code
stringlengths 82
53.2k
| code_codestyle
int64 0
721
| style_context
stringlengths 91
41.9k
| style_context_codestyle
int64 0
699
| label
int64 0
1
|
|---|---|---|---|---|
'''simple docstring'''
from typing import Optional
import torch
import torch.utils.checkpoint
from torch import Tensor, nn
from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss
from ...activations import ACTaFN
from ...file_utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward
from ...modeling_outputs import (
BaseModelOutputWithNoAttention,
BaseModelOutputWithPoolingAndNoAttention,
ImageClassifierOutputWithNoAttention,
)
from ...modeling_utils import PreTrainedModel
from ...utils import logging
from .configuration_regnet import RegNetConfig
_a : Optional[Any] = logging.get_logger(__name__)
# General docstring
_a : Optional[Any] = "RegNetConfig"
# Base docstring
_a : int = "facebook/regnet-y-040"
_a : Tuple = [1, 1_088, 7, 7]
# Image classification docstring
_a : List[str] = "facebook/regnet-y-040"
_a : Any = "tabby, tabby cat"
_a : Tuple = [
"facebook/regnet-y-040",
# See all regnet models at https://huggingface.co/models?filter=regnet
]
class _lowercase ( nn.Module ):
def __init__( self : Optional[int] , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int = 3 , SCREAMING_SNAKE_CASE_ : int = 1 , SCREAMING_SNAKE_CASE_ : int = 1 , SCREAMING_SNAKE_CASE_ : Optional[str] = "relu" , ) -> List[str]:
super().__init__()
__snake_case = nn.Convad(
SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , kernel_size=SCREAMING_SNAKE_CASE_ , stride=SCREAMING_SNAKE_CASE_ , padding=kernel_size // 2 , groups=SCREAMING_SNAKE_CASE_ , bias=SCREAMING_SNAKE_CASE_ , )
__snake_case = nn.BatchNormad(SCREAMING_SNAKE_CASE_ )
__snake_case = ACTaFN[activation] if activation is not None else nn.Identity()
def a ( self : int , SCREAMING_SNAKE_CASE_ : Dict ) -> Dict:
__snake_case = self.convolution(SCREAMING_SNAKE_CASE_ )
__snake_case = self.normalization(SCREAMING_SNAKE_CASE_ )
__snake_case = self.activation(SCREAMING_SNAKE_CASE_ )
return hidden_state
class _lowercase ( nn.Module ):
def __init__( self : Tuple , SCREAMING_SNAKE_CASE_ : RegNetConfig ) -> Dict:
super().__init__()
__snake_case = RegNetConvLayer(
config.num_channels , config.embedding_size , kernel_size=3 , stride=2 , activation=config.hidden_act )
__snake_case = config.num_channels
def a ( self : int , SCREAMING_SNAKE_CASE_ : Optional[Any] ) -> Dict:
__snake_case = pixel_values.shape[1]
if 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.' )
__snake_case = self.embedder(SCREAMING_SNAKE_CASE_ )
return hidden_state
class _lowercase ( nn.Module ):
def __init__( self : Tuple , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int = 2 ) -> List[str]:
super().__init__()
__snake_case = nn.Convad(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , kernel_size=1 , stride=SCREAMING_SNAKE_CASE_ , bias=SCREAMING_SNAKE_CASE_ )
__snake_case = nn.BatchNormad(SCREAMING_SNAKE_CASE_ )
def a ( self : List[Any] , SCREAMING_SNAKE_CASE_ : Tensor ) -> Tensor:
__snake_case = self.convolution(SCREAMING_SNAKE_CASE_ )
__snake_case = self.normalization(SCREAMING_SNAKE_CASE_ )
return hidden_state
class _lowercase ( nn.Module ):
def __init__( self : Dict , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int ) -> int:
super().__init__()
__snake_case = nn.AdaptiveAvgPoolad((1, 1) )
__snake_case = nn.Sequential(
nn.Convad(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , kernel_size=1 ) , nn.ReLU() , nn.Convad(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , kernel_size=1 ) , nn.Sigmoid() , )
def a ( self : Optional[Any] , SCREAMING_SNAKE_CASE_ : Union[str, Any] ) -> List[Any]:
# b c h w -> b c 1 1
__snake_case = self.pooler(SCREAMING_SNAKE_CASE_ )
__snake_case = self.attention(SCREAMING_SNAKE_CASE_ )
__snake_case = hidden_state * attention
return hidden_state
class _lowercase ( nn.Module ):
def __init__( self : List[Any] , SCREAMING_SNAKE_CASE_ : RegNetConfig , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int = 1 ) -> Tuple:
super().__init__()
__snake_case = in_channels != out_channels or stride != 1
__snake_case = max(1 , out_channels // config.groups_width )
__snake_case = (
RegNetShortCut(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , stride=SCREAMING_SNAKE_CASE_ ) if should_apply_shortcut else nn.Identity()
)
__snake_case = nn.Sequential(
RegNetConvLayer(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , kernel_size=1 , activation=config.hidden_act ) , RegNetConvLayer(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , stride=SCREAMING_SNAKE_CASE_ , groups=SCREAMING_SNAKE_CASE_ , activation=config.hidden_act ) , RegNetConvLayer(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , kernel_size=1 , activation=SCREAMING_SNAKE_CASE_ ) , )
__snake_case = ACTaFN[config.hidden_act]
def a ( self : Optional[Any] , SCREAMING_SNAKE_CASE_ : Optional[int] ) -> List[str]:
__snake_case = hidden_state
__snake_case = self.layer(SCREAMING_SNAKE_CASE_ )
__snake_case = self.shortcut(SCREAMING_SNAKE_CASE_ )
hidden_state += residual
__snake_case = self.activation(SCREAMING_SNAKE_CASE_ )
return hidden_state
class _lowercase ( nn.Module ):
def __init__( self : Any , SCREAMING_SNAKE_CASE_ : RegNetConfig , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int = 1 ) -> Optional[Any]:
super().__init__()
__snake_case = in_channels != out_channels or stride != 1
__snake_case = max(1 , out_channels // config.groups_width )
__snake_case = (
RegNetShortCut(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , stride=SCREAMING_SNAKE_CASE_ ) if should_apply_shortcut else nn.Identity()
)
__snake_case = nn.Sequential(
RegNetConvLayer(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , kernel_size=1 , activation=config.hidden_act ) , RegNetConvLayer(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , stride=SCREAMING_SNAKE_CASE_ , groups=SCREAMING_SNAKE_CASE_ , activation=config.hidden_act ) , RegNetSELayer(SCREAMING_SNAKE_CASE_ , reduced_channels=int(round(in_channels / 4 ) ) ) , RegNetConvLayer(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , kernel_size=1 , activation=SCREAMING_SNAKE_CASE_ ) , )
__snake_case = ACTaFN[config.hidden_act]
def a ( self : Optional[int] , SCREAMING_SNAKE_CASE_ : int ) -> Tuple:
__snake_case = hidden_state
__snake_case = self.layer(SCREAMING_SNAKE_CASE_ )
__snake_case = self.shortcut(SCREAMING_SNAKE_CASE_ )
hidden_state += residual
__snake_case = self.activation(SCREAMING_SNAKE_CASE_ )
return hidden_state
class _lowercase ( nn.Module ):
def __init__( self : Optional[int] , SCREAMING_SNAKE_CASE_ : RegNetConfig , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int , SCREAMING_SNAKE_CASE_ : int = 2 , SCREAMING_SNAKE_CASE_ : int = 2 , ) -> Any:
super().__init__()
__snake_case = RegNetXLayer if config.layer_type == 'x' else RegNetYLayer
__snake_case = nn.Sequential(
# downsampling is done in the first layer with stride of 2
layer(
SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , stride=SCREAMING_SNAKE_CASE_ , ) , *[layer(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ ) for _ in range(depth - 1 )] , )
def a ( self : List[Any] , SCREAMING_SNAKE_CASE_ : str ) -> Any:
__snake_case = self.layers(SCREAMING_SNAKE_CASE_ )
return hidden_state
class _lowercase ( nn.Module ):
def __init__( self : int , SCREAMING_SNAKE_CASE_ : RegNetConfig ) -> Union[str, Any]:
super().__init__()
__snake_case = nn.ModuleList([] )
# based on `downsample_in_first_stage`, the first layer of the first stage may or may not downsample the input
self.stages.append(
RegNetStage(
SCREAMING_SNAKE_CASE_ , config.embedding_size , config.hidden_sizes[0] , stride=2 if config.downsample_in_first_stage else 1 , depth=config.depths[0] , ) )
__snake_case = zip(config.hidden_sizes , config.hidden_sizes[1:] )
for (in_channels, out_channels), depth in zip(SCREAMING_SNAKE_CASE_ , config.depths[1:] ):
self.stages.append(RegNetStage(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , depth=SCREAMING_SNAKE_CASE_ ) )
def a ( self : Union[str, Any] , SCREAMING_SNAKE_CASE_ : Tensor , SCREAMING_SNAKE_CASE_ : bool = False , SCREAMING_SNAKE_CASE_ : bool = True ) -> BaseModelOutputWithNoAttention:
__snake_case = () if output_hidden_states else None
for stage_module in self.stages:
if output_hidden_states:
__snake_case = hidden_states + (hidden_state,)
__snake_case = stage_module(SCREAMING_SNAKE_CASE_ )
if output_hidden_states:
__snake_case = hidden_states + (hidden_state,)
if not return_dict:
return tuple(v for v in [hidden_state, hidden_states] if v is not None )
return BaseModelOutputWithNoAttention(last_hidden_state=SCREAMING_SNAKE_CASE_ , hidden_states=SCREAMING_SNAKE_CASE_ )
class _lowercase ( __lowercase ):
_SCREAMING_SNAKE_CASE : Optional[Any] = RegNetConfig
_SCREAMING_SNAKE_CASE : Optional[int] = "regnet"
_SCREAMING_SNAKE_CASE : int = "pixel_values"
_SCREAMING_SNAKE_CASE : Optional[int] = True
def a ( self : Union[str, Any] , SCREAMING_SNAKE_CASE_ : Any ) -> Optional[Any]:
if isinstance(SCREAMING_SNAKE_CASE_ , nn.Convad ):
nn.init.kaiming_normal_(module.weight , mode='fan_out' , nonlinearity='relu' )
elif isinstance(SCREAMING_SNAKE_CASE_ , (nn.BatchNormad, nn.GroupNorm) ):
nn.init.constant_(module.weight , 1 )
nn.init.constant_(module.bias , 0 )
def a ( self : int , SCREAMING_SNAKE_CASE_ : Tuple , SCREAMING_SNAKE_CASE_ : Tuple=False ) -> Tuple:
if isinstance(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ ):
__snake_case = value
_a : Union[str, Any] = R"\n This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass. Use it\n as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and\n behavior.\n\n Parameters:\n config ([`RegNetConfig`]): Model configuration class with all the parameters of the model.\n Initializing with a config file does not load the weights associated with the model, only the\n configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights.\n"
_a : Dict = R"\n Args:\n pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)`):\n Pixel values. Pixel values can be obtained using [`AutoImageProcessor`]. See\n [`ConvNextImageProcessor.__call__`] for details.\n\n output_hidden_states (`bool`, *optional*):\n Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for\n more detail.\n return_dict (`bool`, *optional*):\n Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple.\n"
@add_start_docstrings(
"The bare RegNet model outputting raw features without any specific head on top." , __lowercase , )
# Copied from transformers.models.resnet.modeling_resnet.ResNetModel with RESNET->REGNET,ResNet->RegNet
class _lowercase ( __lowercase ):
def __init__( self : Optional[Any] , SCREAMING_SNAKE_CASE_ : List[Any] ) -> Tuple:
super().__init__(SCREAMING_SNAKE_CASE_ )
__snake_case = config
__snake_case = RegNetEmbeddings(SCREAMING_SNAKE_CASE_ )
__snake_case = RegNetEncoder(SCREAMING_SNAKE_CASE_ )
__snake_case = nn.AdaptiveAvgPoolad((1, 1) )
# Initialize weights and apply final processing
self.post_init()
@add_start_docstrings_to_model_forward(SCREAMING_SNAKE_CASE_ )
@add_code_sample_docstrings(
checkpoint=_CHECKPOINT_FOR_DOC , output_type=SCREAMING_SNAKE_CASE_ , config_class=_CONFIG_FOR_DOC , modality='vision' , expected_output=_EXPECTED_OUTPUT_SHAPE , )
def a ( self : List[Any] , SCREAMING_SNAKE_CASE_ : Tensor , SCREAMING_SNAKE_CASE_ : Optional[bool] = None , SCREAMING_SNAKE_CASE_ : Optional[bool] = None ) -> BaseModelOutputWithPoolingAndNoAttention:
__snake_case = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
__snake_case = return_dict if return_dict is not None else self.config.use_return_dict
__snake_case = self.embedder(SCREAMING_SNAKE_CASE_ )
__snake_case = self.encoder(
SCREAMING_SNAKE_CASE_ , output_hidden_states=SCREAMING_SNAKE_CASE_ , return_dict=SCREAMING_SNAKE_CASE_ )
__snake_case = encoder_outputs[0]
__snake_case = self.pooler(SCREAMING_SNAKE_CASE_ )
if not return_dict:
return (last_hidden_state, pooled_output) + encoder_outputs[1:]
return BaseModelOutputWithPoolingAndNoAttention(
last_hidden_state=SCREAMING_SNAKE_CASE_ , pooler_output=SCREAMING_SNAKE_CASE_ , hidden_states=encoder_outputs.hidden_states , )
@add_start_docstrings(
"\n RegNet Model with an image classification head on top (a linear layer on top of the pooled features), e.g. for\n ImageNet.\n " , __lowercase , )
# Copied from transformers.models.resnet.modeling_resnet.ResNetForImageClassification with RESNET->REGNET,ResNet->RegNet,resnet->regnet
class _lowercase ( __lowercase ):
def __init__( self : Union[str, Any] , SCREAMING_SNAKE_CASE_ : Tuple ) -> Any:
super().__init__(SCREAMING_SNAKE_CASE_ )
__snake_case = config.num_labels
__snake_case = RegNetModel(SCREAMING_SNAKE_CASE_ )
# classification head
__snake_case = nn.Sequential(
nn.Flatten() , nn.Linear(config.hidden_sizes[-1] , config.num_labels ) if config.num_labels > 0 else nn.Identity() , )
# initialize weights and apply final processing
self.post_init()
@add_start_docstrings_to_model_forward(SCREAMING_SNAKE_CASE_ )
@add_code_sample_docstrings(
checkpoint=_IMAGE_CLASS_CHECKPOINT , output_type=SCREAMING_SNAKE_CASE_ , config_class=_CONFIG_FOR_DOC , expected_output=_IMAGE_CLASS_EXPECTED_OUTPUT , )
def a ( self : int , SCREAMING_SNAKE_CASE_ : Optional[torch.FloatTensor] = None , SCREAMING_SNAKE_CASE_ : Optional[torch.LongTensor] = None , SCREAMING_SNAKE_CASE_ : Optional[bool] = None , SCREAMING_SNAKE_CASE_ : Optional[bool] = None , ) -> ImageClassifierOutputWithNoAttention:
__snake_case = return_dict if return_dict is not None else self.config.use_return_dict
__snake_case = self.regnet(SCREAMING_SNAKE_CASE_ , output_hidden_states=SCREAMING_SNAKE_CASE_ , return_dict=SCREAMING_SNAKE_CASE_ )
__snake_case = outputs.pooler_output if return_dict else outputs[1]
__snake_case = self.classifier(SCREAMING_SNAKE_CASE_ )
__snake_case = None
if labels is not None:
if self.config.problem_type is None:
if self.num_labels == 1:
__snake_case = 'regression'
elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
__snake_case = 'single_label_classification'
else:
__snake_case = 'multi_label_classification'
if self.config.problem_type == "regression":
__snake_case = MSELoss()
if self.num_labels == 1:
__snake_case = loss_fct(logits.squeeze() , labels.squeeze() )
else:
__snake_case = loss_fct(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ )
elif self.config.problem_type == "single_label_classification":
__snake_case = CrossEntropyLoss()
__snake_case = loss_fct(logits.view(-1 , self.num_labels ) , labels.view(-1 ) )
elif self.config.problem_type == "multi_label_classification":
__snake_case = BCEWithLogitsLoss()
__snake_case = loss_fct(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ )
if not return_dict:
__snake_case = (logits,) + outputs[2:]
return (loss,) + output if loss is not None else output
return ImageClassifierOutputWithNoAttention(loss=SCREAMING_SNAKE_CASE_ , logits=SCREAMING_SNAKE_CASE_ , hidden_states=outputs.hidden_states )
| 56
|
'''simple docstring'''
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 (lowercase__ : Optional[Any] ) -> List[str]:
"""simple docstring"""
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 _lowercase ( nn.Module ):
def __init__( self : Dict , SCREAMING_SNAKE_CASE_ : nn.Module , SCREAMING_SNAKE_CASE_ : int ) -> str:
super().__init__()
__snake_case = module
__snake_case = nn.Sequential(
nn.Linear(module.in_features , SCREAMING_SNAKE_CASE_ , bias=SCREAMING_SNAKE_CASE_ ) , nn.Linear(SCREAMING_SNAKE_CASE_ , module.out_features , bias=SCREAMING_SNAKE_CASE_ ) , )
__snake_case = (2.0 / (5 * min(module.in_features , module.out_features ))) ** 0.5
nn.init.normal_(self.adapter[0].weight , std=SCREAMING_SNAKE_CASE_ )
nn.init.zeros_(self.adapter[1].weight )
self.adapter.to(module.weight.device )
def a ( self : Optional[Any] , SCREAMING_SNAKE_CASE_ : Any , *SCREAMING_SNAKE_CASE_ : List[Any] , **SCREAMING_SNAKE_CASE_ : List[str] ) -> Union[str, Any]:
return self.module(SCREAMING_SNAKE_CASE_ , *SCREAMING_SNAKE_CASE_ , **SCREAMING_SNAKE_CASE_ ) + self.adapter(SCREAMING_SNAKE_CASE_ )
@require_bitsandbytes
@require_accelerate
@require_torch
@require_torch_gpu
@slow
class _lowercase ( 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
_SCREAMING_SNAKE_CASE : Tuple = "bigscience/bloom-1b7"
# Constant values
_SCREAMING_SNAKE_CASE : Union[str, Any] = 2.109659552692574
_SCREAMING_SNAKE_CASE : Optional[Any] = "Hello my name is"
_SCREAMING_SNAKE_CASE : List[str] = 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" )
_SCREAMING_SNAKE_CASE : Dict = 1_0
def a ( self : Optional[Any] ) -> List[Any]:
# Models and tokenizer
__snake_case = AutoTokenizer.from_pretrained(self.model_name )
class _lowercase ( __lowercase ):
def a ( self : Union[str, Any] ) -> List[str]:
super().setUp()
# Models and tokenizer
__snake_case = AutoModelForCausalLM.from_pretrained(
self.model_name , torch_dtype=torch.floataa , device_map='auto' )
__snake_case = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' )
def a ( self : Optional[Any] ) -> Any:
del self.model_fpaa
del self.model_abit
gc.collect()
torch.cuda.empty_cache()
def a ( self : Optional[Any] ) -> int:
__snake_case = self.model_abit.config
self.assertTrue(hasattr(SCREAMING_SNAKE_CASE_ , 'quantization_config' ) )
__snake_case = config.to_dict()
__snake_case = config.to_diff_dict()
__snake_case = config.to_json_string()
def a ( self : Optional[Any] ) -> str:
from bitsandbytes.nn import Paramsabit
__snake_case = self.model_fpaa.get_memory_footprint()
__snake_case = self.model_abit.get_memory_footprint()
self.assertAlmostEqual(mem_fpaa / mem_abit , self.EXPECTED_RELATIVE_DIFFERENCE )
__snake_case = get_some_linear_layer(self.model_abit )
self.assertTrue(linear.weight.__class__ == Paramsabit )
def a ( self : Union[str, Any] ) -> Optional[Any]:
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(SCREAMING_SNAKE_CASE_ , 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 a ( self : Union[str, Any] ) -> int:
__snake_case = self.tokenizer(self.input_text , return_tensors='pt' )
__snake_case = 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=SCREAMING_SNAKE_CASE_ ) , self.EXPECTED_OUTPUTS )
def a ( self : Optional[Any] ) -> Dict:
__snake_case = BitsAndBytesConfig()
__snake_case = True
__snake_case = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=SCREAMING_SNAKE_CASE_ , device_map='auto' )
__snake_case = self.tokenizer(self.input_text , return_tensors='pt' )
__snake_case = 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=SCREAMING_SNAKE_CASE_ ) , self.EXPECTED_OUTPUTS )
def a ( self : List[Any] ) -> str:
with self.assertRaises(SCREAMING_SNAKE_CASE_ ), tempfile.TemporaryDirectory() as tmpdirname:
self.model_abit.save_pretrained(SCREAMING_SNAKE_CASE_ )
def a ( self : Any ) -> Union[str, Any]:
__snake_case = BitsAndBytesConfig()
with self.assertRaises(SCREAMING_SNAKE_CASE_ ):
__snake_case = AutoModelForCausalLM.from_pretrained(
self.model_name , quantization_config=SCREAMING_SNAKE_CASE_ , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' , bnb_abit_quant_type='nf4' , )
def a ( self : Tuple ) -> Dict:
with self.assertRaises(SCREAMING_SNAKE_CASE_ ):
# Tries with `str`
self.model_abit.to('cpu' )
with self.assertRaises(SCREAMING_SNAKE_CASE_ ):
# Tries with a `dtype``
self.model_abit.to(torch.floataa )
with self.assertRaises(SCREAMING_SNAKE_CASE_ ):
# Tries with a `device`
self.model_abit.to(torch.device('cuda:0' ) )
with self.assertRaises(SCREAMING_SNAKE_CASE_ ):
# Tries with a `device`
self.model_abit.float()
with self.assertRaises(SCREAMING_SNAKE_CASE_ ):
# Tries with a `device`
self.model_abit.half()
# Test if we did not break anything
__snake_case = self.tokenizer(self.input_text , return_tensors='pt' )
__snake_case = self.model_fpaa.to(torch.floataa )
__snake_case = self.model_fpaa.generate(input_ids=encoded_input['input_ids'].to(0 ) , max_new_tokens=10 )
# Check this does not throw an error
__snake_case = self.model_fpaa.to('cpu' )
# Check this does not throw an error
__snake_case = self.model_fpaa.half()
# Check this does not throw an error
__snake_case = self.model_fpaa.float()
def a ( self : Tuple ) -> Union[str, Any]:
__snake_case = AutoModelForSeqaSeqLM.from_pretrained('t5-small' , load_in_abit=SCREAMING_SNAKE_CASE_ , 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 _lowercase ( unittest.TestCase ):
@classmethod
def a ( cls : Union[str, Any] ) -> Dict:
__snake_case = 't5-small'
__snake_case = 'google/flan-t5-small' # flan-t5 uses dense-act instead of dense-relu-dense
__snake_case = AutoTokenizer.from_pretrained(cls.model_name )
__snake_case = 'Translate in German: Hello, my dog is cute'
def a ( self : List[Any] ) -> str:
gc.collect()
torch.cuda.empty_cache()
def a ( self : int ) -> Optional[Any]:
from transformers import TaForConditionalGeneration
__snake_case = TaForConditionalGeneration._keep_in_fpaa_modules
__snake_case = None
# test with `t5-small`
__snake_case = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' )
__snake_case = self.tokenizer(self.input_text , return_tensors='pt' ).to(0 )
__snake_case = model.generate(**SCREAMING_SNAKE_CASE_ )
# test with `flan-t5-small`
__snake_case = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' )
__snake_case = self.tokenizer(self.input_text , return_tensors='pt' ).to(0 )
__snake_case = model.generate(**SCREAMING_SNAKE_CASE_ )
__snake_case = modules
def a ( self : List[str] ) -> Any:
import bitsandbytes as bnb
from transformers import TaForConditionalGeneration
# test with `t5-small`
__snake_case = TaForConditionalGeneration.from_pretrained(self.model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , 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 ) )
__snake_case = self.tokenizer(self.input_text , return_tensors='pt' ).to(0 )
__snake_case = model.generate(**SCREAMING_SNAKE_CASE_ )
# test with `flan-t5-small`
__snake_case = TaForConditionalGeneration.from_pretrained(
self.dense_act_model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' )
__snake_case = self.tokenizer(self.input_text , return_tensors='pt' ).to(0 )
__snake_case = model.generate(**SCREAMING_SNAKE_CASE_ )
class _lowercase ( __lowercase ):
def a ( self : Dict ) -> str:
super().setUp()
# model_name
__snake_case = 'bigscience/bloom-560m'
__snake_case = 't5-small'
# Different types of model
__snake_case = AutoModel.from_pretrained(self.model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' )
# Sequence classification model
__snake_case = AutoModelForSequenceClassification.from_pretrained(
self.model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' )
# CausalLM model
__snake_case = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' )
# Seq2seq model
__snake_case = AutoModelForSeqaSeqLM.from_pretrained(
self.seq_to_seq_name , load_in_abit=SCREAMING_SNAKE_CASE_ , device_map='auto' )
def a ( self : int ) -> Dict:
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 a ( self : Any ) -> Optional[Any]:
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 _lowercase ( __lowercase ):
def a ( self : str ) -> Union[str, Any]:
super().setUp()
def a ( self : Optional[Any] ) -> str:
del self.pipe
gc.collect()
torch.cuda.empty_cache()
def a ( self : Optional[int] ) -> List[str]:
__snake_case = 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
__snake_case = self.pipe(self.input_text )
self.assertIn(pipeline_output[0]['generated_text'] , self.EXPECTED_OUTPUTS )
@require_torch_multi_gpu
class _lowercase ( __lowercase ):
def a ( self : Optional[int] ) -> Union[str, Any]:
super().setUp()
def a ( self : Optional[int] ) -> List[Any]:
__snake_case = AutoModelForCausalLM.from_pretrained(
self.model_name , load_in_abit=SCREAMING_SNAKE_CASE_ , 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
__snake_case = self.tokenizer(self.input_text , return_tensors='pt' )
# Second real batch
__snake_case = 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=SCREAMING_SNAKE_CASE_ ) , self.EXPECTED_OUTPUTS )
class _lowercase ( __lowercase ):
def a ( self : Any ) -> str:
__snake_case = 'facebook/opt-350m'
super().setUp()
def a ( self : int ) -> List[Any]:
if version.parse(importlib.metadata.version('bitsandbytes' ) ) < version.parse('0.37.0' ):
return
# Step 1: freeze all parameters
__snake_case = AutoModelForCausalLM.from_pretrained(self.model_name , load_in_abit=SCREAMING_SNAKE_CASE_ )
self.assertEqual(set(model.hf_device_map.values() ) , {torch.cuda.current_device()} )
for param in model.parameters():
__snake_case = False # freeze the model - train adapters later
if param.ndim == 1:
# cast the small parameters (e.g. layernorm) to fp32 for stability
__snake_case = param.data.to(torch.floataa )
# Step 2: add adapters
for _, module in model.named_modules():
if "OPTAttention" in repr(type(SCREAMING_SNAKE_CASE_ ) ):
__snake_case = LoRALayer(module.q_proj , rank=16 )
__snake_case = LoRALayer(module.k_proj , rank=16 )
__snake_case = LoRALayer(module.v_proj , rank=16 )
# Step 3: dummy batch
__snake_case = self.tokenizer('Test batch ' , return_tensors='pt' ).to(0 )
# Step 4: Check if the gradient is not None
with torch.cuda.amp.autocast():
__snake_case = model.forward(**SCREAMING_SNAKE_CASE_ )
out.logits.norm().backward()
for module in model.modules():
if isinstance(SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ ):
self.assertTrue(module.adapter[1].weight.grad is not None )
self.assertTrue(module.adapter[1].weight.grad.norm().item() > 0 )
elif isinstance(SCREAMING_SNAKE_CASE_ , nn.Embedding ):
self.assertTrue(module.weight.grad is None )
class _lowercase ( __lowercase ):
_SCREAMING_SNAKE_CASE : Union[str, Any] = "gpt2-xl"
_SCREAMING_SNAKE_CASE : Optional[int] = 3.3191854854152187
| 56
| 1
|
"""simple docstring"""
def SCREAMING_SNAKE_CASE ( __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ) -> List[Any]:
if height >= 1:
move_tower(height - 1 , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ )
move_disk(lowerCAmelCase__ , lowerCAmelCase__ )
move_tower(height - 1 , lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ )
def SCREAMING_SNAKE_CASE ( __UpperCAmelCase , __UpperCAmelCase ) -> int:
print("moving disk from" , lowerCAmelCase__ , "to" , lowerCAmelCase__ )
def SCREAMING_SNAKE_CASE ( ) -> List[str]:
SCREAMING_SNAKE_CASE__ = int(input("Height of hanoi: " ).strip() )
move_tower(lowerCAmelCase__ , "A" , "B" , "C" )
if __name__ == "__main__":
main()
| 714
|
"""simple docstring"""
import unittest
from transformers import AlbertConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
AlbertForMaskedLM,
AlbertForMultipleChoice,
AlbertForPreTraining,
AlbertForQuestionAnswering,
AlbertForSequenceClassification,
AlbertForTokenClassification,
AlbertModel,
)
from transformers.models.albert.modeling_albert import ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST
class lowerCamelCase :
'''simple docstring'''
def __init__( self : str , _snake_case : List[str] , _snake_case : List[Any]=13 , _snake_case : List[str]=7 , _snake_case : Dict=True , _snake_case : str=True , _snake_case : Optional[Any]=True , _snake_case : Tuple=True , _snake_case : List[Any]=99 , _snake_case : Dict=16 , _snake_case : Tuple=36 , _snake_case : Optional[int]=6 , _snake_case : Optional[int]=6 , _snake_case : Tuple=6 , _snake_case : Optional[int]=37 , _snake_case : Dict="gelu" , _snake_case : str=0.1 , _snake_case : Tuple=0.1 , _snake_case : List[str]=512 , _snake_case : Any=16 , _snake_case : Optional[int]=2 , _snake_case : Optional[int]=0.02 , _snake_case : Union[str, Any]=3 , _snake_case : int=4 , _snake_case : Optional[int]=None , ) -> List[str]:
SCREAMING_SNAKE_CASE__ = parent
SCREAMING_SNAKE_CASE__ = batch_size
SCREAMING_SNAKE_CASE__ = seq_length
SCREAMING_SNAKE_CASE__ = is_training
SCREAMING_SNAKE_CASE__ = use_input_mask
SCREAMING_SNAKE_CASE__ = use_token_type_ids
SCREAMING_SNAKE_CASE__ = use_labels
SCREAMING_SNAKE_CASE__ = vocab_size
SCREAMING_SNAKE_CASE__ = embedding_size
SCREAMING_SNAKE_CASE__ = hidden_size
SCREAMING_SNAKE_CASE__ = num_hidden_layers
SCREAMING_SNAKE_CASE__ = num_hidden_groups
SCREAMING_SNAKE_CASE__ = num_attention_heads
SCREAMING_SNAKE_CASE__ = intermediate_size
SCREAMING_SNAKE_CASE__ = hidden_act
SCREAMING_SNAKE_CASE__ = hidden_dropout_prob
SCREAMING_SNAKE_CASE__ = attention_probs_dropout_prob
SCREAMING_SNAKE_CASE__ = max_position_embeddings
SCREAMING_SNAKE_CASE__ = type_vocab_size
SCREAMING_SNAKE_CASE__ = type_sequence_label_size
SCREAMING_SNAKE_CASE__ = initializer_range
SCREAMING_SNAKE_CASE__ = num_labels
SCREAMING_SNAKE_CASE__ = num_choices
SCREAMING_SNAKE_CASE__ = scope
def lowerCAmelCase_ ( self : Tuple ) -> Tuple:
SCREAMING_SNAKE_CASE__ = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
SCREAMING_SNAKE_CASE__ = None
if self.use_input_mask:
SCREAMING_SNAKE_CASE__ = random_attention_mask([self.batch_size, self.seq_length] )
SCREAMING_SNAKE_CASE__ = None
if self.use_token_type_ids:
SCREAMING_SNAKE_CASE__ = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
SCREAMING_SNAKE_CASE__ = None
SCREAMING_SNAKE_CASE__ = None
SCREAMING_SNAKE_CASE__ = None
if self.use_labels:
SCREAMING_SNAKE_CASE__ = ids_tensor([self.batch_size] , self.type_sequence_label_size )
SCREAMING_SNAKE_CASE__ = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
SCREAMING_SNAKE_CASE__ = ids_tensor([self.batch_size] , self.num_choices )
SCREAMING_SNAKE_CASE__ = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowerCAmelCase_ ( self : Dict ) -> Union[str, Any]:
return AlbertConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , initializer_range=self.initializer_range , num_hidden_groups=self.num_hidden_groups , )
def lowerCAmelCase_ ( self : int , _snake_case : Tuple , _snake_case : Optional[Any] , _snake_case : Union[str, Any] , _snake_case : int , _snake_case : Union[str, Any] , _snake_case : str , _snake_case : int ) -> Union[str, Any]:
SCREAMING_SNAKE_CASE__ = AlbertModel(config=_snake_case )
model.to(_snake_case )
model.eval()
SCREAMING_SNAKE_CASE__ = model(_snake_case , attention_mask=_snake_case , token_type_ids=_snake_case )
SCREAMING_SNAKE_CASE__ = model(_snake_case , token_type_ids=_snake_case )
SCREAMING_SNAKE_CASE__ = model(_snake_case )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape , (self.batch_size, self.hidden_size) )
def lowerCAmelCase_ ( self : Tuple , _snake_case : Optional[Any] , _snake_case : int , _snake_case : str , _snake_case : List[Any] , _snake_case : Dict , _snake_case : Union[str, Any] , _snake_case : str ) -> Union[str, Any]:
SCREAMING_SNAKE_CASE__ = AlbertForPreTraining(config=_snake_case )
model.to(_snake_case )
model.eval()
SCREAMING_SNAKE_CASE__ = model(
_snake_case , attention_mask=_snake_case , token_type_ids=_snake_case , labels=_snake_case , sentence_order_label=_snake_case , )
self.parent.assertEqual(result.prediction_logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
self.parent.assertEqual(result.sop_logits.shape , (self.batch_size, config.num_labels) )
def lowerCAmelCase_ ( self : Union[str, Any] , _snake_case : int , _snake_case : Any , _snake_case : List[Any] , _snake_case : Any , _snake_case : str , _snake_case : Tuple , _snake_case : List[Any] ) -> Any:
SCREAMING_SNAKE_CASE__ = AlbertForMaskedLM(config=_snake_case )
model.to(_snake_case )
model.eval()
SCREAMING_SNAKE_CASE__ = model(_snake_case , attention_mask=_snake_case , token_type_ids=_snake_case , labels=_snake_case )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def lowerCAmelCase_ ( self : Optional[Any] , _snake_case : Optional[Any] , _snake_case : Optional[Any] , _snake_case : Optional[Any] , _snake_case : Any , _snake_case : Optional[int] , _snake_case : Optional[Any] , _snake_case : Tuple ) -> List[str]:
SCREAMING_SNAKE_CASE__ = AlbertForQuestionAnswering(config=_snake_case )
model.to(_snake_case )
model.eval()
SCREAMING_SNAKE_CASE__ = model(
_snake_case , attention_mask=_snake_case , token_type_ids=_snake_case , start_positions=_snake_case , end_positions=_snake_case , )
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 : Optional[Any] , _snake_case : Optional[Any] , _snake_case : str , _snake_case : Optional[int] , _snake_case : int , _snake_case : List[Any] , _snake_case : List[Any] , _snake_case : str ) -> Optional[int]:
SCREAMING_SNAKE_CASE__ = self.num_labels
SCREAMING_SNAKE_CASE__ = AlbertForSequenceClassification(_snake_case )
model.to(_snake_case )
model.eval()
SCREAMING_SNAKE_CASE__ = model(_snake_case , attention_mask=_snake_case , token_type_ids=_snake_case , labels=_snake_case )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def lowerCAmelCase_ ( self : int , _snake_case : Any , _snake_case : str , _snake_case : List[Any] , _snake_case : Any , _snake_case : Union[str, Any] , _snake_case : Dict , _snake_case : Tuple ) -> Dict:
SCREAMING_SNAKE_CASE__ = self.num_labels
SCREAMING_SNAKE_CASE__ = AlbertForTokenClassification(config=_snake_case )
model.to(_snake_case )
model.eval()
SCREAMING_SNAKE_CASE__ = model(_snake_case , attention_mask=_snake_case , token_type_ids=_snake_case , labels=_snake_case )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def lowerCAmelCase_ ( self : Tuple , _snake_case : str , _snake_case : Any , _snake_case : Any , _snake_case : int , _snake_case : List[Any] , _snake_case : List[str] , _snake_case : int ) -> List[str]:
SCREAMING_SNAKE_CASE__ = self.num_choices
SCREAMING_SNAKE_CASE__ = AlbertForMultipleChoice(config=_snake_case )
model.to(_snake_case )
model.eval()
SCREAMING_SNAKE_CASE__ = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
SCREAMING_SNAKE_CASE__ = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
SCREAMING_SNAKE_CASE__ = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
SCREAMING_SNAKE_CASE__ = model(
_snake_case , attention_mask=_snake_case , token_type_ids=_snake_case , labels=_snake_case , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def lowerCAmelCase_ ( self : Optional[int] ) -> Optional[int]:
SCREAMING_SNAKE_CASE__ = self.prepare_config_and_inputs()
(
(
SCREAMING_SNAKE_CASE__
) , (
SCREAMING_SNAKE_CASE__
) , (
SCREAMING_SNAKE_CASE__
) , (
SCREAMING_SNAKE_CASE__
) , (
SCREAMING_SNAKE_CASE__
) , (
SCREAMING_SNAKE_CASE__
) , (
SCREAMING_SNAKE_CASE__
) ,
) = config_and_inputs
SCREAMING_SNAKE_CASE__ = {"input_ids": input_ids, "token_type_ids": token_type_ids, "attention_mask": input_mask}
return config, inputs_dict
@require_torch
class lowerCamelCase (_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , unittest.TestCase ):
'''simple docstring'''
a = (
(
AlbertModel,
AlbertForPreTraining,
AlbertForMaskedLM,
AlbertForMultipleChoice,
AlbertForSequenceClassification,
AlbertForTokenClassification,
AlbertForQuestionAnswering,
)
if is_torch_available()
else ()
)
a = (
{
"feature-extraction": AlbertModel,
"fill-mask": AlbertForMaskedLM,
"question-answering": AlbertForQuestionAnswering,
"text-classification": AlbertForSequenceClassification,
"token-classification": AlbertForTokenClassification,
"zero-shot": AlbertForSequenceClassification,
}
if is_torch_available()
else {}
)
a = True
def lowerCAmelCase_ ( self : Union[str, Any] , _snake_case : Dict , _snake_case : str , _snake_case : str=False ) -> Dict:
SCREAMING_SNAKE_CASE__ = super()._prepare_for_class(_snake_case , _snake_case , return_labels=_snake_case )
if return_labels:
if model_class in get_values(_snake_case ):
SCREAMING_SNAKE_CASE__ = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) , dtype=torch.long , device=_snake_case )
SCREAMING_SNAKE_CASE__ = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=_snake_case )
return inputs_dict
def lowerCAmelCase_ ( self : Union[str, Any] ) -> List[Any]:
SCREAMING_SNAKE_CASE__ = AlbertModelTester(self )
SCREAMING_SNAKE_CASE__ = ConfigTester(self , config_class=_snake_case , hidden_size=37 )
def lowerCAmelCase_ ( self : List[str] ) -> List[str]:
self.config_tester.run_common_tests()
def lowerCAmelCase_ ( self : Dict ) -> Union[str, Any]:
SCREAMING_SNAKE_CASE__ = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*_snake_case )
def lowerCAmelCase_ ( self : Union[str, Any] ) -> Tuple:
SCREAMING_SNAKE_CASE__ = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_pretraining(*_snake_case )
def lowerCAmelCase_ ( self : Any ) -> Union[str, Any]:
SCREAMING_SNAKE_CASE__ = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*_snake_case )
def lowerCAmelCase_ ( self : Union[str, Any] ) -> Optional[Any]:
SCREAMING_SNAKE_CASE__ = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*_snake_case )
def lowerCAmelCase_ ( self : Optional[Any] ) -> List[Any]:
SCREAMING_SNAKE_CASE__ = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*_snake_case )
def lowerCAmelCase_ ( self : Optional[Any] ) -> Optional[Any]:
SCREAMING_SNAKE_CASE__ = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*_snake_case )
def lowerCAmelCase_ ( self : List[str] ) -> Optional[int]:
SCREAMING_SNAKE_CASE__ = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
SCREAMING_SNAKE_CASE__ = type
self.model_tester.create_and_check_model(*_snake_case )
@slow
def lowerCAmelCase_ ( self : Any ) -> str:
for model_name in ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
SCREAMING_SNAKE_CASE__ = AlbertModel.from_pretrained(_snake_case )
self.assertIsNotNone(_snake_case )
@require_torch
class lowerCamelCase (unittest.TestCase ):
'''simple docstring'''
@slow
def lowerCAmelCase_ ( self : Any ) -> Any:
SCREAMING_SNAKE_CASE__ = AlbertModel.from_pretrained("albert-base-v2" )
SCREAMING_SNAKE_CASE__ = torch.tensor([[0, 345, 232, 328, 740, 140, 1695, 69, 6078, 1588, 2]] )
SCREAMING_SNAKE_CASE__ = torch.tensor([[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]] )
with torch.no_grad():
SCREAMING_SNAKE_CASE__ = model(_snake_case , attention_mask=_snake_case )[0]
SCREAMING_SNAKE_CASE__ = torch.Size((1, 11, 768) )
self.assertEqual(output.shape , _snake_case )
SCREAMING_SNAKE_CASE__ = torch.tensor(
[[[-0.6513, 1.5035, -0.2766], [-0.6515, 1.5046, -0.2780], [-0.6512, 1.5049, -0.2784]]] )
self.assertTrue(torch.allclose(output[:, 1:4, 1:4] , _snake_case , atol=1e-4 ) )
| 538
| 0
|
"""simple docstring"""
from torch import nn
def _snake_case ( _snake_case : int ) -> Any:
'''simple docstring'''
if act_fn in ["swish", "silu"]:
return nn.SiLU()
elif act_fn == "mish":
return nn.Mish()
elif act_fn == "gelu":
return nn.GELU()
else:
raise ValueError(F'''Unsupported activation function: {act_fn}''' )
| 7
|
"""simple docstring"""
from __future__ import annotations
from collections.abc import Generator
def _SCREAMING_SNAKE_CASE () -> Generator[int, None, None]:
'''simple docstring'''
lowercase_ = {}
lowercase_ = 2
while True:
lowercase_ = factor_map.pop(__lowerCAmelCase , __lowerCAmelCase )
if factor:
lowercase_ = factor + prime
while x in factor_map:
x += factor
lowercase_ = factor
else:
lowercase_ = prime
yield prime
prime += 1
def _SCREAMING_SNAKE_CASE (__lowerCAmelCase = 1E10 ) -> int:
'''simple docstring'''
lowercase_ = sieve()
lowercase_ = 1
while True:
lowercase_ = next(__lowerCAmelCase )
if (2 * prime * n) > limit:
return n
# Ignore the next prime as the reminder will be 2.
next(__lowerCAmelCase )
n += 2
if __name__ == "__main__":
print(solution())
| 567
| 0
|
"""simple docstring"""
__snake_case :int = '0.21.0'
from .accelerator import Accelerator
from .big_modeling import (
cpu_offload,
cpu_offload_with_hook,
disk_offload,
dispatch_model,
init_empty_weights,
init_on_device,
load_checkpoint_and_dispatch,
)
from .data_loader import skip_first_batches
from .launchers import debug_launcher, notebook_launcher
from .state import PartialState
from .utils import (
DeepSpeedPlugin,
DistributedDataParallelKwargs,
DistributedType,
FullyShardedDataParallelPlugin,
GradScalerKwargs,
InitProcessGroupKwargs,
find_executable_batch_size,
infer_auto_device_map,
is_rich_available,
load_checkpoint_in_model,
synchronize_rng_states,
)
if is_rich_available():
from .utils import rich
| 700
|
"""simple docstring"""
def lowerCamelCase_ ( UpperCAmelCase_ , UpperCAmelCase_ ) ->bool:
"""simple docstring"""
return numa ^ numa < 0
if __name__ == "__main__":
import doctest
doctest.testmod()
| 374
| 0
|
"""simple docstring"""
import numpy as np
lowerCAmelCase__ = [
['''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 __snake_case :
def __init__( self : Optional[Any] ):
"""simple docstring"""
_lowerCamelCase : List[str] = np.array(__lowerCAmelCase )
def SCREAMING_SNAKE_CASE ( self : List[str] , __lowerCAmelCase : str ):
"""simple docstring"""
_lowerCamelCase , _lowerCamelCase : str = np.where(letter == self.SQUARE )
_lowerCamelCase : int = np.concatenate([indexa + 1, indexa + 1] )
return indexes
def SCREAMING_SNAKE_CASE ( self : List[Any] , __lowerCAmelCase : int , __lowerCAmelCase : int ):
"""simple docstring"""
_lowerCamelCase : Any = self.SQUARE[indexa - 1, indexa - 1]
return letter
def SCREAMING_SNAKE_CASE ( self : str , __lowerCAmelCase : str ):
"""simple docstring"""
_lowerCamelCase : Tuple = message.lower()
_lowerCamelCase : List[Any] = message.replace(''' ''' , '''''' )
_lowerCamelCase : str = message.replace('''j''' , '''i''' )
_lowerCamelCase : Optional[Any] = np.empty((2, len(__lowerCAmelCase )) )
for letter_index in range(len(__lowerCAmelCase ) ):
_lowerCamelCase : int = self.letter_to_numbers(message[letter_index] )
_lowerCamelCase : Dict = numbers[0]
_lowerCamelCase : Optional[int] = numbers[1]
_lowerCamelCase : Tuple = first_step.reshape(2 * len(__lowerCAmelCase ) )
_lowerCamelCase : str = ''''''
for numbers_index in range(len(__lowerCAmelCase ) ):
_lowerCamelCase : Any = int(second_step[numbers_index * 2] )
_lowerCamelCase : Tuple = int(second_step[(numbers_index * 2) + 1] )
_lowerCamelCase : int = self.numbers_to_letter(__lowerCAmelCase , __lowerCAmelCase )
_lowerCamelCase : List[Any] = encoded_message + letter
return encoded_message
def SCREAMING_SNAKE_CASE ( self : str , __lowerCAmelCase : str ):
"""simple docstring"""
_lowerCamelCase : Optional[int] = message.lower()
message.replace(''' ''' , '''''' )
_lowerCamelCase : Tuple = np.empty(2 * len(__lowerCAmelCase ) )
for letter_index in range(len(__lowerCAmelCase ) ):
_lowerCamelCase : Optional[int] = self.letter_to_numbers(message[letter_index] )
_lowerCamelCase : Optional[int] = numbers[0]
_lowerCamelCase : Union[str, Any] = numbers[1]
_lowerCamelCase : Tuple = first_step.reshape((2, len(__lowerCAmelCase )) )
_lowerCamelCase : Union[str, Any] = ''''''
for numbers_index in range(len(__lowerCAmelCase ) ):
_lowerCamelCase : Dict = int(second_step[0, numbers_index] )
_lowerCamelCase : Union[str, Any] = int(second_step[1, numbers_index] )
_lowerCamelCase : Dict = self.numbers_to_letter(__lowerCAmelCase , __lowerCAmelCase )
_lowerCamelCase : Any = decoded_message + letter
return decoded_message
| 83
|
"""simple docstring"""
import logging
import os
from dataclasses import dataclass
from enum import Enum
from typing import List, Optional, Union
from filelock import FileLock
from transformers import PreTrainedTokenizer, is_tf_available, is_torch_available
a__ : Any = logging.getLogger(__name__)
@dataclass
class __magic_name__ :
UpperCamelCase : str
UpperCamelCase : List[str]
UpperCamelCase : Optional[List[str]]
@dataclass
class __magic_name__ :
UpperCamelCase : List[int]
UpperCamelCase : List[int]
UpperCamelCase : Optional[List[int]] = None
UpperCamelCase : Optional[List[int]] = None
class __magic_name__ ( _UpperCamelCase ):
UpperCamelCase : List[str] = "train"
UpperCamelCase : Any = "dev"
UpperCamelCase : str = "test"
class __magic_name__ :
@staticmethod
def _lowerCamelCase ( __magic_name__ , __magic_name__ ):
"""simple docstring"""
raise NotImplementedError
@staticmethod
def _lowerCamelCase ( __magic_name__ ):
"""simple docstring"""
raise NotImplementedError
@staticmethod
def _lowerCamelCase ( __magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ , __magic_name__=False , __magic_name__="[CLS]" , __magic_name__=1 , __magic_name__="[SEP]" , __magic_name__=False , __magic_name__=False , __magic_name__=0 , __magic_name__=0 , __magic_name__=-1_0_0 , __magic_name__=0 , __magic_name__=True , ):
"""simple docstring"""
_lowerCAmelCase = {label: i for i, label in enumerate(__magic_name__ )}
_lowerCAmelCase = []
for ex_index, example in enumerate(__magic_name__ ):
if ex_index % 1_0_0_0_0 == 0:
logger.info('Writing example %d of %d' , __magic_name__ , len(__magic_name__ ) )
_lowerCAmelCase = []
_lowerCAmelCase = []
for word, label in zip(example.words , example.labels ):
_lowerCAmelCase = tokenizer.tokenize(__magic_name__ )
# bert-base-multilingual-cased sometimes output "nothing ([]) when calling tokenize with just a space.
if len(__magic_name__ ) > 0:
tokens.extend(__magic_name__ )
# Use the real label id for the first token of the word, and padding ids for the remaining tokens
label_ids.extend([label_map[label]] + [pad_token_label_id] * (len(__magic_name__ ) - 1) )
# Account for [CLS] and [SEP] with "- 2" and with "- 3" for RoBERTa.
_lowerCAmelCase = tokenizer.num_special_tokens_to_add()
if len(__magic_name__ ) > max_seq_length - special_tokens_count:
_lowerCAmelCase = tokens[: (max_seq_length - special_tokens_count)]
_lowerCAmelCase = label_ids[: (max_seq_length - special_tokens_count)]
# The convention in BERT is:
# (a) For sequence pairs:
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
# (b) For single sequences:
# tokens: [CLS] the dog is hairy . [SEP]
# type_ids: 0 0 0 0 0 0 0
#
# Where "type_ids" are used to indicate whether this is the first
# sequence or the second sequence. The embedding vectors for `type=0` and
# `type=1` were learned during pre-training and are added to the wordpiece
# embedding vector (and position vector). This is not *strictly* necessary
# since the [SEP] token unambiguously separates the sequences, but it makes
# it easier for the model to learn the concept of sequences.
#
# For classification tasks, the first vector (corresponding to [CLS]) is
# used as the "sentence vector". Note that this only makes sense because
# the entire model is fine-tuned.
tokens += [sep_token]
label_ids += [pad_token_label_id]
if sep_token_extra:
# roberta uses an extra separator b/w pairs of sentences
tokens += [sep_token]
label_ids += [pad_token_label_id]
_lowerCAmelCase = [sequence_a_segment_id] * len(__magic_name__ )
if cls_token_at_end:
tokens += [cls_token]
label_ids += [pad_token_label_id]
segment_ids += [cls_token_segment_id]
else:
_lowerCAmelCase = [cls_token] + tokens
_lowerCAmelCase = [pad_token_label_id] + label_ids
_lowerCAmelCase = [cls_token_segment_id] + segment_ids
_lowerCAmelCase = tokenizer.convert_tokens_to_ids(__magic_name__ )
# The mask has 1 for real tokens and 0 for padding tokens. Only real
# tokens are attended to.
_lowerCAmelCase = [1 if mask_padding_with_zero else 0] * len(__magic_name__ )
# Zero-pad up to the sequence length.
_lowerCAmelCase = max_seq_length - len(__magic_name__ )
if pad_on_left:
_lowerCAmelCase = ([pad_token] * padding_length) + input_ids
_lowerCAmelCase = ([0 if mask_padding_with_zero else 1] * padding_length) + input_mask
_lowerCAmelCase = ([pad_token_segment_id] * padding_length) + segment_ids
_lowerCAmelCase = ([pad_token_label_id] * padding_length) + label_ids
else:
input_ids += [pad_token] * padding_length
input_mask += [0 if mask_padding_with_zero else 1] * padding_length
segment_ids += [pad_token_segment_id] * padding_length
label_ids += [pad_token_label_id] * padding_length
assert len(__magic_name__ ) == max_seq_length
assert len(__magic_name__ ) == max_seq_length
assert len(__magic_name__ ) == max_seq_length
assert len(__magic_name__ ) == max_seq_length
if ex_index < 5:
logger.info('*** Example ***' )
logger.info('guid: %s' , example.guid )
logger.info('tokens: %s' , ' '.join([str(__magic_name__ ) for x in tokens] ) )
logger.info('input_ids: %s' , ' '.join([str(__magic_name__ ) for x in input_ids] ) )
logger.info('input_mask: %s' , ' '.join([str(__magic_name__ ) for x in input_mask] ) )
logger.info('segment_ids: %s' , ' '.join([str(__magic_name__ ) for x in segment_ids] ) )
logger.info('label_ids: %s' , ' '.join([str(__magic_name__ ) for x in label_ids] ) )
if "token_type_ids" not in tokenizer.model_input_names:
_lowerCAmelCase = None
features.append(
InputFeatures(
input_ids=__magic_name__ , attention_mask=__magic_name__ , token_type_ids=__magic_name__ , label_ids=__magic_name__ ) )
return features
if is_torch_available():
import torch
from torch import nn
from torch.utils.data import Dataset
class __magic_name__ ( _UpperCamelCase ):
UpperCamelCase : List[InputFeatures]
UpperCamelCase : int = nn.CrossEntropyLoss().ignore_index
def __init__( self , __magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ = None , __magic_name__=False , __magic_name__ = Split.train , ):
"""simple docstring"""
_lowerCAmelCase = os.path.join(
__magic_name__ , 'cached_{}_{}_{}'.format(mode.value , tokenizer.__class__.__name__ , str(__magic_name__ ) ) , )
# Make sure only the first process in distributed training processes the dataset,
# and the others will use the cache.
_lowerCAmelCase = cached_features_file + '.lock'
with FileLock(__magic_name__ ):
if os.path.exists(__magic_name__ ) and not overwrite_cache:
logger.info(F'''Loading features from cached file {cached_features_file}''' )
_lowerCAmelCase = torch.load(__magic_name__ )
else:
logger.info(F'''Creating features from dataset file at {data_dir}''' )
_lowerCAmelCase = token_classification_task.read_examples_from_file(__magic_name__ , __magic_name__ )
# TODO clean up all this to leverage built-in features of tokenizers
_lowerCAmelCase = token_classification_task.convert_examples_to_features(
__magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ , cls_token_at_end=bool(model_type in ['xlnet'] ) , cls_token=tokenizer.cls_token , cls_token_segment_id=2 if model_type in ['xlnet'] else 0 , sep_token=tokenizer.sep_token , sep_token_extra=__magic_name__ , pad_on_left=bool(tokenizer.padding_side == 'left' ) , pad_token=tokenizer.pad_token_id , pad_token_segment_id=tokenizer.pad_token_type_id , pad_token_label_id=self.pad_token_label_id , )
logger.info(F'''Saving features into cached file {cached_features_file}''' )
torch.save(self.features , __magic_name__ )
def __len__( self ):
"""simple docstring"""
return len(self.features )
def __getitem__( self , __magic_name__ ):
"""simple docstring"""
return self.features[i]
if is_tf_available():
import tensorflow as tf
class __magic_name__ :
UpperCamelCase : List[InputFeatures]
UpperCamelCase : int = -100
def __init__( self , __magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ = None , __magic_name__=False , __magic_name__ = Split.train , ):
"""simple docstring"""
_lowerCAmelCase = token_classification_task.read_examples_from_file(__magic_name__ , __magic_name__ )
# TODO clean up all this to leverage built-in features of tokenizers
_lowerCAmelCase = token_classification_task.convert_examples_to_features(
__magic_name__ , __magic_name__ , __magic_name__ , __magic_name__ , cls_token_at_end=bool(model_type in ['xlnet'] ) , cls_token=tokenizer.cls_token , cls_token_segment_id=2 if model_type in ['xlnet'] else 0 , sep_token=tokenizer.sep_token , sep_token_extra=__magic_name__ , pad_on_left=bool(tokenizer.padding_side == 'left' ) , pad_token=tokenizer.pad_token_id , pad_token_segment_id=tokenizer.pad_token_type_id , pad_token_label_id=self.pad_token_label_id , )
def gen():
for ex in self.features:
if ex.token_type_ids is None:
yield (
{"input_ids": ex.input_ids, "attention_mask": ex.attention_mask},
ex.label_ids,
)
else:
yield (
{
"input_ids": ex.input_ids,
"attention_mask": ex.attention_mask,
"token_type_ids": ex.token_type_ids,
},
ex.label_ids,
)
if "token_type_ids" not in tokenizer.model_input_names:
_lowerCAmelCase = tf.data.Dataset.from_generator(
__magic_name__ , ({'input_ids': tf.intaa, 'attention_mask': tf.intaa}, tf.intaa) , (
{'input_ids': tf.TensorShape([None] ), 'attention_mask': tf.TensorShape([None] )},
tf.TensorShape([None] ),
) , )
else:
_lowerCAmelCase = tf.data.Dataset.from_generator(
__magic_name__ , ({'input_ids': tf.intaa, 'attention_mask': tf.intaa, 'token_type_ids': tf.intaa}, tf.intaa) , (
{
'input_ids': tf.TensorShape([None] ),
'attention_mask': tf.TensorShape([None] ),
'token_type_ids': tf.TensorShape([None] ),
},
tf.TensorShape([None] ),
) , )
def _lowerCamelCase ( self ):
"""simple docstring"""
_lowerCAmelCase = self.dataset.apply(tf.data.experimental.assert_cardinality(len(self.features ) ) )
return self.dataset
def __len__( self ):
"""simple docstring"""
return len(self.features )
def __getitem__( self , __magic_name__ ):
"""simple docstring"""
return self.features[i]
| 589
| 0
|
"""simple docstring"""
from math import asin, atan, cos, radians, sin, sqrt, tan
__SCREAMING_SNAKE_CASE =6_37_81_37.0
__SCREAMING_SNAKE_CASE =6_35_67_52.31_42_45
__SCREAMING_SNAKE_CASE =637_8137
def lowercase__( __SCREAMING_SNAKE_CASE : float , __SCREAMING_SNAKE_CASE : float , __SCREAMING_SNAKE_CASE : float , __SCREAMING_SNAKE_CASE : float ):
lowercase_ : Any = (AXIS_A - AXIS_B) / AXIS_A
lowercase_ : Optional[int] = atan((1 - flattening) * tan(radians(__SCREAMING_SNAKE_CASE ) ) )
lowercase_ : int = atan((1 - flattening) * tan(radians(__SCREAMING_SNAKE_CASE ) ) )
lowercase_ : Optional[Any] = radians(__SCREAMING_SNAKE_CASE )
lowercase_ : Union[str, Any] = radians(__SCREAMING_SNAKE_CASE )
# Equation
lowercase_ : str = sin((phi_a - phi_a) / 2 )
lowercase_ : Tuple = sin((lambda_a - lambda_a) / 2 )
# Square both values
sin_sq_phi *= sin_sq_phi
sin_sq_lambda *= sin_sq_lambda
lowercase_ : Optional[Any] = sqrt(sin_sq_phi + (cos(__SCREAMING_SNAKE_CASE ) * cos(__SCREAMING_SNAKE_CASE ) * sin_sq_lambda) )
return 2 * RADIUS * asin(__SCREAMING_SNAKE_CASE )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 477
|
"""simple docstring"""
import copy
from typing import Any, Dict, List, Optional, Union
import numpy as np
import torch
from ...audio_utils import mel_filter_bank, spectrogram, window_function
from ...feature_extraction_sequence_utils import SequenceFeatureExtractor
from ...feature_extraction_utils import BatchFeature
from ...utils import TensorType, logging
__SCREAMING_SNAKE_CASE =logging.get_logger(__name__)
class UpperCamelCase ( lowercase_ ):
lowercase = ['input_features', 'is_longer']
def __init__( self ,__UpperCamelCase=64 ,__UpperCamelCase=4_8000 ,__UpperCamelCase=480 ,__UpperCamelCase=10 ,__UpperCamelCase=1024 ,__UpperCamelCase=0.0 ,__UpperCamelCase=False ,__UpperCamelCase = 0 ,__UpperCamelCase = 1_4000 ,__UpperCamelCase = None ,__UpperCamelCase = "fusion" ,__UpperCamelCase = "repeatpad" ,**__UpperCamelCase ,) -> Optional[int]:
'''simple docstring'''
super().__init__(
feature_size=__UpperCamelCase ,sampling_rate=__UpperCamelCase ,padding_value=__UpperCamelCase ,return_attention_mask=__UpperCamelCase ,**__UpperCamelCase ,)
lowercase_ : Union[str, Any] = top_db
lowercase_ : Any = truncation
lowercase_ : str = padding
lowercase_ : Optional[Any] = fft_window_size
lowercase_ : List[Any] = (fft_window_size >> 1) + 1
lowercase_ : Any = hop_length
lowercase_ : List[Any] = max_length_s
lowercase_ : Any = max_length_s * sampling_rate
lowercase_ : Optional[int] = sampling_rate
lowercase_ : List[Any] = frequency_min
lowercase_ : str = frequency_max
lowercase_ : Any = mel_filter_bank(
num_frequency_bins=self.nb_frequency_bins ,num_mel_filters=__UpperCamelCase ,min_frequency=__UpperCamelCase ,max_frequency=__UpperCamelCase ,sampling_rate=__UpperCamelCase ,norm=__UpperCamelCase ,mel_scale='htk' ,)
lowercase_ : Optional[Any] = mel_filter_bank(
num_frequency_bins=self.nb_frequency_bins ,num_mel_filters=__UpperCamelCase ,min_frequency=__UpperCamelCase ,max_frequency=__UpperCamelCase ,sampling_rate=__UpperCamelCase ,norm='slaney' ,mel_scale='slaney' ,)
def _UpperCAmelCase ( self ) -> Dict[str, Any]:
'''simple docstring'''
lowercase_ : int = copy.deepcopy(self.__dict__ )
lowercase_ : Any = self.__class__.__name__
if "mel_filters" in output:
del output["mel_filters"]
if "mel_filters_slaney" in output:
del output["mel_filters_slaney"]
return output
def _UpperCAmelCase ( self ,__UpperCamelCase ,__UpperCamelCase = None ) -> np.ndarray:
'''simple docstring'''
lowercase_ : Union[str, Any] = spectrogram(
__UpperCamelCase ,window_function(self.fft_window_size ,'hann' ) ,frame_length=self.fft_window_size ,hop_length=self.hop_length ,power=2.0 ,mel_filters=__UpperCamelCase ,log_mel='dB' ,)
return log_mel_spectrogram.T
def _UpperCAmelCase ( self ,__UpperCamelCase ,__UpperCamelCase ,__UpperCamelCase ) -> Any:
'''simple docstring'''
lowercase_ : Dict = np.array_split(list(range(0 ,total_frames - chunk_frames + 1 ) ) ,3 )
if len(ranges[1] ) == 0:
# if the audio is too short, we just use the first chunk
lowercase_ : Optional[int] = [0]
if len(ranges[2] ) == 0:
# if the audio is too short, we just use the first chunk
lowercase_ : List[str] = [0]
# randomly choose index for each part
lowercase_ : List[str] = np.random.choice(ranges[0] )
lowercase_ : Optional[Any] = np.random.choice(ranges[1] )
lowercase_ : Union[str, Any] = np.random.choice(ranges[2] )
lowercase_ : Tuple = mel[idx_front : idx_front + chunk_frames, :]
lowercase_ : str = mel[idx_middle : idx_middle + chunk_frames, :]
lowercase_ : Optional[Any] = mel[idx_back : idx_back + chunk_frames, :]
lowercase_ : Tuple = torch.tensor(mel[None, None, :] )
lowercase_ : Optional[Any] = torch.nn.functional.interpolate(
__UpperCamelCase ,size=[chunk_frames, 64] ,mode='bilinear' ,align_corners=__UpperCamelCase )
lowercase_ : str = mel_shrink[0][0].numpy()
lowercase_ : Optional[int] = np.stack([mel_shrink, mel_chunk_front, mel_chunk_middle, mel_chunk_back] ,axis=0 )
return mel_fusion
def _UpperCAmelCase ( self ,__UpperCamelCase ,__UpperCamelCase ,__UpperCamelCase ,__UpperCamelCase ) -> np.array:
'''simple docstring'''
if waveform.shape[0] > max_length:
if truncation == "rand_trunc":
lowercase_ : Any = True
# random crop to max_length (for compatibility) -> this should be handled by self.pad
lowercase_ : Any = len(__UpperCamelCase ) - max_length
lowercase_ : List[str] = np.random.randint(0 ,overflow + 1 )
lowercase_ : Any = waveform[idx : idx + max_length]
lowercase_ : Optional[int] = self._np_extract_fbank_features(__UpperCamelCase ,self.mel_filters_slaney )[None, :]
elif truncation == "fusion":
lowercase_ : str = self._np_extract_fbank_features(__UpperCamelCase ,self.mel_filters )
lowercase_ : List[str] = max_length // self.hop_length + 1 # the +1 related to how the spectrogram is computed
lowercase_ : Dict = mel.shape[0]
if chunk_frames == total_frames:
# there is a corner case where the audio length is larger than max_length but smaller than max_length+hop_length.
# In this case, we just use the whole audio.
lowercase_ : int = np.stack([mel, mel, mel, mel] ,axis=0 )
lowercase_ : Optional[Any] = False
else:
lowercase_ : int = self._random_mel_fusion(__UpperCamelCase ,__UpperCamelCase ,__UpperCamelCase )
lowercase_ : Tuple = True
else:
raise NotImplementedError(f'''data_truncating {truncation} not implemented''' )
else:
lowercase_ : Dict = False
# only use repeat as a new possible value for padding. you repeat the audio before applying the usual max_length padding
if waveform.shape[0] < max_length:
if padding == "repeat":
lowercase_ : Optional[int] = int(max_length / len(__UpperCamelCase ) )
lowercase_ : Any = np.stack(np.tile(__UpperCamelCase ,n_repeat + 1 ) )[:max_length]
if padding == "repeatpad":
lowercase_ : Union[str, Any] = int(max_length / len(__UpperCamelCase ) )
lowercase_ : int = np.stack(np.tile(__UpperCamelCase ,__UpperCamelCase ) )
lowercase_ : str = np.pad(__UpperCamelCase ,(0, max_length - waveform.shape[0]) ,mode='constant' ,constant_values=0 )
if truncation == "fusion":
lowercase_ : List[str] = self._np_extract_fbank_features(__UpperCamelCase ,self.mel_filters )
lowercase_ : List[str] = np.stack([input_mel, input_mel, input_mel, input_mel] ,axis=0 )
else:
lowercase_ : Any = self._np_extract_fbank_features(__UpperCamelCase ,self.mel_filters_slaney )[None, :]
return input_mel, longer
def __call__( self ,__UpperCamelCase ,__UpperCamelCase = None ,__UpperCamelCase = None ,__UpperCamelCase = None ,__UpperCamelCase = None ,__UpperCamelCase = None ,**__UpperCamelCase ,) -> BatchFeature:
'''simple docstring'''
lowercase_ : Union[str, Any] = truncation if truncation is not None else self.truncation
lowercase_ : List[Any] = padding if padding else self.padding
if sampling_rate is not None:
if sampling_rate != self.sampling_rate:
raise ValueError(
f'''The model corresponding to this feature extractor: {self.__class__.__name__} was trained using a'''
f''' sampling rate of {self.sampling_rate}. Please make sure that the provided `raw_speech` input'''
f''' was sampled with {self.sampling_rate} and not {sampling_rate}.''' )
else:
logger.warning(
'It is strongly recommended to pass the `sampling_rate` argument to this function. '
'Failing to do so can result in silent errors that might be hard to debug.' )
lowercase_ : Tuple = 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}''' )
lowercase_ : List[str] = is_batched_numpy or (
isinstance(__UpperCamelCase ,(list, tuple) ) and (isinstance(raw_speech[0] ,(np.ndarray, tuple, list) ))
)
if is_batched:
lowercase_ : str = [np.asarray(__UpperCamelCase ,dtype=np.floataa ) for speech in raw_speech]
elif not is_batched and not isinstance(__UpperCamelCase ,np.ndarray ):
lowercase_ : Any = np.asarray(__UpperCamelCase ,dtype=np.floataa )
elif isinstance(__UpperCamelCase ,np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ):
lowercase_ : Dict = raw_speech.astype(np.floataa )
# always return batch
if not is_batched:
lowercase_ : int = [np.asarray(__UpperCamelCase )]
# convert to mel spectrogram, truncate and pad if needed.
lowercase_ : List[str] = [
self._get_input_mel(__UpperCamelCase ,max_length if max_length else self.nb_max_samples ,__UpperCamelCase ,__UpperCamelCase )
for waveform in raw_speech
]
lowercase_ : int = []
lowercase_ : int = []
for mel, longer in padded_inputs:
input_mel.append(__UpperCamelCase )
is_longer.append(__UpperCamelCase )
if truncation == "fusion" and sum(__UpperCamelCase ) == 0:
# if no audio is longer than 10s, then randomly select one audio to be longer
lowercase_ : int = np.random.randint(0 ,len(__UpperCamelCase ) )
lowercase_ : Tuple = True
if isinstance(input_mel[0] ,__UpperCamelCase ):
lowercase_ : Union[str, Any] = [np.asarray(__UpperCamelCase ,dtype=np.floataa ) for feature in input_mel]
# is_longer is a list of bool
lowercase_ : Any = [[longer] for longer in is_longer]
lowercase_ : List[Any] = {'input_features': input_mel, 'is_longer': is_longer}
lowercase_ : Union[str, Any] = BatchFeature(__UpperCamelCase )
if return_tensors is not None:
lowercase_ : Any = input_features.convert_to_tensors(__UpperCamelCase )
return input_features
| 477
| 1
|
'''simple docstring'''
import logging
import math
import os
from dataclasses import dataclass, field
from glob import glob
from typing import Optional
from torch.utils.data import ConcatDataset
import transformers
from transformers import (
CONFIG_MAPPING,
MODEL_WITH_LM_HEAD_MAPPING,
AutoConfig,
AutoModelWithLMHead,
AutoTokenizer,
DataCollatorForLanguageModeling,
DataCollatorForPermutationLanguageModeling,
DataCollatorForWholeWordMask,
HfArgumentParser,
LineByLineTextDataset,
LineByLineWithRefDataset,
PreTrainedTokenizer,
TextDataset,
Trainer,
TrainingArguments,
set_seed,
)
from transformers.trainer_utils import is_main_process
__a = logging.getLogger(__name__)
__a = list(MODEL_WITH_LM_HEAD_MAPPING.keys())
__a = tuple(conf.model_type for conf in MODEL_CONFIG_CLASSES)
@dataclass
class UpperCAmelCase_ :
"""simple docstring"""
lowercase = field(
default=_a , metadata={
"help": (
"The model checkpoint for weights initialization. Leave None if you want to train a model from"
" scratch."
)
} , )
lowercase = field(
default=_a , metadata={"help": "If training from scratch, pass a model type from the list: " + ", ".join(_a )} , )
lowercase = field(
default=_a , metadata={"help": "Pretrained config name or path if not the same as model_name"} )
lowercase = field(
default=_a , metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"} )
lowercase = field(
default=_a , metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"} , )
@dataclass
class UpperCAmelCase_ :
"""simple docstring"""
lowercase = field(
default=_a , metadata={"help": "The input training data file (a text file)."} )
lowercase = field(
default=_a , metadata={
"help": (
"The input training data files (multiple files in glob format). "
"Very often splitting large files to smaller files can prevent tokenizer going out of memory"
)
} , )
lowercase = field(
default=_a , metadata={"help": "An optional input evaluation data file to evaluate the perplexity on (a text file)."} , )
lowercase = field(
default=_a , metadata={"help": "An optional input train ref data file for whole word mask in Chinese."} , )
lowercase = field(
default=_a , metadata={"help": "An optional input eval ref data file for whole word mask in Chinese."} , )
lowercase = field(
default=_a , metadata={"help": "Whether distinct lines of text in the dataset are to be handled as distinct sequences."} , )
lowercase = field(
default=_a , metadata={"help": "Train with masked-language modeling loss instead of language modeling."} )
lowercase = field(default=_a , metadata={"help": "Whether ot not to use whole word mask."} )
lowercase = field(
default=0.15 , metadata={"help": "Ratio of tokens to mask for masked language modeling loss"} )
lowercase = field(
default=1 / 6 , metadata={
"help": (
"Ratio of length of a span of masked tokens to surrounding context length for permutation language"
" modeling."
)
} , )
lowercase = field(
default=5 , metadata={"help": "Maximum length of a span of masked tokens for permutation language modeling."} )
lowercase = field(
default=-1 , metadata={
"help": (
"Optional input sequence length after tokenization."
"The training dataset will be truncated in block of this size for training."
"Default to the model max input length for single sentence inputs (take into account special tokens)."
)
} , )
lowercase = field(
default=_a , metadata={"help": "Overwrite the cached training and evaluation sets"} )
def __snake_case( _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase = False , _lowerCAmelCase = None , ) -> Tuple:
def _dataset(_lowerCAmelCase , _lowerCAmelCase=None ):
if args.line_by_line:
if ref_path is not None:
if not args.whole_word_mask or not args.mlm:
raise ValueError("""You need to set world whole masking and mlm to True for Chinese Whole Word Mask""" )
return LineByLineWithRefDataset(
tokenizer=_lowerCAmelCase , file_path=_lowerCAmelCase , block_size=args.block_size , ref_path=_lowerCAmelCase , )
return LineByLineTextDataset(tokenizer=_lowerCAmelCase , file_path=_lowerCAmelCase , block_size=args.block_size )
else:
return TextDataset(
tokenizer=_lowerCAmelCase , file_path=_lowerCAmelCase , block_size=args.block_size , overwrite_cache=args.overwrite_cache , cache_dir=_lowerCAmelCase , )
if evaluate:
return _dataset(args.eval_data_file , args.eval_ref_file )
elif args.train_data_files:
return ConcatDataset([_dataset(_lowerCAmelCase ) for f in glob(args.train_data_files )] )
else:
return _dataset(args.train_data_file , args.train_ref_file )
def __snake_case( ) -> Dict:
# See all possible arguments in src/transformers/training_args.py
# or by passing the --help flag to this script.
# We now keep distinct sets of args, for a cleaner separation of concerns.
snake_case__ : Optional[int] = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments) )
snake_case__ , snake_case__ , snake_case__ : int = parser.parse_args_into_dataclasses()
if data_args.eval_data_file is None and training_args.do_eval:
raise ValueError(
"""Cannot do evaluation without an evaluation data file. Either supply a file to --eval_data_file """
"""or remove the --do_eval argument.""" )
if (
os.path.exists(training_args.output_dir )
and os.listdir(training_args.output_dir )
and training_args.do_train
and not training_args.overwrite_output_dir
):
raise ValueError(
f"Output directory ({training_args.output_dir}) already exists and is not empty. Use"
""" --overwrite_output_dir to overcome.""" )
# Setup logging
logging.basicConfig(
format="""%(asctime)s - %(levelname)s - %(name)s - %(message)s""" , datefmt="""%m/%d/%Y %H:%M:%S""" , level=logging.INFO if training_args.local_rank in [-1, 0] else logging.WARN , )
logger.warning(
"""Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s""" , training_args.local_rank , training_args.device , training_args.n_gpu , bool(training_args.local_rank != -1 ) , training_args.fpaa , )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
logger.info("""Training/evaluation parameters %s""" , _lowerCAmelCase )
# Set seed
set_seed(training_args.seed )
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
if model_args.config_name:
snake_case__ : Any = AutoConfig.from_pretrained(model_args.config_name , cache_dir=model_args.cache_dir )
elif model_args.model_name_or_path:
snake_case__ : Any = AutoConfig.from_pretrained(model_args.model_name_or_path , cache_dir=model_args.cache_dir )
else:
snake_case__ : Tuple = CONFIG_MAPPING[model_args.model_type]()
logger.warning("""You are instantiating a new config instance from scratch.""" )
if model_args.tokenizer_name:
snake_case__ : str = AutoTokenizer.from_pretrained(model_args.tokenizer_name , cache_dir=model_args.cache_dir )
elif model_args.model_name_or_path:
snake_case__ : str = AutoTokenizer.from_pretrained(model_args.model_name_or_path , cache_dir=model_args.cache_dir )
else:
raise ValueError(
"""You are instantiating a new tokenizer from scratch. This is not supported, but you can do it from another"""
""" script, save it,and load it from here, using --tokenizer_name""" )
if model_args.model_name_or_path:
snake_case__ : Union[str, Any] = AutoModelWithLMHead.from_pretrained(
model_args.model_name_or_path , from_tf=bool(""".ckpt""" in model_args.model_name_or_path ) , config=_lowerCAmelCase , cache_dir=model_args.cache_dir , )
else:
logger.info("""Training new model from scratch""" )
snake_case__ : List[Any] = AutoModelWithLMHead.from_config(_lowerCAmelCase )
model.resize_token_embeddings(len(_lowerCAmelCase ) )
if config.model_type in ["bert", "roberta", "distilbert", "camembert"] and not data_args.mlm:
raise ValueError(
"""BERT and RoBERTa-like models do not have LM heads but masked LM heads. They must be run using the"""
"""--mlm flag (masked language modeling).""" )
if data_args.block_size <= 0:
snake_case__ : List[Any] = tokenizer.max_len
# Our input block size will be the max possible for the model
else:
snake_case__ : List[str] = min(data_args.block_size , tokenizer.max_len )
# Get datasets
snake_case__ : int = (
get_dataset(_lowerCAmelCase , tokenizer=_lowerCAmelCase , cache_dir=model_args.cache_dir ) if training_args.do_train else None
)
snake_case__ : Any = (
get_dataset(_lowerCAmelCase , tokenizer=_lowerCAmelCase , evaluate=_lowerCAmelCase , cache_dir=model_args.cache_dir )
if training_args.do_eval
else None
)
if config.model_type == "xlnet":
snake_case__ : str = DataCollatorForPermutationLanguageModeling(
tokenizer=_lowerCAmelCase , plm_probability=data_args.plm_probability , max_span_length=data_args.max_span_length , )
else:
if data_args.mlm and data_args.whole_word_mask:
snake_case__ : List[str] = DataCollatorForWholeWordMask(
tokenizer=_lowerCAmelCase , mlm_probability=data_args.mlm_probability )
else:
snake_case__ : Optional[int] = DataCollatorForLanguageModeling(
tokenizer=_lowerCAmelCase , mlm=data_args.mlm , mlm_probability=data_args.mlm_probability )
# Initialize our Trainer
snake_case__ : Optional[Any] = Trainer(
model=_lowerCAmelCase , args=_lowerCAmelCase , data_collator=_lowerCAmelCase , train_dataset=_lowerCAmelCase , eval_dataset=_lowerCAmelCase , prediction_loss_only=_lowerCAmelCase , )
# Training
if training_args.do_train:
snake_case__ : Any = (
model_args.model_name_or_path
if model_args.model_name_or_path is not None and os.path.isdir(model_args.model_name_or_path )
else None
)
trainer.train(model_path=_lowerCAmelCase )
trainer.save_model()
# For convenience, we also re-save the tokenizer to the same directory,
# so that you can share your model easily on huggingface.co/models =)
if trainer.is_world_master():
tokenizer.save_pretrained(training_args.output_dir )
# Evaluation
snake_case__ : Union[str, Any] = {}
if training_args.do_eval:
logger.info("""*** Evaluate ***""" )
snake_case__ : Dict = trainer.evaluate()
snake_case__ : Dict = math.exp(eval_output["""eval_loss"""] )
snake_case__ : str = {"""perplexity""": perplexity}
snake_case__ : Any = os.path.join(training_args.output_dir , """eval_results_lm.txt""" )
if trainer.is_world_master():
with open(_lowerCAmelCase , """w""" ) as writer:
logger.info("""***** Eval results *****""" )
for key in sorted(result.keys() ):
logger.info(""" %s = %s""" , _lowerCAmelCase , str(result[key] ) )
writer.write("""%s = %s\n""" % (key, str(result[key] )) )
results.update(_lowerCAmelCase )
return results
def __snake_case( _lowerCAmelCase ) -> Tuple:
# For xla_spawn (TPUs)
main()
if __name__ == "__main__":
main()
| 374
|
'''simple docstring'''
import unittest
import torch
from torch import nn
from diffusers.models.activations import get_activation
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def lowerCamelCase ( self : Union[str, Any] ):
snake_case__ : int = get_activation("""swish""" )
self.assertIsInstance(snake_case_ , nn.SiLU )
self.assertEqual(act(torch.tensor(-100 , dtype=torch.floataa ) ).item() , 0 )
self.assertNotEqual(act(torch.tensor(-1 , dtype=torch.floataa ) ).item() , 0 )
self.assertEqual(act(torch.tensor(0 , dtype=torch.floataa ) ).item() , 0 )
self.assertEqual(act(torch.tensor(20 , dtype=torch.floataa ) ).item() , 20 )
def lowerCamelCase ( self : Optional[Any] ):
snake_case__ : int = get_activation("""silu""" )
self.assertIsInstance(snake_case_ , nn.SiLU )
self.assertEqual(act(torch.tensor(-100 , dtype=torch.floataa ) ).item() , 0 )
self.assertNotEqual(act(torch.tensor(-1 , dtype=torch.floataa ) ).item() , 0 )
self.assertEqual(act(torch.tensor(0 , dtype=torch.floataa ) ).item() , 0 )
self.assertEqual(act(torch.tensor(20 , dtype=torch.floataa ) ).item() , 20 )
def lowerCamelCase ( self : Dict ):
snake_case__ : Union[str, Any] = get_activation("""mish""" )
self.assertIsInstance(snake_case_ , nn.Mish )
self.assertEqual(act(torch.tensor(-200 , dtype=torch.floataa ) ).item() , 0 )
self.assertNotEqual(act(torch.tensor(-1 , dtype=torch.floataa ) ).item() , 0 )
self.assertEqual(act(torch.tensor(0 , dtype=torch.floataa ) ).item() , 0 )
self.assertEqual(act(torch.tensor(20 , dtype=torch.floataa ) ).item() , 20 )
def lowerCamelCase ( self : Union[str, Any] ):
snake_case__ : int = get_activation("""gelu""" )
self.assertIsInstance(snake_case_ , nn.GELU )
self.assertEqual(act(torch.tensor(-100 , dtype=torch.floataa ) ).item() , 0 )
self.assertNotEqual(act(torch.tensor(-1 , dtype=torch.floataa ) ).item() , 0 )
self.assertEqual(act(torch.tensor(0 , dtype=torch.floataa ) ).item() , 0 )
self.assertEqual(act(torch.tensor(20 , dtype=torch.floataa ) ).item() , 20 )
| 374
| 1
|
import argparse
import torch
from transformers import (
EncodecConfig,
EncodecFeatureExtractor,
EncodecModel,
logging,
)
# checkpoints downloaded from:
# https://dl.fbaipublicfiles.com/encodec/v0/encodec_24khz-d7cc33bc.th
# https://huggingface.co/facebook/musicgen-small/resolve/main/compression_state_dict.bin
# https://dl.fbaipublicfiles.com/encodec/v0/encodec_48khz-7e698e3e.th
logging.set_verbosity_info()
_a = logging.get_logger("transformers.models.encodec")
_a = {
'''quantizer.vq.layers.*._codebook.inited''': '''quantizer.layers.*.codebook.inited''',
'''quantizer.vq.layers.*._codebook.cluster_size''': '''quantizer.layers.*.codebook.cluster_size''',
'''quantizer.vq.layers.*._codebook.embed''': '''quantizer.layers.*.codebook.embed''',
'''quantizer.vq.layers.*._codebook.embed_avg''': '''quantizer.layers.*.codebook.embed_avg''',
}
_a = {
'''encoder.model.0.conv.conv''': '''encoder.layers.0.conv''',
'''encoder.model.1.block.1.conv.conv''': '''encoder.layers.1.block.1.conv''',
'''encoder.model.1.block.3.conv.conv''': '''encoder.layers.1.block.3.conv''',
'''encoder.model.1.shortcut.conv.conv''': '''encoder.layers.1.shortcut.conv''',
'''encoder.model.3.conv.conv''': '''encoder.layers.3.conv''',
'''encoder.model.4.block.1.conv.conv''': '''encoder.layers.4.block.1.conv''',
'''encoder.model.4.block.3.conv.conv''': '''encoder.layers.4.block.3.conv''',
'''encoder.model.4.shortcut.conv.conv''': '''encoder.layers.4.shortcut.conv''',
'''encoder.model.6.conv.conv''': '''encoder.layers.6.conv''',
'''encoder.model.7.block.1.conv.conv''': '''encoder.layers.7.block.1.conv''',
'''encoder.model.7.block.3.conv.conv''': '''encoder.layers.7.block.3.conv''',
'''encoder.model.7.shortcut.conv.conv''': '''encoder.layers.7.shortcut.conv''',
'''encoder.model.9.conv.conv''': '''encoder.layers.9.conv''',
'''encoder.model.10.block.1.conv.conv''': '''encoder.layers.10.block.1.conv''',
'''encoder.model.10.block.3.conv.conv''': '''encoder.layers.10.block.3.conv''',
'''encoder.model.10.shortcut.conv.conv''': '''encoder.layers.10.shortcut.conv''',
'''encoder.model.12.conv.conv''': '''encoder.layers.12.conv''',
'''encoder.model.13.lstm''': '''encoder.layers.13.lstm''',
'''encoder.model.15.conv.conv''': '''encoder.layers.15.conv''',
}
_a = {
'''encoder.model.0.conv.norm''': '''encoder.layers.0.norm''',
'''encoder.model.1.block.1.conv.norm''': '''encoder.layers.1.block.1.norm''',
'''encoder.model.1.block.3.conv.norm''': '''encoder.layers.1.block.3.norm''',
'''encoder.model.1.shortcut.conv.norm''': '''encoder.layers.1.shortcut.norm''',
'''encoder.model.3.conv.norm''': '''encoder.layers.3.norm''',
'''encoder.model.4.block.1.conv.norm''': '''encoder.layers.4.block.1.norm''',
'''encoder.model.4.block.3.conv.norm''': '''encoder.layers.4.block.3.norm''',
'''encoder.model.4.shortcut.conv.norm''': '''encoder.layers.4.shortcut.norm''',
'''encoder.model.6.conv.norm''': '''encoder.layers.6.norm''',
'''encoder.model.7.block.1.conv.norm''': '''encoder.layers.7.block.1.norm''',
'''encoder.model.7.block.3.conv.norm''': '''encoder.layers.7.block.3.norm''',
'''encoder.model.7.shortcut.conv.norm''': '''encoder.layers.7.shortcut.norm''',
'''encoder.model.9.conv.norm''': '''encoder.layers.9.norm''',
'''encoder.model.10.block.1.conv.norm''': '''encoder.layers.10.block.1.norm''',
'''encoder.model.10.block.3.conv.norm''': '''encoder.layers.10.block.3.norm''',
'''encoder.model.10.shortcut.conv.norm''': '''encoder.layers.10.shortcut.norm''',
'''encoder.model.12.conv.norm''': '''encoder.layers.12.norm''',
'''encoder.model.15.conv.norm''': '''encoder.layers.15.norm''',
}
_a = {
'''decoder.model.0.conv.conv''': '''decoder.layers.0.conv''',
'''decoder.model.1.lstm''': '''decoder.layers.1.lstm''',
'''decoder.model.3.convtr.convtr''': '''decoder.layers.3.conv''',
'''decoder.model.4.block.1.conv.conv''': '''decoder.layers.4.block.1.conv''',
'''decoder.model.4.block.3.conv.conv''': '''decoder.layers.4.block.3.conv''',
'''decoder.model.4.shortcut.conv.conv''': '''decoder.layers.4.shortcut.conv''',
'''decoder.model.6.convtr.convtr''': '''decoder.layers.6.conv''',
'''decoder.model.7.block.1.conv.conv''': '''decoder.layers.7.block.1.conv''',
'''decoder.model.7.block.3.conv.conv''': '''decoder.layers.7.block.3.conv''',
'''decoder.model.7.shortcut.conv.conv''': '''decoder.layers.7.shortcut.conv''',
'''decoder.model.9.convtr.convtr''': '''decoder.layers.9.conv''',
'''decoder.model.10.block.1.conv.conv''': '''decoder.layers.10.block.1.conv''',
'''decoder.model.10.block.3.conv.conv''': '''decoder.layers.10.block.3.conv''',
'''decoder.model.10.shortcut.conv.conv''': '''decoder.layers.10.shortcut.conv''',
'''decoder.model.12.convtr.convtr''': '''decoder.layers.12.conv''',
'''decoder.model.13.block.1.conv.conv''': '''decoder.layers.13.block.1.conv''',
'''decoder.model.13.block.3.conv.conv''': '''decoder.layers.13.block.3.conv''',
'''decoder.model.13.shortcut.conv.conv''': '''decoder.layers.13.shortcut.conv''',
'''decoder.model.15.conv.conv''': '''decoder.layers.15.conv''',
}
_a = {
'''decoder.model.0.conv.norm''': '''decoder.layers.0.norm''',
'''decoder.model.3.convtr.norm''': '''decoder.layers.3.norm''',
'''decoder.model.4.block.1.conv.norm''': '''decoder.layers.4.block.1.norm''',
'''decoder.model.4.block.3.conv.norm''': '''decoder.layers.4.block.3.norm''',
'''decoder.model.4.shortcut.conv.norm''': '''decoder.layers.4.shortcut.norm''',
'''decoder.model.6.convtr.norm''': '''decoder.layers.6.norm''',
'''decoder.model.7.block.1.conv.norm''': '''decoder.layers.7.block.1.norm''',
'''decoder.model.7.block.3.conv.norm''': '''decoder.layers.7.block.3.norm''',
'''decoder.model.7.shortcut.conv.norm''': '''decoder.layers.7.shortcut.norm''',
'''decoder.model.9.convtr.norm''': '''decoder.layers.9.norm''',
'''decoder.model.10.block.1.conv.norm''': '''decoder.layers.10.block.1.norm''',
'''decoder.model.10.block.3.conv.norm''': '''decoder.layers.10.block.3.norm''',
'''decoder.model.10.shortcut.conv.norm''': '''decoder.layers.10.shortcut.norm''',
'''decoder.model.12.convtr.norm''': '''decoder.layers.12.norm''',
'''decoder.model.13.block.1.conv.norm''': '''decoder.layers.13.block.1.norm''',
'''decoder.model.13.block.3.conv.norm''': '''decoder.layers.13.block.3.norm''',
'''decoder.model.13.shortcut.conv.norm''': '''decoder.layers.13.shortcut.norm''',
'''decoder.model.15.conv.norm''': '''decoder.layers.15.norm''',
}
_a = {
**MAPPING_QUANTIZER,
**MAPPING_ENCODER,
**MAPPING_DECODER,
}
_a = {
**MAPPING_QUANTIZER,
**MAPPING_ENCODER,
**MAPPING_ENCODER_48K,
**MAPPING_DECODER,
**MAPPING_DECODER_48K,
}
_a = []
_a = []
def lowerCAmelCase__(__snake_case ,__snake_case ,__snake_case ,__snake_case ,__snake_case ) -> str:
'''simple docstring'''
for attribute in key.split('''.''' ):
lowerCamelCase__ = getattr(_A ,_A )
if weight_type is not None:
lowerCamelCase__ = getattr(_A ,_A ).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 == "weight_ih_l0":
lowerCamelCase__ = value
elif weight_type == "weight_hh_l0":
lowerCamelCase__ = value
elif weight_type == "bias_ih_l0":
lowerCamelCase__ = value
elif weight_type == "bias_hh_l0":
lowerCamelCase__ = value
elif weight_type == "weight_ih_l1":
lowerCamelCase__ = value
elif weight_type == "weight_hh_l1":
lowerCamelCase__ = value
elif weight_type == "bias_ih_l1":
lowerCamelCase__ = value
elif weight_type == "bias_hh_l1":
lowerCamelCase__ = value
else:
lowerCamelCase__ = value
logger.info(F'{key + ("." + weight_type if weight_type is not None else "")} was initialized from {full_name}.' )
def lowerCAmelCase__(__snake_case ,__snake_case ) -> Optional[Any]:
'''simple docstring'''
for key in ignore_keys:
if key.endswith('''.*''' ):
if name.startswith(key[:-1] ):
return True
elif ".*." in key:
lowerCamelCase__ , lowerCamelCase__ = key.split('''.*.''' )
if prefix in name and suffix in name:
return True
elif key in name:
return True
return False
def lowerCAmelCase__(__snake_case ,__snake_case ,__snake_case ) -> List[str]:
'''simple docstring'''
lowerCamelCase__ = []
if model_name == "encodec_24khz" or "encodec_32khz":
lowerCamelCase__ = MAPPING_24K
elif model_name == "encodec_48khz":
lowerCamelCase__ = MAPPING_48K
else:
raise ValueError(F'Unsupported model: {model_name}' )
for name, value in orig_dict.items():
if should_ignore(_A ,_A ):
logger.info(F'{name} was ignored' )
continue
lowerCamelCase__ = False
for key, mapped_key in MAPPING.items():
if "*" in key:
lowerCamelCase__ , lowerCamelCase__ = key.split('''.*.''' )
if prefix in name and suffix in name:
lowerCamelCase__ = suffix
if key in name:
# HACK otherwise .embed gets initialized with .embed_avg too
if key.endswith('''embed''' ) and name.endswith('''embed_avg''' ):
continue
lowerCamelCase__ = True
if "*" in mapped_key:
lowerCamelCase__ = name.split(_A )[0].split('''.''' )[-2]
lowerCamelCase__ = mapped_key.replace('''*''' ,_A )
if "weight_g" in name:
lowerCamelCase__ = '''weight_g'''
elif "weight_v" in name:
lowerCamelCase__ = '''weight_v'''
elif "weight_ih_l0" in name:
lowerCamelCase__ = '''weight_ih_l0'''
elif "weight_hh_l0" in name:
lowerCamelCase__ = '''weight_hh_l0'''
elif "bias_ih_l0" in name:
lowerCamelCase__ = '''bias_ih_l0'''
elif "bias_hh_l0" in name:
lowerCamelCase__ = '''bias_hh_l0'''
elif "weight_ih_l1" in name:
lowerCamelCase__ = '''weight_ih_l1'''
elif "weight_hh_l1" in name:
lowerCamelCase__ = '''weight_hh_l1'''
elif "bias_ih_l1" in name:
lowerCamelCase__ = '''bias_ih_l1'''
elif "bias_hh_l1" in name:
lowerCamelCase__ = '''bias_hh_l1'''
elif "bias" in name:
lowerCamelCase__ = '''bias'''
elif "weight" in name:
lowerCamelCase__ = '''weight'''
elif "running_mean" in name:
lowerCamelCase__ = '''running_mean'''
elif "running_var" in name:
lowerCamelCase__ = '''running_var'''
elif "num_batches_tracked" in name:
lowerCamelCase__ = '''num_batches_tracked'''
else:
lowerCamelCase__ = None
set_recursively(_A ,_A ,_A ,_A ,_A )
continue
if not is_used:
unused_weights.append(_A )
logger.warning(F'Unused weights: {unused_weights}' )
@torch.no_grad()
def lowerCAmelCase__(__snake_case ,__snake_case ,__snake_case ,__snake_case=None ,__snake_case=None ,) -> Tuple:
'''simple docstring'''
if config_path is not None:
lowerCamelCase__ = EncodecConfig.from_pretrained(_A )
else:
lowerCamelCase__ = EncodecConfig()
if model_name == "encodec_24khz":
pass # config is already correct
elif model_name == "encodec_32khz":
lowerCamelCase__ = [8, 5, 4, 4]
lowerCamelCase__ = [2.2]
lowerCamelCase__ = 64
lowerCamelCase__ = 32000
lowerCamelCase__ = 2048
lowerCamelCase__ = False
lowerCamelCase__ = False
lowerCamelCase__ = False
elif model_name == "encodec_48khz":
lowerCamelCase__ = [8, 5, 4, 2]
lowerCamelCase__ = [3.0, 6.0, 1_2.0, 2_4.0]
lowerCamelCase__ = 48000
lowerCamelCase__ = 2
lowerCamelCase__ = False
lowerCamelCase__ = '''time_group_norm'''
lowerCamelCase__ = True
lowerCamelCase__ = 1.0
lowerCamelCase__ = 0.0_1
else:
raise ValueError(F'Unknown model name: {model_name}' )
lowerCamelCase__ = EncodecModel(_A )
lowerCamelCase__ = EncodecFeatureExtractor(
feature_size=config.audio_channels ,sampling_rate=config.sampling_rate ,chunk_length_s=config.chunk_length_s ,overlap=config.overlap ,)
feature_extractor.save_pretrained(_A )
lowerCamelCase__ = torch.load(_A )
if "best_state" in original_checkpoint:
# we might have a training state saved, in which case discard the yaml results and just retain the weights
lowerCamelCase__ = original_checkpoint['''best_state''']
recursively_load_weights(_A ,_A ,_A )
model.save_pretrained(_A )
if repo_id:
print('''Pushing to the hub...''' )
feature_extractor.push_to_hub(_A )
model.push_to_hub(_A )
if __name__ == "__main__":
_a = argparse.ArgumentParser()
parser.add_argument(
"--model",
default="encodec_24khz",
type=str,
help="The model to convert. Should be one of \'encodec_24khz\', \'encodec_32khz\', \'encodec_48khz\'.",
)
parser.add_argument("--checkpoint_path", required=True, default=None, type=str, help="Path to original checkpoint")
parser.add_argument("--config_path", default=None, type=str, help="Path to hf config.json of model to convert")
parser.add_argument(
"--pytorch_dump_folder_path", required=True, default=None, type=str, help="Path to the output PyTorch model."
)
parser.add_argument(
"--push_to_hub", default=None, type=str, help="Where to upload the converted model on the 🤗 hub."
)
_a = parser.parse_args()
convert_checkpoint(
args.model,
args.checkpoint_path,
args.pytorch_dump_folder_path,
args.config_path,
args.push_to_hub,
)
| 719
|
# Usage:
# ./gen-card-allenai-wmt16.py
import os
from pathlib import Path
def lowerCAmelCase__(__snake_case ,__snake_case ,__snake_case ,__snake_case ) -> Any:
'''simple docstring'''
lowerCamelCase__ = {
'''en''': '''Machine learning is great, isn\'t it?''',
'''ru''': '''Машинное обучение - это здорово, не так ли?''',
'''de''': '''Maschinelles Lernen ist großartig, nicht wahr?''',
}
# BLUE scores as follows:
# "pair": [fairseq, transformers]
lowerCamelCase__ = {
'''wmt16-en-de-dist-12-1''': [2_8.3, 2_7.5_2],
'''wmt16-en-de-dist-6-1''': [2_7.4, 2_7.1_1],
'''wmt16-en-de-12-1''': [2_6.9, 2_5.7_5],
}
lowerCamelCase__ = F'{src_lang}-{tgt_lang}'
lowerCamelCase__ = F'\n---\nlanguage:\n- {src_lang}\n- {tgt_lang}\nthumbnail:\ntags:\n- translation\n- wmt16\n- allenai\nlicense: apache-2.0\ndatasets:\n- wmt16\nmetrics:\n- bleu\n---\n\n# FSMT\n\n## Model description\n\nThis is a ported version of fairseq-based [wmt16 transformer](https://github.com/jungokasai/deep-shallow/) for {src_lang}-{tgt_lang}.\n\nFor more details, please, see [Deep Encoder, Shallow Decoder: Reevaluating the Speed-Quality Tradeoff in Machine Translation](https://arxiv.org/abs/2006.10369).\n\nAll 3 models are available:\n\n* [wmt16-en-de-dist-12-1](https://huggingface.co/allenai/wmt16-en-de-dist-12-1)\n* [wmt16-en-de-dist-6-1](https://huggingface.co/allenai/wmt16-en-de-dist-6-1)\n* [wmt16-en-de-12-1](https://huggingface.co/allenai/wmt16-en-de-12-1)\n\n\n## Intended uses & limitations\n\n#### How to use\n\n```python\nfrom transformers import FSMTForConditionalGeneration, FSMTTokenizer\nmname = "allenai/{model_name}"\ntokenizer = FSMTTokenizer.from_pretrained(mname)\nmodel = FSMTForConditionalGeneration.from_pretrained(mname)\n\ninput = "{texts[src_lang]}"\ninput_ids = tokenizer.encode(input, return_tensors="pt")\noutputs = model.generate(input_ids)\ndecoded = tokenizer.decode(outputs[0], skip_special_tokens=True)\nprint(decoded) # {texts[tgt_lang]}\n\n```\n\n#### Limitations and bias\n\n\n## Training data\n\nPretrained weights were left identical to the original model released by allenai. For more details, please, see the [paper](https://arxiv.org/abs/2006.10369).\n\n## Eval results\n\nHere are the BLEU scores:\n\nmodel | fairseq | transformers\n-------|---------|----------\n{model_name} | {scores[model_name][0]} | {scores[model_name][1]}\n\nThe score is slightly below the score reported in the paper, as the researchers don\'t use `sacrebleu` and measure the score on tokenized outputs. `transformers` score was measured using `sacrebleu` on detokenized outputs.\n\nThe score was calculated using this code:\n\n```bash\ngit clone https://github.com/huggingface/transformers\ncd transformers\nexport PAIR={pair}\nexport DATA_DIR=data/$PAIR\nexport SAVE_DIR=data/$PAIR\nexport BS=8\nexport NUM_BEAMS=5\nmkdir -p $DATA_DIR\nsacrebleu -t wmt16 -l $PAIR --echo src > $DATA_DIR/val.source\nsacrebleu -t wmt16 -l $PAIR --echo ref > $DATA_DIR/val.target\necho $PAIR\nPYTHONPATH="src:examples/seq2seq" python examples/seq2seq/run_eval.py allenai/{model_name} $DATA_DIR/val.source $SAVE_DIR/test_translations.txt --reference_path $DATA_DIR/val.target --score_path $SAVE_DIR/test_bleu.json --bs $BS --task translation --num_beams $NUM_BEAMS\n```\n\n## Data Sources\n\n- [training, etc.](http://www.statmt.org/wmt16/)\n- [test set](http://matrix.statmt.org/test_sets/newstest2016.tgz?1504722372)\n\n\n### BibTeX entry and citation info\n\n```\n@misc{{kasai2020deep,\n title={{Deep Encoder, Shallow Decoder: Reevaluating the Speed-Quality Tradeoff in Machine Translation}},\n author={{Jungo Kasai and Nikolaos Pappas and Hao Peng and James Cross and Noah A. Smith}},\n year={{2020}},\n eprint={{2006.10369}},\n archivePrefix={{arXiv}},\n primaryClass={{cs.CL}}\n}}\n```\n\n'
model_card_dir.mkdir(parents=__snake_case ,exist_ok=__snake_case )
lowerCamelCase__ = os.path.join(__snake_case ,'''README.md''' )
print(F'Generating {path}' )
with open(__snake_case ,'''w''' ,encoding='''utf-8''' ) as f:
f.write(__snake_case )
# make sure we are under the root of the project
_a = Path(__file__).resolve().parent.parent.parent
_a = repo_dir / "model_cards"
for model_name in ["wmt16-en-de-dist-12-1", "wmt16-en-de-dist-6-1", "wmt16-en-de-12-1"]:
_a = model_cards_dir / "allenai" / model_name
write_model_card(model_card_dir, src_lang="en", tgt_lang="de", model_name=model_name)
| 29
| 0
|
from __future__ import annotations
def _A ( lowerCAmelCase_ : list , lowerCAmelCase_ : int | None = None , lowerCAmelCase_ : int | None = None ):
"""simple docstring"""
if start is None:
lowerCAmelCase__ = 0
if end is None:
lowerCAmelCase__ = len(lowerCAmelCase_ ) - 1
if start >= end:
return
lowerCAmelCase__ = (start + end) // 2
slowsort(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
slowsort(lowerCAmelCase_ , mid + 1 , lowerCAmelCase_ )
if sequence[end] < sequence[mid]:
lowerCAmelCase__ , lowerCAmelCase__ = sequence[mid], sequence[end]
slowsort(lowerCAmelCase_ , lowerCAmelCase_ , end - 1 )
if __name__ == "__main__":
from doctest import testmod
testmod()
| 61
|
'''simple docstring'''
import inspect
import unittest
class lowerCAmelCase__ ( unittest.TestCase ):
"""simple docstring"""
def __lowerCAmelCase ( self : Dict ) -> Dict:
'''simple docstring'''
try:
import diffusers # noqa: F401
except ImportError:
assert False
def __lowerCAmelCase ( self : int ) -> str:
'''simple docstring'''
import diffusers
from diffusers.dependency_versions_table import deps
a__ : Optional[int] = inspect.getmembers(A__ , inspect.isclass )
for cls_name, cls_module in all_classes:
if "dummy_" in cls_module.__module__:
for backend in cls_module._backends:
if backend == "k_diffusion":
a__ : int = '''k-diffusion'''
elif backend == "invisible_watermark":
a__ : int = '''invisible-watermark'''
assert backend in deps, F'{backend} is not in the deps table!'
| 688
| 0
|
# Copyright 2023 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ..models.auto import AutoModelForVisionaSeq
from ..utils import requires_backends
from .base import PipelineTool
if TYPE_CHECKING:
from PIL import Image
class __lowerCAmelCase ( UpperCAmelCase_ ):
"""simple docstring"""
A__ : str = "Salesforce/blip-image-captioning-base"
A__ : Any = (
"This is a tool that generates a description of an image. It takes an input named `image` which should be the "
"image to caption, and returns a text that contains the description in English."
)
A__ : Optional[Any] = "image_captioner"
A__ : Dict = AutoModelForVisionaSeq
A__ : int = ["image"]
A__ : int = ["text"]
def __init__( self : Tuple , *_snake_case : Optional[int] , **_snake_case : str ):
"""simple docstring"""
requires_backends(self , ['vision'] )
super().__init__(*_snake_case , **_snake_case )
def _a ( self : List[Any] , _snake_case : "Image" ):
"""simple docstring"""
return self.pre_processor(images=_snake_case , return_tensors='pt' )
def _a ( self : Any , _snake_case : Optional[int] ):
"""simple docstring"""
return self.model.generate(**_snake_case )
def _a ( self : Dict , _snake_case : Union[str, Any] ):
"""simple docstring"""
return self.pre_processor.batch_decode(_snake_case , skip_special_tokens=_snake_case )[0].strip()
| 52
|
from typing import Dict
from .base import GenericTensor, Pipeline
class __lowerCAmelCase ( UpperCAmelCase_ ):
"""simple docstring"""
def _a ( self : Any , _snake_case : str=None , _snake_case : Dict=None , _snake_case : Any=None , **_snake_case : str ):
"""simple docstring"""
if tokenize_kwargs is None:
A__ = {}
if truncation is not None:
if "truncation" in tokenize_kwargs:
raise ValueError(
'truncation parameter defined twice (given as keyword argument as well as in tokenize_kwargs)' )
A__ = truncation
A__ = tokenize_kwargs
A__ = {}
if return_tensors is not None:
A__ = return_tensors
return preprocess_params, {}, postprocess_params
def _a ( self : Any , _snake_case : Dict , **_snake_case : Optional[Any] ):
"""simple docstring"""
A__ = self.framework
A__ = self.tokenizer(_snake_case , return_tensors=_snake_case , **_snake_case )
return model_inputs
def _a ( self : List[Any] , _snake_case : Dict ):
"""simple docstring"""
A__ = self.model(**_snake_case )
return model_outputs
def _a ( self : Optional[Any] , _snake_case : List[Any] , _snake_case : str=False ):
"""simple docstring"""
if return_tensors:
return model_outputs[0]
if self.framework == "pt":
return model_outputs[0].tolist()
elif self.framework == "tf":
return model_outputs[0].numpy().tolist()
def __call__( self : Dict , *_snake_case : int , **_snake_case : List[str] ):
"""simple docstring"""
return super().__call__(*_snake_case , **_snake_case )
| 52
| 1
|
import os
import time
import pytest
from datasets.utils.filelock import FileLock, Timeout
def SCREAMING_SNAKE_CASE ( lowercase_ ) -> Dict:
"""simple docstring"""
A__ = FileLock(str(tmpdir / '''foo.lock''' ) )
A__ = FileLock(str(tmpdir / '''foo.lock''' ) )
A__ = 0.01
with locka.acquire():
with pytest.raises(lowercase_ ):
A__ = time.time()
locka.acquire(lowercase_ )
assert time.time() - _start > timeout
def SCREAMING_SNAKE_CASE ( lowercase_ ) -> str:
"""simple docstring"""
A__ = '''a''' * 1_000 + '''.lock'''
A__ = FileLock(str(tmpdir / filename ) )
assert locka._lock_file.endswith('''.lock''' )
assert not locka._lock_file.endswith(lowercase_ )
assert len(os.path.basename(locka._lock_file ) ) <= 255
A__ = FileLock(tmpdir / filename )
with locka.acquire():
with pytest.raises(lowercase_ ):
locka.acquire(0 )
| 87
|
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
_lowerCamelCase : Union[str, Any] = logging.get_logger(__name__)
def SCREAMING_SNAKE_CASE ( lowercase_ , lowercase_ , lowercase_ , lowercase_=None , lowercase_=None ) -> Dict:
"""simple docstring"""
if "." in tensor_name:
A__ = tensor_name.split('''.''' )
for split in splits[:-1]:
A__ = getattr(lowercase_ , lowercase_ )
if new_module is None:
raise ValueError(f"""{module} has no attribute {split}.""" )
A__ = new_module
A__ = 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}.""" )
A__ = tensor_name in module._buffers
A__ = getattr(lowercase_ , lowercase_ )
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}.""" )
A__ = False
A__ = False
if is_buffer or not is_bitsandbytes_available():
A__ = False
A__ = False
else:
A__ = hasattr(bnb.nn , '''Params4bit''' ) and isinstance(module._parameters[tensor_name] , bnb.nn.Paramsabit )
A__ = isinstance(module._parameters[tensor_name] , bnb.nn.IntaParams )
if is_abit or is_abit:
A__ = module._parameters[tensor_name]
if param.device.type != "cuda":
if value is None:
A__ = old_value.to(lowercase_ )
elif isinstance(lowercase_ , torch.Tensor ):
A__ = value.to('''cpu''' )
if value.dtype == torch.inta:
A__ = 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:
A__ = torch.tensor(lowercase_ , 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 , lowercase_ ) and fpaa_statistics is None:
A__ = new_value.T
A__ = old_value.__dict__
if is_abit:
A__ = bnb.nn.IntaParams(lowercase_ , requires_grad=lowercase_ , **lowercase_ ).to(lowercase_ )
elif is_abit:
A__ = bnb.nn.Paramsabit(lowercase_ , requires_grad=lowercase_ , **lowercase_ ).to(lowercase_ )
A__ = new_value
if fpaa_statistics is not None:
setattr(module.weight , '''SCB''' , fpaa_statistics.to(lowercase_ ) )
else:
if value is None:
A__ = old_value.to(lowercase_ )
elif isinstance(lowercase_ , torch.Tensor ):
A__ = value.to(lowercase_ )
else:
A__ = torch.tensor(lowercase_ , device=lowercase_ )
if is_buffer:
A__ = new_value
else:
A__ = nn.Parameter(lowercase_ , requires_grad=old_value.requires_grad )
A__ = new_value
def SCREAMING_SNAKE_CASE ( lowercase_ , lowercase_=None , lowercase_=None , lowercase_=None , lowercase_=False ) -> Dict:
"""simple docstring"""
for name, module in model.named_children():
if current_key_name is None:
A__ = []
current_key_name.append(lowercase_ )
if (isinstance(lowercase_ , nn.Linear ) or isinstance(lowercase_ , lowercase_ )) 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(lowercase_ ) for key in modules_to_not_convert ):
with init_empty_weights():
if isinstance(lowercase_ , lowercase_ ):
A__ , A__ = module.weight.shape
else:
A__ = module.in_features
A__ = module.out_features
if quantization_config.quantization_method() == "llm_int8":
A__ = bnb.nn.LinearabitLt(
lowercase_ , lowercase_ , module.bias is not None , has_fpaa_weights=quantization_config.llm_inta_has_fpaa_weight , threshold=quantization_config.llm_inta_threshold , )
A__ = True
else:
if (
quantization_config.llm_inta_skip_modules is not None
and name in quantization_config.llm_inta_skip_modules
):
pass
else:
A__ = bnb.nn.Linearabit(
lowercase_ , lowercase_ , 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 , )
A__ = True
# Store the module class in case we need to transpose the weight later
A__ = type(lowercase_ )
# Force requires grad to False to avoid unexpected errors
model._modules[name].requires_grad_(lowercase_ )
if len(list(module.children() ) ) > 0:
A__ , A__ = _replace_with_bnb_linear(
lowercase_ , lowercase_ , lowercase_ , lowercase_ , has_been_replaced=lowercase_ , )
# Remove the last key for recursion
current_key_name.pop(-1 )
return model, has_been_replaced
def SCREAMING_SNAKE_CASE ( lowercase_ , lowercase_=None , lowercase_=None , lowercase_=None ) -> Tuple:
"""simple docstring"""
A__ = ['''lm_head'''] if modules_to_not_convert is None else modules_to_not_convert
A__ , A__ = _replace_with_bnb_linear(
lowercase_ , lowercase_ , lowercase_ , lowercase_ )
if not has_been_replaced:
logger.warning(
'''You are loading your model in 8bit or 4bit but no linear modules were found in your model.'''
''' Please double check your model architecture, or submit an issue on github if you think this is'''
''' a bug.''' )
return model
def SCREAMING_SNAKE_CASE ( *lowercase_ , **lowercase_ ) -> Dict:
"""simple docstring"""
warnings.warn(
'''`replace_8bit_linear` will be deprecated in a future version, please use `replace_with_bnb_linear` instead''' , lowercase_ , )
return replace_with_bnb_linear(*lowercase_ , **lowercase_ )
def SCREAMING_SNAKE_CASE ( *lowercase_ , **lowercase_ ) -> Optional[Any]:
"""simple docstring"""
warnings.warn(
'''`set_module_8bit_tensor_to_device` will be deprecated in a future version, please use `set_module_quantized_tensor_to_device` instead''' , lowercase_ , )
return set_module_quantized_tensor_to_device(*lowercase_ , **lowercase_ )
def SCREAMING_SNAKE_CASE ( lowercase_ ) -> List[str]:
"""simple docstring"""
A__ = deepcopy(lowercase_ ) # this has 0 cost since it is done inside `init_empty_weights` context manager`
tied_model.tie_weights()
A__ = find_tied_parameters(lowercase_ )
# For compatibility with Accelerate < 0.18
if isinstance(lowercase_ , lowercase_ ):
A__ = sum(list(tied_params.values() ) , [] ) + list(tied_params.keys() )
else:
A__ = sum(lowercase_ , [] )
A__ = len(lowercase_ ) > 0
# Check if it is a base model
A__ = not hasattr(lowercase_ , model.base_model_prefix )
# Ignore this for base models (BertModel, GPT2Model, etc.)
if (not has_tied_params) and is_base_model:
return []
# otherwise they have an attached head
A__ = list(model.named_children() )
A__ = [list_modules[-1][0]]
# add last module together with tied weights
A__ = set(lowercase_ ) - set(lowercase_ )
A__ = list(set(lowercase_ ) ) + list(lowercase_ )
# remove ".weight" from the keys
A__ = ['''.weight''', '''.bias''']
A__ = []
for name in list_untouched:
for name_to_remove in names_to_remove:
if name_to_remove in name:
A__ = name.replace(lowercase_ , '''''' )
filtered_module_names.append(lowercase_ )
return filtered_module_names
| 87
| 1
|
from itertools import product
from cva import COLOR_BGR2GRAY, cvtColor, imread, imshow, waitKey
from numpy import dot, exp, mgrid, pi, ravel, square, uinta, zeros
def a__ ( snake_case , snake_case ):
"""simple docstring"""
__SCREAMING_SNAKE_CASE : int = k_size // 2
__SCREAMING_SNAKE_CASE : List[str] = mgrid[0 - center : k_size - center, 0 - center : k_size - center]
__SCREAMING_SNAKE_CASE : Dict = 1 / (2 * pi * sigma) * exp(-(square(snake_case ) + square(snake_case )) / (2 * square(snake_case )) )
return g
def a__ ( snake_case , snake_case , snake_case ):
"""simple docstring"""
__SCREAMING_SNAKE_CASE : str = image.shape[0], image.shape[1]
# dst image height and width
__SCREAMING_SNAKE_CASE : str = height - k_size + 1
__SCREAMING_SNAKE_CASE : int = width - k_size + 1
# im2col, turn the k_size*k_size pixels into a row and np.vstack all rows
__SCREAMING_SNAKE_CASE : Any = zeros((dst_height * dst_width, k_size * k_size) )
__SCREAMING_SNAKE_CASE : List[str] = 0
for i, j in product(range(snake_case ) , range(snake_case ) ):
__SCREAMING_SNAKE_CASE : int = ravel(image[i : i + k_size, j : j + k_size] )
__SCREAMING_SNAKE_CASE : Tuple = window
row += 1
# turn the kernel into shape(k*k, 1)
__SCREAMING_SNAKE_CASE : int = gen_gaussian_kernel(snake_case , snake_case )
__SCREAMING_SNAKE_CASE : Dict = ravel(snake_case )
# reshape and get the dst image
__SCREAMING_SNAKE_CASE : str = dot(snake_case , snake_case ).reshape(snake_case , snake_case ).astype(snake_case )
return dst
if __name__ == "__main__":
# read original image
lowercase_ = imread(R"""../image_data/lena.jpg""")
# turn image in gray scale value
lowercase_ = cvtColor(img, COLOR_BGR2GRAY)
# get values with two different mask size
lowercase_ = gaussian_filter(gray, 3, sigma=1)
lowercase_ = gaussian_filter(gray, 5, sigma=0.8)
# show result images
imshow("""gaussian filter with 3x3 mask""", gaussianaxa)
imshow("""gaussian filter with 5x5 mask""", gaussianaxa)
waitKey()
| 720
|
import functools
def a__ ( snake_case , snake_case ):
"""simple docstring"""
__SCREAMING_SNAKE_CASE : Tuple = len(snake_case )
__SCREAMING_SNAKE_CASE : Optional[int] = len(snake_case )
@functools.cache
def min_distance(snake_case , snake_case ) -> int:
# if first word index is overflow - delete all from the second word
if indexa >= len_worda:
return len_worda - indexa
# if second word index is overflow - delete all from the first word
if indexa >= len_worda:
return len_worda - indexa
__SCREAMING_SNAKE_CASE : Union[str, Any] = int(worda[indexa] != worda[indexa] ) # current letters not identical
return min(
1 + min_distance(indexa + 1 , snake_case ) , 1 + min_distance(snake_case , indexa + 1 ) , diff + min_distance(indexa + 1 , indexa + 1 ) , )
return min_distance(0 , 0 )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 131
| 0
|
from collections import UserDict
from typing import Union
import numpy as np
import requests
from ..utils import (
add_end_docstrings,
logging,
)
from .audio_classification import ffmpeg_read
from .base import PIPELINE_INIT_ARGS, Pipeline
lowerCAmelCase = logging.get_logger(__name__)
@add_end_docstrings(_UpperCamelCase )
class lowerCamelCase ( _UpperCamelCase ):
def __init__( self , **lowercase__):
super().__init__(**lowercase__)
if self.framework != "pt":
raise ValueError(F"The {self.__class__} is only available in PyTorch.")
# No specific FOR_XXX available yet
def __call__( self , lowercase__ , **lowercase__):
return super().__call__(lowercase__ , **lowercase__)
def A( self , **lowercase__):
__UpperCAmelCase : Optional[int] = {}
if "candidate_labels" in kwargs:
__UpperCAmelCase : List[str] = kwargs['''candidate_labels''']
if "hypothesis_template" in kwargs:
__UpperCAmelCase : Union[str, Any] = kwargs['''hypothesis_template''']
return preprocess_params, {}, {}
def A( self , lowercase__ , lowercase__=None , lowercase__="This is a sound of {}."):
if isinstance(lowercase__ , lowercase__):
if audio.startswith('''http://''') or audio.startswith('''https://'''):
# We need to actually check for a real protocol, otherwise it's impossible to use a local file
# like http_huggingface_co.png
__UpperCAmelCase : Any = requests.get(lowercase__).content
else:
with open(lowercase__ , '''rb''') as f:
__UpperCAmelCase : Union[str, Any] = f.read()
if isinstance(lowercase__ , lowercase__):
__UpperCAmelCase : Any = ffmpeg_read(lowercase__ , self.feature_extractor.sampling_rate)
if not isinstance(lowercase__ , np.ndarray):
raise ValueError('''We expect a numpy ndarray as input''')
if len(audio.shape) != 1:
raise ValueError('''We expect a single channel audio input for ZeroShotAudioClassificationPipeline''')
__UpperCAmelCase : str = self.feature_extractor(
[audio] , sampling_rate=self.feature_extractor.sampling_rate , return_tensors='''pt''')
__UpperCAmelCase : Tuple = candidate_labels
__UpperCAmelCase : str = [hypothesis_template.format(lowercase__) for x in candidate_labels]
__UpperCAmelCase : Optional[int] = self.tokenizer(lowercase__ , return_tensors=self.framework , padding=lowercase__)
__UpperCAmelCase : Optional[Any] = [text_inputs]
return inputs
def A( self , lowercase__):
__UpperCAmelCase : List[str] = model_inputs.pop('''candidate_labels''')
__UpperCAmelCase : str = model_inputs.pop('''text_inputs''')
if isinstance(text_inputs[0] , lowercase__):
__UpperCAmelCase : Dict = text_inputs[0]
else:
# Batching case.
__UpperCAmelCase : Any = text_inputs[0][0]
__UpperCAmelCase : Any = self.model(**lowercase__ , **lowercase__)
__UpperCAmelCase : Dict = {
'''candidate_labels''': candidate_labels,
'''logits''': outputs.logits_per_audio,
}
return model_outputs
def A( self , lowercase__):
__UpperCAmelCase : str = model_outputs.pop('''candidate_labels''')
__UpperCAmelCase : Tuple = model_outputs['''logits'''][0]
if self.framework == "pt":
__UpperCAmelCase : Dict = logits.softmax(dim=0)
__UpperCAmelCase : Union[str, Any] = probs.tolist()
else:
raise ValueError('''`tf` framework not supported.''')
