAmelCase )
# Check attention is always last and order is fine
__lowerCamelCase = True
__lowerCamelCase = True
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
self.assertEqual(out_len + 1 , len(__UpperCAmelCase ) )
__lowerCamelCase = outputs.attentions
self.assertEqual(len(__UpperCAmelCase ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
def lowerCamelCase ( self ):
'''simple docstring'''
def check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ):
__lowerCamelCase = model_class(__UpperCAmelCase )
model.to(__UpperCAmelCase )
model.eval()
with torch.no_grad():
__lowerCamelCase = model(**self._prepare_for_class(__UpperCAmelCase , __UpperCAmelCase ) )
__lowerCamelCase = outputs.hidden_states
__lowerCamelCase = self.model_tester.num_hidden_layers + 1
self.assertEqual(len(__UpperCAmelCase ) , __UpperCAmelCase )
__lowerCamelCase = self.model_tester.seq_length
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [seq_length, self.model_tester.hidden_size] , )
__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(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__lowerCamelCase = True
check_hidden_states_output(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
def a__ ( ):
__lowerCamelCase = hf_hub_download(
repo_id='''hf-internal-testing/spaghetti-video''' ,filename='''eating_spaghetti.npy''' ,repo_type='''dataset''' )
__lowerCamelCase = np.load(_UpperCamelCase )
return list(_UpperCamelCase )
@require_torch
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
@cached_property
def lowerCamelCase ( self ):
'''simple docstring'''
# logits were tested with a different mean and std, so we use the same here
return (
VideoMAEImageProcessor(image_mean=[0.5, 0.5, 0.5] , image_std=[0.5, 0.5, 0.5] )
if is_vision_available()
else None
)
@slow
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = TimesformerForVideoClassification.from_pretrained('''facebook/timesformer-base-finetuned-k400''' ).to(
__UpperCAmelCase )
__lowerCamelCase = self.default_image_processor
__lowerCamelCase = prepare_video()
__lowerCamelCase = image_processor(video[:8] , return_tensors='''pt''' ).to(__UpperCAmelCase )
# forward pass
with torch.no_grad():
__lowerCamelCase = model(**__UpperCAmelCase )
# verify the logits
__lowerCamelCase = torch.Size((1, 400) )
self.assertEqual(outputs.logits.shape , __UpperCAmelCase )
__lowerCamelCase = torch.tensor([-0.3_016, -0.7_713, -0.4_205] ).to(__UpperCAmelCase )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , __UpperCAmelCase , atol=1E-4 ) )
| 330 |
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a_ = {"""configuration_timm_backbone""": ["""TimmBackboneConfig"""]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a_ = ["""TimmBackbone"""]
if TYPE_CHECKING:
from .configuration_timm_backbone import TimmBackboneConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_timm_backbone import TimmBackbone
else:
import sys
a_ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 330 | 1 |
import argparse
import json
import os
import fairseq
import torch
from fairseq.data import Dictionary
from transformers import (
WavaVecaConformerConfig,
WavaVecaConformerForCTC,
WavaVecaConformerForPreTraining,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaProcessor,
logging,
)
logging.set_verbosity_info()
a_ = logging.get_logger(__name__)
a_ = {
"""post_extract_proj""": """feature_projection.projection""",
"""encoder.pos_conv.0""": """encoder.pos_conv_embed.conv""",
"""self_attn.linear_k""": """encoder.layers.*.self_attn.linear_k""",
"""self_attn.linear_v""": """encoder.layers.*.self_attn.linear_v""",
"""self_attn.linear_q""": """encoder.layers.*.self_attn.linear_q""",
"""self_attn.pos_bias_u""": """encoder.layers.*.self_attn.pos_bias_u""",
"""self_attn.pos_bias_v""": """encoder.layers.*.self_attn.pos_bias_v""",
"""self_attn.linear_out""": """encoder.layers.*.self_attn.linear_out""",
"""self_attn.linear_pos""": """encoder.layers.*.self_attn.linear_pos""",
"""self_attn.rotary_emb""": """encoder.embed_positions""",
"""self_attn_layer_norm""": """encoder.layers.*.self_attn_layer_norm""",
"""conv_module.pointwise_conv1""": """encoder.layers.*.conv_module.pointwise_conv1""",
"""conv_module.pointwise_conv2""": """encoder.layers.*.conv_module.pointwise_conv2""",
"""conv_module.depthwise_conv""": """encoder.layers.*.conv_module.depthwise_conv""",
"""conv_module.batch_norm""": """encoder.layers.*.conv_module.batch_norm""",
"""conv_module.layer_norm""": """encoder.layers.*.conv_module.layer_norm""",
"""ffn1.w_1""": """encoder.layers.*.ffn1.intermediate_dense""",
"""ffn1.w_2""": """encoder.layers.*.ffn1.output_dense""",
"""ffn1.layer_norm""": """encoder.layers.*.ffn1_layer_norm""",
"""ffn2.w_1""": """encoder.layers.*.ffn2.intermediate_dense""",
"""ffn2.w_2""": """encoder.layers.*.ffn2.output_dense""",
"""ffn2.layer_norm""": """encoder.layers.*.ffn2_layer_norm""",
"""final_layer_norm""": """encoder.layers.*.final_layer_norm""",
"""encoder.layer_norm""": """encoder.layer_norm""",
"""w2v_model.layer_norm""": """feature_projection.layer_norm""",
"""quantizer.weight_proj""": """quantizer.weight_proj""",
"""quantizer.vars""": """quantizer.codevectors""",