__UpperCAmelCase : Dict = [
{'''score''': score, '''label''': candidate_label}
for score, candidate_label in sorted(zip(lowercase__ , lowercase__) , key=lambda lowercase__: -x[0])
]
return result
| 462
|
from dataclasses import dataclass
from typing import Optional, Tuple, Union
import numpy as np
import torch
from ..configuration_utils import ConfigMixin, register_to_config
from ..utils import BaseOutput, randn_tensor
from .scheduling_utils import SchedulerMixin
@dataclass
class lowerCamelCase ( _UpperCamelCase ):
_lowerCAmelCase : torch.FloatTensor
_lowerCAmelCase : torch.FloatTensor
_lowerCAmelCase : Optional[torch.FloatTensor] = None
class lowerCamelCase ( _UpperCamelCase , _UpperCamelCase ):
_lowerCAmelCase : Tuple = 2
@register_to_config
def __init__( self , lowercase__ = 0.0_2 , lowercase__ = 1_0_0 , lowercase__ = 1.0_0_7 , lowercase__ = 8_0 , lowercase__ = 0.0_5 , lowercase__ = 5_0 , ):
# standard deviation of the initial noise distribution
__UpperCAmelCase : Union[str, Any] = sigma_max
# setable values
__UpperCAmelCase : int = None
__UpperCAmelCase : np.IntTensor = None
__UpperCAmelCase : torch.FloatTensor = None # sigma(t_i)
def A( self , lowercase__ , lowercase__ = None):
return sample
def A( self , lowercase__ , lowercase__ = None):
__UpperCAmelCase : int = num_inference_steps
__UpperCAmelCase : List[str] = np.arange(0 , self.num_inference_steps)[::-1].copy()
__UpperCAmelCase : Any = torch.from_numpy(lowercase__).to(lowercase__)
__UpperCAmelCase : Union[str, Any] = [
(
self.config.sigma_max**2
* (self.config.sigma_min**2 / self.config.sigma_max**2) ** (i / (num_inference_steps - 1))
)
for i in self.timesteps
]
__UpperCAmelCase : Tuple = torch.tensor(lowercase__ , dtype=torch.floataa , device=lowercase__)
def A( self , lowercase__ , lowercase__ , lowercase__ = None):
if self.config.s_min <= sigma <= self.config.s_max:
__UpperCAmelCase : int = min(self.config.s_churn / self.num_inference_steps , 2**0.5 - 1)
else:
__UpperCAmelCase : int = 0
# sample eps ~ N(0, S_noise^2 * I)
__UpperCAmelCase : List[str] = self.config.s_noise * randn_tensor(sample.shape , generator=lowercase__).to(sample.device)
__UpperCAmelCase : Optional[int] = sigma + gamma * sigma
__UpperCAmelCase : List[Any] = sample + ((sigma_hat**2 - sigma**2) ** 0.5 * eps)
return sample_hat, sigma_hat
def A( self , lowercase__ , lowercase__ , lowercase__ , lowercase__ , lowercase__ = True , ):
__UpperCAmelCase : str = sample_hat + sigma_hat * model_output
__UpperCAmelCase : Tuple = (sample_hat - pred_original_sample) / sigma_hat
__UpperCAmelCase : str = sample_hat + (sigma_prev - sigma_hat) * derivative
if not return_dict:
return (sample_prev, derivative)
return KarrasVeOutput(
prev_sample=lowercase__ , derivative=lowercase__ , pred_original_sample=lowercase__)
def A( self , lowercase__ , lowercase__ , lowercase__ , lowercase__ , lowercase__ , lowercase__ , lowercase__ = True , ):
__UpperCAmelCase : Any = sample_prev + sigma_prev * model_output
__UpperCAmelCase : List[str] = (sample_prev - pred_original_sample) / sigma_prev
__UpperCAmelCase : Union[str, Any] = sample_hat + (sigma_prev - sigma_hat) * (0.5 * derivative + 0.5 * derivative_corr)
if not return_dict:
return (sample_prev, derivative)
return KarrasVeOutput(
prev_sample=lowercase__ , derivative=lowercase__ , pred_original_sample=lowercase__)
def A( self , lowercase__ , lowercase__ , lowercase__):
raise NotImplementedError()
| 462
| 1
|
"""simple docstring"""
import importlib
import os
import fsspec
import pytest
from fsspec import register_implementation
from fsspec.registry import _registry as _fsspec_registry
from datasets.filesystems import COMPRESSION_FILESYSTEMS, HfFileSystem, extract_path_from_uri, is_remote_filesystem
from .utils import require_lza, require_zstandard
def lowerCAmelCase_ ( SCREAMING_SNAKE_CASE__ ) -> Tuple:
assert "mock" in _fsspec_registry
assert "bz2" in _fsspec_registry
def lowerCAmelCase_ ( ) -> Optional[int]:
assert "mock" not in _fsspec_registry
assert "bz2" in _fsspec_registry
def lowerCAmelCase_ ( ) -> Optional[int]:
a_ : Tuple = "mock-s3-bucket"
a_ : Dict = F"""s3://{mock_bucket}"""
a_ : Any = extract_path_from_uri(SCREAMING_SNAKE_CASE__ )
assert dataset_path.startswith("s3://" ) is False
a_ : Tuple = "./local/path"
a_ : str = extract_path_from_uri(SCREAMING_SNAKE_CASE__ )
assert dataset_path == new_dataset_path
def lowerCAmelCase_ ( SCREAMING_SNAKE_CASE__ ) -> Tuple:
a_ : int = is_remote_filesystem(SCREAMING_SNAKE_CASE__ )
assert is_remote is True
a_ : Union[str, Any] = fsspec.filesystem("file" )
a_ : Optional[Any] = is_remote_filesystem(SCREAMING_SNAKE_CASE__ )
assert is_remote is False
@pytest.mark.parametrize("compression_fs_class", SCREAMING_SNAKE_CASE__ )
def lowerCAmelCase_ ( SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__ ) -> List[Any]:
a_ : str = {"gzip": gz_file, "xz": xz_file, "zstd": zstd_file, "bz2": bza_file, "lz4": lza_file}
a_ : Tuple = input_paths[compression_fs_class.protocol]
if input_path is None:
a_ : int = F"""for '{compression_fs_class.protocol}' compression protocol, """
if compression_fs_class.protocol == "lz4":
reason += require_lza.kwargs["reason"]
elif compression_fs_class.protocol == "zstd":
reason += require_zstandard.kwargs["reason"]
pytest.skip(SCREAMING_SNAKE_CASE__ )
a_ : str = fsspec.filesystem(compression_fs_class.protocol, fo=SCREAMING_SNAKE_CASE__ )
assert isinstance(SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__ )
a_ : List[Any] = os.path.basename(SCREAMING_SNAKE_CASE__ )
a_ : Any = expected_filename[: expected_filename.rindex("." )]
assert fs.glob("*" ) == [expected_filename]
with fs.open(SCREAMING_SNAKE_CASE__, "r", encoding="utf-8" ) as f, open(SCREAMING_SNAKE_CASE__, encoding="utf-8" ) as expected_file:
assert f.read() == expected_file.read()
@pytest.mark.parametrize("protocol", ["zip", "gzip"] )
def lowerCAmelCase_ ( SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__ ) -> Optional[int]:
a_ : Dict = {"zip": zip_jsonl_path, "gzip": jsonl_gz_path}
a_ : Dict = compressed_file_paths[protocol]
a_ : Any = "dataset.jsonl"
a_ : str = F"""{protocol}://{member_file_path}::{compressed_file_path}"""
a_ : Dict = fsspec.get_fs_token_paths(SCREAMING_SNAKE_CASE__ )
assert fs.isfile(SCREAMING_SNAKE_CASE__ )
assert not fs.isfile("non_existing_" + member_file_path )
@pytest.mark.integration
def lowerCAmelCase_ ( SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__ ) -> List[str]:
a_ : Optional[Any] = hf_api.dataset_info(SCREAMING_SNAKE_CASE__, token=SCREAMING_SNAKE_CASE__ )
a_ : int = HfFileSystem(repo_info=SCREAMING_SNAKE_CASE__, token=SCREAMING_SNAKE_CASE__ )
assert sorted(hffs.glob("*" ) ) == [".gitattributes", "data"]
assert hffs.isdir("data" )
assert hffs.isfile(".gitattributes" ) and hffs.isfile("data/text_data.txt" )
with open(SCREAMING_SNAKE_CASE__ ) as f:
assert hffs.open("data/text_data.txt", "r" ).read() == f.read()
def lowerCAmelCase_ ( ) -> str:
a_ : Any = "bz2"
# Import module
import datasets.filesystems
# Overwrite protocol and reload
register_implementation(SCREAMING_SNAKE_CASE__, SCREAMING_SNAKE_CASE__, clobber=SCREAMING_SNAKE_CASE__ )
with pytest.warns(SCREAMING_SNAKE_CASE__ ) as warning_info:
importlib.reload(datasets.filesystems )
assert len(SCREAMING_SNAKE_CASE__ ) == 1
assert (
str(warning_info[0].message )
== F"""A filesystem protocol was already set for {protocol} and will be overwritten."""
)
| 720
|
"""simple docstring"""
from collections import OrderedDict
from typing import Any, Mapping, Optional
from ... import PreTrainedTokenizer
from ...configuration_utils import PretrainedConfig
from ...file_utils import TensorType, is_torch_available
from ...onnx import OnnxConfig, OnnxConfigWithPast, OnnxSeqaSeqConfigWithPast
from ...onnx.utils import compute_effective_axis_dimension
from ...utils import logging
SCREAMING_SNAKE_CASE_ = logging.get_logger(__name__)
SCREAMING_SNAKE_CASE_ = {
"""facebook/blenderbot_small-90M""": """https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/config.json""",
# See all BlenderbotSmall models at https://huggingface.co/models?filter=blenderbot_small
}
class snake_case_ ( a_ ):
__lowerCAmelCase = "blenderbot-small"
__lowerCAmelCase = ["past_key_values"]
__lowerCAmelCase = {"num_attention_heads": "encoder_attention_heads", "hidden_size": "d_model"}
def __init__( self , a_=5_0_2_6_5 , a_=5_1_2 , a_=8 , a_=2_0_4_8 , a_=1_6 , a_=8 , a_=2_0_4_8 , a_=1_6 , a_=0.0 , a_=0.0 , a_=True , a_=True , a_="gelu" , a_=5_1_2 , a_=0.1 , a_=0.0 , a_=0.0 , a_=0.02 , a_=1 , a_=False , a_=0 , a_=1 , a_=2 , a_=2 , **a_ , ):
a_ : int = vocab_size
a_ : Any = max_position_embeddings
a_ : Optional[int] = d_model
a_ : Tuple = encoder_ffn_dim
a_ : List[Any] = encoder_layers
a_ : Optional[int] = encoder_attention_heads
a_ : Optional[int] = decoder_ffn_dim
a_ : List[str] = decoder_layers
a_ : Dict = decoder_attention_heads
a_ : List[str] = dropout
a_ : List[Any] = attention_dropout
a_ : List[str] = activation_dropout
a_ : Optional[Any] = activation_function
a_ : List[Any] = init_std
a_ : int = encoder_layerdrop
a_ : Optional[int] = decoder_layerdrop
a_ : List[str] = use_cache
a_ : Optional[int] = encoder_layers
a_ : Optional[int] = scale_embedding # scale factor will be sqrt(d_model) if True
super().__init__(
pad_token_id=a_ , bos_token_id=a_ , eos_token_id=a_ , is_encoder_decoder=a_ , decoder_start_token_id=a_ , forced_eos_token_id=a_ , **a_ , )
class snake_case_ ( a_ ):
@property
def snake_case_ ( self ):
if self.task in ["default", "seq2seq-lm"]:
a_ : Optional[int] = OrderedDict(
[
("input_ids", {0: "batch", 1: "encoder_sequence"}),
("attention_mask", {0: "batch", 1: "encoder_sequence"}),
] )
if self.use_past:
a_ : Tuple = {0: "batch"}
a_ : int = {0: "batch", 1: "past_decoder_sequence + sequence"}
else:
a_ : List[str] = {0: "batch", 1: "decoder_sequence"}
a_ : Optional[int] = {0: "batch", 1: "decoder_sequence"}
if self.use_past:
self.fill_with_past_key_values_(a_ , direction="inputs" )
elif self.task == "causal-lm":
# TODO: figure this case out.
a_ : Dict = OrderedDict(
[
("input_ids", {0: "batch", 1: "encoder_sequence"}),
("attention_mask", {0: "batch", 1: "encoder_sequence"}),
] )
if self.use_past:
a_ , a_ : Optional[int] = self.num_layers
for i in range(a_ ):
a_ : Union[str, Any] = {0: "batch", 2: "past_sequence + sequence"}
a_ : List[Any] = {0: "batch", 2: "past_sequence + sequence"}
else:
a_ : List[str] = OrderedDict(
[
("input_ids", {0: "batch", 1: "encoder_sequence"}),
("attention_mask", {0: "batch", 1: "encoder_sequence"}),
("decoder_input_ids", {0: "batch", 1: "decoder_sequence"}),
("decoder_attention_mask", {0: "batch", 1: "decoder_sequence"}),
] )
return common_inputs
@property
def snake_case_ ( self ):
if self.task in ["default", "seq2seq-lm"]:
a_ : List[Any] = super().outputs
else:
a_ : Tuple = super(a_ , self ).outputs
if self.use_past:
a_ , a_ : Dict = self.num_layers
for i in range(a_ ):
a_ : List[Any] = {0: "batch", 2: "past_sequence + sequence"}
a_ : List[Any] = {0: "batch", 2: "past_sequence + sequence"}
return common_outputs
def snake_case_ ( self , a_ , a_ = -1 , a_ = -1 , a_ = False , a_ = None , ):
a_ : List[str] = self._generate_dummy_inputs_for_sequence_classification_and_question_answering(
a_ , a_ , a_ , a_ , a_ )
# Generate decoder inputs
a_ : Optional[int] = seq_length if not self.use_past else 1
a_ : str = self._generate_dummy_inputs_for_sequence_classification_and_question_answering(
a_ , a_ , a_ , a_ , a_ )
a_ : int = {F"""decoder_{name}""": tensor for name, tensor in decoder_inputs.items()}
a_ : Tuple = dict(**a_ , **a_ )
if self.use_past:
if not is_torch_available():
raise ValueError("Cannot generate dummy past_keys inputs without PyTorch installed." )
else:
import torch
a_ , a_ : Optional[int] = common_inputs["input_ids"].shape
a_ : str = common_inputs["decoder_input_ids"].shape[1]
a_ , a_ : Dict = self.num_attention_heads
a_ : List[str] = (
batch,
num_encoder_attention_heads,
encoder_seq_length,
self._config.hidden_size // num_encoder_attention_heads,
)
a_ : Optional[Any] = decoder_seq_length + 3
a_ : str = (
batch,
num_decoder_attention_heads,
decoder_past_length,
self._config.hidden_size // num_decoder_attention_heads,
)
a_ : Dict = torch.cat(
[common_inputs["decoder_attention_mask"], torch.ones(a_ , a_ )] , dim=1 )
a_ : Optional[int] = []
# If the number of encoder and decoder layers are present in the model configuration, both are considered
a_ , a_ : Tuple = self.num_layers
a_ : str = min(a_ , a_ )
a_ : Dict = max(a_ , a_ ) - min_num_layers
a_ : Union[str, Any] = "encoder" if num_encoder_layers > num_decoder_layers else "decoder"
for _ in range(a_ ):
common_inputs["past_key_values"].append(
(
torch.zeros(a_ ),
torch.zeros(a_ ),
torch.zeros(a_ ),
torch.zeros(a_ ),
) )
# TODO: test this.
a_ : Optional[int] = encoder_shape if remaining_side_name == "encoder" else decoder_shape
for _ in range(a_ , a_ ):
common_inputs["past_key_values"].append((torch.zeros(a_ ), torch.zeros(a_ )) )
return common_inputs
def snake_case_ ( self , a_ , a_ = -1 , a_ = -1 , a_ = False , a_ = None , ):
a_ : Optional[int] = self._generate_dummy_inputs_for_sequence_classification_and_question_answering(
a_ , a_ , a_ , a_ , a_ )
if self.use_past:
if not is_torch_available():
raise ValueError("Cannot generate dummy past_keys inputs without PyTorch installed." )
else:
import torch
a_ , a_ : Dict = common_inputs["input_ids"].shape
# Not using the same length for past_key_values
a_ : int = seqlen + 2
a_ , a_ : Optional[int] = self.num_layers
a_ , a_ : Optional[int] = self.num_attention_heads
a_ : Optional[Any] = (
batch,
num_encoder_attention_heads,
past_key_values_length,
self._config.hidden_size // num_encoder_attention_heads,
)
a_ : str = common_inputs["attention_mask"].dtype
a_ : Tuple = torch.cat(
[common_inputs["attention_mask"], torch.ones(a_ , a_ , dtype=a_ )] , dim=1 )
a_ : Optional[int] = [
(torch.zeros(a_ ), torch.zeros(a_ )) for _ in range(a_ )
]
return common_inputs
def snake_case_ ( self , a_ , a_ = -1 , a_ = -1 , a_ = False , a_ = None , ):
# Copied from OnnxConfig.generate_dummy_inputs
# Did not use super(OnnxConfigWithPast, self).generate_dummy_inputs for code clarity.
# If dynamic axis (-1) we forward with a fixed dimension of 2 samples to avoid optimizations made by ONNX
a_ : Optional[Any] = compute_effective_axis_dimension(
a_ , fixed_dimension=OnnxConfig.default_fixed_batch , num_token_to_add=0 )
# If dynamic axis (-1) we forward with a fixed dimension of 8 tokens to avoid optimizations made by ONNX
a_ : Tuple = tokenizer.num_special_tokens_to_add(a_ )
a_ : Union[str, Any] = compute_effective_axis_dimension(
a_ , fixed_dimension=OnnxConfig.default_fixed_sequence , num_token_to_add=a_ )
# Generate dummy inputs according to compute batch and sequence
a_ : Union[str, Any] = [" ".join([tokenizer.unk_token] ) * seq_length] * batch_size
a_ : str = dict(tokenizer(a_ , return_tensors=a_ ) )
return common_inputs
def snake_case_ ( self , a_ , a_ = -1 , a_ = -1 , a_ = False , a_ = None , ):
if self.task in ["default", "seq2seq-lm"]:
a_ : List[str] = self._generate_dummy_inputs_for_default_and_seqaseq_lm(
a_ , batch_size=a_ , seq_length=a_ , is_pair=a_ , framework=a_ )
elif self.task == "causal-lm":
a_ : List[Any] = self._generate_dummy_inputs_for_causal_lm(
a_ , batch_size=a_ , seq_length=a_ , is_pair=a_ , framework=a_ )
else:
a_ : Union[str, Any] = self._generate_dummy_inputs_for_sequence_classification_and_question_answering(
a_ , batch_size=a_ , seq_length=a_ , is_pair=a_ , framework=a_ )
return common_inputs
def snake_case_ ( self , a_ , a_ , a_ , a_ ):
if self.task in ["default", "seq2seq-lm"]:
a_ : Optional[int] = super()._flatten_past_key_values_(a_ , a_ , a_ , a_ )
else:
a_ : int = super(a_ , self )._flatten_past_key_values_(
a_ , a_ , a_ , a_ )
| 370
| 0
|
_UpperCAmelCase : Tuple = 8.3144598
def SCREAMING_SNAKE_CASE ( _UpperCAmelCase , _UpperCAmelCase ) -> float:
if temperature < 0:
raise Exception('Temperature cannot be less than 0 K' )
if molar_mass <= 0:
raise Exception('Molar mass cannot be less than or equal to 0 kg/mol' )
else:
return (3 * UNIVERSAL_GAS_CONSTANT * temperature / molar_mass) ** 0.5
if __name__ == "__main__":
import doctest
# run doctest
doctest.testmod()
# example
_UpperCAmelCase : List[str] = 3_00
_UpperCAmelCase : List[Any] = 28
_UpperCAmelCase : List[Any] = rms_speed_of_molecule(temperature, molar_mass)
print(F"""Vrms of Nitrogen gas at 300 K is {vrms} m/s""")
| 295
|
from __future__ import annotations
_UpperCAmelCase : Tuple = """#"""
class lowerCAmelCase :
def __init__( self : Optional[Any] ) -> None:
lowerCamelCase__ : dict = {}
def A_ ( self : Optional[int] , UpperCAmelCase : str ) -> None:
lowerCamelCase__ : int = self._trie
for char in text:
if char not in trie:
lowerCamelCase__ : Tuple = {}
lowerCamelCase__ : Dict = trie[char]
lowerCamelCase__ : Optional[int] = True
def A_ ( self : Optional[int] , UpperCAmelCase : str ) -> tuple | list:
lowerCamelCase__ : Tuple = self._trie
for char in prefix:
if char in trie:
lowerCamelCase__ : List[Any] = trie[char]
else:
return []
return self._elements(UpperCAmelCase )
def A_ ( self : List[str] , UpperCAmelCase : dict ) -> tuple:
lowerCamelCase__ : Optional[Any] = []
for c, v in d.items():
lowerCamelCase__ : Optional[Any] = [' '] if c == END else [(c + s) for s in self._elements(UpperCAmelCase )]
result.extend(UpperCAmelCase )
return tuple(UpperCAmelCase )
_UpperCAmelCase : str = Trie()
_UpperCAmelCase : Tuple = ("""depart""", """detergent""", """daring""", """dog""", """deer""", """deal""")
for word in words:
trie.insert_word(word)
def SCREAMING_SNAKE_CASE ( _UpperCAmelCase ) -> tuple:
lowerCamelCase__ : str = trie.find_word(_UpperCAmelCase )
return tuple(string + word for word in suffixes )
def SCREAMING_SNAKE_CASE ( ) -> None:
print(autocomplete_using_trie('de' ) )
if __name__ == "__main__":
import doctest
doctest.testmod()
main()
| 295
| 1
|
from argparse import ArgumentParser, Namespace
from typing import Any, List, Optional
from ..pipelines import Pipeline, get_supported_tasks, pipeline
from ..utils import logging
from . import BaseTransformersCLICommand
try:
from fastapi import Body, FastAPI, HTTPException
from fastapi.routing import APIRoute
from pydantic import BaseModel
from starlette.responses import JSONResponse
from uvicorn import run
__A : Tuple = True
except (ImportError, AttributeError):
__A : Optional[Any] = object
def __UpperCamelCase ( *_A : Tuple , **_A : Tuple ) ->Tuple:
"""simple docstring"""
pass
__A : List[Any] = False
__A : List[Any] = logging.get_logger('transformers-cli/serving')
def __UpperCamelCase ( _A : Namespace ) ->int:
"""simple docstring"""
lowerCamelCase_ =pipeline(
task=args.task , model=args.model if args.model else None , config=args.config , tokenizer=args.tokenizer , device=args.device , )
return ServeCommand(lowerCamelCase_ , args.host , args.port , args.workers )
class _SCREAMING_SNAKE_CASE ( lowerCamelCase__):
_UpperCamelCase:Any = 42
class _SCREAMING_SNAKE_CASE ( lowerCamelCase__):
_UpperCamelCase:List[str] = 42
_UpperCamelCase:Dict = 42
class _SCREAMING_SNAKE_CASE ( lowerCamelCase__):
_UpperCamelCase:Optional[Any] = 42
class _SCREAMING_SNAKE_CASE ( lowerCamelCase__):
_UpperCamelCase:Dict = 42
class _SCREAMING_SNAKE_CASE ( lowerCamelCase__):
@staticmethod
def _snake_case ( _SCREAMING_SNAKE_CASE )-> Optional[int]:
lowerCamelCase_ =parser.add_parser(
"""serve""" , help="""CLI tool to run inference requests through REST and GraphQL endpoints.""" )
serve_parser.add_argument(
"""--task""" , type=__lowerCamelCase , choices=get_supported_tasks() , help="""The task to run the pipeline on""" , )
serve_parser.add_argument("""--host""" , type=__lowerCamelCase , default="""localhost""" , help="""Interface the server will listen on.""" )
serve_parser.add_argument("""--port""" , type=__lowerCamelCase , default=8888 , help="""Port the serving will listen to.""" )
serve_parser.add_argument("""--workers""" , type=__lowerCamelCase , default=1 , help="""Number of http workers""" )
serve_parser.add_argument("""--model""" , type=__lowerCamelCase , help="""Model\'s name or path to stored model.""" )
serve_parser.add_argument("""--config""" , type=__lowerCamelCase , help="""Model\'s config name or path to stored model.""" )
serve_parser.add_argument("""--tokenizer""" , type=__lowerCamelCase , help="""Tokenizer name to use.""" )
serve_parser.add_argument(
"""--device""" , type=__lowerCamelCase , default=-1 , help="""Indicate the device to run onto, -1 indicates CPU, >= 0 indicates GPU (default: -1)""" , )
serve_parser.set_defaults(func=__lowerCamelCase )
def __init__( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )-> List[str]:
lowerCamelCase_ =pipeline
lowerCamelCase_ =host
lowerCamelCase_ =port
lowerCamelCase_ =workers
if not _serve_dependencies_installed:
raise RuntimeError(
"""Using serve command requires FastAPI and uvicorn. """
"""Please install transformers with [serving]: pip install \"transformers[serving]\"."""
"""Or install FastAPI and uvicorn separately.""" )
else:
logger.info(f'Serving model over {host}:{port}' )
lowerCamelCase_ =FastAPI(
routes=[
APIRoute(
"""/""" , self.model_info , response_model=__lowerCamelCase , response_class=__lowerCamelCase , methods=["""GET"""] , ),
APIRoute(
"""/tokenize""" , self.tokenize , response_model=__lowerCamelCase , response_class=__lowerCamelCase , methods=["""POST"""] , ),
APIRoute(
"""/detokenize""" , self.detokenize , response_model=__lowerCamelCase , response_class=__lowerCamelCase , methods=["""POST"""] , ),
APIRoute(
"""/forward""" , self.forward , response_model=__lowerCamelCase , response_class=__lowerCamelCase , methods=["""POST"""] , ),
] , timeout=600 , )
def _snake_case ( self )-> Optional[Any]:
run(self._app , host=self.host , port=self.port , workers=self.workers )
def _snake_case ( self )-> List[str]:
return ServeModelInfoResult(infos=vars(self._pipeline.model.config ) )
def _snake_case ( self , _SCREAMING_SNAKE_CASE = Body(__lowerCamelCase , embed=__lowerCamelCase ) , _SCREAMING_SNAKE_CASE = Body(__lowerCamelCase , embed=__lowerCamelCase ) )-> Any:
try:
lowerCamelCase_ =self._pipeline.tokenizer.tokenize(__lowerCamelCase )
if return_ids:
lowerCamelCase_ =self._pipeline.tokenizer.convert_tokens_to_ids(__lowerCamelCase )
return ServeTokenizeResult(tokens=__lowerCamelCase , tokens_ids=__lowerCamelCase )
else:
return ServeTokenizeResult(tokens=__lowerCamelCase )
except Exception as e:
raise HTTPException(status_code=500 , detail={"""model""": """""", """error""": str(__lowerCamelCase )} )
def _snake_case ( self , _SCREAMING_SNAKE_CASE = Body(__lowerCamelCase , embed=__lowerCamelCase ) , _SCREAMING_SNAKE_CASE = Body(__lowerCamelCase , embed=__lowerCamelCase ) , _SCREAMING_SNAKE_CASE = Body(__lowerCamelCase , embed=__lowerCamelCase ) , )-> List[str]:
try:
lowerCamelCase_ =self._pipeline.tokenizer.decode(__lowerCamelCase , __lowerCamelCase , __lowerCamelCase )
return ServeDeTokenizeResult(model="""""" , text=__lowerCamelCase )
except Exception as e:
raise HTTPException(status_code=500 , detail={"""model""": """""", """error""": str(__lowerCamelCase )} )
async def _snake_case ( self , _SCREAMING_SNAKE_CASE=Body(__lowerCamelCase , embed=__lowerCamelCase ) )-> Optional[Any]:
if len(__lowerCamelCase ) == 0:
return ServeForwardResult(output=[] , attention=[] )
try:
# Forward through the model
lowerCamelCase_ =self._pipeline(__lowerCamelCase )
return ServeForwardResult(output=__lowerCamelCase )
except Exception as e:
raise HTTPException(500 , {"""error""": str(__lowerCamelCase )} )
| 701
|
from collections import OrderedDict
from typing import Mapping
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
__A : int = {
'albert-base-v1': 'https://huggingface.co/albert-base-v1/resolve/main/config.json',
'albert-large-v1': 'https://huggingface.co/albert-large-v1/resolve/main/config.json',
'albert-xlarge-v1': 'https://huggingface.co/albert-xlarge-v1/resolve/main/config.json',
'albert-xxlarge-v1': 'https://huggingface.co/albert-xxlarge-v1/resolve/main/config.json',
'albert-base-v2': 'https://huggingface.co/albert-base-v2/resolve/main/config.json',
'albert-large-v2': 'https://huggingface.co/albert-large-v2/resolve/main/config.json',
'albert-xlarge-v2': 'https://huggingface.co/albert-xlarge-v2/resolve/main/config.json',
'albert-xxlarge-v2': 'https://huggingface.co/albert-xxlarge-v2/resolve/main/config.json',
}
class _SCREAMING_SNAKE_CASE ( lowerCAmelCase__):
_UpperCamelCase:Any = "albert"
def __init__( self , _SCREAMING_SNAKE_CASE=3_0000 , _SCREAMING_SNAKE_CASE=128 , _SCREAMING_SNAKE_CASE=4096 , _SCREAMING_SNAKE_CASE=12 , _SCREAMING_SNAKE_CASE=1 , _SCREAMING_SNAKE_CASE=64 , _SCREAMING_SNAKE_CASE=1_6384 , _SCREAMING_SNAKE_CASE=1 , _SCREAMING_SNAKE_CASE="gelu_new" , _SCREAMING_SNAKE_CASE=0 , _SCREAMING_SNAKE_CASE=0 , _SCREAMING_SNAKE_CASE=512 , _SCREAMING_SNAKE_CASE=2 , _SCREAMING_SNAKE_CASE=0.0_2 , _SCREAMING_SNAKE_CASE=1E-12 , _SCREAMING_SNAKE_CASE=0.1 , _SCREAMING_SNAKE_CASE="absolute" , _SCREAMING_SNAKE_CASE=0 , _SCREAMING_SNAKE_CASE=2 , _SCREAMING_SNAKE_CASE=3 , **_SCREAMING_SNAKE_CASE , )-> Optional[int]:
super().__init__(pad_token_id=_SCREAMING_SNAKE_CASE , bos_token_id=_SCREAMING_SNAKE_CASE , eos_token_id=_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )
lowerCamelCase_ =vocab_size
lowerCamelCase_ =embedding_size
lowerCamelCase_ =hidden_size
lowerCamelCase_ =num_hidden_layers
lowerCamelCase_ =num_hidden_groups
lowerCamelCase_ =num_attention_heads
lowerCamelCase_ =inner_group_num
lowerCamelCase_ =hidden_act
lowerCamelCase_ =intermediate_size
lowerCamelCase_ =hidden_dropout_prob
lowerCamelCase_ =attention_probs_dropout_prob
lowerCamelCase_ =max_position_embeddings
lowerCamelCase_ =type_vocab_size
lowerCamelCase_ =initializer_range
lowerCamelCase_ =layer_norm_eps
lowerCamelCase_ =classifier_dropout_prob
lowerCamelCase_ =position_embedding_type
class _SCREAMING_SNAKE_CASE ( lowerCAmelCase__):
@property
def _snake_case ( self )-> Mapping[str, Mapping[int, str]]:
if self.task == "multiple-choice":
lowerCamelCase_ ={0: """batch""", 1: """choice""", 2: """sequence"""}
else:
lowerCamelCase_ ={0: """batch""", 1: """sequence"""}
return OrderedDict(
[
("""input_ids""", dynamic_axis),
("""attention_mask""", dynamic_axis),
("""token_type_ids""", dynamic_axis),
] )
| 75
| 0
|
"""simple docstring"""
import unittest
from transformers import is_torch_available, is_vision_available
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
if is_torch_available():
import torch
from transformers import AutoModelForImageClassification
if is_vision_available():
from transformers import AutoImageProcessor
@require_torch
@require_vision
class lowerCAmelCase__ ( unittest.TestCase ):
'''simple docstring'''
@slow
def _lowerCAmelCase ( self : str ) -> Any:
"""simple docstring"""
SCREAMING_SNAKE_CASE : int = AutoImageProcessor.from_pretrained('microsoft/dit-base-finetuned-rvlcdip' )
SCREAMING_SNAKE_CASE : List[Any] = AutoModelForImageClassification.from_pretrained('microsoft/dit-base-finetuned-rvlcdip' )
model.to(_SCREAMING_SNAKE_CASE )
from datasets import load_dataset
SCREAMING_SNAKE_CASE : List[str] = load_dataset('nielsr/rvlcdip-demo' )
SCREAMING_SNAKE_CASE : str = dataset['train'][0]['image'].convert('RGB' )
SCREAMING_SNAKE_CASE : Tuple = image_processor(_SCREAMING_SNAKE_CASE , return_tensors='pt' ).to(_SCREAMING_SNAKE_CASE )
# forward pass
with torch.no_grad():
SCREAMING_SNAKE_CASE : List[Any] = model(**_SCREAMING_SNAKE_CASE )
SCREAMING_SNAKE_CASE : Union[str, Any] = outputs.logits
SCREAMING_SNAKE_CASE : int = torch.Size((1, 16) )
self.assertEqual(logits.shape , _SCREAMING_SNAKE_CASE )
SCREAMING_SNAKE_CASE : List[Any] = torch.tensor(
[-0.4_1_5_8, -0.4_0_9_2, -0.4_3_4_7] , device=_SCREAMING_SNAKE_CASE , dtype=torch.float , )
self.assertTrue(torch.allclose(logits[0, :3] , _SCREAMING_SNAKE_CASE , atol=1E-4 ) )
| 265
|
"""simple docstring"""
import io
import json
import unittest
from parameterized import parameterized
from transformers import FSMTForConditionalGeneration, FSMTTokenizer
from transformers.testing_utils import get_tests_dir, require_torch, slow, torch_device
from utils import calculate_bleu
SCREAMING_SNAKE_CASE__ = get_tests_dir() + "/test_data/fsmt/fsmt_val_data.json"
with io.open(filename, "r", encoding="utf-8") as f:
SCREAMING_SNAKE_CASE__ = json.load(f)
@require_torch
class lowercase ( unittest.TestCase ):
def _snake_case ( self , lowercase ) -> Tuple:
return FSMTTokenizer.from_pretrained(lowercase )
def _snake_case ( self , lowercase ) -> Dict:
lowerCAmelCase = FSMTForConditionalGeneration.from_pretrained(lowercase ).to(lowercase )
if torch_device == "cuda":
model.half()
return model
@parameterized.expand(
[
["""en-ru""", 26.0],
["""ru-en""", 22.0],
["""en-de""", 22.0],
["""de-en""", 29.0],
] )
@slow
def _snake_case ( self , lowercase , lowercase ) -> Dict:
# note: this test is not testing the best performance since it only evals a small batch
# but it should be enough to detect a regression in the output quality
lowerCAmelCase = f'facebook/wmt19-{pair}'
lowerCAmelCase = self.get_tokenizer(lowercase )
lowerCAmelCase = self.get_model(lowercase )
lowerCAmelCase = bleu_data[pair]["""src"""]
lowerCAmelCase = bleu_data[pair]["""tgt"""]
lowerCAmelCase = tokenizer(lowercase , return_tensors="""pt""" , truncation=lowercase , padding="""longest""" ).to(lowercase )
lowerCAmelCase = model.generate(
input_ids=batch.input_ids , num_beams=8 , )
lowerCAmelCase = tokenizer.batch_decode(
lowercase , skip_special_tokens=lowercase , clean_up_tokenization_spaces=lowercase )
lowerCAmelCase = calculate_bleu(lowercase , lowercase )
print(lowercase )
self.assertGreaterEqual(scores["""bleu"""] , lowercase )
| 532
| 0
|
"""simple docstring"""
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_torch_available,
)
lowerCAmelCase_: List[str] = {
"configuration_mega": ["MEGA_PRETRAINED_CONFIG_ARCHIVE_MAP", "MegaConfig", "MegaOnnxConfig"],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
lowerCAmelCase_: Optional[int] = [
"MEGA_PRETRAINED_MODEL_ARCHIVE_LIST",
"MegaForCausalLM",
"MegaForMaskedLM",
"MegaForMultipleChoice",
"MegaForQuestionAnswering",
"MegaForSequenceClassification",
"MegaForTokenClassification",
"MegaModel",
"MegaPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_mega import MEGA_PRETRAINED_CONFIG_ARCHIVE_MAP, MegaConfig, MegaOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_mega import (
MEGA_PRETRAINED_MODEL_ARCHIVE_LIST,
MegaForCausalLM,
MegaForMaskedLM,
MegaForMultipleChoice,
MegaForQuestionAnswering,
MegaForSequenceClassification,
MegaForTokenClassification,
MegaModel,
MegaPreTrainedModel,
)
else:
import sys
lowerCAmelCase_: Optional[Any] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| 668
|
"""simple docstring"""
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
lowerCAmelCase_: Union[str, Any] = logging.get_logger(__name__)
class a__ ( _a ):
snake_case_ = ["audio_values", "audio_mask"]
def __init__( self, _UpperCAmelCase=2048, _UpperCAmelCase=1, _UpperCAmelCase=[16, 16], _UpperCAmelCase=128, _UpperCAmelCase=4_4100, _UpperCAmelCase=86, _UpperCAmelCase=2048, _UpperCAmelCase=0.0, **_UpperCAmelCase, ):
'''simple docstring'''
super().__init__(
feature_size=_UpperCAmelCase, sampling_rate=_UpperCAmelCase, padding_value=_UpperCAmelCase, **_UpperCAmelCase, )
lowercase__ = spectrogram_length
lowercase__ = num_channels
lowercase__ = patch_size
lowercase__ = feature_size // self.patch_size[1]
lowercase__ = n_fft
lowercase__ = sampling_rate // hop_length_to_sampling_rate
lowercase__ = sampling_rate
lowercase__ = padding_value
lowercase__ = 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 snake_case__ ( self, _UpperCAmelCase ):
'''simple docstring'''
lowercase__ = 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, )
lowercase__ = log_spec[:, :-1]
lowercase__ = log_spec - 20.0
lowercase__ = 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." )
lowercase__ = 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}''' )
lowercase__ = is_batched_numpy or (
isinstance(_UpperCAmelCase, (list, tuple) ) and (isinstance(raw_speech[0], (np.ndarray, tuple, list) ))
)
if is_batched:
lowercase__ = [np.asarray([speech], dtype=np.floataa ).T for speech in raw_speech]
elif not is_batched and not isinstance(_UpperCAmelCase, np.ndarray ):
lowercase__ = np.asarray(_UpperCAmelCase, dtype=np.floataa )
elif isinstance(_UpperCAmelCase, np.ndarray ) and raw_speech.dtype is np.dtype(np.floataa ):
lowercase__ = raw_speech.astype(np.floataa )
# always return batch
if not is_batched:
lowercase__ = [np.asarray([raw_speech] ).T]
# Convert audio signals to log mel spectrograms, truncate by time axis
lowercase__ = [
self._np_extract_fbank_features(waveform.squeeze() ).T[: self.spectrogram_length] for waveform in raw_speech
]
if isinstance(audio_features[0], _UpperCAmelCase ):
lowercase__ = [np.asarray(_UpperCAmelCase, dtype=np.floataa ) for feature in audio_features]
# Create audio attention mask
lowercase__ = 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:
lowercase__ = [
(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
]
lowercase__ = np.array(_UpperCAmelCase ).astype(np.floataa )
# convert into correct format for padding
lowercase__ = max_patch_len // self.freq_len * self.patch_size[0] # The maximum audio size in a batch
lowercase__ = np.ones([len(_UpperCAmelCase ), 1, max_time_len, self.feature_size] ).astype(np.floataa )
lowercase__ = padded_audio_features * self.padding_value
for i in range(len(_UpperCAmelCase ) ):
lowercase__ = audio_features[i]
lowercase__ = feature
# return as BatchFeature
if return_attention_mask:
lowercase__ = {"audio_values": padded_audio_features, "audio_mask": audio_mask}
else:
lowercase__ = {"audio_values": padded_audio_features}
lowercase__ = BatchFeature(data=_UpperCAmelCase, tensor_type=_UpperCAmelCase )
return encoded_inputs
| 668
| 1
|
"""simple docstring"""
import argparse
import os.path as osp
import re
import torch
from safetensors.torch import load_file, save_file
# =================#
# UNet Conversion #
# =================#
UpperCamelCase = [
# (stable-diffusion, HF Diffusers)
('time_embed.0.weight', 'time_embedding.linear_1.weight'),
('time_embed.0.bias', 'time_embedding.linear_1.bias'),
('time_embed.2.weight', 'time_embedding.linear_2.weight'),
('time_embed.2.bias', 'time_embedding.linear_2.bias'),
('input_blocks.0.0.weight', 'conv_in.weight'),
('input_blocks.0.0.bias', 'conv_in.bias'),
('out.0.weight', 'conv_norm_out.weight'),
('out.0.bias', 'conv_norm_out.bias'),
('out.2.weight', 'conv_out.weight'),
('out.2.bias', 'conv_out.bias'),
]
UpperCamelCase = [
# (stable-diffusion, HF Diffusers)
('in_layers.0', 'norm1'),
('in_layers.2', 'conv1'),
('out_layers.0', 'norm2'),
('out_layers.3', 'conv2'),
('emb_layers.1', 'time_emb_proj'),
('skip_connection', 'conv_shortcut'),
]
UpperCamelCase = []
# hardcoded number of downblocks and resnets/attentions...
# would need smarter logic for other networks.
for i in range(4):
# loop over downblocks/upblocks
for j in range(2):
# loop over resnets/attentions for downblocks
UpperCamelCase = F'down_blocks.{i}.resnets.{j}.'
UpperCamelCase = F'input_blocks.{3*i + j + 1}.0.'
unet_conversion_map_layer.append((sd_down_res_prefix, hf_down_res_prefix))
if i < 3:
# no attention layers in down_blocks.3
UpperCamelCase = F'down_blocks.{i}.attentions.{j}.'
UpperCamelCase = F'input_blocks.{3*i + j + 1}.1.'
unet_conversion_map_layer.append((sd_down_atn_prefix, hf_down_atn_prefix))
for j in range(3):
# loop over resnets/attentions for upblocks
UpperCamelCase = F'up_blocks.{i}.resnets.{j}.'
UpperCamelCase = F'output_blocks.{3*i + j}.0.'
unet_conversion_map_layer.append((sd_up_res_prefix, hf_up_res_prefix))
if i > 0:
# no attention layers in up_blocks.0
UpperCamelCase = F'up_blocks.{i}.attentions.{j}.'
UpperCamelCase = F'output_blocks.{3*i + j}.1.'
unet_conversion_map_layer.append((sd_up_atn_prefix, hf_up_atn_prefix))
if i < 3:
# no downsample in down_blocks.3
UpperCamelCase = F'down_blocks.{i}.downsamplers.0.conv.'
UpperCamelCase = F'input_blocks.{3*(i+1)}.0.op.'
unet_conversion_map_layer.append((sd_downsample_prefix, hf_downsample_prefix))
# no upsample in up_blocks.3
UpperCamelCase = F'up_blocks.{i}.upsamplers.0.'
UpperCamelCase = F'output_blocks.{3*i + 2}.{1 if i == 0 else 2}.'
unet_conversion_map_layer.append((sd_upsample_prefix, hf_upsample_prefix))
UpperCamelCase = 'mid_block.attentions.0.'
UpperCamelCase = 'middle_block.1.'
unet_conversion_map_layer.append((sd_mid_atn_prefix, hf_mid_atn_prefix))
for j in range(2):
UpperCamelCase = F'mid_block.resnets.{j}.'
UpperCamelCase = F'middle_block.{2*j}.'
unet_conversion_map_layer.append((sd_mid_res_prefix, hf_mid_res_prefix))
def lowerCAmelCase_ (_SCREAMING_SNAKE_CASE :Any ) -> List[str]:
# buyer beware: this is a *brittle* function,
# and correct output requires that all of these pieces interact in
# the exact order in which I have arranged them.
a_ : Tuple = {k: k for k in unet_state_dict.keys()}
for sd_name, hf_name in unet_conversion_map:
a_ : Tuple = sd_name
for k, v in mapping.items():
if "resnets" in k:
for sd_part, hf_part in unet_conversion_map_resnet:
a_ : List[Any] = v.replace(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
a_ : Tuple = v
for k, v in mapping.items():
for sd_part, hf_part in unet_conversion_map_layer:
a_ : List[Any] = v.replace(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
a_ : str = v
a_ : List[str] = {v: unet_state_dict[k] for k, v in mapping.items()}
return new_state_dict
# ================#
# VAE Conversion #
# ================#
UpperCamelCase = [
# (stable-diffusion, HF Diffusers)
('nin_shortcut', 'conv_shortcut'),
('norm_out', 'conv_norm_out'),
('mid.attn_1.', 'mid_block.attentions.0.'),
]
for i in range(4):
# down_blocks have two resnets
for j in range(2):
UpperCamelCase = F'encoder.down_blocks.{i}.resnets.{j}.'
UpperCamelCase = F'encoder.down.{i}.block.{j}.'
vae_conversion_map.append((sd_down_prefix, hf_down_prefix))
if i < 3:
UpperCamelCase = F'down_blocks.{i}.downsamplers.0.'
UpperCamelCase = F'down.{i}.downsample.'
vae_conversion_map.append((sd_downsample_prefix, hf_downsample_prefix))
UpperCamelCase = F'up_blocks.{i}.upsamplers.0.'
UpperCamelCase = F'up.{3-i}.upsample.'
vae_conversion_map.append((sd_upsample_prefix, hf_upsample_prefix))
# up_blocks have three resnets
# also, up blocks in hf are numbered in reverse from sd
for j in range(3):
UpperCamelCase = F'decoder.up_blocks.{i}.resnets.{j}.'
UpperCamelCase = F'decoder.up.{3-i}.block.{j}.'
vae_conversion_map.append((sd_up_prefix, hf_up_prefix))
# this part accounts for mid blocks in both the encoder and the decoder
for i in range(2):
UpperCamelCase = F'mid_block.resnets.{i}.'
UpperCamelCase = F'mid.block_{i+1}.'
vae_conversion_map.append((sd_mid_res_prefix, hf_mid_res_prefix))
UpperCamelCase = [
# (stable-diffusion, HF Diffusers)
('norm.', 'group_norm.'),
('q.', 'query.'),
('k.', 'key.'),
('v.', 'value.'),
('proj_out.', 'proj_attn.'),
]
def lowerCAmelCase_ (_SCREAMING_SNAKE_CASE :Tuple ) -> Dict:
# convert HF linear weights to SD conv2d weights
return w.reshape(*w.shape , 1 , 1 )
def lowerCAmelCase_ (_SCREAMING_SNAKE_CASE :int ) -> Optional[Any]:
a_ : Optional[int] = {k: k for k in vae_state_dict.keys()}
for k, v in mapping.items():
for sd_part, hf_part in vae_conversion_map:
a_ : List[Any] = v.replace(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
a_ : List[str] = v
for k, v in mapping.items():
if "attentions" in k:
for sd_part, hf_part in vae_conversion_map_attn:
a_ : Optional[Any] = v.replace(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
a_ : str = v
a_ : str = {v: vae_state_dict[k] for k, v in mapping.items()}
a_ : Union[str, Any] = ["q", "k", "v", "proj_out"]
for k, v in new_state_dict.items():
for weight_name in weights_to_convert:
if F'''mid.attn_1.{weight_name}.weight''' in k:
print(F'''Reshaping {k} for SD format''' )
a_ : Tuple = reshape_weight_for_sd(_SCREAMING_SNAKE_CASE )
return new_state_dict
# =========================#
# Text Encoder Conversion #
# =========================#
UpperCamelCase = [
# (stable-diffusion, HF Diffusers)
('resblocks.', 'text_model.encoder.layers.'),
('ln_1', 'layer_norm1'),
('ln_2', 'layer_norm2'),
('.c_fc.', '.fc1.'),
('.c_proj.', '.fc2.'),
('.attn', '.self_attn'),
('ln_final.', 'transformer.text_model.final_layer_norm.'),
('token_embedding.weight', 'transformer.text_model.embeddings.token_embedding.weight'),
('positional_embedding', 'transformer.text_model.embeddings.position_embedding.weight'),
]
UpperCamelCase = {re.escape(x[1]): x[0] for x in textenc_conversion_lst}
UpperCamelCase = re.compile('|'.join(protected.keys()))
# Ordering is from https://github.com/pytorch/pytorch/blob/master/test/cpp/api/modules.cpp
UpperCamelCase = {'q': 0, 'k': 1, 'v': 2}
def lowerCAmelCase_ (_SCREAMING_SNAKE_CASE :List[str] ) -> List[str]:
a_ : List[Any] = {}
a_ : Tuple = {}
a_ : Tuple = {}
for k, v in text_enc_dict.items():
if (
k.endswith(".self_attn.q_proj.weight" )
or k.endswith(".self_attn.k_proj.weight" )
or k.endswith(".self_attn.v_proj.weight" )
):
a_ : List[str] = k[: -len(".q_proj.weight" )]
a_ : Optional[int] = k[-len("q_proj.weight" )]
if k_pre not in capture_qkv_weight:
a_ : Tuple = [None, None, None]
a_ : Dict = v
continue
if (
k.endswith(".self_attn.q_proj.bias" )
or k.endswith(".self_attn.k_proj.bias" )
or k.endswith(".self_attn.v_proj.bias" )
):
a_ : Tuple = k[: -len(".q_proj.bias" )]
a_ : int = k[-len("q_proj.bias" )]
if k_pre not in capture_qkv_bias:
a_ : int = [None, None, None]
a_ : Optional[Any] = v
continue
a_ : str = textenc_pattern.sub(lambda _SCREAMING_SNAKE_CASE : protected[re.escape(m.group(0 ) )] , _SCREAMING_SNAKE_CASE )
a_ : int = v
for k_pre, tensors in capture_qkv_weight.items():
if None in tensors:
raise Exception("CORRUPTED MODEL: one of the q-k-v values for the text encoder was missing" )
a_ : List[str] = textenc_pattern.sub(lambda _SCREAMING_SNAKE_CASE : protected[re.escape(m.group(0 ) )] , _SCREAMING_SNAKE_CASE )
a_ : List[Any] = torch.cat(_SCREAMING_SNAKE_CASE )
for k_pre, tensors in capture_qkv_bias.items():
if None in tensors:
raise Exception("CORRUPTED MODEL: one of the q-k-v values for the text encoder was missing" )
a_ : int = textenc_pattern.sub(lambda _SCREAMING_SNAKE_CASE : protected[re.escape(m.group(0 ) )] , _SCREAMING_SNAKE_CASE )
a_ : Union[str, Any] = torch.cat(_SCREAMING_SNAKE_CASE )
return new_state_dict
def lowerCAmelCase_ (_SCREAMING_SNAKE_CASE :Optional[Any] ) -> List[str]:
return text_enc_dict
if __name__ == "__main__":
UpperCamelCase = argparse.ArgumentParser()
parser.add_argument('--model_path', default=None, type=str, required=True, help='Path to the model to convert.')
parser.add_argument('--checkpoint_path', default=None, type=str, required=True, help='Path to the output model.')
parser.add_argument('--half', action='store_true', help='Save weights in half precision.')
parser.add_argument(
'--use_safetensors', action='store_true', help='Save weights use safetensors, default is ckpt.'
)
UpperCamelCase = parser.parse_args()
assert args.model_path is not None, "Must provide a model path!"
assert args.checkpoint_path is not None, "Must provide a checkpoint path!"
# Path for safetensors
UpperCamelCase = osp.join(args.model_path, 'unet', 'diffusion_pytorch_model.safetensors')
UpperCamelCase = osp.join(args.model_path, 'vae', 'diffusion_pytorch_model.safetensors')
UpperCamelCase = osp.join(args.model_path, 'text_encoder', 'model.safetensors')
# Load models from safetensors if it exists, if it doesn't pytorch
if osp.exists(unet_path):
UpperCamelCase = load_file(unet_path, device='cpu')
else:
UpperCamelCase = osp.join(args.model_path, 'unet', 'diffusion_pytorch_model.bin')
UpperCamelCase = torch.load(unet_path, map_location='cpu')
if osp.exists(vae_path):
UpperCamelCase = load_file(vae_path, device='cpu')
else:
UpperCamelCase = osp.join(args.model_path, 'vae', 'diffusion_pytorch_model.bin')
UpperCamelCase = torch.load(vae_path, map_location='cpu')
if osp.exists(text_enc_path):
UpperCamelCase = load_file(text_enc_path, device='cpu')
else:
UpperCamelCase = osp.join(args.model_path, 'text_encoder', 'pytorch_model.bin')
UpperCamelCase = torch.load(text_enc_path, map_location='cpu')
# Convert the UNet model
UpperCamelCase = convert_unet_state_dict(unet_state_dict)
UpperCamelCase = {'model.diffusion_model.' + k: v for k, v in unet_state_dict.items()}
# Convert the VAE model
UpperCamelCase = convert_vae_state_dict(vae_state_dict)
UpperCamelCase = {'first_stage_model.' + k: v for k, v in vae_state_dict.items()}
# Easiest way to identify v2.0 model seems to be that the text encoder (OpenCLIP) is deeper
UpperCamelCase = 'text_model.encoder.layers.22.layer_norm2.bias' in text_enc_dict
if is_vaa_model:
# Need to add the tag 'transformer' in advance so we can knock it out from the final layer-norm
UpperCamelCase = {'transformer.' + k: v for k, v in text_enc_dict.items()}
UpperCamelCase = convert_text_enc_state_dict_vaa(text_enc_dict)
UpperCamelCase = {'cond_stage_model.model.' + k: v for k, v in text_enc_dict.items()}
else:
UpperCamelCase = convert_text_enc_state_dict(text_enc_dict)
UpperCamelCase = {'cond_stage_model.transformer.' + k: v for k, v in text_enc_dict.items()}
# Put together new checkpoint
UpperCamelCase = {**unet_state_dict, **vae_state_dict, **text_enc_dict}
if args.half:
UpperCamelCase = {k: v.half() for k, v in state_dict.items()}
if args.use_safetensors:
save_file(state_dict, args.checkpoint_path)
else:
UpperCamelCase = {'state_dict': state_dict}
torch.save(state_dict, args.checkpoint_path)
| 473
|
"""simple docstring"""
from __future__ import annotations
from math import pi, sqrt
def lowerCAmelCase_ (_SCREAMING_SNAKE_CASE :float , _SCREAMING_SNAKE_CASE :float ) -> tuple:
if inductance <= 0:
raise ValueError("Inductance cannot be 0 or negative" )
elif capacitance <= 0:
raise ValueError("Capacitance cannot be 0 or negative" )
else:
return (
"Resonant frequency",
float(1 / (2 * pi * (sqrt(inductance * capacitance ))) ),
)
if __name__ == "__main__":
import doctest
doctest.testmod()
| 473
| 1
|
# This is the module that test_patching.py uses to test patch_submodule()
import os # noqa: this is just for tests
import os as renamed_os # noqa: this is just for tests
from os import path # noqa: this is just for tests
from os import path as renamed_path # noqa: this is just for tests
from os.path import join # noqa: this is just for tests
from os.path import join as renamed_join # noqa: this is just for tests
_lowerCamelCase : Dict = open # noqa: we just need to have a builtin inside this module to test it properly
| 407
|
from typing import Optional
from urllib.parse import quote
import huggingface_hub as hfh
from packaging import version
def _lowerCAmelCase ( __magic_name__ :str , __magic_name__ :str , __magic_name__ :Optional[str] = None ):
if version.parse(hfh.__version__ ).release < version.parse('''0.11.0''' ).release:
# old versions of hfh don't url-encode the file path
UpperCAmelCase_ = quote(__magic_name__ )
return hfh.hf_hub_url(__magic_name__ , __magic_name__ , repo_type='''dataset''' , revision=__magic_name__ )
| 407
| 1
|
'''simple docstring'''
import math
from datetime import datetime, timedelta
def A_ ( _lowerCAmelCase : int ):
"""simple docstring"""
_lowerCamelCase : List[str] = year % 19
_lowerCamelCase : Dict = year % 4
_lowerCamelCase : str = year % 7
_lowerCamelCase : Any = math.floor(year / 100 )
_lowerCamelCase : Union[str, Any] = math.floor((13 + 8 * leap_day_inhibits) / 25 )
_lowerCamelCase : str = leap_day_inhibits / 4
_lowerCamelCase : Tuple = (
15 - lunar_orbit_correction + leap_day_inhibits - leap_day_reinstall_number
) % 30
_lowerCamelCase : int = (4 + leap_day_inhibits - leap_day_reinstall_number) % 7
# days to be added to March 21
_lowerCamelCase : str = (19 * metonic_cycle + secular_moon_shift) % 30
# PHM -> Paschal Full Moon
_lowerCamelCase : Any = (
2 * julian_leap_year
+ 4 * non_leap_year
+ 6 * days_to_add
+ century_starting_point
) % 7
if days_to_add == 29 and days_from_phm_to_sunday == 6:
return datetime(_lowerCAmelCase , 4 , 19 )
elif days_to_add == 28 and days_from_phm_to_sunday == 6:
return datetime(_lowerCAmelCase , 4 , 18 )
else:
return datetime(_lowerCAmelCase , 3 , 22 ) + timedelta(
days=int(days_to_add + days_from_phm_to_sunday ) )
if __name__ == "__main__":
for year in (1994, 2000, 2010, 2021, 2023):
UpperCAmelCase_ : Any = 'will be' if year > datetime.now().year else 'was'
print(f'''Easter in {year} {tense} {gauss_easter(year)}''')
| 44
|
'''simple docstring'''
from typing import List, Optional, Union
from ...configuration_utils import PretrainedConfig
from ...utils import logging
lowerCAmelCase_ : Union[str, Any] = logging.get_logger(__name__)
lowerCAmelCase_ : Optional[int] = {
"huggingface/informer-tourism-monthly": (
"https://huggingface.co/huggingface/informer-tourism-monthly/resolve/main/config.json"
),
# See all Informer models at https://huggingface.co/models?filter=informer
}
class UpperCamelCase__ ( __lowerCAmelCase ):
lowerCAmelCase__ : List[str] = "informer"
lowerCAmelCase__ : List[str] = {
"hidden_size": "d_model",
"num_attention_heads": "encoder_attention_heads",
"num_hidden_layers": "encoder_layers",
}
def __init__( self : Optional[Any] , lowerCamelCase : Optional[int] = None , lowerCamelCase : Optional[int] = None , lowerCamelCase : str = "student_t" , lowerCamelCase : str = "nll" , lowerCamelCase : int = 1 , lowerCamelCase : List[int] = None , lowerCamelCase : Optional[Union[str, bool]] = "mean" , lowerCamelCase : int = 0 , lowerCamelCase : int = 0 , lowerCamelCase : int = 0 , lowerCamelCase : int = 0 , lowerCamelCase : Optional[List[int]] = None , lowerCamelCase : Optional[List[int]] = None , lowerCamelCase : int = 6_4 , lowerCamelCase : int = 3_2 , lowerCamelCase : int = 3_2 , lowerCamelCase : int = 2 , lowerCamelCase : int = 2 , lowerCamelCase : int = 2 , lowerCamelCase : int = 2 , lowerCamelCase : bool = True , lowerCamelCase : str = "gelu" , lowerCamelCase : float = 0.05 , lowerCamelCase : float = 0.1 , lowerCamelCase : float = 0.1 , lowerCamelCase : float = 0.1 , lowerCamelCase : float = 0.1 , lowerCamelCase : int = 1_0_0 , lowerCamelCase : float = 0.02 , lowerCamelCase : Union[str, Any]=True , lowerCamelCase : str = "prob" , lowerCamelCase : int = 5 , lowerCamelCase : bool = True , **lowerCamelCase : Dict , ):
'''simple docstring'''
# time series specific configuration
a__ = prediction_length
a__ = context_length or prediction_length
a__ = distribution_output
a__ = loss
a__ = input_size
a__ = num_time_features
a__ = lags_sequence if lags_sequence is not None else [1, 2, 3, 4, 5, 6, 7]
a__ = scaling
a__ = num_dynamic_real_features
a__ = num_static_real_features
a__ = num_static_categorical_features
# set cardinality
if cardinality and num_static_categorical_features > 0:
if len(lowerCamelCase ) != num_static_categorical_features:
raise ValueError(
"The cardinality should be a list of the same length as `num_static_categorical_features`" )
a__ = cardinality
else:
a__ = [0]
# set embedding_dimension
if embedding_dimension and num_static_categorical_features > 0:
if len(lowerCamelCase ) != num_static_categorical_features:
raise ValueError(
"The embedding dimension should be a list of the same length as `num_static_categorical_features`" )
a__ = embedding_dimension
else:
a__ = [min(5_0 , (cat + 1) // 2 ) for cat in self.cardinality]
a__ = num_parallel_samples
# Transformer architecture configuration
a__ = input_size * len(self.lags_sequence ) + self._number_of_features
a__ = d_model
a__ = encoder_attention_heads
a__ = decoder_attention_heads
a__ = encoder_ffn_dim
a__ = decoder_ffn_dim
a__ = encoder_layers
a__ = decoder_layers
a__ = dropout
a__ = attention_dropout
a__ = activation_dropout
a__ = encoder_layerdrop
a__ = decoder_layerdrop
a__ = activation_function
a__ = init_std
a__ = use_cache
# Informer
a__ = attention_type
a__ = sampling_factor
a__ = distil
super().__init__(is_encoder_decoder=lowerCamelCase , **lowerCamelCase )
@property
def __a ( self : Optional[int] ):
'''simple docstring'''
return (
sum(self.embedding_dimension )
+ self.num_dynamic_real_features
+ self.num_time_features
+ self.num_static_real_features
+ self.input_size * 2 # the log1p(abs(loc)) and log(scale) features
)
| 489
| 0
|
"""simple docstring"""
def A_ ( _lowercase ):
'''simple docstring'''
snake_case_ :list[list[int]] = [[0 for _ in range(_lowercase )] for _ in range(m + 1 )]
for i in range(m + 1 ):
snake_case_ :Optional[int] = 1
for n in range(m + 1 ):
for k in range(1, _lowercase ):
memo[n][k] += memo[n][k - 1]
if n - k > 0:
memo[n][k] += memo[n - k - 1][k]
return memo[m][m - 1]
if __name__ == "__main__":
import sys
if len(sys.argv) == 1:
try:
__a = int(input("Enter a number: ").strip())
print(partition(n))
except ValueError:
print("Please enter a number.")
else:
try:
__a = int(sys.argv[1])
print(partition(n))
except ValueError:
print("Please pass a number.")
| 711
|
"""simple docstring"""
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
__a = {"configuration_ibert": ["IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP", "IBertConfig", "IBertOnnxConfig"]}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
__a = [
"IBERT_PRETRAINED_MODEL_ARCHIVE_LIST",
"IBertForMaskedLM",
"IBertForMultipleChoice",
"IBertForQuestionAnswering",
"IBertForSequenceClassification",
"IBertForTokenClassification",
"IBertModel",
"IBertPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_ibert import IBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, IBertConfig, IBertOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_ibert import (
IBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
IBertForMaskedLM,
IBertForMultipleChoice,
IBertForQuestionAnswering,
IBertForSequenceClassification,
IBertForTokenClassification,
IBertModel,
IBertPreTrainedModel,
)
else:
import sys
__a = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| 310
| 0
|
import itertools
from dataclasses import dataclass
from typing import Any, Callable, Dict, List, Optional, Union
import pandas as pd
import pyarrow as pa
import datasets
import datasets.config
from datasets.features.features import require_storage_cast
from datasets.table import table_cast
from datasets.utils.py_utils import Literal
__magic_name__ : List[Any] = datasets.utils.logging.get_logger(__name__)
__magic_name__ : List[Any] = ["""names""", """prefix"""]
__magic_name__ : Tuple = ["""warn_bad_lines""", """error_bad_lines""", """mangle_dupe_cols"""]
__magic_name__ : Any = ["""encoding_errors""", """on_bad_lines"""]
__magic_name__ : List[Any] = ["""date_format"""]
@dataclass
class SCREAMING_SNAKE_CASE__ (datasets.BuilderConfig ):
lowercase_ : str = ","
lowercase_ : Optional[str] = None
lowercase_ : Optional[Union[int, List[int], str]] = "infer"
lowercase_ : Optional[List[str]] = None
lowercase_ : Optional[List[str]] = None
lowercase_ : Optional[Union[int, str, List[int], List[str]]] = None
lowercase_ : Optional[Union[List[int], List[str]]] = None
lowercase_ : Optional[str] = None
lowercase_ : bool = True
lowercase_ : Optional[Literal["c", "python", "pyarrow"]] = None
lowercase_ : Dict[Union[int, str], Callable[[Any], Any]] = None
lowercase_ : Optional[list] = None
lowercase_ : Optional[list] = None
lowercase_ : bool = False
lowercase_ : Optional[Union[int, List[int]]] = None
lowercase_ : Optional[int] = None
lowercase_ : Optional[Union[str, List[str]]] = None
lowercase_ : bool = True
lowercase_ : bool = True
lowercase_ : bool = False
lowercase_ : bool = True
lowercase_ : Optional[str] = None
lowercase_ : str = "."
lowercase_ : Optional[str] = None
lowercase_ : str = '"'
lowercase_ : int = 0
lowercase_ : Optional[str] = None
lowercase_ : Optional[str] = None
lowercase_ : Optional[str] = None
lowercase_ : Optional[str] = None
lowercase_ : bool = True
lowercase_ : bool = True
lowercase_ : int = 0
lowercase_ : bool = True
lowercase_ : bool = False
lowercase_ : Optional[str] = None
lowercase_ : int = 10000
lowercase_ : Optional[datasets.Features] = None
lowercase_ : Optional[str] = "strict"
lowercase_ : Literal["error", "warn", "skip"] = "error"
lowercase_ : Optional[str] = None
def A__ ( self : List[Any] ):
"""simple docstring"""
if self.delimiter is not None:
lowerCAmelCase__ = self.delimiter
if self.column_names is not None:
lowerCAmelCase__ = self.column_names
@property
def A__ ( self : Any ):
"""simple docstring"""
lowerCAmelCase__ = {
"sep": self.sep,
"header": self.header,
"names": self.names,
"index_col": self.index_col,
"usecols": self.usecols,
"prefix": self.prefix,
"mangle_dupe_cols": self.mangle_dupe_cols,
"engine": self.engine,
"converters": self.converters,
"true_values": self.true_values,
"false_values": self.false_values,
"skipinitialspace": self.skipinitialspace,
"skiprows": self.skiprows,
"nrows": self.nrows,
"na_values": self.na_values,
"keep_default_na": self.keep_default_na,
"na_filter": self.na_filter,
"verbose": self.verbose,
"skip_blank_lines": self.skip_blank_lines,
"thousands": self.thousands,
"decimal": self.decimal,
"lineterminator": self.lineterminator,
"quotechar": self.quotechar,
"quoting": self.quoting,
"escapechar": self.escapechar,
"comment": self.comment,
"encoding": self.encoding,
"dialect": self.dialect,
"error_bad_lines": self.error_bad_lines,
"warn_bad_lines": self.warn_bad_lines,
"skipfooter": self.skipfooter,
"doublequote": self.doublequote,
"memory_map": self.memory_map,
"float_precision": self.float_precision,
"chunksize": self.chunksize,
"encoding_errors": self.encoding_errors,
"on_bad_lines": self.on_bad_lines,
"date_format": self.date_format,
}
# some kwargs must not be passed if they don't have a default value
# some others are deprecated and we can also not pass them if they are the default value
for pd_read_csv_parameter in _PANDAS_READ_CSV_NO_DEFAULT_PARAMETERS + _PANDAS_READ_CSV_DEPRECATED_PARAMETERS:
if pd_read_csv_kwargs[pd_read_csv_parameter] == getattr(CsvConfig() , __UpperCAmelCase ):
del pd_read_csv_kwargs[pd_read_csv_parameter]
# Remove 2.0 new arguments
if not (datasets.config.PANDAS_VERSION.major >= 2):
for pd_read_csv_parameter in _PANDAS_READ_CSV_NEW_2_0_0_PARAMETERS:
del pd_read_csv_kwargs[pd_read_csv_parameter]
# Remove 1.3 new arguments
if not (datasets.config.PANDAS_VERSION.major >= 1 and datasets.config.PANDAS_VERSION.minor >= 3):
for pd_read_csv_parameter in _PANDAS_READ_CSV_NEW_1_3_0_PARAMETERS:
del pd_read_csv_kwargs[pd_read_csv_parameter]
return pd_read_csv_kwargs
class SCREAMING_SNAKE_CASE__ (datasets.ArrowBasedBuilder ):
lowercase_ : Any = CsvConfig
def A__ ( self : str ):
"""simple docstring"""
return datasets.DatasetInfo(features=self.config.features )
def A__ ( self : Tuple , __lowerCamelCase : str ):
"""simple docstring"""
if not self.config.data_files:
raise ValueError(F"""At least one data file must be specified, but got data_files={self.config.data_files}""" )
lowerCAmelCase__ = dl_manager.download_and_extract(self.config.data_files )
if isinstance(__UpperCAmelCase , (str, list, tuple) ):
lowerCAmelCase__ = data_files
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
lowerCAmelCase__ = [files]
lowerCAmelCase__ = [dl_manager.iter_files(__UpperCAmelCase ) for file in files]
return [datasets.SplitGenerator(name=datasets.Split.TRAIN , gen_kwargs={'''files''': files} )]
lowerCAmelCase__ = []
for split_name, files in data_files.items():
if isinstance(__UpperCAmelCase , __UpperCAmelCase ):
lowerCAmelCase__ = [files]
lowerCAmelCase__ = [dl_manager.iter_files(__UpperCAmelCase ) for file in files]
splits.append(datasets.SplitGenerator(name=__UpperCAmelCase , gen_kwargs={'''files''': files} ) )
return splits
def A__ ( self : List[str] , __lowerCamelCase : Any ):
"""simple docstring"""
if self.config.features is not None:
lowerCAmelCase__ = self.config.features.arrow_schema
if all(not require_storage_cast(__UpperCAmelCase ) for feature in self.config.features.values() ):
# cheaper cast
lowerCAmelCase__ = pa.Table.from_arrays([pa_table[field.name] for field in schema] , schema=__UpperCAmelCase )
else:
# more expensive cast; allows str <-> int/float or str to Audio for example
lowerCAmelCase__ = table_cast(__UpperCAmelCase , __UpperCAmelCase )
return pa_table
def A__ ( self : Union[str, Any] , __lowerCamelCase : Dict ):
"""simple docstring"""
lowerCAmelCase__ = self.config.features.arrow_schema if self.config.features else None
# dtype allows reading an int column as str
lowerCAmelCase__ = (
{
name: dtype.to_pandas_dtype() if not require_storage_cast(__UpperCAmelCase ) else object
for name, dtype, feature in zip(schema.names , schema.types , self.config.features.values() )
}
if schema is not None
else None
)
for file_idx, file in enumerate(itertools.chain.from_iterable(__UpperCAmelCase ) ):
lowerCAmelCase__ = pd.read_csv(__UpperCAmelCase , iterator=__UpperCAmelCase , dtype=__UpperCAmelCase , **self.config.pd_read_csv_kwargs )
try:
for batch_idx, df in enumerate(__UpperCAmelCase ):
lowerCAmelCase__ = pa.Table.from_pandas(__UpperCAmelCase )
# Uncomment for debugging (will print the Arrow table size and elements)
# logger.warning(f"pa_table: {pa_table} num rows: {pa_table.num_rows}")
# logger.warning('\n'.join(str(pa_table.slice(i, 1).to_pydict()) for i in range(pa_table.num_rows)))
yield (file_idx, batch_idx), self._cast_table(__UpperCAmelCase )
except ValueError as e:
logger.error(F"""Failed to read file '{file}' with error {type(__UpperCAmelCase )}: {e}""" )
raise
| 615
|
def lowerCamelCase_ ( _lowercase ) -> int:
if not isinstance(_lowercase , _lowercase ):
raise TypeError("only integers accepted as input" )
else:
__A : int = str(abs(_lowercase ) )
__A : List[str] = [list(_lowercase ) for char in range(len(_lowercase ) )]
for index in range(len(_lowercase ) ):
num_transpositions[index].pop(_lowercase )
return max(
int("".join(list(_lowercase ) ) ) for transposition in num_transpositions )
if __name__ == "__main__":
__import__('doctest').testmod()
| 520
| 0
|
'''simple docstring'''
import json
import os
import tempfile
import unittest
import unittest.mock as mock
from pathlib import Path
from requests.exceptions import HTTPError
from transformers.utils import (
CONFIG_NAME,
FLAX_WEIGHTS_NAME,
TF2_WEIGHTS_NAME,
TRANSFORMERS_CACHE,
WEIGHTS_NAME,
cached_file,
get_file_from_repo,
has_file,
)
a_ : int = 'hf-internal-testing/tiny-random-bert'
a_ : Union[str, Any] = os.path.join(TRANSFORMERS_CACHE, 'models--hf-internal-testing--tiny-random-bert')
a_ : List[str] = '9b8c223d42b2188cb49d29af482996f9d0f3e5a6'
class lowerCamelCase__ ( unittest.TestCase):
"""simple docstring"""
def _a (self ):
'''simple docstring'''
lowerCamelCase = cached_file(UpperCamelCase_ , UpperCamelCase_ )
# Should have downloaded the file in here
self.assertTrue(os.path.isdir(UpperCamelCase_ ) )
# Cache should contain at least those three subfolders:
for subfolder in ["blobs", "refs", "snapshots"]:
self.assertTrue(os.path.isdir(os.path.join(UpperCamelCase_ , UpperCamelCase_ ) ) )
with open(os.path.join(UpperCamelCase_ , "refs" , "main" ) ) as f:
lowerCamelCase = f.read()
self.assertEqual(UpperCamelCase_ , os.path.join(UpperCamelCase_ , "snapshots" , UpperCamelCase_ , UpperCamelCase_ ) )
self.assertTrue(os.path.isfile(UpperCamelCase_ ) )
# File is cached at the same place the second time.
lowerCamelCase = cached_file(UpperCamelCase_ , UpperCamelCase_ )
self.assertEqual(UpperCamelCase_ , UpperCamelCase_ )
# Using a specific revision to test the full commit hash.
lowerCamelCase = cached_file(UpperCamelCase_ , UpperCamelCase_ , revision="9b8c223" )
self.assertEqual(UpperCamelCase_ , os.path.join(UpperCamelCase_ , "snapshots" , UpperCamelCase_ , UpperCamelCase_ ) )
def _a (self ):
'''simple docstring'''
with self.assertRaisesRegex(UpperCamelCase_ , "is not a valid model identifier" ):
lowerCamelCase = cached_file("tiny-random-bert" , UpperCamelCase_ )
with self.assertRaisesRegex(UpperCamelCase_ , "is not a valid git identifier" ):
lowerCamelCase = cached_file(UpperCamelCase_ , UpperCamelCase_ , revision="aaaa" )
with self.assertRaisesRegex(UpperCamelCase_ , "does not appear to have a file named" ):
lowerCamelCase = cached_file(UpperCamelCase_ , "conf" )
def _a (self ):
'''simple docstring'''
with self.assertRaisesRegex(UpperCamelCase_ , "does not appear to have a file named" ):
lowerCamelCase = cached_file(UpperCamelCase_ , "conf" )
with open(os.path.join(UpperCamelCase_ , "refs" , "main" ) ) as f:
lowerCamelCase = f.read()
self.assertTrue(os.path.isfile(os.path.join(UpperCamelCase_ , ".no_exist" , UpperCamelCase_ , "conf" ) ) )
lowerCamelCase = cached_file(UpperCamelCase_ , "conf" , _raise_exceptions_for_missing_entries=UpperCamelCase_ )
self.assertIsNone(UpperCamelCase_ )
lowerCamelCase = cached_file(UpperCamelCase_ , "conf" , local_files_only=UpperCamelCase_ , _raise_exceptions_for_missing_entries=UpperCamelCase_ )
self.assertIsNone(UpperCamelCase_ )
lowerCamelCase = mock.Mock()
lowerCamelCase = 5_00
lowerCamelCase = {}
lowerCamelCase = HTTPError
lowerCamelCase = {}
# Under the mock environment we get a 500 error when trying to reach the tokenizer.
with mock.patch("requests.Session.request" , return_value=UpperCamelCase_ ) as mock_head:
lowerCamelCase = cached_file(UpperCamelCase_ , "conf" , _raise_exceptions_for_connection_errors=UpperCamelCase_ )
self.assertIsNone(UpperCamelCase_ )
# This check we did call the fake head request
mock_head.assert_called()
def _a (self ):
'''simple docstring'''
self.assertTrue(has_file("hf-internal-testing/tiny-bert-pt-only" , UpperCamelCase_ ) )
self.assertFalse(has_file("hf-internal-testing/tiny-bert-pt-only" , UpperCamelCase_ ) )
self.assertFalse(has_file("hf-internal-testing/tiny-bert-pt-only" , UpperCamelCase_ ) )
def _a (self ):
'''simple docstring'''
self.assertIsNone(get_file_from_repo("bert-base-cased" , "ahah.txt" ) )
# The function raises if the repository does not exist.
with self.assertRaisesRegex(UpperCamelCase_ , "is not a valid model identifier" ):
get_file_from_repo("bert-base-case" , UpperCamelCase_ )
# The function raises if the revision does not exist.
with self.assertRaisesRegex(UpperCamelCase_ , "is not a valid git identifier" ):
get_file_from_repo("bert-base-cased" , UpperCamelCase_ , revision="ahaha" )
lowerCamelCase = get_file_from_repo("bert-base-cased" , UpperCamelCase_ )
# The name is the cached name which is not very easy to test, so instead we load the content.
lowerCamelCase = json.loads(open(UpperCamelCase_ , "r" ).read() )
self.assertEqual(config["hidden_size"] , 7_68 )
def _a (self ):
'''simple docstring'''
with tempfile.TemporaryDirectory() as tmp_dir:
lowerCamelCase = Path(UpperCamelCase_ ) / "a.txt"
filename.touch()
self.assertEqual(get_file_from_repo(UpperCamelCase_ , "a.txt" ) , str(UpperCamelCase_ ) )
self.assertIsNone(get_file_from_repo(UpperCamelCase_ , "b.txt" ) )
| 714
|
import unittest
from transformers.testing_utils import CaptureStdout
from transformers.tools.python_interpreter import evaluate
def __lowercase( UpperCAmelCase__ ):
"""simple docstring"""
return x + 2
class lowerCamelCase__ ( unittest.TestCase):
"""simple docstring"""
def _a (self ):
'''simple docstring'''
lowerCamelCase = "x = 3"
lowerCamelCase = {}
lowerCamelCase = evaluate(__a , {} , state=__a )
assert result == 3
self.assertDictEqual(__a , {"x": 3} )
lowerCamelCase = "x = y"
lowerCamelCase = {"y": 5}
lowerCamelCase = evaluate(__a , {} , state=__a )
# evaluate returns the value of the last assignment.
assert result == 5
self.assertDictEqual(__a , {"x": 5, "y": 5} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "y = add_two(x)"
lowerCamelCase = {"x": 3}
lowerCamelCase = evaluate(__a , {"add_two": add_two} , state=__a )
assert result == 5
self.assertDictEqual(__a , {"x": 3, "y": 5} )
# Won't work without the tool
with CaptureStdout() as out:
lowerCamelCase = evaluate(__a , {} , state=__a )
assert result is None
assert "tried to execute add_two" in out.out
def _a (self ):
'''simple docstring'''
lowerCamelCase = "x = 3"
lowerCamelCase = {}
lowerCamelCase = evaluate(__a , {} , state=__a )
assert result == 3
self.assertDictEqual(__a , {"x": 3} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "test_dict = {'x': x, 'y': add_two(x)}"
lowerCamelCase = {"x": 3}
lowerCamelCase = evaluate(__a , {"add_two": add_two} , state=__a )
self.assertDictEqual(__a , {"x": 3, "y": 5} )
self.assertDictEqual(__a , {"x": 3, "test_dict": {"x": 3, "y": 5}} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "x = 3\ny = 5"
lowerCamelCase = {}
lowerCamelCase = evaluate(__a , {} , state=__a )
# evaluate returns the value of the last assignment.
assert result == 5
self.assertDictEqual(__a , {"x": 3, "y": 5} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "text = f'This is x: {x}.'"
lowerCamelCase = {"x": 3}
lowerCamelCase = evaluate(__a , {} , state=__a )
# evaluate returns the value of the last assignment.
assert result == "This is x: 3."