"""project_q""": """project_q""",
"""final_proj""": """project_hid""",
"""w2v_encoder.proj""": """lm_head""",
"""mask_emb""": """masked_spec_embed""",
}
a_ = [
"""lm_head""",
"""quantizer.weight_proj""",
"""quantizer.codevectors""",
"""project_q""",
"""project_hid""",
]
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : str ,_UpperCamelCase : List[Any] ,_UpperCamelCase : Optional[Any] ):
for attribute in key.split('''.''' ):
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase )
if weight_type is not None:
__lowerCamelCase = getattr(_UpperCamelCase ,_UpperCamelCase ).shape
else:
__lowerCamelCase = hf_pointer.shape
if hf_shape != value.shape:
raise ValueError(
F"""Shape of hf {key + "." + weight_type if weight_type is not None else ""} is {hf_shape}, but should be"""
F""" {value.shape} for {full_name}""" )
if weight_type == "weight":
__lowerCamelCase = value
elif weight_type == "weight_g":
__lowerCamelCase = value
elif weight_type == "weight_v":
__lowerCamelCase = value
elif weight_type == "bias":
__lowerCamelCase = value
elif weight_type == "running_mean":
__lowerCamelCase = value
elif weight_type == "running_var":
__lowerCamelCase = value
elif weight_type == "num_batches_tracked":
__lowerCamelCase = value
elif weight_type == "inv_freq":
__lowerCamelCase = value
else:
__lowerCamelCase = value
logger.info(F"""{key + "." + weight_type if weight_type is not None else ""} was initialized from {full_name}.""" )
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : str ,_UpperCamelCase : Dict ):
__lowerCamelCase = []
__lowerCamelCase = fairseq_model.state_dict()
__lowerCamelCase = hf_model.wavaveca_conformer.feature_extractor
for name, value in fairseq_dict.items():
__lowerCamelCase = False
if "conv_layers" in name:
load_conv_layer(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,hf_model.config.feat_extract_norm == '''group''' ,)
__lowerCamelCase = True
else:
for key, mapped_key in MAPPING.items():
__lowerCamelCase = '''wav2vec2_conformer.''' + mapped_key if mapped_key not in TOP_LEVEL_KEYS else mapped_key
if key in name or key.split('''w2v_model.''' )[-1] == name.split('''.''' )[0]:
__lowerCamelCase = True
if "*" in mapped_key:
__lowerCamelCase = name.split(_UpperCamelCase )[0].split('''.''' )[-2]
__lowerCamelCase = mapped_key.replace('''*''' ,_UpperCamelCase )
if "pos_bias_u" in name:
__lowerCamelCase = None
elif "pos_bias_v" in name:
__lowerCamelCase = None
elif "weight_g" in name:
__lowerCamelCase = '''weight_g'''
elif "weight_v" in name:
__lowerCamelCase = '''weight_v'''
elif "bias" in name:
__lowerCamelCase = '''bias'''
elif "weight" in name:
# TODO: don't match quantizer.weight_proj
__lowerCamelCase = '''weight'''
elif "running_mean" in name:
__lowerCamelCase = '''running_mean'''
elif "inv_freq" in name:
__lowerCamelCase = '''inv_freq'''
elif "running_var" in name:
__lowerCamelCase = '''running_var'''
elif "num_batches_tracked" in name:
__lowerCamelCase = '''num_batches_tracked'''
else:
__lowerCamelCase = None
set_recursively(_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
continue
if not is_used:
unused_weights.append(_UpperCamelCase )
logger.warning(F"""Unused weights: {unused_weights}""" )
def a__ ( _UpperCamelCase : List[Any] ,_UpperCamelCase : int ,_UpperCamelCase : Any ,_UpperCamelCase : Union[str, Any] ,_UpperCamelCase : Union[str, Any] ):
__lowerCamelCase = full_name.split('''conv_layers.''' )[-1]
__lowerCamelCase = name.split('''.''' )
__lowerCamelCase = int(items[0] )
__lowerCamelCase = int(items[1] )
if type_id == 0:
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].conv.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].conv.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract conv layer {layer_id} was initialized from {full_name}.""" )
elif (type_id == 2 and not use_group_norm) or (type_id == 2 and layer_id == 0 and use_group_norm):
if "bias" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
elif "weight" in name:
if value.shape != feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape:
raise ValueError(
F"""{full_name} has size {value.shape}, but"""
F""" {feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape} was found.""" )
__lowerCamelCase = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(_UpperCamelCase )
@torch.no_grad()
def a__ ( _UpperCamelCase : Optional[int] ,_UpperCamelCase : Optional[Any] ,_UpperCamelCase : Optional[Any]=None ,_UpperCamelCase : Tuple=None ,_UpperCamelCase : Dict=True ):
if config_path is not None:
__lowerCamelCase = WavaVecaConformerConfig.from_pretrained(_UpperCamelCase ,hidden_act='''swish''' )
else:
__lowerCamelCase = WavaVecaConformerConfig()
if "rope" in checkpoint_path:
__lowerCamelCase = '''rotary'''
if is_finetuned:
if dict_path:
__lowerCamelCase = Dictionary.load(_UpperCamelCase )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
__lowerCamelCase = target_dict.pad_index
__lowerCamelCase = target_dict.bos_index
__lowerCamelCase = target_dict.eos_index
__lowerCamelCase = len(target_dict.symbols )
__lowerCamelCase = os.path.join(_UpperCamelCase ,'''vocab.json''' )
if not os.path.isdir(_UpperCamelCase ):
logger.error('''--pytorch_dump_folder_path ({}) should be a directory'''.format(_UpperCamelCase ) )