self.assertDictEqual(__a , {"x": 3, "text": "This is x: 3."} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "if x <= 3:\n y = 2\nelse:\n y = 5"
lowerCamelCase = {"x": 3}
lowerCamelCase = evaluate(__a , {} , state=__a )
# evaluate returns the value of the last assignment.
assert result == 2
self.assertDictEqual(__a , {"x": 3, "y": 2} )
lowerCamelCase = {"x": 8}
lowerCamelCase = evaluate(__a , {} , state=__a )
# evaluate returns the value of the last assignment.
assert result == 5
self.assertDictEqual(__a , {"x": 8, "y": 5} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "test_list = [x, add_two(x)]"
lowerCamelCase = {"x": 3}
lowerCamelCase = evaluate(__a , {"add_two": add_two} , state=__a )
self.assertListEqual(__a , [3, 5] )
self.assertDictEqual(__a , {"x": 3, "test_list": [3, 5]} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "y = x"
lowerCamelCase = {"x": 3}
lowerCamelCase = evaluate(__a , {} , state=__a )
assert result == 3
self.assertDictEqual(__a , {"x": 3, "y": 3} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "test_list = [x, add_two(x)]\ntest_list[1]"
lowerCamelCase = {"x": 3}
lowerCamelCase = evaluate(__a , {"add_two": add_two} , state=__a )
assert result == 5
self.assertDictEqual(__a , {"x": 3, "test_list": [3, 5]} )
lowerCamelCase = "test_dict = {'x': x, 'y': add_two(x)}\ntest_dict['y']"
lowerCamelCase = {"x": 3}
lowerCamelCase = evaluate(__a , {"add_two": add_two} , state=__a )
assert result == 5
self.assertDictEqual(__a , {"x": 3, "test_dict": {"x": 3, "y": 5}} )
def _a (self ):
'''simple docstring'''
lowerCamelCase = "x = 0\nfor i in range(3):\n x = i"
lowerCamelCase = {}
lowerCamelCase = evaluate(__a , {"range": range} , state=__a )
assert result == 2
self.assertDictEqual(__a , {"x": 2, "i": 2} )
| 484
| 0
|
import math
import tensorflow as tf
from packaging import version
def _UpperCAmelCase ( A ):
'''simple docstring'''
UpperCAmelCase__ =tf.convert_to_tensor(A )
UpperCAmelCase__ =0.5 * (1.0 + tf.math.erf(x / tf.cast(tf.sqrt(2.0 ) , x.dtype ) ))
return x * cdf
def _UpperCAmelCase ( A ):
'''simple docstring'''
UpperCAmelCase__ =tf.convert_to_tensor(A )
UpperCAmelCase__ =tf.cast(math.pi , x.dtype )
UpperCAmelCase__ =tf.cast(0.04_47_15 , x.dtype )
UpperCAmelCase__ =0.5 * (1.0 + tf.tanh(tf.sqrt(2.0 / pi ) * (x + coeff * tf.pow(A , 3 )) ))
return x * cdf
def _UpperCAmelCase ( A ):
'''simple docstring'''
UpperCAmelCase__ =tf.convert_to_tensor(A )
return x * tf.tanh(tf.math.softplus(A ) )
def _UpperCAmelCase ( A ):
'''simple docstring'''
UpperCAmelCase__ =tf.convert_to_tensor(A )
UpperCAmelCase__ =tf.cast(0.04_47_15 , x.dtype )
UpperCAmelCase__ =tf.cast(0.79_78_84_56_08 , x.dtype )
return 0.5 * x * (1.0 + tf.tanh(x * coeffa * (1.0 + coeffa * x * x) ))
def _UpperCAmelCase ( A ):
'''simple docstring'''
UpperCAmelCase__ =tf.convert_to_tensor(A )
UpperCAmelCase__ =tf.cast(1.7_02 , x.dtype )
return x * tf.math.sigmoid(coeff * x )
def _UpperCAmelCase ( A ):
'''simple docstring'''
return tf.clip_by_value(_gelu(A ) , -10 , 10 )
def _UpperCAmelCase ( A , A=-1 ):
'''simple docstring'''
UpperCAmelCase__ , UpperCAmelCase__ =tf.split(A , 2 , axis=A )
return a * tf.math.sigmoid(A )
if version.parse(tf.version.VERSION) >= version.parse('2.4'):
def _UpperCAmelCase ( A ):
'''simple docstring'''
return tf.keras.activations.gelu(A , approximate=A )
UpperCamelCase_ = tf.keras.activations.gelu
UpperCamelCase_ = approximate_gelu_wrap
else:
UpperCamelCase_ = _gelu
UpperCamelCase_ = _gelu_new
UpperCamelCase_ = {
'gelu': gelu,
'gelu_10': gelu_aa,
'gelu_fast': gelu_fast,
'gelu_new': gelu_new,
'glu': glu,
'mish': mish,
'quick_gelu': quick_gelu,
'relu': tf.keras.activations.relu,
'sigmoid': tf.keras.activations.sigmoid,
'silu': tf.keras.activations.swish,
'swish': tf.keras.activations.swish,
'tanh': tf.keras.activations.tanh,
}
def _UpperCAmelCase ( A ):
'''simple docstring'''
if activation_string in ACTaFN:
return ACTaFN[activation_string]
else:
raise KeyError(F"""function {activation_string} not found in ACT2FN mapping {list(ACTaFN.keys() )}""" )
| 625
|
import argparse
import shutil
from pathlib import Path
from tqdm import tqdm
from transformers import AutoTokenizer
def _UpperCAmelCase ( A , A , A , A=1024 ):
'''simple docstring'''
UpperCAmelCase__ , UpperCAmelCase__ =[], []
UpperCAmelCase__ =list(zip(A , A ) )
UpperCAmelCase__ , UpperCAmelCase__ =sorted_examples[0]
def is_too_big(A ):
return tok(A , return_tensors="pt" ).input_ids.shape[1] > max_tokens
for src, tgt in tqdm(sorted_examples[1:] ):
UpperCAmelCase__ =new_src + " " + src
UpperCAmelCase__ =new_tgt + " " + tgt
if is_too_big(A ) or is_too_big(A ): # cant fit, finalize example
finished_src.append(A )
finished_tgt.append(A )
UpperCAmelCase__ , UpperCAmelCase__ =src, tgt
else: # can fit, keep adding
UpperCAmelCase__ , UpperCAmelCase__ =cand_src, cand_tgt
# cleanup
if new_src:
assert new_tgt
finished_src.append(A )
finished_tgt.append(A )
return finished_src, finished_tgt
def _UpperCAmelCase ( A , A , A , A ):
'''simple docstring'''
UpperCAmelCase__ =Path(A )
save_path.mkdir(exist_ok=A )
for split in ["train"]:
UpperCAmelCase__ , UpperCAmelCase__ =data_dir / F"""{split}.source""", data_dir / F"""{split}.target"""
UpperCAmelCase__ =[x.rstrip() for x in Path(A ).open().readlines()]
UpperCAmelCase__ =[x.rstrip() for x in Path(A ).open().readlines()]
UpperCAmelCase__ , UpperCAmelCase__ =pack_examples(A , A , A , A )
print(F"""packed {split} split from {len(A )} examples -> {len(A )}.""" )
Path(save_path / F"""{split}.source""" ).open("w" ).write("\n".join(A ) )
Path(save_path / F"""{split}.target""" ).open("w" ).write("\n".join(A ) )
for split in ["val", "test"]:
UpperCAmelCase__ , UpperCAmelCase__ =data_dir / F"""{split}.source""", data_dir / F"""{split}.target"""
shutil.copyfile(A , save_path / F"""{split}.source""" )
shutil.copyfile(A , save_path / F"""{split}.target""" )
def _UpperCAmelCase ( ):
'''simple docstring'''
UpperCAmelCase__ =argparse.ArgumentParser()
parser.add_argument("--tok_name" , type=A , help="like facebook/bart-large-cnn,t5-base, etc." )
parser.add_argument("--max_seq_len" , type=A , default=128 )
parser.add_argument("--data_dir" , type=A )
parser.add_argument("--save_path" , type=A )
UpperCAmelCase__ =parser.parse_args()
UpperCAmelCase__ =AutoTokenizer.from_pretrained(args.tok_name )
return pack_data_dir(A , Path(args.data_dir ) , args.max_seq_len , args.save_path )
if __name__ == "__main__":
packer_cli()
| 625
| 1
|
import pytest
from datasets.splits import SplitDict, SplitInfo
from datasets.utils.py_utils import asdict
@pytest.mark.parametrize(
'''split_dict''' ,[
SplitDict(),
SplitDict({'''train''': SplitInfo(name='''train''' ,num_bytes=1337 ,num_examples=42 ,dataset_name='''my_dataset''')}),
SplitDict({'''train''': SplitInfo(name='''train''' ,num_bytes=1337 ,num_examples=42)}),
SplitDict({'''train''': SplitInfo()}),
] ,)
def lowerCamelCase( a__):
_SCREAMING_SNAKE_CASE =split_dict._to_yaml_list()
assert len(a__) == len(a__)
_SCREAMING_SNAKE_CASE =SplitDict._from_yaml_list(a__)
for split_name, split_info in split_dict.items():
# dataset_name field is deprecated, and is therefore not part of the YAML dump
_SCREAMING_SNAKE_CASE =None
# the split name of split_dict takes over the name of the split info object
_SCREAMING_SNAKE_CASE =split_name
assert split_dict == reloaded
@pytest.mark.parametrize(
'''split_info''' ,[SplitInfo(), SplitInfo(dataset_name=a__), SplitInfo(dataset_name='''my_dataset''')])
def lowerCamelCase( a__):
# For backward compatibility, we need asdict(split_dict) to return split info dictrionaries with the "dataset_name"
# field even if it's deprecated. This way old versionso of `datasets` can still reload dataset_infos.json files
_SCREAMING_SNAKE_CASE =asdict(SplitDict({'''train''': split_info}))
assert "dataset_name" in split_dict_asdict["train"]
assert split_dict_asdict["train"]["dataset_name"] == split_info.dataset_name
| 191
|
snake_case_ : dict[tuple[int, int, int], int] = {}
def lowerCamelCase( a__ ,a__ ,a__):
# if we are absent twice, or late 3 consecutive days,
# no further prize strings are possible
if late == 3 or absent == 2:
return 0
# if we have no days left, and have not failed any other rules,
# we have a prize string
if days == 0:
return 1
# No easy solution, so now we need to do the recursive calculation
# First, check if the combination is already in the cache, and
# if yes, return the stored value from there since we already
# know the number of possible prize strings from this point on
_SCREAMING_SNAKE_CASE =(days, absent, late)
if key in cache:
return cache[key]
# now we calculate the three possible ways that can unfold from
# this point on, depending on our attendance today
# 1) if we are late (but not absent), the "absent" counter stays as
# it is, but the "late" counter increases by one
_SCREAMING_SNAKE_CASE =_calculate(days - 1 ,a__ ,late + 1)
# 2) if we are absent, the "absent" counter increases by 1, and the
# "late" counter resets to 0
_SCREAMING_SNAKE_CASE =_calculate(days - 1 ,absent + 1 ,0)
# 3) if we are on time, this resets the "late" counter and keeps the
# absent counter
_SCREAMING_SNAKE_CASE =_calculate(days - 1 ,a__ ,0)
_SCREAMING_SNAKE_CASE =state_late + state_absent + state_ontime
_SCREAMING_SNAKE_CASE =prizestrings
return prizestrings
def lowerCamelCase( a__ = 30):
return _calculate(a__ ,absent=0 ,late=0)
if __name__ == "__main__":
print(solution())
| 191
| 1
|
'''simple docstring'''
import os
import pytest
from transformers.dynamic_module_utils import get_imports
__snake_case: Dict = "\nimport os\n"
__snake_case: Dict = "\ndef foo():\n import os\n return False\n"
__snake_case: List[Any] = "\ndef foo():\n def bar():\n if True:\n import os\n return False\n return bar()\n"
__snake_case: Any = "\nimport os\n\ntry:\n import bar\nexcept ImportError:\n raise ValueError()\n"
__snake_case: str = "\nimport os\n\ndef foo():\n try:\n import bar\n except ImportError:\n raise ValueError()\n"
__snake_case: str = "\nimport os\n\ntry:\n import bar\nexcept (ImportError, AttributeError):\n raise ValueError()\n"
__snake_case: Dict = "\nimport os\n\ntry:\n import bar\nexcept ImportError as e:\n raise ValueError()\n"
__snake_case: Tuple = "\nimport os\n\ntry:\n import bar\nexcept:\n raise ValueError()\n"
__snake_case: Optional[int] = "\nimport os\n\ntry:\n import bar\n import baz\nexcept ImportError:\n raise ValueError()\n"
__snake_case: List[str] = "\nimport os\n\ntry:\n import bar\n import baz\nexcept ImportError:\n x = 1\n raise ValueError()\n"
__snake_case: int = [
TOP_LEVEL_IMPORT,
IMPORT_IN_FUNCTION,
DEEPLY_NESTED_IMPORT,
TOP_LEVEL_TRY_IMPORT,
GENERIC_EXCEPT_IMPORT,
MULTILINE_TRY_IMPORT,
MULTILINE_BOTH_IMPORT,
MULTIPLE_EXCEPTS_IMPORT,
EXCEPT_AS_IMPORT,
TRY_IMPORT_IN_FUNCTION,
]
@pytest.mark.parametrize("""case""" , __SCREAMING_SNAKE_CASE )
def _snake_case ( A_ : List[str] , A_ : Dict ):
"""simple docstring"""
a_ : Optional[int] = os.path.join(__SCREAMING_SNAKE_CASE , """test_file.py""" )
with open(__SCREAMING_SNAKE_CASE , """w""" ) as _tmp_file:
_tmp_file.write(__SCREAMING_SNAKE_CASE )
a_ : int = get_imports(__SCREAMING_SNAKE_CASE )
assert parsed_imports == ["os"]
| 577
|
'''simple docstring'''
import itertools
import random
import unittest
import numpy as np
from transformers import BatchFeature, SpeechTaFeatureExtractor
from transformers.testing_utils import require_torch
from transformers.utils.import_utils import is_torch_available
from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
if is_torch_available():
import torch
SCREAMING_SNAKE_CASE_ = random.Random()
def __lowercase ( __SCREAMING_SNAKE_CASE , __SCREAMING_SNAKE_CASE=1.0 , __SCREAMING_SNAKE_CASE=None , __SCREAMING_SNAKE_CASE=None ) -> List[str]:
"""simple docstring"""
if rng is None:
__a = global_rng
__a = []
for batch_idx in range(shape[0] ):
values.append([] )
for _ in range(shape[1] ):
values[-1].append(rng.random() * scale )
return values
@require_torch
class lowerCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __init__( self : Optional[int] , SCREAMING_SNAKE_CASE__ : Optional[int] , SCREAMING_SNAKE_CASE__ : List[Any]=7 , SCREAMING_SNAKE_CASE__ : Any=4_0_0 , SCREAMING_SNAKE_CASE__ : Union[str, Any]=2_0_0_0 , SCREAMING_SNAKE_CASE__ : List[str]=1 , SCREAMING_SNAKE_CASE__ : List[str]=0.0 , SCREAMING_SNAKE_CASE__ : Any=1_6_0_0_0 , SCREAMING_SNAKE_CASE__ : List[Any]=True , SCREAMING_SNAKE_CASE__ : Union[str, Any]=8_0 , SCREAMING_SNAKE_CASE__ : Tuple=1_6 , SCREAMING_SNAKE_CASE__ : Dict=6_4 , SCREAMING_SNAKE_CASE__ : Optional[int]="hann_window" , SCREAMING_SNAKE_CASE__ : List[Any]=8_0 , SCREAMING_SNAKE_CASE__ : int=7_6_0_0 , SCREAMING_SNAKE_CASE__ : str=1E-10 , SCREAMING_SNAKE_CASE__ : Union[str, Any]=True , ):
'''simple docstring'''
__a = parent
__a = batch_size
__a = min_seq_length
__a = max_seq_length
__a = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
__a = feature_size
__a = padding_value
__a = sampling_rate
__a = do_normalize
__a = num_mel_bins
__a = hop_length
__a = win_length
__a = win_function
__a = fmin
__a = fmax
__a = mel_floor
__a = return_attention_mask
def __a ( self : Tuple ):
'''simple docstring'''
return {
"feature_size": self.feature_size,
"padding_value": self.padding_value,
"sampling_rate": self.sampling_rate,
"do_normalize": self.do_normalize,
"num_mel_bins": self.num_mel_bins,
"hop_length": self.hop_length,
"win_length": self.win_length,
"win_function": self.win_function,
"fmin": self.fmin,
"fmax": self.fmax,
"mel_floor": self.mel_floor,
"return_attention_mask": self.return_attention_mask,
}
def __a ( self : Dict , SCREAMING_SNAKE_CASE__ : Any=False , SCREAMING_SNAKE_CASE__ : Optional[int]=False ):
'''simple docstring'''
def _flatten(SCREAMING_SNAKE_CASE__ : str ):
return list(itertools.chain(*SCREAMING_SNAKE_CASE__ ) )
if equal_length:
__a = floats_list((self.batch_size, self.max_seq_length) )
else:
# make sure that inputs increase in size
__a = [
_flatten(floats_list((x, self.feature_size) ) )
for x in range(self.min_seq_length , self.max_seq_length , self.seq_length_diff )
]
if numpify:
__a = [np.asarray(SCREAMING_SNAKE_CASE__ ) for x in speech_inputs]
return speech_inputs
def __a ( self : List[str] , SCREAMING_SNAKE_CASE__ : str=False , SCREAMING_SNAKE_CASE__ : int=False ):
'''simple docstring'''
if equal_length:
__a = [floats_list((self.max_seq_length, self.num_mel_bins) ) for _ in range(self.batch_size )]
else:
# make sure that inputs increase in size
__a = [
floats_list((x, self.num_mel_bins) )
for x in range(self.min_seq_length , self.max_seq_length , self.seq_length_diff )
]
if numpify:
__a = [np.asarray(SCREAMING_SNAKE_CASE__ ) for x in speech_inputs]
return speech_inputs
@require_torch
class lowerCAmelCase_ ( snake_case__ , unittest.TestCase ):
"""simple docstring"""
a_ :List[Any] =SpeechTaFeatureExtractor
def __a ( self : List[Any] ):
'''simple docstring'''
__a = SpeechTaFeatureExtractionTester(self )
def __a ( self : Any , SCREAMING_SNAKE_CASE__ : Dict ):
'''simple docstring'''
self.assertTrue(np.all(np.mean(SCREAMING_SNAKE_CASE__ , axis=0 ) < 1E-3 ) )
self.assertTrue(np.all(np.abs(np.var(SCREAMING_SNAKE_CASE__ , axis=0 ) - 1 ) < 1E-3 ) )
def __a ( self : Dict ):
'''simple docstring'''
__a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
# create three inputs of length 800, 1000, and 1200
__a = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
__a = [np.asarray(SCREAMING_SNAKE_CASE__ ) for speech_input in speech_inputs]
# Test not batched input
__a = feat_extract(speech_inputs[0] , return_tensors="""np""" ).input_values
__a = feat_extract(np_speech_inputs[0] , return_tensors="""np""" ).input_values
self.assertTrue(np.allclose(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ , atol=1E-3 ) )
# Test batched
__a = feat_extract(SCREAMING_SNAKE_CASE__ , return_tensors="""np""" ).input_values
__a = feat_extract(SCREAMING_SNAKE_CASE__ , return_tensors="""np""" ).input_values
for enc_seq_a, enc_seq_a in zip(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ ):
self.assertTrue(np.allclose(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ , atol=1E-3 ) )
def __a ( self : Dict ):
'''simple docstring'''
__a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__a = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
__a = ["""longest""", """max_length""", """do_not_pad"""]
__a = [None, 1_6_0_0, None]
for max_length, padding in zip(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ ):
__a = feat_extract(SCREAMING_SNAKE_CASE__ , padding=SCREAMING_SNAKE_CASE__ , max_length=SCREAMING_SNAKE_CASE__ , return_tensors="""np""" )
__a = processed.input_values
self._check_zero_mean_unit_variance(input_values[0][:8_0_0] )
self.assertTrue(input_values[0][8_0_0:].sum() < 1E-6 )
self._check_zero_mean_unit_variance(input_values[1][:1_0_0_0] )
self.assertTrue(input_values[0][1_0_0_0:].sum() < 1E-6 )
self._check_zero_mean_unit_variance(input_values[2][:1_2_0_0] )
def __a ( self : List[str] ):
'''simple docstring'''
__a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__a = range(8_0_0 , 1_4_0_0 , 2_0_0 )
__a = [floats_list((1, x) )[0] for x in lengths]
__a = ["""longest""", """max_length""", """do_not_pad"""]
__a = [None, 1_6_0_0, None]
for max_length, padding in zip(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ ):
__a = feat_extract(SCREAMING_SNAKE_CASE__ , max_length=SCREAMING_SNAKE_CASE__ , padding=SCREAMING_SNAKE_CASE__ )
__a = processed.input_values
self._check_zero_mean_unit_variance(input_values[0][:8_0_0] )
self._check_zero_mean_unit_variance(input_values[1][:1_0_0_0] )
self._check_zero_mean_unit_variance(input_values[2][:1_2_0_0] )
def __a ( self : List[Any] ):
'''simple docstring'''
__a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__a = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
__a = feat_extract(
SCREAMING_SNAKE_CASE__ , truncation=SCREAMING_SNAKE_CASE__ , max_length=1_0_0_0 , padding="""max_length""" , return_tensors="""np""" )
__a = processed.input_values
self._check_zero_mean_unit_variance(input_values[0, :8_0_0] )
self._check_zero_mean_unit_variance(input_values[1] )
self._check_zero_mean_unit_variance(input_values[2] )
def __a ( self : Union[str, Any] ):
'''simple docstring'''
__a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__a = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
__a = feat_extract(
SCREAMING_SNAKE_CASE__ , truncation=SCREAMING_SNAKE_CASE__ , max_length=1_0_0_0 , padding="""longest""" , return_tensors="""np""" )
__a = processed.input_values
self._check_zero_mean_unit_variance(input_values[0, :8_0_0] )
self._check_zero_mean_unit_variance(input_values[1, :1_0_0_0] )
self._check_zero_mean_unit_variance(input_values[2] )
# make sure that if max_length < longest -> then pad to max_length
self.assertTrue(input_values.shape == (3, 1_0_0_0) )
__a = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
__a = feat_extract(
SCREAMING_SNAKE_CASE__ , truncation=SCREAMING_SNAKE_CASE__ , max_length=2_0_0_0 , padding="""longest""" , return_tensors="""np""" )
__a = processed.input_values
self._check_zero_mean_unit_variance(input_values[0, :8_0_0] )
self._check_zero_mean_unit_variance(input_values[1, :1_0_0_0] )
self._check_zero_mean_unit_variance(input_values[2] )
# make sure that if max_length > longest -> then pad to longest
self.assertTrue(input_values.shape == (3, 1_2_0_0) )
def __a ( self : Any ):
'''simple docstring'''
__a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__a = np.random.rand(1_0_0 ).astype(np.floataa )
__a = np_speech_inputs.tolist()
for inputs in [py_speech_inputs, np_speech_inputs]:
__a = feature_extractor.pad([{"""input_values""": inputs}] , return_tensors="""np""" )
self.assertTrue(np_processed.input_values.dtype == np.floataa )
__a = feature_extractor.pad([{"""input_values""": inputs}] , return_tensors="""pt""" )
self.assertTrue(pt_processed.input_values.dtype == torch.floataa )
def __a ( self : Dict ):
'''simple docstring'''
__a = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
# create three inputs of length 800, 1000, and 1200
__a = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
__a = [np.asarray(SCREAMING_SNAKE_CASE__ ) for speech_input in speech_inputs]
# Test feature size
__a = feature_extractor(audio_target=SCREAMING_SNAKE_CASE__ , padding=SCREAMING_SNAKE_CASE__ , return_tensors="""np""" ).input_values
self.assertTrue(input_values.ndim == 3 )
self.assertTrue(input_values.shape[-1] == feature_extractor.num_mel_bins )
# Test not batched input
__a = feature_extractor(speech_inputs[0] , return_tensors="""np""" ).input_values
__a = feature_extractor(np_speech_inputs[0] , return_tensors="""np""" ).input_values
self.assertTrue(np.allclose(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ , atol=1E-3 ) )
# Test batched
__a = feature_extractor(SCREAMING_SNAKE_CASE__ , return_tensors="""np""" ).input_values
__a = feature_extractor(SCREAMING_SNAKE_CASE__ , return_tensors="""np""" ).input_values
for enc_seq_a, enc_seq_a in zip(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ ):
self.assertTrue(np.allclose(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ , atol=1E-3 ) )
# Test 2-D numpy arrays are batched.
__a = [floats_list((1, x) )[0] for x in (8_0_0, 8_0_0, 8_0_0)]
__a = np.asarray(SCREAMING_SNAKE_CASE__ )
__a = feature_extractor(SCREAMING_SNAKE_CASE__ , return_tensors="""np""" ).input_values
__a = feature_extractor(SCREAMING_SNAKE_CASE__ , return_tensors="""np""" ).input_values
for enc_seq_a, enc_seq_a in zip(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ ):
self.assertTrue(np.allclose(SCREAMING_SNAKE_CASE__ , SCREAMING_SNAKE_CASE__ , atol=1E-3 ) )
def __a ( self : Optional[Any] ):
'''simple docstring'''
__a = self.feat_extract_tester.prepare_inputs_for_target()
__a = self.feature_extraction_class(**self.feat_extract_dict )
__a = feat_extract.model_input_names[0]
__a = BatchFeature({input_name: speech_inputs} )
self.assertTrue(all(len(SCREAMING_SNAKE_CASE__ ) == len(SCREAMING_SNAKE_CASE__ ) for x, y in zip(SCREAMING_SNAKE_CASE__ , processed_features[input_name] ) ) )
__a = self.feat_extract_tester.prepare_inputs_for_target(equal_length=SCREAMING_SNAKE_CASE__ )
__a = BatchFeature({input_name: speech_inputs} , tensor_type="""np""" )
__a = processed_features[input_name]
if len(batch_features_input.shape ) < 3:
__a = batch_features_input[:, :, None]
self.assertTrue(
batch_features_input.shape
== (self.feat_extract_tester.batch_size, len(speech_inputs[0] ), self.feat_extract_tester.num_mel_bins) )
@require_torch
def __a ( self : Any ):
'''simple docstring'''
__a = self.feat_extract_tester.prepare_inputs_for_target(equal_length=SCREAMING_SNAKE_CASE__ )
__a = self.feature_extraction_class(**self.feat_extract_dict )
__a = feat_extract.model_input_names[0]
__a = BatchFeature({input_name: speech_inputs} , tensor_type="""pt""" )
__a = processed_features[input_name]
if len(batch_features_input.shape ) < 3:
__a = batch_features_input[:, :, None]
self.assertTrue(
batch_features_input.shape
== (self.feat_extract_tester.batch_size, len(speech_inputs[0] ), self.feat_extract_tester.num_mel_bins) )
@require_torch
def __a ( self : int ):
'''simple docstring'''
__a = self.feature_extraction_class(**self.feat_extract_dict )
__a = self.feat_extract_tester.prepare_inputs_for_target()
__a = feat_extract.model_input_names[0]
__a = BatchFeature({input_name: speech_inputs} )
__a = feat_extract.num_mel_bins # hack!
__a = feat_extract.pad(SCREAMING_SNAKE_CASE__ , padding="""longest""" , return_tensors="""np""" )[input_name]
__a = feat_extract.pad(SCREAMING_SNAKE_CASE__ , padding="""longest""" , return_tensors="""pt""" )[input_name]
self.assertTrue(abs(input_np.astype(np.floataa ).sum() - input_pt.numpy().astype(np.floataa ).sum() ) < 1E-2 )
def __a ( self : Any ):
'''simple docstring'''
__a = self.feat_extract_dict
__a = True
__a = self.feature_extraction_class(**SCREAMING_SNAKE_CASE__ )
__a = self.feat_extract_tester.prepare_inputs_for_target()
__a = [len(SCREAMING_SNAKE_CASE__ ) for x in speech_inputs]
__a = feat_extract.model_input_names[0]
__a = BatchFeature({input_name: speech_inputs} )
__a = feat_extract.num_mel_bins # hack!
__a = feat_extract.pad(SCREAMING_SNAKE_CASE__ , padding="""longest""" , return_tensors="""np""" )
self.assertIn("""attention_mask""" , SCREAMING_SNAKE_CASE__ )
self.assertListEqual(list(processed.attention_mask.shape ) , list(processed[input_name].shape[:2] ) )
self.assertListEqual(processed.attention_mask.sum(-1 ).tolist() , SCREAMING_SNAKE_CASE__ )
def __a ( self : Tuple ):
'''simple docstring'''
__a = self.feat_extract_dict
__a = True
__a = self.feature_extraction_class(**SCREAMING_SNAKE_CASE__ )
__a = self.feat_extract_tester.prepare_inputs_for_target()
__a = [len(SCREAMING_SNAKE_CASE__ ) for x in speech_inputs]
__a = feat_extract.model_input_names[0]
__a = BatchFeature({input_name: speech_inputs} )
__a = min(SCREAMING_SNAKE_CASE__ )
__a = feat_extract.num_mel_bins # hack!
__a = feat_extract.pad(
SCREAMING_SNAKE_CASE__ , padding="""max_length""" , max_length=SCREAMING_SNAKE_CASE__ , truncation=SCREAMING_SNAKE_CASE__ , return_tensors="""np""" )
self.assertIn("""attention_mask""" , SCREAMING_SNAKE_CASE__ )
self.assertListEqual(
list(processed_pad.attention_mask.shape ) , [processed_pad[input_name].shape[0], max_length] )
self.assertListEqual(
processed_pad.attention_mask[:, :max_length].sum(-1 ).tolist() , [max_length for x in speech_inputs] )
def __a ( self : Optional[int] , SCREAMING_SNAKE_CASE__ : List[str] ):
'''simple docstring'''
from datasets import load_dataset
__a = load_dataset("""hf-internal-testing/librispeech_asr_dummy""" , """clean""" , split="""validation""" )
# automatic decoding with librispeech
__a = ds.sort("""id""" ).select(range(SCREAMING_SNAKE_CASE__ ) )[:num_samples]["""audio"""]
return [x["array"] for x in speech_samples]
def __a ( self : List[str] ):
'''simple docstring'''
__a = torch.tensor(
[2.38_04E-03, 2.07_52E-03, 1.98_36E-03, 2.10_57E-03, 1.61_74E-03,
3.05_18E-04, 9.15_53E-05, 3.35_69E-04, 9.76_56E-04, 1.83_11E-03,
2.01_42E-03, 2.10_57E-03, 1.73_95E-03, 4.57_76E-04, -3.96_73E-04,
4.57_76E-04, 1.00_71E-03, 9.15_53E-05, 4.88_28E-04, 1.15_97E-03,
7.32_42E-04, 9.46_04E-04, 1.80_05E-03, 1.83_11E-03, 8.85_01E-04,
4.27_25E-04, 4.88_28E-04, 7.32_42E-04, 1.09_86E-03, 2.10_57E-03] )
# fmt: on
__a = self._load_datasamples(1 )
__a = SpeechTaFeatureExtractor()
__a = feature_extractor(SCREAMING_SNAKE_CASE__ , return_tensors="""pt""" ).input_values
self.assertEquals(input_values.shape , (1, 9_3_6_8_0) )
self.assertTrue(torch.allclose(input_values[0, :3_0] , SCREAMING_SNAKE_CASE__ , atol=1E-6 ) )
def __a ( self : Optional[int] ):
'''simple docstring'''
__a = torch.tensor(
[-2.6_8_7_0, -3.0_1_0_4, -3.1_3_5_6, -3.5_3_5_2, -3.0_0_4_4, -3.0_3_5_3, -3.4_7_1_9, -3.6_7_7_7,
-3.1_5_2_0, -2.9_4_3_5, -2.6_5_5_3, -2.8_7_9_5, -2.9_9_4_4, -2.5_9_2_1, -3.0_2_7_9, -3.0_3_8_6,
-3.0_8_6_4, -3.1_2_9_1, -3.2_3_5_3, -2.7_4_4_4, -2.6_8_3_1, -2.7_2_8_7, -3.1_7_6_1, -3.1_5_7_1,
-3.2_7_2_6, -3.0_5_8_2, -3.1_0_0_7, -3.4_5_3_3, -3.4_6_9_5, -3.0_9_9_8] )
# fmt: on
__a = self._load_datasamples(1 )
__a = SpeechTaFeatureExtractor()
__a = feature_extractor(audio_target=SCREAMING_SNAKE_CASE__ , return_tensors="""pt""" ).input_values
self.assertEquals(input_values.shape , (1, 3_6_6, 8_0) )
self.assertTrue(torch.allclose(input_values[0, 0, :3_0] , SCREAMING_SNAKE_CASE__ , atol=1E-4 ) )
| 582
| 0
|
def A_ ( _lowerCAmelCase ) -> bool:
UpperCamelCase : Optional[int] = n ** (1 / 3)
return (val * val * val) == n
if __name__ == "__main__":
print(perfect_cube(27))
print(perfect_cube(4))
| 38
|
from __future__ import annotations
from random import random
from typing import Generic, TypeVar
__lowerCamelCase : Dict = TypeVar("""KT""")
__lowerCamelCase : Dict = TypeVar("""VT""")
class A__ ( Generic[KT, VT] ):
def __init__( self , A_ = "root" , A_ = None ):
'''simple docstring'''
UpperCamelCase : int = key
UpperCamelCase : List[Any] = value
UpperCamelCase : list[Node[KT, VT]] = []
def __repr__( self ):
'''simple docstring'''
return F"""Node({self.key}: {self.value})"""
@property
def __UpperCamelCase( self ):
'''simple docstring'''
return len(self.forward )
class A__ ( Generic[KT, VT] ):
def __init__( self , A_ = 0.5 , A_ = 16 ):
'''simple docstring'''
UpperCamelCase : Node[KT, VT] = Node[KT, VT]()
UpperCamelCase : List[Any] = 0
UpperCamelCase : Union[str, Any] = p
UpperCamelCase : List[str] = max_level
def __str__( self ):
'''simple docstring'''
UpperCamelCase : int = list(self )
if len(A_ ) == 0:
return F"""SkipList(level={self.level})"""
UpperCamelCase : str = max((len(str(A_ ) ) for item in items) , default=4 )
UpperCamelCase : Dict = max(A_ , 4 ) + 4
UpperCamelCase : str = self.head
UpperCamelCase : List[Any] = []
UpperCamelCase : int = node.forward.copy()
lines.append(F"""[{node.key}]""".ljust(A_ , "-" ) + "* " * len(A_ ) )
lines.append(" " * label_size + "| " * len(A_ ) )
while len(node.forward ) != 0:
UpperCamelCase : Union[str, Any] = node.forward[0]
lines.append(
F"""[{node.key}]""".ljust(A_ , "-" )
+ " ".join(str(n.key ) if n.key == node.key else "|" for n in forwards ) )
lines.append(" " * label_size + "| " * len(A_ ) )
UpperCamelCase : Tuple = node.forward
lines.append("None".ljust(A_ ) + "* " * len(A_ ) )
return F"""SkipList(level={self.level})\n""" + "\n".join(A_ )
def __iter__( self ):
'''simple docstring'''
UpperCamelCase : Union[str, Any] = self.head
while len(node.forward ) != 0:
yield node.forward[0].key
UpperCamelCase : Union[str, Any] = node.forward[0]
def __UpperCamelCase( self ):
'''simple docstring'''
UpperCamelCase : Tuple = 1
while random() < self.p and level < self.max_level:
level += 1
return level
def __UpperCamelCase( self , A_ ):
'''simple docstring'''
UpperCamelCase : List[str] = []
UpperCamelCase : List[Any] = self.head
for i in reversed(range(self.level ) ):
# i < node.level - When node level is lesser than `i` decrement `i`.
# node.forward[i].key < key - Jumping to node with key value higher
# or equal to searched key would result
# in skipping searched key.
while i < node.level and node.forward[i].key < key:
UpperCamelCase : str = node.forward[i]
# Each leftmost node (relative to searched node) will potentially have to
# be updated.
update_vector.append(A_ )
update_vector.reverse() # Note that we were inserting values in reverse order.
# len(node.forward) != 0 - If current node doesn't contain any further
# references then searched key is not present.
# node.forward[0].key == key - Next node key should be equal to search key
# if key is present.
if len(node.forward ) != 0 and node.forward[0].key == key:
return node.forward[0], update_vector
else:
return None, update_vector
def __UpperCamelCase( self , A_ ):
'''simple docstring'''
UpperCamelCase , UpperCamelCase : str = self._locate_node(A_ )
if node is not None:
for i, update_node in enumerate(A_ ):
# Remove or replace all references to removed node.
if update_node.level > i and update_node.forward[i].key == key:
if node.level > i:
UpperCamelCase : Tuple = node.forward[i]
else:
UpperCamelCase : List[Any] = update_node.forward[:i]
def __UpperCamelCase( self , A_ , A_ ):
'''simple docstring'''
UpperCamelCase , UpperCamelCase : Optional[int] = self._locate_node(A_ )
if node is not None:
UpperCamelCase : Union[str, Any] = value
else:
UpperCamelCase : Dict = self.random_level()
if level > self.level:
# After level increase we have to add additional nodes to head.
for _ in range(self.level - 1 , A_ ):
update_vector.append(self.head )
UpperCamelCase : Optional[int] = level
UpperCamelCase : Dict = Node(A_ , A_ )
for i, update_node in enumerate(update_vector[:level] ):
# Change references to pass through new node.
if update_node.level > i:
new_node.forward.append(update_node.forward[i] )
if update_node.level < i + 1:
update_node.forward.append(A_ )
else:
UpperCamelCase : List[Any] = new_node
def __UpperCamelCase( self , A_ ):
'''simple docstring'''
UpperCamelCase , UpperCamelCase : Union[str, Any] = self._locate_node(A_ )
if node is not None:
return node.value
return None
def A_ ( ) -> List[Any]:
UpperCamelCase : int = SkipList()
skip_list.insert("Key1" , 3 )
skip_list.insert("Key2" , 12 )
skip_list.insert("Key3" , 41 )
skip_list.insert("Key4" , -19 )
UpperCamelCase : Optional[int] = skip_list.head
UpperCamelCase : List[str] = {}
while node.level != 0:
UpperCamelCase : str = node.forward[0]
UpperCamelCase : Optional[int] = node.value
assert len(_lowerCAmelCase ) == 4
assert all_values["Key1"] == 3
assert all_values["Key2"] == 12
assert all_values["Key3"] == 41
assert all_values["Key4"] == -19
def A_ ( ) -> List[Any]:
UpperCamelCase : Optional[int] = SkipList()
skip_list.insert("Key1" , 10 )
skip_list.insert("Key1" , 12 )
skip_list.insert("Key5" , 7 )
skip_list.insert("Key7" , 10 )
skip_list.insert("Key10" , 5 )
skip_list.insert("Key7" , 7 )
skip_list.insert("Key5" , 5 )
skip_list.insert("Key10" , 10 )
UpperCamelCase : Dict = skip_list.head
UpperCamelCase : Tuple = {}
while node.level != 0:
UpperCamelCase : List[str] = node.forward[0]
UpperCamelCase : Dict = node.value
if len(_lowerCAmelCase ) != 4:
print()
assert len(_lowerCAmelCase ) == 4
assert all_values["Key1"] == 12
assert all_values["Key7"] == 7
assert all_values["Key5"] == 5
assert all_values["Key10"] == 10
def A_ ( ) -> List[Any]:
UpperCamelCase : List[Any] = SkipList()
assert skip_list.find("Some key" ) is None
def A_ ( ) -> Tuple:
UpperCamelCase : Optional[int] = SkipList()
skip_list.insert("Key2" , 20 )
assert skip_list.find("Key2" ) == 20
skip_list.insert("Some Key" , 10 )
skip_list.insert("Key2" , 8 )
skip_list.insert("V" , 13 )
assert skip_list.find("Y" ) is None
assert skip_list.find("Key2" ) == 8
assert skip_list.find("Some Key" ) == 10
assert skip_list.find("V" ) == 13
def A_ ( ) -> Dict:
UpperCamelCase : Optional[int] = SkipList()
skip_list.delete("Some key" )
assert len(skip_list.head.forward ) == 0
def A_ ( ) -> Dict:
UpperCamelCase : List[Any] = SkipList()
skip_list.insert("Key1" , 12 )
skip_list.insert("V" , 13 )
skip_list.insert("X" , 14 )
skip_list.insert("Key2" , 15 )
skip_list.delete("V" )
skip_list.delete("Key2" )
assert skip_list.find("V" ) is None
assert skip_list.find("Key2" ) is None
def A_ ( ) -> List[str]:
UpperCamelCase : int = SkipList()
skip_list.insert("Key1" , 12 )
skip_list.insert("V" , 13 )
skip_list.insert("X" , 14 )
skip_list.insert("Key2" , 15 )
skip_list.delete("V" )
assert skip_list.find("V" ) is None
assert skip_list.find("X" ) == 14
assert skip_list.find("Key1" ) == 12
assert skip_list.find("Key2" ) == 15
skip_list.delete("X" )
assert skip_list.find("V" ) is None
assert skip_list.find("X" ) is None
assert skip_list.find("Key1" ) == 12
assert skip_list.find("Key2" ) == 15
skip_list.delete("Key1" )
assert skip_list.find("V" ) is None
assert skip_list.find("X" ) is None
assert skip_list.find("Key1" ) is None
assert skip_list.find("Key2" ) == 15
skip_list.delete("Key2" )
assert skip_list.find("V" ) is None
assert skip_list.find("X" ) is None
assert skip_list.find("Key1" ) is None
assert skip_list.find("Key2" ) is None
def A_ ( ) -> List[Any]:
UpperCamelCase : List[Any] = SkipList()
skip_list.insert("Key1" , 12 )
skip_list.insert("V" , 13 )
skip_list.insert("X" , 142 )
skip_list.insert("Key2" , 15 )
skip_list.delete("X" )
def traverse_keys(_lowerCAmelCase ):
yield node.key
for forward_node in node.forward:
yield from traverse_keys(_lowerCAmelCase )
assert len(set(traverse_keys(skip_list.head ) ) ) == 4
def A_ ( ) -> Union[str, Any]:
def is_sorted(_lowerCAmelCase ):
return all(next_item >= item for item, next_item in zip(_lowerCAmelCase , lst[1:] ) )
UpperCamelCase : int = SkipList()
for i in range(10 ):
skip_list.insert(_lowerCAmelCase , _lowerCAmelCase )
assert is_sorted(list(_lowerCAmelCase ) )
skip_list.delete(5 )
skip_list.delete(8 )
skip_list.delete(2 )
assert is_sorted(list(_lowerCAmelCase ) )
skip_list.insert(-12 , -12 )
skip_list.insert(77 , 77 )
assert is_sorted(list(_lowerCAmelCase ) )
def A_ ( ) -> Tuple:
for _ in range(100 ):
# Repeat test 100 times due to the probabilistic nature of skip list
# random values == random bugs
test_insert()
test_insert_overrides_existing_value()
test_searching_empty_list_returns_none()
test_search()
test_deleting_item_from_empty_list_do_nothing()
test_deleted_items_are_not_founded_by_find_method()
test_delete_removes_only_given_key()
test_delete_doesnt_leave_dead_nodes()
test_iter_always_yields_sorted_values()
def A_ ( ) -> List[str]:
UpperCamelCase : Optional[int] = SkipList()
skip_list.insert(2 , "2" )
skip_list.insert(4 , "4" )
skip_list.insert(6 , "4" )
skip_list.insert(4 , "5" )
skip_list.insert(8 , "4" )
skip_list.insert(9 , "4" )
skip_list.delete(4 )
print(_lowerCAmelCase )
if __name__ == "__main__":
import doctest
doctest.testmod()
main()
| 38
| 1
|
from math import log
from scipy.constants import Boltzmann, physical_constants
SCREAMING_SNAKE_CASE__ : Optional[int] = 3_00 # TEMPERATURE (unit = K)
def _lowerCamelCase ( __lowerCamelCase , __lowerCamelCase , __lowerCamelCase , ) -> float:
'''simple docstring'''
if donor_conc <= 0:
raise ValueError("""Donor concentration should be positive""" )
elif acceptor_conc <= 0:
raise ValueError("""Acceptor concentration should be positive""" )
elif intrinsic_conc <= 0:
raise ValueError("""Intrinsic concentration should be positive""" )
elif donor_conc <= intrinsic_conc:
raise ValueError(
"""Donor concentration should be greater than intrinsic concentration""" )
elif acceptor_conc <= intrinsic_conc:
raise ValueError(
"""Acceptor concentration should be greater than intrinsic concentration""" )
else:
return (
Boltzmann
* T
* log((donor_conc * acceptor_conc) / intrinsic_conc**2 )
/ physical_constants["electron volt"][0]
)
if __name__ == "__main__":
import doctest
doctest.testmod()
| 79
|
import json
import os
import tempfile
import unittest
import numpy as np
from datasets import load_dataset
from transformers.testing_utils import require_torch, require_vision, slow
from transformers.utils import is_torch_available, is_vision_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin
if is_torch_available():
import torch
if is_vision_available():
from PIL import Image
from transformers import ImageGPTImageProcessor
class UpperCAmelCase_ ( unittest.TestCase ):
def __init__( self , _lowerCAmelCase , _lowerCAmelCase=7 , _lowerCAmelCase=3 , _lowerCAmelCase=18 , _lowerCAmelCase=30 , _lowerCAmelCase=400 , _lowerCAmelCase=True , _lowerCAmelCase=None , _lowerCAmelCase=True , ):
UpperCAmelCase__ : List[str] = size if size is not None else {"""height""": 18, """width""": 18}
UpperCAmelCase__ : Union[str, Any] = parent
UpperCAmelCase__ : int = batch_size
UpperCAmelCase__ : Tuple = num_channels
UpperCAmelCase__ : Dict = image_size
UpperCAmelCase__ : List[Any] = min_resolution
UpperCAmelCase__ : str = max_resolution
UpperCAmelCase__ : Union[str, Any] = do_resize
UpperCAmelCase__ : Tuple = size
UpperCAmelCase__ : int = do_normalize
def __UpperCAmelCase ( self ):
return {
# here we create 2 clusters for the sake of simplicity
"clusters": np.asarray(
[
[0.8_8_6_6_4_4_3_6_3_4_0_3_3_2_0_3, 0.6_6_1_8_8_2_9_3_6_9_5_4_4_9_8_3, 0.3_8_9_1_7_4_6_4_0_1_7_8_6_8_0_4],
[-0.6_0_4_2_5_5_9_1_4_6_8_8_1_1_0_4, -0.0_2_2_9_5_0_0_8_8_6_0_5_2_8_4_6_9, 0.5_4_2_3_7_9_7_3_6_9_0_0_3_2_9_6],
] ),
"do_resize": self.do_resize,
"size": self.size,
"do_normalize": self.do_normalize,
}
@require_torch
@require_vision
class UpperCAmelCase_ ( __lowerCamelCase , unittest.TestCase ):
__lowerCamelCase = ImageGPTImageProcessor if is_vision_available() else None
def __UpperCAmelCase ( self ):
UpperCAmelCase__ : Any = ImageGPTImageProcessingTester(self )
@property
def __UpperCAmelCase ( self ):
return self.image_processor_tester.prepare_image_processor_dict()
def __UpperCAmelCase ( self ):
UpperCAmelCase__ : str = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(_lowerCAmelCase , """clusters""" ) )
self.assertTrue(hasattr(_lowerCAmelCase , """do_resize""" ) )
self.assertTrue(hasattr(_lowerCAmelCase , """size""" ) )
self.assertTrue(hasattr(_lowerCAmelCase , """do_normalize""" ) )
def __UpperCAmelCase ( self ):
UpperCAmelCase__ : List[Any] = self.image_processing_class.from_dict(self.image_processor_dict )
self.assertEqual(image_processor.size , {"""height""": 18, """width""": 18} )
UpperCAmelCase__ : Optional[Any] = self.image_processing_class.from_dict(self.image_processor_dict , size=42 )
self.assertEqual(image_processor.size , {"""height""": 42, """width""": 42} )
def __UpperCAmelCase ( self ):
UpperCAmelCase__ : Optional[int] = self.image_processing_class(**self.image_processor_dict )
UpperCAmelCase__ : Optional[int] = json.loads(image_processor.to_json_string() )
for key, value in self.image_processor_dict.items():
if key == "clusters":
self.assertTrue(np.array_equal(_lowerCAmelCase , obj[key] ) )
else:
self.assertEqual(obj[key] , _lowerCAmelCase )
def __UpperCAmelCase ( self ):
UpperCAmelCase__ : Optional[int] = self.image_processing_class(**self.image_processor_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
UpperCAmelCase__ : Union[str, Any] = os.path.join(_lowerCAmelCase , """image_processor.json""" )
image_processor_first.to_json_file(_lowerCAmelCase )
UpperCAmelCase__ : Optional[Any] = self.image_processing_class.from_json_file(_lowerCAmelCase ).to_dict()
UpperCAmelCase__ : Dict = image_processor_first.to_dict()
for key, value in image_processor_first.items():
if key == "clusters":
self.assertTrue(np.array_equal(_lowerCAmelCase , image_processor_second[key] ) )
else:
self.assertEqual(image_processor_first[key] , _lowerCAmelCase )
def __UpperCAmelCase ( self ):
UpperCAmelCase__ : str = self.image_processing_class(**self.image_processor_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
image_processor_first.save_pretrained(_lowerCAmelCase )
UpperCAmelCase__ : List[Any] = self.image_processing_class.from_pretrained(_lowerCAmelCase ).to_dict()
UpperCAmelCase__ : Tuple = image_processor_first.to_dict()
for key, value in image_processor_first.items():
if key == "clusters":
self.assertTrue(np.array_equal(_lowerCAmelCase , image_processor_second[key] ) )
else:
self.assertEqual(image_processor_first[key] , _lowerCAmelCase )
@unittest.skip("""ImageGPT requires clusters at initialization""" )
def __UpperCAmelCase ( self ):
pass
def _lowerCamelCase ( ) -> Tuple:
'''simple docstring'''
UpperCAmelCase__ : Any = load_dataset("""hf-internal-testing/fixtures_image_utils""" , split="""test""" )
UpperCAmelCase__ : Dict = Image.open(dataset[4]["""file"""] )
UpperCAmelCase__ : Optional[Any] = Image.open(dataset[5]["""file"""] )
UpperCAmelCase__ : List[Any] = [imagea, imagea]
return images
@require_vision
@require_torch
class UpperCAmelCase_ ( unittest.TestCase ):
@slow
def __UpperCAmelCase ( self ):
UpperCAmelCase__ : Tuple = ImageGPTImageProcessor.from_pretrained("""openai/imagegpt-small""" )
UpperCAmelCase__ : int = prepare_images()
# test non-batched
UpperCAmelCase__ : List[str] = image_processing(images[0] , return_tensors="""pt""" )
self.assertIsInstance(encoding.input_ids , torch.LongTensor )
self.assertEqual(encoding.input_ids.shape , (1, 1024) )
UpperCAmelCase__ : List[Any] = [306, 191, 191]
self.assertEqual(encoding.input_ids[0, :3].tolist() , _lowerCAmelCase )
# test batched
UpperCAmelCase__ : List[str] = image_processing(_lowerCAmelCase , return_tensors="""pt""" )
self.assertIsInstance(encoding.input_ids , torch.LongTensor )
self.assertEqual(encoding.input_ids.shape , (2, 1024) )
UpperCAmelCase__ : Any = [303, 13, 13]
self.assertEqual(encoding.input_ids[1, -3:].tolist() , _lowerCAmelCase )
| 79
| 1
|
"""simple docstring"""
from math import loga
def lowercase ( _SCREAMING_SNAKE_CASE : Optional[int] ):
'''simple docstring'''
if a < 0:
raise ValueError('''Input value must be a positive integer''' )
elif isinstance(_lowerCAmelCase , _lowerCAmelCase ):
raise TypeError('''Input value must be a \'int\' type''' )
return 0 if (a == 0) else int(loga(a & -a ) )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 716
|
"""simple docstring"""
import argparse
import json
import requests
import timm
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import AutoImageProcessor, SwinConfig, SwinForImageClassification
def lowercase ( _SCREAMING_SNAKE_CASE : Optional[Any] ):
'''simple docstring'''
_UpperCAmelCase = SwinConfig()
_UpperCAmelCase = swin_name.split('''_''' )
_UpperCAmelCase = name_split[1]
_UpperCAmelCase = int(name_split[4] )
_UpperCAmelCase = int(name_split[3][-1] )
if model_size == "tiny":
_UpperCAmelCase = 96
_UpperCAmelCase = (2, 2, 6, 2)
_UpperCAmelCase = (3, 6, 12, 24)
elif model_size == "small":
_UpperCAmelCase = 96
_UpperCAmelCase = (2, 2, 18, 2)
_UpperCAmelCase = (3, 6, 12, 24)
elif model_size == "base":
_UpperCAmelCase = 128
_UpperCAmelCase = (2, 2, 18, 2)
_UpperCAmelCase = (4, 8, 16, 32)
else:
_UpperCAmelCase = 192
_UpperCAmelCase = (2, 2, 18, 2)
_UpperCAmelCase = (6, 12, 24, 48)
if "in22k" in swin_name:
_UpperCAmelCase = 2_1841
else:
_UpperCAmelCase = 1000
_UpperCAmelCase = '''huggingface/label-files'''
_UpperCAmelCase = '''imagenet-1k-id2label.json'''
_UpperCAmelCase = json.load(open(hf_hub_download(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , repo_type='''dataset''' ) , '''r''' ) )
_UpperCAmelCase = {int(_SCREAMING_SNAKE_CASE ): v for k, v in idalabel.items()}
_UpperCAmelCase = idalabel
_UpperCAmelCase = {v: k for k, v in idalabel.items()}
_UpperCAmelCase = img_size
_UpperCAmelCase = num_classes
_UpperCAmelCase = embed_dim
_UpperCAmelCase = depths
_UpperCAmelCase = num_heads
_UpperCAmelCase = window_size
return config
def lowercase ( _SCREAMING_SNAKE_CASE : int ):
'''simple docstring'''
if "patch_embed.proj" in name:
_UpperCAmelCase = name.replace('''patch_embed.proj''' , '''embeddings.patch_embeddings.projection''' )
if "patch_embed.norm" in name:
_UpperCAmelCase = name.replace('''patch_embed.norm''' , '''embeddings.norm''' )
if "layers" in name:
_UpperCAmelCase = '''encoder.''' + name
if "attn.proj" in name:
_UpperCAmelCase = name.replace('''attn.proj''' , '''attention.output.dense''' )
if "attn" in name:
_UpperCAmelCase = name.replace('''attn''' , '''attention.self''' )
if "norm1" in name:
_UpperCAmelCase = name.replace('''norm1''' , '''layernorm_before''' )
if "norm2" in name:
_UpperCAmelCase = name.replace('''norm2''' , '''layernorm_after''' )
if "mlp.fc1" in name:
_UpperCAmelCase = name.replace('''mlp.fc1''' , '''intermediate.dense''' )
if "mlp.fc2" in name:
_UpperCAmelCase = name.replace('''mlp.fc2''' , '''output.dense''' )
if name == "norm.weight":
_UpperCAmelCase = '''layernorm.weight'''
if name == "norm.bias":
_UpperCAmelCase = '''layernorm.bias'''
if "head" in name:
_UpperCAmelCase = name.replace('''head''' , '''classifier''' )
else:
_UpperCAmelCase = '''swin.''' + name
return name
def lowercase ( _SCREAMING_SNAKE_CASE : Union[str, Any] , _SCREAMING_SNAKE_CASE : Dict ):
'''simple docstring'''
for key in orig_state_dict.copy().keys():
_UpperCAmelCase = orig_state_dict.pop(_SCREAMING_SNAKE_CASE )
if "mask" in key:
continue
elif "qkv" in key:
_UpperCAmelCase = key.split('''.''' )
_UpperCAmelCase = int(key_split[1] )
_UpperCAmelCase = int(key_split[3] )
_UpperCAmelCase = model.swin.encoder.layers[layer_num].blocks[block_num].attention.self.all_head_size
if "weight" in key:
_UpperCAmelCase = val[:dim, :]
_UpperCAmelCase = val[
dim : dim * 2, :
]
_UpperCAmelCase = val[-dim:, :]
else:
_UpperCAmelCase = val[
:dim
]
_UpperCAmelCase = val[
dim : dim * 2
]
_UpperCAmelCase = val[
-dim:
]
else:
_UpperCAmelCase = val
return orig_state_dict
def lowercase ( _SCREAMING_SNAKE_CASE : int , _SCREAMING_SNAKE_CASE : int ):
'''simple docstring'''
_UpperCAmelCase = timm.create_model(_SCREAMING_SNAKE_CASE , pretrained=_SCREAMING_SNAKE_CASE )
timm_model.eval()
_UpperCAmelCase = get_swin_config(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = SwinForImageClassification(_SCREAMING_SNAKE_CASE )
model.eval()
_UpperCAmelCase = convert_state_dict(timm_model.state_dict() , _SCREAMING_SNAKE_CASE )
model.load_state_dict(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = '''http://images.cocodataset.org/val2017/000000039769.jpg'''
_UpperCAmelCase = AutoImageProcessor.from_pretrained('''microsoft/{}'''.format(swin_name.replace('''_''' , '''-''' ) ) )
_UpperCAmelCase = Image.open(requests.get(_SCREAMING_SNAKE_CASE , stream=_SCREAMING_SNAKE_CASE ).raw )
_UpperCAmelCase = image_processor(images=_SCREAMING_SNAKE_CASE , return_tensors='''pt''' )
_UpperCAmelCase = timm_model(inputs['''pixel_values'''] )
_UpperCAmelCase = model(**_SCREAMING_SNAKE_CASE ).logits
assert torch.allclose(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , atol=1E-3 )
print(f'Saving model {swin_name} to {pytorch_dump_folder_path}' )
model.save_pretrained(_SCREAMING_SNAKE_CASE )
print(f'Saving image processor to {pytorch_dump_folder_path}' )
image_processor.save_pretrained(_SCREAMING_SNAKE_CASE )
if __name__ == "__main__":
__A : int = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--swin_name",
default="swin_tiny_patch4_window7_224",
type=str,
help="Name of the Swin timm model you'd like to convert.",
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, help="Path to the output PyTorch model directory."
)
__A : Tuple = parser.parse_args()
convert_swin_checkpoint(args.swin_name, args.pytorch_dump_folder_path)
| 95
| 0
|
def _SCREAMING_SNAKE_CASE ( __snake_case ) -> Any:
_UpperCAmelCase = len(a_ )
while cur > 1:
# Find the maximum number in arr
_UpperCAmelCase = arr.index(max(arr[0:cur] ) )
# Reverse from 0 to mi
_UpperCAmelCase = arr[mi::-1] + arr[mi + 1 : len(a_ )]
# Reverse whole list
_UpperCAmelCase = arr[cur - 1 :: -1] + arr[cur : len(a_ )]
cur -= 1
return arr
if __name__ == "__main__":
__a: Dict = input('''Enter numbers separated by a comma:\n''').strip()
__a: Optional[Any] = [int(item) for item in user_input.split(''',''')]
print(pancake_sort(unsorted))
| 108
|
'''simple docstring'''
import json
import os
from datetime import date
from pathlib import Path
from tabulate import DataRow, TableFormat, tabulate
UpperCamelCase =TableFormat(
lineabove=None,
linebelowheader=None,
linebetweenrows=None,
linebelow=None,
headerrow=DataRow("", "|", "|"),
datarow=DataRow("", "|", "|"),
padding=1,
with_header_hide=None,
)
UpperCamelCase =[]
UpperCamelCase =[]
UpperCamelCase ={"type": "section", "text": {"type": "plain_text", "text": "No failed tests! 🤗", "emoji": True}}
UpperCamelCase =[
{
"type": "header",
"text": {
"type": "plain_text",
"text": f"🤗 Accelerate nightly {os.environ.get('TEST_TYPE', '')} test results",
"emoji": True,
},
}
]
UpperCamelCase =0
for log in Path().glob("*.log"):
UpperCamelCase =0
with open(log, "r") as f:
for line in f:
UpperCamelCase =json.loads(line)
if line.get("nodeid", "") != "":
UpperCamelCase =line["nodeid"]
if line.get("duration", None) is not None:
UpperCamelCase =f"{line['duration']:.4f}"
if line.get("outcome", "") == "failed":
section_num_failed += 1
failed.append([test, duration, log.name.split("_")[0]])
total_num_failed += 1
group_info.append([str(log), section_num_failed, failed])
UpperCamelCase =[]
log.unlink()
UpperCamelCase =""
UpperCamelCase =[]
if total_num_failed > 0:
for name, num_failed, failed_tests in group_info:
if num_failed > 0:
if num_failed == 1:
message += f"*{name[1:]}: {num_failed} failed test*\n"
else:
message += f"*{name[1:]}: {num_failed} failed tests*\n"
UpperCamelCase =[]
UpperCamelCase ={}
for test in failed_tests:
UpperCamelCase =test[0].split("::")
UpperCamelCase =data[0].split("/")[-1]
if data[0] not in filesafailed:
UpperCamelCase =[data[1:]]
else:
filesafailed[data[0]] += [data[1:]]
failed_table.append(data)
UpperCamelCase =[test[0] for test in failed_table]
UpperCamelCase =list(set(files))
# Count number of instances in failed_tests
UpperCamelCase =[]
for file in individual_files:
table.append([file, len(filesafailed[file])])
UpperCamelCase =tabulate(
table,
headers=["Test Location", "Num Failed"],
tablefmt=hf_table_format,
stralign="right",
)
message += f"\n```\n{failed_table}\n```"
all_filesafailed.append(filesafailed)
if len(message) > 3000:
UpperCamelCase ="Too many failed tests, please see the full report in the Action results."
UpperCamelCase =len(err) + 10
UpperCamelCase =message[: 3000 - offset] + f"\n...\n```\n{err}"
print(f"### {message}")
else:
UpperCamelCase ="No failed tests! 🤗"
print(f"## {message}")
payload.append(no_error_payload)
if os.environ.get("TEST_TYPE", "") != "":
from slack_sdk import WebClient
UpperCamelCase =WebClient(token=os.environ["SLACK_API_TOKEN"])
if message != "No failed tests! 🤗":
UpperCamelCase ={
"type": "section",
"text": {
"type": "mrkdwn",
"text": message,
},
}
payload.append(md_report)
UpperCamelCase ={
"type": "section",
"text": {
"type": "mrkdwn",
"text": "*For more details:*",
},
"accessory": {
"type": "button",
"text": {
"type": "plain_text",
"text": "Check Action results",
"emoji": True,
},
"url": f"https://github.com/{os.environ['GITHUB_REPOSITORY']}/actions/runs/{os.environ['GITHUB_RUN_ID']}",
},
}
payload.append(action_button)
UpperCamelCase ={
"type": "context",
"elements": [
{
"type": "plain_text",
"text": f"Nightly {os.environ.get('TEST_TYPE')} test results for {date.today()}",
}
],
}
payload.append(date_report)
UpperCamelCase =client.chat_postMessage(channel="#accelerate-ci-daily", text=message, blocks=payload)
UpperCamelCase =response.data["ts"]
for failed_file in all_filesafailed:
for test_location, test_failures in failed_file.items():
# Keep only the first instance of the test name
UpperCamelCase =""
for i, row in enumerate(test_failures):
if row[0] != test_class:
UpperCamelCase =row[0]
else:
UpperCamelCase =""
UpperCamelCase ={
"type": "section",
"text": {
"type": "mrkdwn",
"text": f"Test location: {test_location}\n```\n{tabulate(test_failures, headers=['Class', 'Test'], tablefmt=hf_table_format, stralign='right')}\n```",
},
}
client.chat_postMessage(
channel="#accelerate-ci-daily",
thread_ts=ts,
blocks=[payload],
)
| 208
| 0
|
from typing import Dict, List, Optional, Tuple, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
flip_channel_order,
get_resize_output_image_size,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_torch_available, is_torch_tensor, is_vision_available, logging
if is_vision_available():
import PIL
if is_torch_available():
import torch
_snake_case : List[Any] = logging.get_logger(__name__)
class a (_lowerCAmelCase ):
"""simple docstring"""
__UpperCAmelCase : List[str] = ["pixel_values"]
def __init__( self : Optional[int] , lowerCamelCase : bool = True , lowerCamelCase : Dict[str, int] = None , lowerCamelCase : PILImageResampling = PILImageResampling.BILINEAR , lowerCamelCase : bool = True , lowerCamelCase : Union[int, float] = 1 / 255 , lowerCamelCase : bool = True , lowerCamelCase : Dict[str, int] = None , lowerCamelCase : bool = True , **lowerCamelCase : Optional[Any] , ) -> None:
super().__init__(**lowerCamelCase )
__snake_case : Optional[Any] = size if size is not None else {"shortest_edge": 224}
__snake_case : List[str] = get_size_dict(lowerCamelCase , default_to_square=lowerCamelCase )
__snake_case : Union[str, Any] = crop_size if crop_size is not None else {"height": 256, "width": 256}
__snake_case : Dict = get_size_dict(lowerCamelCase , param_name="crop_size" )
__snake_case : Union[str, Any] = do_resize
__snake_case : Optional[Any] = size
__snake_case : int = resample
__snake_case : Tuple = do_rescale
__snake_case : Any = rescale_factor
__snake_case : int = do_center_crop
__snake_case : Union[str, Any] = crop_size
__snake_case : Optional[Any] = do_flip_channel_order
def __snake_case ( self : int , lowerCamelCase : np.ndarray , lowerCamelCase : Dict[str, int] , lowerCamelCase : PILImageResampling = PIL.Image.BILINEAR , lowerCamelCase : Optional[Union[str, ChannelDimension]] = None , **lowerCamelCase : Dict , ) -> np.ndarray:
__snake_case : Union[str, Any] = get_size_dict(lowerCamelCase , default_to_square=lowerCamelCase )
if "shortest_edge" not in size:
raise ValueError(F'The `size` dictionary must contain the key `shortest_edge`. Got {size.keys()}' )
__snake_case : List[str] = get_resize_output_image_size(lowerCamelCase , size=size["shortest_edge"] , default_to_square=lowerCamelCase )
return resize(lowerCamelCase , size=lowerCamelCase , resample=lowerCamelCase , data_format=lowerCamelCase , **lowerCamelCase )
def __snake_case ( self : Any , lowerCamelCase : np.ndarray , lowerCamelCase : Dict[str, int] , lowerCamelCase : Optional[Union[str, ChannelDimension]] = None , **lowerCamelCase : Dict , ) -> np.ndarray:
__snake_case : Optional[int] = get_size_dict(lowerCamelCase )
if "height" not in size or "width" not in size:
raise ValueError(F'The `size` dictionary must contain the keys `height` and `width`. Got {size.keys()}' )
return center_crop(lowerCamelCase , size=(size["height"], size["width"]) , data_format=lowerCamelCase , **lowerCamelCase )
def __snake_case ( self : List[Any] , lowerCamelCase : np.ndarray , lowerCamelCase : Union[int, float] , lowerCamelCase : Optional[Union[str, ChannelDimension]] = None , **lowerCamelCase : Optional[int] , ) -> Union[str, Any]:
return rescale(lowerCamelCase , scale=lowerCamelCase , data_format=lowerCamelCase , **lowerCamelCase )
def __snake_case ( self : Tuple , lowerCamelCase : np.ndarray , lowerCamelCase : Optional[Union[str, ChannelDimension]] = None ) -> np.ndarray:
return flip_channel_order(lowerCamelCase , data_format=lowerCamelCase )
def __snake_case ( self : Any , lowerCamelCase : ImageInput , lowerCamelCase : bool = None , lowerCamelCase : Dict[str, int] = None , lowerCamelCase : PILImageResampling = None , lowerCamelCase : bool = None , lowerCamelCase : float = None , lowerCamelCase : bool = None , lowerCamelCase : Dict[str, int] = None , lowerCamelCase : bool = None , lowerCamelCase : Optional[Union[str, TensorType]] = None , lowerCamelCase : ChannelDimension = ChannelDimension.FIRST , **lowerCamelCase : str , ) -> PIL.Image.Image:
__snake_case : str = do_resize if do_resize is not None else self.do_resize
__snake_case : int = resample if resample is not None else self.resample
__snake_case : Optional[Any] = do_rescale if do_rescale is not None else self.do_rescale
__snake_case : Dict = rescale_factor if rescale_factor is not None else self.rescale_factor
__snake_case : int = do_center_crop if do_center_crop is not None else self.do_center_crop
__snake_case : Optional[int] = (
do_flip_channel_order if do_flip_channel_order is not None else self.do_flip_channel_order
)
__snake_case : str = size if size is not None else self.size
__snake_case : List[str] = get_size_dict(lowerCamelCase , default_to_square=lowerCamelCase )
__snake_case : Union[str, Any] = crop_size if crop_size is not None else self.crop_size
__snake_case : Optional[Any] = get_size_dict(lowerCamelCase , param_name="crop_size" )
__snake_case : Optional[Any] = make_list_of_images(lowerCamelCase )
if not valid_images(lowerCamelCase ):
raise ValueError(
"Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, "
"torch.Tensor, tf.Tensor or jax.ndarray." )
if do_resize and size is None:
raise ValueError("Size must be specified if do_resize is True." )
if do_rescale and rescale_factor is None:
raise ValueError("Rescale factor must be specified if do_rescale is True." )
if do_center_crop and crop_size is None:
raise ValueError("Crop size must be specified if do_center_crop is True." )
# All transformations expect numpy arrays.
__snake_case : List[str] = [to_numpy_array(lowerCamelCase ) for image in images]
if do_resize:
__snake_case : Optional[Any] = [self.resize(image=lowerCamelCase , size=lowerCamelCase , resample=lowerCamelCase ) for image in images]
if do_center_crop:
__snake_case : Tuple = [self.center_crop(image=lowerCamelCase , size=lowerCamelCase ) for image in images]
if do_rescale:
__snake_case : List[Any] = [self.rescale(image=lowerCamelCase , scale=lowerCamelCase ) for image in images]
# the pretrained checkpoints assume images are BGR, not RGB
if do_flip_channel_order:
__snake_case : Optional[Any] = [self.flip_channel_order(image=lowerCamelCase ) for image in images]
__snake_case : List[str] = [to_channel_dimension_format(lowerCamelCase , lowerCamelCase ) for image in images]
__snake_case : List[Any] = {"pixel_values": images}
return BatchFeature(data=lowerCamelCase , tensor_type=lowerCamelCase )
def __snake_case ( self : Optional[Any] , lowerCamelCase : List[str] , lowerCamelCase : List[Tuple] = None ) -> Optional[Any]:
__snake_case : Tuple = outputs.logits
# Resize logits and compute semantic segmentation maps
if target_sizes is not None:
if len(lowerCamelCase ) != len(lowerCamelCase ):
raise ValueError(
"Make sure that you pass in as many target sizes as the batch dimension of the logits" )
if is_torch_tensor(lowerCamelCase ):
__snake_case : List[str] = target_sizes.numpy()
__snake_case : Optional[Any] = []
for idx in range(len(lowerCamelCase ) ):
__snake_case : Tuple = torch.nn.functional.interpolate(
logits[idx].unsqueeze(dim=0 ) , size=target_sizes[idx] , mode="bilinear" , align_corners=lowerCamelCase )
__snake_case : List[str] = resized_logits[0].argmax(dim=0 )
semantic_segmentation.append(lowerCamelCase )
else:
__snake_case : Optional[int] = logits.argmax(dim=1 )
__snake_case : List[Any] = [semantic_segmentation[i] for i in range(semantic_segmentation.shape[0] )]
return semantic_segmentation
| 203
|
import inspect
import unittest
from math import floor
from transformers import CvtConfig
from transformers.file_utils import cached_property, is_torch_available, is_vision_available
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import CvtForImageClassification, CvtModel
from transformers.models.cvt.modeling_cvt import CVT_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import AutoImageProcessor
class a (_lowerCAmelCase ):
"""simple docstring"""
def __snake_case ( self : str ) -> Dict:
__snake_case : Any = self.config_class(**self.inputs_dict )
self.parent.assertTrue(hasattr(lowerCamelCase , "embed_dim" ) )
self.parent.assertTrue(hasattr(lowerCamelCase , "num_heads" ) )
class a :
"""simple docstring"""
def __init__( self : str , lowerCamelCase : Union[str, Any] , lowerCamelCase : Any=13 , lowerCamelCase : Any=64 , lowerCamelCase : int=3 , lowerCamelCase : Tuple=[16, 48, 96] , lowerCamelCase : Optional[int]=[1, 3, 6] , lowerCamelCase : List[str]=[1, 2, 10] , lowerCamelCase : Any=[7, 3, 3] , lowerCamelCase : Any=[4, 2, 2] , lowerCamelCase : Optional[Any]=[2, 1, 1] , lowerCamelCase : str=[2, 2, 2] , lowerCamelCase : Dict=[False, False, True] , lowerCamelCase : Dict=[0.0, 0.0, 0.0] , lowerCamelCase : Optional[Any]=0.02 , lowerCamelCase : Tuple=1E-12 , lowerCamelCase : Union[str, Any]=True , lowerCamelCase : Union[str, Any]=True , lowerCamelCase : List[str]=2 , ) -> Optional[int]:
__snake_case : Tuple = parent
__snake_case : List[str] = batch_size
__snake_case : Optional[int] = image_size
__snake_case : Optional[int] = patch_sizes
__snake_case : Union[str, Any] = patch_stride
__snake_case : int = patch_padding
__snake_case : Optional[Any] = is_training
__snake_case : Optional[Any] = use_labels
__snake_case : Tuple = num_labels
__snake_case : Union[str, Any] = num_channels
__snake_case : Tuple = embed_dim
__snake_case : List[str] = num_heads
__snake_case : Dict = stride_kv
__snake_case : Optional[int] = depth
__snake_case : List[Any] = cls_token
__snake_case : List[Any] = attention_drop_rate
__snake_case : Any = initializer_range
__snake_case : str = layer_norm_eps
def __snake_case ( self : List[str] ) -> str:
__snake_case : str = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
__snake_case : Tuple = None
if self.use_labels:
__snake_case : Tuple = ids_tensor([self.batch_size] , self.num_labels )
__snake_case : Optional[Any] = self.get_config()
return config, pixel_values, labels
def __snake_case ( self : Tuple ) -> Optional[Any]:
return CvtConfig(
image_size=self.image_size , num_labels=self.num_labels , num_channels=self.num_channels , embed_dim=self.embed_dim , num_heads=self.num_heads , patch_sizes=self.patch_sizes , patch_padding=self.patch_padding , patch_stride=self.patch_stride , stride_kv=self.stride_kv , depth=self.depth , cls_token=self.cls_token , attention_drop_rate=self.attention_drop_rate , initializer_range=self.initializer_range , )
def __snake_case ( self : Optional[int] , lowerCamelCase : Union[str, Any] , lowerCamelCase : str , lowerCamelCase : str ) -> Tuple:
__snake_case : int = CvtModel(config=lowerCamelCase )
model.to(lowerCamelCase )
model.eval()
__snake_case : List[str] = model(lowerCamelCase )
__snake_case : List[Any] = (self.image_size, self.image_size)
__snake_case , __snake_case : List[str] = image_size[0], image_size[1]
for i in range(len(self.depth ) ):
__snake_case : int = floor(((height + 2 * self.patch_padding[i] - self.patch_sizes[i]) / self.patch_stride[i]) + 1 )
__snake_case : Union[str, Any] = floor(((width + 2 * self.patch_padding[i] - self.patch_sizes[i]) / self.patch_stride[i]) + 1 )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.embed_dim[-1], height, width) )
def __snake_case ( self : List[str] , lowerCamelCase : Optional[int] , lowerCamelCase : Dict , lowerCamelCase : int ) -> Optional[Any]:
__snake_case : str = self.num_labels
__snake_case : Any = CvtForImageClassification(lowerCamelCase )
model.to(lowerCamelCase )
model.eval()
__snake_case : List[Any] = model(lowerCamelCase , labels=lowerCamelCase )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def __snake_case ( self : Union[str, Any] ) -> List[Any]:
__snake_case : List[Any] = self.prepare_config_and_inputs()
__snake_case , __snake_case , __snake_case : str = config_and_inputs
__snake_case : List[str] = {"pixel_values": pixel_values}
return config, inputs_dict
@require_torch
class a (_lowerCAmelCase , _lowerCAmelCase , unittest.TestCase ):
"""simple docstring"""
__UpperCAmelCase : List[str] = (CvtModel, CvtForImageClassification) if is_torch_available() else ()
__UpperCAmelCase : List[str] = (
{"feature-extraction": CvtModel, "image-classification": CvtForImageClassification}
if is_torch_available()
else {}
)
__UpperCAmelCase : List[str] = False
__UpperCAmelCase : Tuple = False
__UpperCAmelCase : List[Any] = False
__UpperCAmelCase : Optional[int] = False
__UpperCAmelCase : List[Any] = False
def __snake_case ( self : str ) -> List[str]:
__snake_case : Tuple = CvtModelTester(self )
__snake_case : Tuple = ConfigTester(self , config_class=lowerCamelCase , has_text_modality=lowerCamelCase , hidden_size=37 )
def __snake_case ( self : Union[str, Any] ) -> Union[str, Any]:
self.create_and_test_config_common_properties()
self.config_tester.create_and_test_config_to_json_string()
self.config_tester.create_and_test_config_to_json_file()
self.config_tester.create_and_test_config_from_and_save_pretrained()
self.config_tester.create_and_test_config_with_num_labels()
self.config_tester.check_config_can_be_init_without_params()
self.config_tester.check_config_arguments_init()
def __snake_case ( self : Tuple ) -> Union[str, Any]:
return
@unittest.skip(reason="Cvt does not output attentions" )
def __snake_case ( self : Optional[int] ) -> List[str]:
pass
@unittest.skip(reason="Cvt does not use inputs_embeds" )
def __snake_case ( self : List[str] ) -> Any:
pass
@unittest.skip(reason="Cvt does not support input and output embeddings" )
def __snake_case ( self : int ) -> List[Any]:
pass
def __snake_case ( self : int ) -> int:
__snake_case , __snake_case : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__snake_case : str = model_class(lowerCamelCase )
__snake_case : List[Any] = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
__snake_case : List[str] = [*signature.parameters.keys()]
__snake_case : Dict = ["pixel_values"]
self.assertListEqual(arg_names[:1] , lowerCamelCase )
def __snake_case ( self : str ) -> Dict:
__snake_case : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCamelCase )
def __snake_case ( self : int ) -> Optional[Any]:
def check_hidden_states_output(lowerCamelCase : Dict , lowerCamelCase : Dict , lowerCamelCase : Optional[Any] ):
__snake_case : Any = model_class(lowerCamelCase )
model.to(lowerCamelCase )
model.eval()
with torch.no_grad():
__snake_case : List[str] = model(**self._prepare_for_class(lowerCamelCase , lowerCamelCase ) )
__snake_case : List[Any] = outputs.hidden_states
__snake_case : Optional[int] = len(self.model_tester.depth )
self.assertEqual(len(lowerCamelCase ) , lowerCamelCase )
# verify the first hidden states (first block)
self.assertListEqual(
list(hidden_states[0].shape[-3:] ) , [
self.model_tester.embed_dim[0],
self.model_tester.image_size // 4,
self.model_tester.image_size // 4,
] , )
__snake_case , __snake_case : List[str] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
__snake_case : Union[str, Any] = True
check_hidden_states_output(lowerCamelCase , lowerCamelCase , lowerCamelCase )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
__snake_case : List[str] = True
check_hidden_states_output(lowerCamelCase , lowerCamelCase , lowerCamelCase )
def __snake_case ( self : Optional[int] ) -> Optional[Any]:
__snake_case : Any = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowerCamelCase )
@unittest.skip("Will be fixed soon by reducing the size of the model used for common tests." )
def __snake_case ( self : List[str] ) -> Optional[int]:
pass
@slow
def __snake_case ( self : Tuple ) -> List[str]:
for model_name in CVT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
__snake_case : int = CvtModel.from_pretrained(lowerCamelCase )
self.assertIsNotNone(lowerCamelCase )
def lowerCAmelCase_ ( ):
__snake_case : Any = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png" )
return image
@require_torch
@require_vision
class a (unittest.TestCase ):
"""simple docstring"""
@cached_property
def __snake_case ( self : Optional[Any] ) -> Union[str, Any]:
return AutoImageProcessor.from_pretrained(CVT_PRETRAINED_MODEL_ARCHIVE_LIST[0] )
@slow
def __snake_case ( self : int ) -> Tuple:
__snake_case : Optional[Any] = CvtForImageClassification.from_pretrained(CVT_PRETRAINED_MODEL_ARCHIVE_LIST[0] ).to(lowerCamelCase )
__snake_case : Optional[int] = self.default_image_processor
__snake_case : Union[str, Any] = prepare_img()
__snake_case : Dict = image_processor(images=lowerCamelCase , return_tensors="pt" ).to(lowerCamelCase )
# forward pass
with torch.no_grad():
__snake_case : Any = model(**lowerCamelCase )
# verify the logits
__snake_case : Any = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape , lowerCamelCase )
__snake_case : Optional[Any] = torch.tensor([0.92_85, 0.90_15, -0.31_50] ).to(lowerCamelCase )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , lowerCamelCase , atol=1E-4 ) )
| 203
| 1
|
from typing import TYPE_CHECKING
# rely on isort to merge the imports
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
SCREAMING_SNAKE_CASE :Optional[Any] = {
'configuration_autoformer': [
'AUTOFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP',
'AutoformerConfig',
],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
SCREAMING_SNAKE_CASE :str = [
'AUTOFORMER_PRETRAINED_MODEL_ARCHIVE_LIST',
'AutoformerForPrediction',
'AutoformerModel',
'AutoformerPreTrainedModel',
]
if TYPE_CHECKING:
from .configuration_autoformer import (
AUTOFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
AutoformerConfig,
)
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_autoformer import (
AUTOFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
AutoformerForPrediction,
AutoformerModel,
AutoformerPreTrainedModel,
)
else:
import sys
SCREAMING_SNAKE_CASE :int = _LazyModule(__name__, globals()['__file__'], _import_structure, module_spec=__spec__)
| 55
|
import numpy as np
class lowerCamelCase_ :
def __init__( self : Dict ):
'''simple docstring'''
a = (0, 0)
a = None
a = 0
a = 0
a = 0
def __eq__( self : Optional[int] ,__lowerCamelCase : Optional[int] ):
'''simple docstring'''
return self.position == cell.position
def SCREAMING_SNAKE_CASE_ ( self : Tuple ):
'''simple docstring'''
print(self.position )
class lowerCamelCase_ :
def __init__( self : List[str] ,__lowerCamelCase : List[Any]=(5, 5) ):
'''simple docstring'''
a = np.zeros(__lowerCamelCase )
a = world_size[0]
a = world_size[1]
def SCREAMING_SNAKE_CASE_ ( self : Union[str, Any] ):
'''simple docstring'''
print(self.w )
def SCREAMING_SNAKE_CASE_ ( self : List[Any] ,__lowerCamelCase : Union[str, Any] ):
'''simple docstring'''
a = [
(-1, -1),
(-1, 0),
(-1, 1),
(0, -1),
(0, 1),
(1, -1),
(1, 0),
(1, 1),
]
a = cell.position[0]
a = cell.position[1]
a = []
for n in neughbour_cord:
a = current_x + n[0]
a = current_y + n[1]
if 0 <= x < self.world_x_limit and 0 <= y < self.world_y_limit:
a = Cell()
a = (x, y)
a = cell
neighbours.append(__lowerCamelCase )
return neighbours
def SCREAMING_SNAKE_CASE__ ( snake_case_, snake_case_, snake_case_ ) -> str:
"""simple docstring"""
a = []
a = []
_open.append(snake_case_ )
while _open:
a = np.argmin([n.f for n in _open] )
a = _open[min_f]
_closed.append(_open.pop(snake_case_ ) )
if current == goal:
break
for n in world.get_neigbours(snake_case_ ):
for c in _closed:
if c == n:
continue
a = current.g + 1
a , a = n.position
a , a = goal.position
a = (ya - ya) ** 2 + (xa - xa) ** 2
a = n.h + n.g
for c in _open:
if c == n and c.f < n.f:
continue
_open.append(snake_case_ )
a = []
while current.parent is not None:
path.append(current.position )
a = current.parent
path.append(current.position )
return path[::-1]
if __name__ == "__main__":
UpperCamelCase__ : List[str] = Gridworld()
# Start position and goal
UpperCamelCase__ : Optional[int] = Cell()
UpperCamelCase__ : List[str] = (0, 0)
UpperCamelCase__ : Union[str, Any] = Cell()
UpperCamelCase__ : List[str] = (4, 4)
print(F"path from {start.position} to {goal.position}")
UpperCamelCase__ : Union[str, Any] = astar(world, start, goal)