return
os.makedirs(_UpperCamelCase ,exist_ok=_UpperCamelCase )
__lowerCamelCase = target_dict.indices
# fairseq has the <pad> and <s> switched
__lowerCamelCase = 0
__lowerCamelCase = 1
with open(_UpperCamelCase ,'''w''' ,encoding='''utf-8''' ) as vocab_handle:
json.dump(_UpperCamelCase ,_UpperCamelCase )
__lowerCamelCase = WavaVecaCTCTokenizer(
_UpperCamelCase ,unk_token=target_dict.unk_word ,pad_token=target_dict.pad_word ,bos_token=target_dict.bos_word ,eos_token=target_dict.eos_word ,word_delimiter_token='''|''' ,do_lower_case=_UpperCamelCase ,)
__lowerCamelCase = True if config.feat_extract_norm == '''layer''' else False
__lowerCamelCase = WavaVecaFeatureExtractor(
feature_size=1 ,sampling_rate=1_60_00 ,padding_value=0 ,do_normalize=_UpperCamelCase ,return_attention_mask=_UpperCamelCase ,)
__lowerCamelCase = WavaVecaProcessor(feature_extractor=_UpperCamelCase ,tokenizer=_UpperCamelCase )
processor.save_pretrained(_UpperCamelCase )
__lowerCamelCase = WavaVecaConformerForCTC(_UpperCamelCase )
else:
__lowerCamelCase = WavaVecaConformerForPreTraining(_UpperCamelCase )
if is_finetuned:
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
__lowerCamelCase = argparse.Namespace(task='''audio_pretraining''' )
__lowerCamelCase = fairseq.tasks.setup_task(_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] ,task=_UpperCamelCase )
__lowerCamelCase = model[0].eval()
recursively_load_weights(_UpperCamelCase ,_UpperCamelCase ,not is_finetuned )
hf_wavavec.save_pretrained(_UpperCamelCase )
if __name__ == "__main__":
a_ = argparse.ArgumentParser()
parser.add_argument("""--pytorch_dump_folder_path""", default=None, type=str, help="""Path to the output PyTorch model.""")
parser.add_argument("""--checkpoint_path""", default=None, type=str, help="""Path to fairseq checkpoint""")
parser.add_argument("""--dict_path""", default=None, type=str, help="""Path to dict of fine-tuned model""")
parser.add_argument("""--config_path""", default=None, type=str, help="""Path to hf config.json of model to convert""")
parser.add_argument(
"""--not_finetuned""", action="""store_true""", help="""Whether the model to convert is a fine-tuned model or not"""
)
a_ = parser.parse_args()
convert_wavaveca_conformer_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 330 |
import random
import unittest
from torch.utils.data import BatchSampler, DataLoader, IterableDataset
from accelerate import Accelerator
from accelerate.data_loader import (
BatchSamplerShard,
DataLoaderDispatcher,
DataLoaderShard,
IterableDatasetShard,
SkipBatchSampler,
SkipDataLoader,
skip_first_batches,
)
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase=0.01 , __UpperCAmelCase=1000 ):
'''simple docstring'''
__lowerCamelCase = p_stop
__lowerCamelCase = max_length
def __iter__( self ):
'''simple docstring'''
__lowerCamelCase = 0
__lowerCamelCase = False
while not stop and count < self.max_length:
yield count
count += 1
__lowerCamelCase = random.random() < self.p_stop
class __lowerCAmelCase ( unittest.TestCase ):
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = [
BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
for i in range(2 )
]
__lowerCamelCase = [list(__UpperCAmelCase ) for batch_sampler_shard in batch_sampler_shards]
if not split_batches:
self.assertListEqual([len(__UpperCAmelCase ) for shard in batch_sampler_shards] , [len(__UpperCAmelCase ) for e in expected] )
self.assertListEqual(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [0, 1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 0]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [1, 2, 3]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1, 0]], [[1, 0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [0, 1]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 0]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [1, 2]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], [[0, 1]]]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of total batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21, 22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=3 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is a round multiple of batch size but not total batch size.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but has a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19, 20]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17], [21]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size but and has not a multiple of
# num_processes batch.
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14], [18, 19]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(20 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1, 2], [6, 7, 8], [12, 13, 14]],
[[3, 4, 5], [9, 10, 11], [15, 16, 17]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=3 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
# Check the shards when the dataset is a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19], [22, 23]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(24 ) , batch_size=4 , drop_last=__UpperCAmelCase )
# Expected shouldn't change
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size.
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20, 21]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(22 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is not a round multiple of batch size or num_processes.
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17], [20]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(21 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [
[[0, 1], [4, 5], [8, 9], [12, 13], [16, 17]],
[[2, 3], [6, 7], [10, 11], [14, 15], [18, 19]],
]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
# Check the shards when the dataset is very small.
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[[0, 1]], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
__lowerCamelCase = BatchSampler(range(2 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = [[], []]
self.check_batch_sampler_shards(__UpperCAmelCase , __UpperCAmelCase , split_batches=__UpperCAmelCase , even_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = [[0, 1, 2], [3, 4], [5, 6, 7, 8], [9, 10, 11], [12, 13]]
__lowerCamelCase = [BatchSamplerShard(__UpperCAmelCase , 2 , __UpperCAmelCase , even_batches=__UpperCAmelCase ) for i in range(2 )]
self.assertEqual(len(batch_sampler_shards[0] ) , 3 )
self.assertEqual(len(batch_sampler_shards[1] ) , 2 )
self.assertListEqual(list(batch_sampler_shards[0] ) , [[0, 1, 2], [5, 6, 7, 8], [12, 13]] )
self.assertListEqual(list(batch_sampler_shards[1] ) , [[3, 4], [9, 10, 11]] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase=False , __UpperCAmelCase=2 , __UpperCAmelCase=False ):
'''simple docstring'''
random.seed(__UpperCAmelCase )
__lowerCamelCase = list(__UpperCAmelCase )
__lowerCamelCase = [
IterableDatasetShard(
__UpperCAmelCase , batch_size=__UpperCAmelCase , drop_last=__UpperCAmelCase , num_processes=__UpperCAmelCase , process_index=__UpperCAmelCase , split_batches=__UpperCAmelCase , )
for i in range(__UpperCAmelCase )
]
__lowerCamelCase = []
for iterable_dataset_shard in iterable_dataset_shards:
# Since our random iterable dataset will be... random... we need to use a seed to get reproducible results.
random.seed(__UpperCAmelCase )
iterable_dataset_lists.append(list(__UpperCAmelCase ) )
__lowerCamelCase = batch_size // num_processes if split_batches else batch_size
# All iterable dataset shard should have the same length, a round multiple of shard_batch_size
__lowerCamelCase = iterable_dataset_lists[0]
for l in iterable_dataset_lists[1:]:
self.assertEqual(len(__UpperCAmelCase ) , len(__UpperCAmelCase ) )
self.assertTrue(len(__UpperCAmelCase ) % shard_batch_size == 0 )
__lowerCamelCase = []
for idx in range(0 , len(__UpperCAmelCase ) , __UpperCAmelCase ):
for l in iterable_dataset_lists:
observed += l[idx : idx + shard_batch_size]
if not drop_last:
while len(__UpperCAmelCase ) < len(__UpperCAmelCase ):
reference += reference
self.assertListEqual(__UpperCAmelCase , reference[: len(__UpperCAmelCase )] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = 42
__lowerCamelCase = RandomIterableDataset()
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
# Edge case with a very small dataset
__lowerCamelCase = RandomIterableDataset(max_length=2 )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
self.check_iterable_dataset_shards(__UpperCAmelCase , __UpperCAmelCase , batch_size=4 , drop_last=__UpperCAmelCase , split_batches=__UpperCAmelCase )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = BatchSampler(range(16 ) , batch_size=4 , drop_last=__UpperCAmelCase )
__lowerCamelCase = SkipBatchSampler(__UpperCAmelCase , 2 )
self.assertListEqual(list(__UpperCAmelCase ) , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = SkipDataLoader(list(range(16 ) ) , batch_size=4 , skip_batches=2 )
self.assertListEqual([t.tolist() for t in dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoader(list(range(16 ) ) , batch_size=4 )
__lowerCamelCase = skip_first_batches(__UpperCAmelCase , num_batches=2 )
self.assertListEqual([t.tolist() for t in new_dataloader] , [[8, 9, 10, 11], [12, 13, 14, 15]] )
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = DataLoaderShard(list(range(16 ) ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
def lowerCamelCase ( self ):
'''simple docstring'''
Accelerator()
__lowerCamelCase = DataLoaderDispatcher(range(16 ) , batch_size=4 )
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
# Test it also works on the second iteration
for idx, _ in enumerate(__UpperCAmelCase ):
self.assertEqual(dataloader.end_of_dataloader , idx == 3 )
| 330 | 1 |
import numpy
# List of input, output pairs
a_ = (
((5, 2, 3), 15),
((6, 5, 9), 25),
((11, 12, 13), 41),
((1, 1, 1), 8),
((11, 12, 13), 41),
)
a_ = (((515, 22, 13), 555), ((61, 35, 49), 150))
a_ = [2, 4, 1, 5]
a_ = len(train_data)
a_ = 0.0_09
def a__ ( _UpperCamelCase : Tuple ,_UpperCamelCase : int="train" ):
return calculate_hypothesis_value(_UpperCamelCase ,_UpperCamelCase ) - output(
_UpperCamelCase ,_UpperCamelCase )
def a__ ( _UpperCamelCase : Tuple ):
__lowerCamelCase = 0
for i in range(len(_UpperCamelCase ) - 1 ):
hyp_val += data_input_tuple[i] * parameter_vector[i + 1]
hyp_val += parameter_vector[0]
return hyp_val
def a__ ( _UpperCamelCase : Any ,_UpperCamelCase : Optional[Any] ):
if data_set == "train":
return train_data[example_no][1]
elif data_set == "test":
return test_data[example_no][1]
return None
def a__ ( _UpperCamelCase : Dict ,_UpperCamelCase : List[str] ):
if data_set == "train":
return _hypothesis_value(train_data[example_no][0] )
elif data_set == "test":
return _hypothesis_value(test_data[example_no][0] )
return None
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : List[Any]=m ):
__lowerCamelCase = 0
for i in range(_UpperCamelCase ):
if index == -1:
summation_value += _error(_UpperCamelCase )
else:
summation_value += _error(_UpperCamelCase ) * train_data[i][0][index]
return summation_value
def a__ ( _UpperCamelCase : Optional[Any] ):
__lowerCamelCase = summation_of_cost_derivative(_UpperCamelCase ,_UpperCamelCase ) / m
return cost_derivative_value
def a__ ( ):
global parameter_vector
# Tune these values to set a tolerance value for predicted output
__lowerCamelCase = 0.000_002
__lowerCamelCase = 0
__lowerCamelCase = 0
while True:
j += 1
__lowerCamelCase = [0, 0, 0, 0]
for i in range(0 ,len(_UpperCamelCase ) ):
__lowerCamelCase = get_cost_derivative(i - 1 )
__lowerCamelCase = (
parameter_vector[i] - LEARNING_RATE * cost_derivative
)
if numpy.allclose(
_UpperCamelCase ,_UpperCamelCase ,atol=_UpperCamelCase ,rtol=_UpperCamelCase ,):
break
__lowerCamelCase = temp_parameter_vector
print(('''Number of iterations:''', j) )
def a__ ( ):
for i in range(len(_UpperCamelCase ) ):
print(('''Actual output value:''', output(_UpperCamelCase ,'''test''' )) )
print(('''Hypothesis output:''', calculate_hypothesis_value(_UpperCamelCase ,'''test''' )) )
if __name__ == "__main__":
run_gradient_descent()
print("""\nTesting gradient descent for a linear hypothesis function.\n""")
test_gradient_descent()
| 330 |
def a__ ( _UpperCamelCase : int ):
__lowerCamelCase = n ** (1 / 3)
return (val * val * val) == n
if __name__ == "__main__":
print(perfect_cube(27))
print(perfect_cube(4))
| 330 | 1 |
from abc import ABC, abstractmethod
from typing import List, Optional
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self ):
'''simple docstring'''
# test for the above condition
self.test()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = 0
__lowerCamelCase = False
while not completed:
if counter == 1:
self.reset()
__lowerCamelCase = self.advance()
if not self.does_advance(__UpperCAmelCase ):
raise Exception(
'''Custom Constraint is not defined correctly. self.does_advance(self.advance()) must be true.''' )
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = self.update(__UpperCAmelCase )
counter += 1
if counter > 10000:
raise Exception('''update() does not fulfill the constraint.''' )
if self.remaining() != 0:
raise Exception('''Custom Constraint is not defined correctly.''' )
@abstractmethod
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError(
F"""{self.__class__} is an abstract class. Only classes inheriting this class can be called.""" )
@abstractmethod
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
raise NotImplementedError(
F"""{self.__class__} is an abstract class. Only classes inheriting this class can be called.""" )
@abstractmethod
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
raise NotImplementedError(
F"""{self.__class__} is an abstract class. Only classes inheriting this class can be called.""" )
@abstractmethod
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError(
F"""{self.__class__} is an abstract class. Only classes inheriting this class can be called.""" )
@abstractmethod
def lowerCamelCase ( self ):
'''simple docstring'''
raise NotImplementedError(
F"""{self.__class__} is an abstract class. Only classes inheriting this class can be called.""" )
@abstractmethod
def lowerCamelCase ( self , __UpperCAmelCase=False ):
'''simple docstring'''
raise NotImplementedError(
F"""{self.__class__} is an abstract class. Only classes inheriting this class can be called.""" )
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
super(__UpperCAmelCase , self ).__init__()
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or len(__UpperCAmelCase ) == 0:
raise ValueError(F"""`token_ids` has to be a non-empty list, but is {token_ids}.""" )
if any((not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or token_id < 0) for token_id in token_ids ):
raise ValueError(F"""Each list in `token_ids` has to be a list of positive integers, but is {token_ids}.""" )
__lowerCamelCase = token_ids
__lowerCamelCase = len(self.token_ids )
__lowerCamelCase = -1 # the index of the currently fulfilled step
__lowerCamelCase = False
def lowerCamelCase ( self ):
'''simple docstring'''
if self.completed:
return None
return self.token_ids[self.fulfilled_idx + 1]
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError(F"""`token_id` has to be an `int`, but is {token_id} of type {type(__UpperCAmelCase )}""" )
if self.completed:
return False
return token_id == self.token_ids[self.fulfilled_idx + 1]
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError(F"""`token_id` has to be an `int`, but is {token_id} of type {type(__UpperCAmelCase )}""" )
__lowerCamelCase = False
__lowerCamelCase = False
__lowerCamelCase = False
if self.does_advance(__UpperCAmelCase ):
self.fulfilled_idx += 1
__lowerCamelCase = True
if self.fulfilled_idx == (self.seqlen - 1):
__lowerCamelCase = True
__lowerCamelCase = completed
else:
# failed to make progress.
__lowerCamelCase = True
self.reset()
return stepped, completed, reset
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = False
__lowerCamelCase = 0
def lowerCamelCase ( self ):
'''simple docstring'''
return self.seqlen - (self.fulfilled_idx + 1)
def lowerCamelCase ( self , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = PhrasalConstraint(self.token_ids )
if stateful:
__lowerCamelCase = self.seqlen
__lowerCamelCase = self.fulfilled_idx
__lowerCamelCase = self.completed
return new_constraint
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = max([len(__UpperCAmelCase ) for one in nested_token_ids] )
__lowerCamelCase = {}
for token_ids in nested_token_ids:
__lowerCamelCase = root
for tidx, token_id in enumerate(__UpperCAmelCase ):
if token_id not in level:
__lowerCamelCase = {}
__lowerCamelCase = level[token_id]
if no_subsets and self.has_subsets(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError(
'''Each list in `nested_token_ids` can\'t be a complete subset of another list, but is'''
F""" {nested_token_ids}.""" )
__lowerCamelCase = root
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.trie
for current_token in current_seq:
__lowerCamelCase = start[current_token]
__lowerCamelCase = list(start.keys() )
return next_tokens
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.next_tokens(__UpperCAmelCase )
return len(__UpperCAmelCase ) == 0
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = list(root.values() )
if len(__UpperCAmelCase ) == 0:
return 1
else:
return sum([self.count_leaves(__UpperCAmelCase ) for nn in next_nodes] )
def lowerCamelCase ( self , __UpperCAmelCase , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = self.count_leaves(__UpperCAmelCase )
return len(__UpperCAmelCase ) != leaf_count
class __lowerCAmelCase ( lowerCAmelCase__ ):
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
super(__UpperCAmelCase , self ).__init__()
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or len(__UpperCAmelCase ) == 0:
raise ValueError(F"""`nested_token_ids` has to be a non-empty list, but is {nested_token_ids}.""" )
if any(not isinstance(__UpperCAmelCase , __UpperCAmelCase ) for token_ids in nested_token_ids ):
raise ValueError(F"""`nested_token_ids` has to be a list of lists, but is {nested_token_ids}.""" )
if any(
any((not isinstance(__UpperCAmelCase , __UpperCAmelCase ) or token_id < 0) for token_id in token_ids )
for token_ids in nested_token_ids ):
raise ValueError(
F"""Each list in `nested_token_ids` has to be a list of positive integers, but is {nested_token_ids}.""" )
__lowerCamelCase = DisjunctiveTrie(__UpperCAmelCase )
__lowerCamelCase = nested_token_ids
__lowerCamelCase = self.trie.max_height
__lowerCamelCase = []
__lowerCamelCase = False
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = self.trie.next_tokens(self.current_seq )
if len(__UpperCAmelCase ) == 0:
return None
else:
return token_list
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError(F"""`token_id` is supposed to be type `int`, but is {token_id} of type {type(__UpperCAmelCase )}""" )
__lowerCamelCase = self.trie.next_tokens(self.current_seq )
return token_id in next_tokens
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError(F"""`token_id` is supposed to be type `int`, but is {token_id} of type {type(__UpperCAmelCase )}""" )
__lowerCamelCase = False
__lowerCamelCase = False
__lowerCamelCase = False
if self.does_advance(__UpperCAmelCase ):
self.current_seq.append(__UpperCAmelCase )
__lowerCamelCase = True
else:
__lowerCamelCase = True
self.reset()
__lowerCamelCase = self.trie.reached_leaf(self.current_seq )
__lowerCamelCase = completed
return stepped, completed, reset
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = False
__lowerCamelCase = []
def lowerCamelCase ( self ):
'''simple docstring'''
if self.completed:
# since this can be completed without reaching max height
return 0
else:
return self.seqlen - len(self.current_seq )
def lowerCamelCase ( self , __UpperCAmelCase=False ):
'''simple docstring'''
__lowerCamelCase = DisjunctiveConstraint(self.token_ids )
if stateful:
__lowerCamelCase = self.seqlen
__lowerCamelCase = self.current_seq
__lowerCamelCase = self.completed
return new_constraint
class __lowerCAmelCase :
def __init__( self , __UpperCAmelCase ):
'''simple docstring'''
__lowerCamelCase = constraints
# max # of steps required to fulfill a given constraint
__lowerCamelCase = max([c.seqlen for c in constraints] )
__lowerCamelCase = len(__UpperCAmelCase )
__lowerCamelCase = False
self.init_state()
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = []
__lowerCamelCase = None
__lowerCamelCase = [constraint.copy(stateful=__UpperCAmelCase ) for constraint in self.constraints]
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = 0
if self.inprogress_constraint:
# extra points for having a constraint mid-fulfilled
add += self.max_seqlen - self.inprogress_constraint.remaining()
return (len(self.complete_constraints ) * self.max_seqlen) + add
def lowerCamelCase ( self ):
'''simple docstring'''
__lowerCamelCase = []
if self.inprogress_constraint is None:
for constraint in self.pending_constraints: # "pending" == "unfulfilled yet"
__lowerCamelCase = constraint.advance()
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
token_list.append(__UpperCAmelCase )
elif isinstance(__UpperCAmelCase , __UpperCAmelCase ):
token_list.extend(__UpperCAmelCase )
else:
__lowerCamelCase = self.inprogress_constraint.advance()
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
token_list.append(__UpperCAmelCase )
elif isinstance(__UpperCAmelCase , __UpperCAmelCase ):
token_list.extend(__UpperCAmelCase )
if len(__UpperCAmelCase ) == 0:
return None
else:
return token_list
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
self.init_state()
if token_ids is not None:
for token in token_ids:
# completes or steps **one** constraint
__lowerCamelCase ,__lowerCamelCase = self.add(__UpperCAmelCase )
# the entire list of constraints are fulfilled
if self.completed:
break
def lowerCamelCase ( self , __UpperCAmelCase ):
'''simple docstring'''
if not isinstance(__UpperCAmelCase , __UpperCAmelCase ):
raise ValueError(F"""`token_id` should be an `int`, but is `{token_id}`.""" )
__lowerCamelCase ,__lowerCamelCase = False, False
if self.completed:
__lowerCamelCase = True
__lowerCamelCase = False
return complete, stepped
if self.inprogress_constraint is not None:
# In the middle of fulfilling a constraint. If the `token_id` *does* makes an incremental progress to current
# job, simply update the state
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = self.inprogress_constraint.update(__UpperCAmelCase )
if reset:
# 1. If the next token breaks the progress, then we must restart.
# e.g. constraint = "I love pies" and sequence so far is "I love" but `token_id` == "books".
# But that doesn't mean we self.init_state(), since we only reset the state for this particular
# constraint, not the full list of constraints.
self.pending_constraints.append(self.inprogress_constraint.copy(stateful=__UpperCAmelCase ) )
__lowerCamelCase = None
if complete:
# 2. If the next token completes the constraint, move it to completed list, set
# inprogress to None. If there are no pending constraints either, then this full list of constraints
# is complete.
self.complete_constraints.append(self.inprogress_constraint )
__lowerCamelCase = None
if len(self.pending_constraints ) == 0:
# we're done!
__lowerCamelCase = True
else:
# Not in the middle of fulfilling a constraint. So does this `token_id` helps us step towards any of our list
# of constraints?
for cidx, pending_constraint in enumerate(self.pending_constraints ):
if pending_constraint.does_advance(__UpperCAmelCase ):
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = pending_constraint.update(__UpperCAmelCase )
if not stepped:
raise Exception(
'''`constraint.update(token_id)` is not yielding incremental progress, '''
'''even though `constraint.does_advance(token_id)` is true.''' )
if complete:
self.complete_constraints.append(__UpperCAmelCase )
__lowerCamelCase = None
if not complete and stepped:
__lowerCamelCase = pending_constraint
if complete or stepped:
# If we made any progress at all, then it's at least not a "pending constraint".
__lowerCamelCase = (
self.pending_constraints[:cidx] + self.pending_constraints[cidx + 1 :]
)
if len(self.pending_constraints ) == 0 and self.inprogress_constraint is None:
# If there's no longer any pending after this and no inprogress either, then we must be
# complete.
__lowerCamelCase = True
break # prevent accidentally stepping through multiple constraints with just one token.
return complete, stepped
def lowerCamelCase ( self , __UpperCAmelCase=True ):
'''simple docstring'''
__lowerCamelCase = ConstraintListState(self.constraints ) # we actually never though self.constraints objects
# throughout this process. So it's at initialization state.
if stateful:
__lowerCamelCase = [
constraint.copy(stateful=__UpperCAmelCase ) for constraint in self.complete_constraints
]
if self.inprogress_constraint is not None:
__lowerCamelCase = self.inprogress_constraint.copy(stateful=__UpperCAmelCase )
__lowerCamelCase = [constraint.copy() for constraint in self.pending_constraints]
return new_state
| 330 |
import argparse
import os
# New Code #
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
from accelerate.utils import find_executable_batch_size
########################################################################
# This is a fully working simple example to use Accelerate,
# specifically showcasing how to ensure out-of-memory errors never
# interrupt training, and builds off the `nlp_example.py` script.
#
# This example trains a Bert base model on GLUE MRPC
# in any of the following settings (with the same script):
# - single CPU or single GPU
# - multi GPUS (using PyTorch distributed mode)
# - (multi) TPUs
# - fp16 (mixed-precision) or fp32 (normal precision)
#
# New additions from the base script can be found quickly by
# looking for the # New Code # tags
#
# To run it in each of these various modes, follow the instructions
# in the readme for examples:
# https://github.com/huggingface/accelerate/tree/main/examples
#
########################################################################
a_ = 16
a_ = 32
def a__ ( _UpperCamelCase : Accelerator ,_UpperCamelCase : int = 16 ):
__lowerCamelCase = AutoTokenizer.from_pretrained('''bert-base-cased''' )
__lowerCamelCase = load_dataset('''glue''' ,'''mrpc''' )
def tokenize_function(_UpperCamelCase : Optional[Any] ):
# max_length=None => use the model max length (it's actually the default)
__lowerCamelCase = tokenizer(examples['''sentence1'''] ,examples['''sentence2'''] ,truncation=_UpperCamelCase ,max_length=_UpperCamelCase )
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(
_UpperCamelCase ,batched=_UpperCamelCase ,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(_UpperCamelCase : Any ):
# On TPU it's best to pad everything to the same length or training will be very slow.
__lowerCamelCase = 1_28 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(
_UpperCamelCase ,padding='''longest''' ,max_length=_UpperCamelCase ,pad_to_multiple_of=_UpperCamelCase ,return_tensors='''pt''' ,)
# Instantiate dataloaders.
__lowerCamelCase = DataLoader(
tokenized_datasets['''train'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
__lowerCamelCase = DataLoader(
tokenized_datasets['''validation'''] ,shuffle=_UpperCamelCase ,collate_fn=_UpperCamelCase ,batch_size=_UpperCamelCase )
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
a_ = mocked_dataloaders # noqa: F811
def a__ ( _UpperCamelCase : str ,_UpperCamelCase : str ):
# For testing only
if os.environ.get('''TESTING_MOCKED_DATALOADERS''' ,_UpperCamelCase ) == "1":
__lowerCamelCase = 2
# Initialize accelerator
__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'''] )
__lowerCamelCase = evaluate.load('''glue''' ,'''mrpc''' )
# New Code #
# We now can define an inner training loop function. It should take a batch size as the only parameter,
# and build the dataloaders in there.
# It also gets our decorator
@find_executable_batch_size(starting_batch_size=_UpperCamelCase )
def inner_training_loop(_UpperCamelCase : Union[str, Any] ):
# And now just move everything below under this function
# We need to bring in the Accelerator object from earlier
nonlocal accelerator
# And reset all of its attributes that could hold onto any memory:
accelerator.free_memory()
# Then we can declare the model, optimizer, and everything else:
set_seed(_UpperCamelCase )
# 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=_UpperCamelCase )
# 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=_UpperCamelCase )
__lowerCamelCase ,__lowerCamelCase = get_dataloaders(_UpperCamelCase ,_UpperCamelCase )
# Instantiate scheduler
__lowerCamelCase = get_linear_schedule_with_warmup(
optimizer=_UpperCamelCase ,num_warmup_steps=1_00 ,num_training_steps=(len(_UpperCamelCase ) * num_epochs) ,)
# Prepare everything
# There is no specific order to remember, we just need to unpack the objects in the same order we gave them to the
# prepare method.
__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase ,__lowerCamelCase = accelerator.prepare(
_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase ,_UpperCamelCase )
# Now we train the model
for epoch in range(_UpperCamelCase ):
model.train()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.loss
accelerator.backward(_UpperCamelCase )
optimizer.step()
lr_scheduler.step()
optimizer.zero_grad()
model.eval()
for step, batch in enumerate(_UpperCamelCase ):
# We could avoid this line since we set the accelerator with `device_placement=True`.
batch.to(accelerator.device )
with torch.no_grad():
__lowerCamelCase = model(**_UpperCamelCase )
__lowerCamelCase = outputs.logits.argmax(dim=-1 )
__lowerCamelCase ,__lowerCamelCase = accelerator.gather_for_metrics((predictions, batch['''labels''']) )
metric.add_batch(
predictions=_UpperCamelCase ,references=_UpperCamelCase ,)
__lowerCamelCase = metric.compute()
# Use accelerator.print to print only on the main process.
accelerator.print(F"""epoch {epoch}:""" ,_UpperCamelCase )
# New Code #
# And call it at the end with no arguments
# Note: You could also refactor this outside of your training loop function
inner_training_loop()
def a__ ( ):
__lowerCamelCase = argparse.ArgumentParser(description='''Simple example of training script.''' )
parser.add_argument(
'''--mixed_precision''' ,type=_UpperCamelCase ,default=_UpperCamelCase ,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.''' )
__lowerCamelCase = parser.parse_args()
__lowerCamelCase = {'''lr''': 2e-5, '''num_epochs''': 3, '''seed''': 42, '''batch_size''': 16}
training_function(_UpperCamelCase ,_UpperCamelCase )
if __name__ == "__main__":
main()
| 330 | 1 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.