# Just for visual reasons.
for i in s:
UpperCamelCase__ : Union[str, Any] = 1
print(world.w)
| 387
| 0
|
'''simple docstring'''
from __future__ import annotations
from collections.abc import Iterator
from typing import Any
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self , lowerCAmelCase_ ):
'''simple docstring'''
a_ : Any = data
a_ : Node | None = None
class _UpperCAmelCase :
"""simple docstring"""
def __init__( self ):
'''simple docstring'''
a_ : Optional[int] = None
a_ : Optional[Any] = None
def __iter__( self ):
'''simple docstring'''
a_ : Optional[int] = self.head
while self.head:
yield node.data
a_ : Optional[int] = node.next
if node == self.head:
break
def __len__( self ):
'''simple docstring'''
return sum(1 for _ in self )
def __repr__( self ):
'''simple docstring'''
return "->".join(str(lowerCAmelCase_ ) for item in iter(self ) )
def _lowerCAmelCase ( self , lowerCAmelCase_ ):
'''simple docstring'''
self.insert_nth(len(self ) , lowerCAmelCase_ )
def _lowerCAmelCase ( self , lowerCAmelCase_ ):
'''simple docstring'''
self.insert_nth(0 , lowerCAmelCase_ )
def _lowerCAmelCase ( self , lowerCAmelCase_ , lowerCAmelCase_ ):
'''simple docstring'''
if index < 0 or index > len(self ):
raise IndexError("""list index out of range.""" )
a_ : List[Any] = Node(lowerCAmelCase_ )
if self.head is None:
a_ : int = new_node # first node points itself
a_ : int = new_node
elif index == 0: # insert at head
a_ : int = self.head
a_ : Union[str, Any] = new_node
else:
a_ : Optional[Any] = self.head
for _ in range(index - 1 ):
a_ : List[Any] = temp.next
a_ : Tuple = temp.next
a_ : int = new_node
if index == len(self ) - 1: # insert at tail
a_ : Optional[Any] = new_node
def _lowerCAmelCase ( self ):
'''simple docstring'''
return self.delete_nth(0 )
def _lowerCAmelCase ( self ):
'''simple docstring'''
return self.delete_nth(len(self ) - 1 )
def _lowerCAmelCase ( self , lowerCAmelCase_ = 0 ):
'''simple docstring'''
if not 0 <= index < len(self ):
raise IndexError("""list index out of range.""" )
a_ : int = self.head
if self.head == self.tail: # just one node
a_ : Tuple = None
elif index == 0: # delete head node
a_ : int = self.tail.next.next
a_ : str = self.head.next
else:
a_ : Dict = self.head
for _ in range(index - 1 ):
a_ : Union[str, Any] = temp.next
a_ : Dict = temp.next
a_ : Dict = temp.next.next
if index == len(self ) - 1: # delete at tail
a_ : int = temp
return delete_node.data
def _lowerCAmelCase ( self ):
'''simple docstring'''
return len(self ) == 0
def _snake_case ( ):
"""simple docstring"""
a_ : List[Any] = CircularLinkedList()
assert len(A_ ) == 0
assert circular_linked_list.is_empty() is True
assert str(A_ ) == ""
try:
circular_linked_list.delete_front()
raise AssertionError # This should not happen
except IndexError:
assert True # This should happen
try:
circular_linked_list.delete_tail()
raise AssertionError # This should not happen
except IndexError:
assert True # This should happen
try:
circular_linked_list.delete_nth(-1 )
raise AssertionError
except IndexError:
assert True
try:
circular_linked_list.delete_nth(0 )
raise AssertionError
except IndexError:
assert True
assert circular_linked_list.is_empty() is True
for i in range(5 ):
assert len(A_ ) == i
circular_linked_list.insert_nth(A_ , i + 1 )
assert str(A_ ) == "->".join(str(A_ ) for i in range(1 , 6 ) )
circular_linked_list.insert_tail(6 )
assert str(A_ ) == "->".join(str(A_ ) for i in range(1 , 7 ) )
circular_linked_list.insert_head(0 )
assert str(A_ ) == "->".join(str(A_ ) for i in range(0 , 7 ) )
assert circular_linked_list.delete_front() == 0
assert circular_linked_list.delete_tail() == 6
assert str(A_ ) == "->".join(str(A_ ) for i in range(1 , 6 ) )
assert circular_linked_list.delete_nth(2 ) == 3
circular_linked_list.insert_nth(2 , 3 )
assert str(A_ ) == "->".join(str(A_ ) for i in range(1 , 6 ) )
assert circular_linked_list.is_empty() is False
if __name__ == "__main__":
import doctest
doctest.testmod()
| 460
|
'''simple docstring'''
from dataclasses import dataclass
from enum import Enum
from typing import List, Optional, Union
import numpy as np
import PIL
from PIL import Image
from ...utils import BaseOutput, is_torch_available, is_transformers_available
@dataclass
class _UpperCAmelCase ( lowerCAmelCase__ ):
"""simple docstring"""
a_ = 42
a_ = 42
if is_transformers_available() and is_torch_available():
from .pipeline_semantic_stable_diffusion import SemanticStableDiffusionPipeline
| 460
| 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()
snake_case : Any = logging.get_logger(__name__)
snake_case : Optional[Any] = {
'''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''',
}
snake_case : Tuple = [
'''lm_head''',
'''quantizer.weight_proj''',
'''quantizer.codevectors''',
'''project_q''',
'''project_hid''',
]
def __lowerCamelCase ( UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : Optional[int] , UpperCAmelCase_ : str , UpperCAmelCase_ : Any , UpperCAmelCase_ : List[Any] ):
"""simple docstring"""
for attribute in key.split('''.''' ):
a :Union[str, Any] = getattr(UpperCAmelCase_ , UpperCAmelCase_ )
if weight_type is not None:
a :Dict = getattr(UpperCAmelCase_ , UpperCAmelCase_ ).shape
else:
a :List[Any] = 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":
a :Union[str, Any] = value
elif weight_type == "weight_g":
a :Optional[Any] = value
elif weight_type == "weight_v":
a :List[Any] = value
elif weight_type == "bias":
a :Tuple = value
elif weight_type == "running_mean":
a :Any = value
elif weight_type == "running_var":
a :int = value
elif weight_type == "num_batches_tracked":
a :Any = value
elif weight_type == "inv_freq":
a :Tuple = value
else:
a :Optional[int] = value
logger.info(F'''{key + '.' + weight_type if weight_type is not None else ''} was initialized from {full_name}.''' )
def __lowerCamelCase ( UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : List[str] ):
"""simple docstring"""
a :Tuple = []
a :Any = fairseq_model.state_dict()
a :List[Any] = hf_model.wavaveca_conformer.feature_extractor
for name, value in fairseq_dict.items():
a :Dict = False
if "conv_layers" in name:
load_conv_layer(
UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ , hf_model.config.feat_extract_norm == '''group''' , )
a :int = True
else:
for key, mapped_key in MAPPING.items():
a :Tuple = '''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]:
a :Union[str, Any] = True
if "*" in mapped_key:
a :Any = name.split(UpperCAmelCase_ )[0].split('''.''' )[-2]
a :Union[str, Any] = mapped_key.replace('''*''' , UpperCAmelCase_ )
if "pos_bias_u" in name:
a :List[Any] = None
elif "pos_bias_v" in name:
a :Optional[Any] = None
elif "weight_g" in name:
a :Any = '''weight_g'''
elif "weight_v" in name:
a :int = '''weight_v'''
elif "bias" in name:
a :Tuple = '''bias'''
elif "weight" in name:
# TODO: don't match quantizer.weight_proj
a :Union[str, Any] = '''weight'''
elif "running_mean" in name:
a :Dict = '''running_mean'''
elif "inv_freq" in name:
a :Any = '''inv_freq'''
elif "running_var" in name:
a :Dict = '''running_var'''
elif "num_batches_tracked" in name:
a :List[str] = '''num_batches_tracked'''
else:
a :Optional[int] = 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 __lowerCamelCase ( UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Dict , UpperCAmelCase_ : int , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Dict ):
"""simple docstring"""
a :Dict = full_name.split('''conv_layers.''' )[-1]
a :Optional[int] = name.split('''.''' )
a :Dict = int(items[0] )
a :Dict = 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.''' )
a :Optional[Any] = 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.''' )
a :Union[str, Any] = 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.''' )
a :Optional[Any] = 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.''' )
a :Optional[int] = 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 __lowerCamelCase ( UpperCAmelCase_ : Any , UpperCAmelCase_ : Tuple , UpperCAmelCase_ : Union[str, Any]=None , UpperCAmelCase_ : List[str]=None , UpperCAmelCase_ : Tuple=True ):
"""simple docstring"""
if config_path is not None:
a :Union[str, Any] = WavaVecaConformerConfig.from_pretrained(UpperCAmelCase_ , hidden_act='''swish''' )
else:
a :List[Any] = WavaVecaConformerConfig()
if "rope" in checkpoint_path:
a :Tuple = '''rotary'''
if is_finetuned:
if dict_path:
a :Any = Dictionary.load(UpperCAmelCase_ )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
a :str = target_dict.pad_index
a :List[str] = target_dict.bos_index
a :str = target_dict.eos_index
a :Optional[Any] = len(target_dict.symbols )
a :Dict = 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_ )
a :List[str] = target_dict.indices
# fairseq has the <pad> and <s> switched
a :int = 0
a :Any = 1
with open(UpperCAmelCase_ , '''w''' , encoding='''utf-8''' ) as vocab_handle:
json.dump(UpperCAmelCase_ , UpperCAmelCase_ )
a :Optional[int] = 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_ , )
a :List[str] = True if config.feat_extract_norm == '''layer''' else False
a :str = WavaVecaFeatureExtractor(
feature_size=1 , sampling_rate=1_6000 , padding_value=0 , do_normalize=UpperCAmelCase_ , return_attention_mask=UpperCAmelCase_ , )
a :List[str] = WavaVecaProcessor(feature_extractor=UpperCAmelCase_ , tokenizer=UpperCAmelCase_ )
processor.save_pretrained(UpperCAmelCase_ )
a :Optional[int] = WavaVecaConformerForCTC(UpperCAmelCase_ )
else:
a :List[str] = WavaVecaConformerForPreTraining(UpperCAmelCase_ )
if is_finetuned:
a , a , a :Dict = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] , arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
a :Optional[int] = argparse.Namespace(task='''audio_pretraining''' )
a :Optional[int] = fairseq.tasks.setup_task(UpperCAmelCase_ )
a , a , a :List[Any] = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] , task=UpperCAmelCase_ )
a :Tuple = model[0].eval()
recursively_load_weights(UpperCAmelCase_ , UpperCAmelCase_ , not is_finetuned )
hf_wavavec.save_pretrained(UpperCAmelCase_ )
if __name__ == "__main__":
snake_case : Union[str, Any] = 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'''
)
snake_case : int = 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
)
| 445
|
import json
import os
import shutil
import tempfile
import unittest
from transformers import BatchEncoding, CanineTokenizer
from transformers.testing_utils import require_tokenizers, require_torch
from transformers.tokenization_utils import AddedToken
from transformers.utils import cached_property
from ...test_tokenization_common import TokenizerTesterMixin
class _snake_case ( _snake_case , unittest.TestCase ):
SCREAMING_SNAKE_CASE__ = CanineTokenizer
SCREAMING_SNAKE_CASE__ = False
def SCREAMING_SNAKE_CASE__ ( self ):
super().setUp()
a :List[str] = CanineTokenizer()
tokenizer.save_pretrained(self.tmpdirname )
@cached_property
def SCREAMING_SNAKE_CASE__ ( self ):
return CanineTokenizer.from_pretrained('''google/canine-s''' )
def SCREAMING_SNAKE_CASE__ ( self , **_lowerCamelCase ):
a :Dict = self.tokenizer_class.from_pretrained(self.tmpdirname , **_lowerCamelCase )
a :List[Any] = 1024
return tokenizer
@require_torch
def SCREAMING_SNAKE_CASE__ ( self ):
a :Any = self.canine_tokenizer
a :Tuple = ['''Life is like a box of chocolates.''', '''You never know what you\'re gonna get.''']
# fmt: off
a :Any = [5_7344, 76, 105, 102, 101, 32, 105, 115, 32, 108, 105, 107, 101, 32, 97, 32, 98, 111, 120, 32, 111, 102, 32, 99, 104, 111, 99, 111, 108, 97, 116, 101, 115, 46, 5_7345, 0, 0, 0, 0]
# fmt: on
a :Dict = tokenizer(_lowerCamelCase , padding=_lowerCamelCase , return_tensors='''pt''' )
self.assertIsInstance(_lowerCamelCase , _lowerCamelCase )
a :str = list(batch.input_ids.numpy()[0] )
self.assertListEqual(_lowerCamelCase , _lowerCamelCase )
self.assertEqual((2, 39) , batch.input_ids.shape )
self.assertEqual((2, 39) , batch.attention_mask.shape )
@require_torch
def SCREAMING_SNAKE_CASE__ ( self ):
a :List[Any] = self.canine_tokenizer
a :Optional[int] = ['''Once there was a man.''', '''He wrote a test in HuggingFace Tranformers.''']
a :Dict = tokenizer(_lowerCamelCase , padding=_lowerCamelCase , return_tensors='''pt''' )
# check if input_ids, attention_mask and token_type_ids are returned
self.assertIn('''input_ids''' , _lowerCamelCase )
self.assertIn('''attention_mask''' , _lowerCamelCase )
self.assertIn('''token_type_ids''' , _lowerCamelCase )
@require_torch
def SCREAMING_SNAKE_CASE__ ( self ):
a :Any = self.canine_tokenizer
a :str = [
'''What\'s the weater?''',
'''It\'s about 25 degrees.''',
]
a :List[str] = tokenizer(
text_target=_lowerCamelCase , max_length=32 , padding='''max_length''' , truncation=_lowerCamelCase , return_tensors='''pt''' )
self.assertEqual(32 , targets['''input_ids'''].shape[1] )
def SCREAMING_SNAKE_CASE__ ( self ):
# safety check on max_len default value so we are sure the test works
a :Tuple = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(F'''{tokenizer.__class__.__name__}''' ):
self.assertNotEqual(tokenizer.model_max_length , 42 )
# Now let's start the test
a :Tuple = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(F'''{tokenizer.__class__.__name__}''' ):
# Isolate this from the other tests because we save additional tokens/etc
a :List[str] = tempfile.mkdtemp()
a :Any = ''' He is very happy, UNwant\u00E9d,running'''
a :Dict = tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
tokenizer.save_pretrained(_lowerCamelCase )
a :Optional[Any] = tokenizer.__class__.from_pretrained(_lowerCamelCase )
a :str = after_tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
self.assertListEqual(_lowerCamelCase , _lowerCamelCase )
shutil.rmtree(_lowerCamelCase )
a :str = self.get_tokenizers(model_max_length=42 )
for tokenizer in tokenizers:
with self.subTest(F'''{tokenizer.__class__.__name__}''' ):
# Isolate this from the other tests because we save additional tokens/etc
a :Any = tempfile.mkdtemp()
a :Optional[Any] = ''' He is very happy, UNwant\u00E9d,running'''
a :Optional[int] = tokenizer.additional_special_tokens
# We can add a new special token for Canine as follows:
a :str = chr(0Xe_0_0_7 )
additional_special_tokens.append(_lowerCamelCase )
tokenizer.add_special_tokens({'''additional_special_tokens''': additional_special_tokens} )
a :Tuple = tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
tokenizer.save_pretrained(_lowerCamelCase )
a :int = tokenizer.__class__.from_pretrained(_lowerCamelCase )
a :Dict = after_tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
self.assertListEqual(_lowerCamelCase , _lowerCamelCase )
self.assertIn(_lowerCamelCase , after_tokenizer.additional_special_tokens )
self.assertEqual(after_tokenizer.model_max_length , 42 )
a :str = tokenizer.__class__.from_pretrained(_lowerCamelCase , model_max_length=43 )
self.assertEqual(tokenizer.model_max_length , 43 )
shutil.rmtree(_lowerCamelCase )
def SCREAMING_SNAKE_CASE__ ( self ):
a :Tuple = self.get_tokenizers(do_lower_case=_lowerCamelCase )
for tokenizer in tokenizers:
with self.subTest(F'''{tokenizer.__class__.__name__}''' ):
a , a :Tuple = self.get_clean_sequence(_lowerCamelCase )
# a special token for Canine can be defined as follows:
a :Tuple = 0Xe_0_0_5
a :Optional[int] = chr(_lowerCamelCase )
tokenizer.add_special_tokens({'''cls_token''': special_token} )
a :Optional[int] = tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
self.assertEqual(len(_lowerCamelCase ) , 1 )
a :List[Any] = tokenizer.decode(ids + encoded_special_token , clean_up_tokenization_spaces=_lowerCamelCase )
a :List[str] = tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
a :Tuple = tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
a :Optional[Any] = tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
self.assertEqual(_lowerCamelCase , input_encoded + special_token_id )
a :Any = tokenizer.decode(_lowerCamelCase , skip_special_tokens=_lowerCamelCase )
self.assertTrue(special_token not in decoded )
def SCREAMING_SNAKE_CASE__ ( self ):
a :int = self.get_tokenizers(do_lower_case=_lowerCamelCase )
for tokenizer in tokenizers:
with self.subTest(F'''{tokenizer.__class__.__name__}''' ):
a :int = chr(0Xe_0_0_5 )
a :str = chr(0Xe_0_0_6 )
# `add_tokens` method stores special tokens only in `tokenizer.unique_no_split_tokens`. (in tokenization_utils.py)
tokenizer.add_tokens([SPECIAL_TOKEN_1] , special_tokens=_lowerCamelCase )
# `add_special_tokens` method stores special tokens in `tokenizer.additional_special_tokens`,
# which also occur in `tokenizer.all_special_tokens`. (in tokenization_utils_base.py)
tokenizer.add_special_tokens({'''additional_special_tokens''': [SPECIAL_TOKEN_2]} )
a :Optional[int] = tokenizer.tokenize(_lowerCamelCase )
a :Optional[Any] = tokenizer.tokenize(_lowerCamelCase )
self.assertEqual(len(_lowerCamelCase ) , 1 )
self.assertEqual(len(_lowerCamelCase ) , 1 )
self.assertEqual(token_a[0] , _lowerCamelCase )
self.assertEqual(token_a[0] , _lowerCamelCase )
@require_tokenizers
def SCREAMING_SNAKE_CASE__ ( self ):
a :Optional[Any] = self.get_tokenizers(do_lower_case=_lowerCamelCase )
for tokenizer in tokenizers:
with self.subTest(F'''{tokenizer.__class__.__name__}''' ):
# a special token for Canine can be defined as follows:
a :Optional[int] = 0Xe_0_0_6
a :List[str] = chr(_lowerCamelCase )
a :Optional[int] = AddedToken(_lowerCamelCase , lstrip=_lowerCamelCase )
tokenizer.add_special_tokens({'''additional_special_tokens''': [new_token]} )
with tempfile.TemporaryDirectory() as tmp_dir_name:
tokenizer.save_pretrained(_lowerCamelCase )
tokenizer.from_pretrained(_lowerCamelCase )
def SCREAMING_SNAKE_CASE__ ( self ):
a :int = []
if self.test_slow_tokenizer:
tokenizer_list.append((self.tokenizer_class, self.get_tokenizer()) )
if self.test_rust_tokenizer:
tokenizer_list.append((self.rust_tokenizer_class, self.get_rust_tokenizer()) )
for tokenizer_class, tokenizer_utils in tokenizer_list:
with tempfile.TemporaryDirectory() as tmp_dir:
tokenizer_utils.save_pretrained(_lowerCamelCase )
with open(os.path.join(_lowerCamelCase , '''special_tokens_map.json''' ) , encoding='''utf-8''' ) as json_file:
a :Optional[Any] = json.load(_lowerCamelCase )
with open(os.path.join(_lowerCamelCase , '''tokenizer_config.json''' ) , encoding='''utf-8''' ) as json_file:
a :Tuple = json.load(_lowerCamelCase )
# a special token for Canine can be defined as follows:
a :int = 0Xe_0_0_6
a :Optional[Any] = chr(_lowerCamelCase )
a :Union[str, Any] = [new_token_a]
a :Optional[int] = [new_token_a]
with open(os.path.join(_lowerCamelCase , '''special_tokens_map.json''' ) , '''w''' , encoding='''utf-8''' ) as outfile:
json.dump(_lowerCamelCase , _lowerCamelCase )
with open(os.path.join(_lowerCamelCase , '''tokenizer_config.json''' ) , '''w''' , encoding='''utf-8''' ) as outfile:
json.dump(_lowerCamelCase , _lowerCamelCase )
# the following checks allow us to verify that our test works as expected, i.e. that the tokenizer takes
# into account the new value of additional_special_tokens given in the "tokenizer_config.json" and
# "special_tokens_map.json" files
a :str = tokenizer_class.from_pretrained(_lowerCamelCase , extra_ids=0 )
self.assertIn(_lowerCamelCase , tokenizer_without_change_in_init.additional_special_tokens )
# self.assertIn("an_additional_special_token",tokenizer_without_change_in_init.get_vocab()) # ByT5Tokenization no vocab
self.assertEqual(
[new_token_a] , tokenizer_without_change_in_init.convert_ids_to_tokens(
tokenizer_without_change_in_init.convert_tokens_to_ids([new_token_a] ) ) , )
a :Optional[int] = 0Xe_0_0_7
a :Any = chr(_lowerCamelCase )
# Now we test that we can change the value of additional_special_tokens in the from_pretrained
a :Tuple = [AddedToken(_lowerCamelCase , lstrip=_lowerCamelCase )]
a :Optional[Any] = tokenizer_class.from_pretrained(
_lowerCamelCase , additional_special_tokens=_lowerCamelCase , extra_ids=0 )
self.assertIn(_lowerCamelCase , tokenizer.additional_special_tokens )
# self.assertIn(new_token_2,tokenizer.get_vocab()) # ByT5Tokenization no vocab
self.assertEqual(
[new_token_a] , tokenizer.convert_ids_to_tokens(tokenizer.convert_tokens_to_ids([new_token_a] ) ) )
@require_tokenizers
def SCREAMING_SNAKE_CASE__ ( self ):
a :List[Any] = self.get_tokenizers(do_lower_case=_lowerCamelCase )
for tokenizer in tokenizers:
with self.subTest(F'''{tokenizer.__class__.__name__}''' ):
a :List[str] = '''hello world'''
if self.space_between_special_tokens:
a :Optional[Any] = '''[CLS] hello world [SEP]'''
else:
a :Tuple = input
a :Any = tokenizer.encode(_lowerCamelCase , add_special_tokens=_lowerCamelCase )
a :List[Any] = tokenizer.decode(_lowerCamelCase , spaces_between_special_tokens=self.space_between_special_tokens )
self.assertIn(_lowerCamelCase , [output, output.lower()] )
def SCREAMING_SNAKE_CASE__ ( self ):
a :int = self.get_tokenizers()
for tokenizer in tokenizers:
with self.subTest(F'''{tokenizer.__class__.__name__}''' ):
a :Any = [
'''bos_token''',
'''eos_token''',
'''unk_token''',
'''sep_token''',
'''pad_token''',
'''cls_token''',
'''mask_token''',
]
a :str = '''a'''
a :List[Any] = ord(_lowerCamelCase )
for attr in attributes_list:
setattr(_lowerCamelCase , attr + '''_id''' , _lowerCamelCase )
self.assertEqual(getattr(_lowerCamelCase , _lowerCamelCase ) , _lowerCamelCase )
self.assertEqual(getattr(_lowerCamelCase , attr + '''_id''' ) , _lowerCamelCase )
setattr(_lowerCamelCase , attr + '''_id''' , _lowerCamelCase )
self.assertEqual(getattr(_lowerCamelCase , _lowerCamelCase ) , _lowerCamelCase )
self.assertEqual(getattr(_lowerCamelCase , attr + '''_id''' ) , _lowerCamelCase )
setattr(_lowerCamelCase , '''additional_special_tokens_ids''' , [] )
self.assertListEqual(getattr(_lowerCamelCase , '''additional_special_tokens''' ) , [] )
self.assertListEqual(getattr(_lowerCamelCase , '''additional_special_tokens_ids''' ) , [] )
a :List[Any] = 0Xe_0_0_6
a :str = chr(_lowerCamelCase )
setattr(_lowerCamelCase , '''additional_special_tokens_ids''' , [additional_special_token_id] )
self.assertListEqual(getattr(_lowerCamelCase , '''additional_special_tokens''' ) , [additional_special_token] )
self.assertListEqual(getattr(_lowerCamelCase , '''additional_special_tokens_ids''' ) , [additional_special_token_id] )
def SCREAMING_SNAKE_CASE__ ( self ):
pass
def SCREAMING_SNAKE_CASE__ ( self ):
pass
def SCREAMING_SNAKE_CASE__ ( self ):
pass
def SCREAMING_SNAKE_CASE__ ( self ):
pass
def SCREAMING_SNAKE_CASE__ ( self ):
pass
def SCREAMING_SNAKE_CASE__ ( self ):
pass
def SCREAMING_SNAKE_CASE__ ( self ):
pass
def SCREAMING_SNAKE_CASE__ ( self ):
pass
| 445
| 1
|
"""simple docstring"""
import numpy as np
import torch
import torch.nn as nn
from transformers import CLIPConfig, CLIPVisionModelWithProjection, PreTrainedModel
from ...utils import logging
snake_case = logging.get_logger(__name__)
class UpperCAmelCase ( __SCREAMING_SNAKE_CASE ):
A__ : Optional[int] = CLIPConfig
A__ : List[str] = ['''CLIPEncoderLayer''']
def __init__( self : List[Any] , __lowerCamelCase : CLIPConfig ):
"""simple docstring"""
super().__init__(__lowerCamelCase )
_snake_case = CLIPVisionModelWithProjection(config.vision_config )
_snake_case = nn.Linear(config.vision_config.projection_dim , 1 )
_snake_case = nn.Linear(config.vision_config.projection_dim , 1 )
@torch.no_grad()
def __UpperCAmelCase ( self : List[Any] , __lowerCamelCase : Tuple , __lowerCamelCase : Tuple , __lowerCamelCase : Optional[int]=0.5 , __lowerCamelCase : Any=0.5 ):
"""simple docstring"""
_snake_case = self.vision_model(__lowerCamelCase )[0]
_snake_case = self.p_head(__lowerCamelCase )
_snake_case = nsfw_detected.flatten()
_snake_case = nsfw_detected > p_threshold
_snake_case = nsfw_detected.tolist()
if any(__lowerCamelCase ):
logger.warning(
'''Potential NSFW content was detected in one or more images. A black image will be returned instead.'''
''' Try again with a different prompt and/or seed.''' )
for idx, nsfw_detected_ in enumerate(__lowerCamelCase ):
if nsfw_detected_:
_snake_case = np.zeros(images[idx].shape )
_snake_case = self.w_head(__lowerCamelCase )
_snake_case = watermark_detected.flatten()
_snake_case = watermark_detected > w_threshold
_snake_case = watermark_detected.tolist()
if any(__lowerCamelCase ):
logger.warning(
'''Potential watermarked content was detected in one or more images. A black image will be returned instead.'''
''' Try again with a different prompt and/or seed.''' )
for idx, watermark_detected_ in enumerate(__lowerCamelCase ):
if watermark_detected_:
_snake_case = np.zeros(images[idx].shape )
return images, nsfw_detected, watermark_detected
| 404
|
"""simple docstring"""
def snake_case ( lowerCAmelCase_ , lowerCAmelCase_ ) -> bool:
return numa ^ numa < 0
if __name__ == "__main__":
import doctest
doctest.testmod()
| 404
| 1
|
import os
from tempfile import TemporaryDirectory
from unittest import TestCase
import pytest
from absl.testing import parameterized
from datasets import config
from datasets.arrow_reader import HF_GCP_BASE_URL
from datasets.builder import DatasetBuilder
from datasets.dataset_dict import IterableDatasetDict
from datasets.iterable_dataset import IterableDataset
from datasets.load import dataset_module_factory, import_main_class
from datasets.utils.file_utils import cached_path
_lowerCAmelCase = [
{"dataset": "wikipedia", "config_name": "20220301.de"},
{"dataset": "wikipedia", "config_name": "20220301.en"},
{"dataset": "wikipedia", "config_name": "20220301.fr"},
{"dataset": "wikipedia", "config_name": "20220301.frr"},
{"dataset": "wikipedia", "config_name": "20220301.it"},
{"dataset": "wikipedia", "config_name": "20220301.simple"},
{"dataset": "snli", "config_name": "plain_text"},
{"dataset": "eli5", "config_name": "LFQA_reddit"},
{"dataset": "wiki40b", "config_name": "en"},
{"dataset": "wiki_dpr", "config_name": "psgs_w100.nq.compressed"},
{"dataset": "wiki_dpr", "config_name": "psgs_w100.nq.no_index"},
{"dataset": "wiki_dpr", "config_name": "psgs_w100.multiset.no_index"},
{"dataset": "natural_questions", "config_name": "default"},
]
def _snake_case ( __snake_case=True ):
if with_config:
return [
{
"testcase_name": d["dataset"] + "/" + d["config_name"],
"dataset": d["dataset"],
"config_name": d["config_name"],
}
for d in DATASETS_ON_HF_GCP
]
else:
return [
{"testcase_name": dataset, "dataset": dataset} for dataset in {d["dataset"] for d in DATASETS_ON_HF_GCP}
]
@parameterized.named_parameters(list_datasets_on_hf_gcp_parameters(with_config=__lowercase ) )
class lowerCAmelCase_ ( __lowercase ):
UpperCAmelCase = None
UpperCAmelCase = None
def UpperCamelCase_ ( self : Any , _A : Union[str, Any] , _A : str ):
with TemporaryDirectory() as tmp_dir:
_UpperCamelCase = dataset_module_factory(_A , cache_dir=_A )
_UpperCamelCase = import_main_class(dataset_module.module_path , dataset=_A )
_UpperCamelCase = builder_cls(
cache_dir=_A , config_name=_A , hash=dataset_module.hash , )
_UpperCamelCase = '''/'''.join(
[
HF_GCP_BASE_URL,
builder_instance._relative_data_dir(with_hash=_A ).replace(os.sep , '''/''' ),
config.DATASET_INFO_FILENAME,
] )
_UpperCamelCase = cached_path(_A , cache_dir=_A )
self.assertTrue(os.path.exists(_A ) )
@pytest.mark.integration
def _snake_case ( __snake_case ):
_UpperCamelCase = tmp_path_factory.mktemp('''test_hf_gcp''' ) / '''test_wikipedia_simple'''
_UpperCamelCase = dataset_module_factory('''wikipedia''' , cache_dir=__snake_case )
_UpperCamelCase = import_main_class(dataset_module.module_path )
_UpperCamelCase = builder_cls(
cache_dir=__snake_case , config_name='''20220301.frr''' , hash=dataset_module.hash , )
# use the HF cloud storage, not the original download_and_prepare that uses apache-beam
_UpperCamelCase = None
builder_instance.download_and_prepare()
_UpperCamelCase = builder_instance.as_dataset()
assert ds
@pytest.mark.integration
def _snake_case ( __snake_case ):
_UpperCamelCase = dataset_module_factory('''wikipedia''' , cache_dir=__snake_case )
_UpperCamelCase = import_main_class(dataset_module.module_path , dataset=__snake_case )
_UpperCamelCase = builder_cls(
cache_dir=__snake_case , config_name='''20220301.frr''' , hash=dataset_module.hash , )
_UpperCamelCase = builder_instance.as_streaming_dataset()
assert ds
assert isinstance(__snake_case , __snake_case )
assert "train" in ds
assert isinstance(ds['''train'''] , __snake_case )
assert next(iter(ds['''train'''] ) )
| 10
|
class UpperCAmelCase_ :
def __init__( self , _lowerCAmelCase ):
# we need a list not a string, so do something to change the type
UpperCAmelCase__ : Dict = arr.split(""",""" )
def __UpperCAmelCase ( self ):
UpperCAmelCase__ : Tuple = [int(self.array[0] )] * len(self.array )
UpperCAmelCase__ : List[str] = [int(self.array[0] )] * len(self.array )
for i in range(1 , len(self.array ) ):
UpperCAmelCase__ : Tuple = max(
int(self.array[i] ) + sum_value[i - 1] , int(self.array[i] ) )
UpperCAmelCase__ : Union[str, Any] = max(sum_value[i] , rear[i - 1] )
return rear[len(self.array ) - 1]
if __name__ == "__main__":
SCREAMING_SNAKE_CASE__ : Tuple = input("""please input some numbers:""")
SCREAMING_SNAKE_CASE__ : Dict = SubArray(whole_array)
SCREAMING_SNAKE_CASE__ : Dict = array.solve_sub_array()
print(("""the results is:""", re))
| 79
| 0
|
'''simple docstring'''
import inspect
import unittest
import warnings
from transformers import DeiTConfig
from transformers.models.auto import get_values
from transformers.testing_utils import (
require_accelerate,
require_torch,
require_torch_gpu,
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_IMAGE_CLASSIFICATION_MAPPING,
MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
MODEL_MAPPING,
DeiTForImageClassification,
DeiTForImageClassificationWithTeacher,
DeiTForMaskedImageModeling,
DeiTModel,
)
from transformers.models.deit.modeling_deit import DEIT_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from PIL import Image
from transformers import DeiTImageProcessor
class SCREAMING_SNAKE_CASE__ :
def __init__( self , lowercase__ , lowercase__=13 , lowercase__=30 , lowercase__=2 , lowercase__=3 , lowercase__=True , lowercase__=True , lowercase__=32 , lowercase__=5 , lowercase__=4 , lowercase__=37 , lowercase__="gelu" , lowercase__=0.1 , lowercase__=0.1 , lowercase__=10 , lowercase__=0.02 , lowercase__=3 , lowercase__=None , lowercase__=2 , ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[Any] = parent
SCREAMING_SNAKE_CASE_ : List[Any] = batch_size
SCREAMING_SNAKE_CASE_ : Any = image_size
SCREAMING_SNAKE_CASE_ : List[str] = patch_size
SCREAMING_SNAKE_CASE_ : Dict = num_channels
SCREAMING_SNAKE_CASE_ : Any = is_training
SCREAMING_SNAKE_CASE_ : int = use_labels
SCREAMING_SNAKE_CASE_ : int = hidden_size
SCREAMING_SNAKE_CASE_ : int = num_hidden_layers
SCREAMING_SNAKE_CASE_ : int = num_attention_heads
SCREAMING_SNAKE_CASE_ : Tuple = intermediate_size
SCREAMING_SNAKE_CASE_ : Any = hidden_act
SCREAMING_SNAKE_CASE_ : Dict = hidden_dropout_prob
SCREAMING_SNAKE_CASE_ : List[Any] = attention_probs_dropout_prob
SCREAMING_SNAKE_CASE_ : List[Any] = type_sequence_label_size
SCREAMING_SNAKE_CASE_ : Any = initializer_range
SCREAMING_SNAKE_CASE_ : List[Any] = scope
SCREAMING_SNAKE_CASE_ : Any = encoder_stride
# in DeiT, the seq length equals the number of patches + 2 (we add 2 for the [CLS] and distilation tokens)
SCREAMING_SNAKE_CASE_ : int = (image_size // patch_size) ** 2
SCREAMING_SNAKE_CASE_ : Tuple = num_patches + 2
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[int] = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
SCREAMING_SNAKE_CASE_ : Any = None
if self.use_labels:
SCREAMING_SNAKE_CASE_ : Any = ids_tensor([self.batch_size] , self.type_sequence_label_size )
SCREAMING_SNAKE_CASE_ : Tuple = self.get_config()
return config, pixel_values, labels
def __lowerCamelCase ( self ):
"""simple docstring"""
return DeiTConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , is_decoder=lowercase__ , initializer_range=self.initializer_range , encoder_stride=self.encoder_stride , )
def __lowerCamelCase ( self , lowercase__ , lowercase__ , lowercase__ ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : List[str] = DeiTModel(config=lowercase__ )
model.to(lowercase__ )
model.eval()
SCREAMING_SNAKE_CASE_ : List[Any] = model(lowercase__ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def __lowerCamelCase ( self , lowercase__ , lowercase__ , lowercase__ ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[int] = DeiTForMaskedImageModeling(config=lowercase__ )
model.to(lowercase__ )
model.eval()
SCREAMING_SNAKE_CASE_ : Optional[int] = model(lowercase__ )
self.parent.assertEqual(
result.reconstruction.shape , (self.batch_size, self.num_channels, self.image_size, self.image_size) )
# test greyscale images
SCREAMING_SNAKE_CASE_ : str = 1
SCREAMING_SNAKE_CASE_ : List[str] = DeiTForMaskedImageModeling(lowercase__ )
model.to(lowercase__ )
model.eval()
SCREAMING_SNAKE_CASE_ : int = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
SCREAMING_SNAKE_CASE_ : List[Any] = model(lowercase__ )
self.parent.assertEqual(result.reconstruction.shape , (self.batch_size, 1, self.image_size, self.image_size) )
def __lowerCamelCase ( self , lowercase__ , lowercase__ , lowercase__ ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[int] = self.type_sequence_label_size
SCREAMING_SNAKE_CASE_ : int = DeiTForImageClassification(lowercase__ )
model.to(lowercase__ )
model.eval()
SCREAMING_SNAKE_CASE_ : Any = model(lowercase__ , labels=lowercase__ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.type_sequence_label_size) )
# test greyscale images
SCREAMING_SNAKE_CASE_ : Any = 1
SCREAMING_SNAKE_CASE_ : Union[str, Any] = DeiTForImageClassification(lowercase__ )
model.to(lowercase__ )
model.eval()
SCREAMING_SNAKE_CASE_ : Optional[Any] = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
SCREAMING_SNAKE_CASE_ : Dict = model(lowercase__ , labels=lowercase__ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.type_sequence_label_size) )
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Union[str, Any] = self.prepare_config_and_inputs()
(
(
SCREAMING_SNAKE_CASE_
), (
SCREAMING_SNAKE_CASE_
), (
SCREAMING_SNAKE_CASE_
),
) : str = config_and_inputs
SCREAMING_SNAKE_CASE_ : Tuple = {"pixel_values": pixel_values}
return config, inputs_dict
@require_torch
class SCREAMING_SNAKE_CASE__ ( _UpperCAmelCase,_UpperCAmelCase,unittest.TestCase ):
_A = (
(
DeiTModel,
DeiTForImageClassification,
DeiTForImageClassificationWithTeacher,
DeiTForMaskedImageModeling,
)
if is_torch_available()
else ()
)
_A = (
{
"feature-extraction": DeiTModel,
"image-classification": (DeiTForImageClassification, DeiTForImageClassificationWithTeacher),
}
if is_torch_available()
else {}
)
_A = False
_A = False
_A = False
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : List[str] = DeiTModelTester(self )
SCREAMING_SNAKE_CASE_ : List[str] = ConfigTester(self , config_class=lowercase__ , has_text_modality=lowercase__ , hidden_size=37 )
def __lowerCamelCase ( self ):
"""simple docstring"""
self.config_tester.run_common_tests()
@unittest.skip(reason="DeiT does not use inputs_embeds" )
def __lowerCamelCase ( self ):
"""simple docstring"""
pass
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_, SCREAMING_SNAKE_CASE_ : int = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
SCREAMING_SNAKE_CASE_ : Optional[int] = model_class(lowercase__ )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
SCREAMING_SNAKE_CASE_ : List[str] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowercase__ , nn.Linear ) )
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_, SCREAMING_SNAKE_CASE_ : Optional[Any] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
SCREAMING_SNAKE_CASE_ : List[str] = model_class(lowercase__ )
SCREAMING_SNAKE_CASE_ : List[Any] = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
SCREAMING_SNAKE_CASE_ : str = [*signature.parameters.keys()]
SCREAMING_SNAKE_CASE_ : Any = ["pixel_values"]
self.assertListEqual(arg_names[:1] , lowercase__ )
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowercase__ )
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_image_modeling(*lowercase__ )
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Tuple = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_image_classification(*lowercase__ )
def __lowerCamelCase ( self , lowercase__ , lowercase__ , lowercase__=False ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Tuple = super()._prepare_for_class(lowercase__ , lowercase__ , return_labels=lowercase__ )
if return_labels:
if model_class.__name__ == "DeiTForImageClassificationWithTeacher":
del inputs_dict["labels"]
return inputs_dict
def __lowerCamelCase ( self ):
"""simple docstring"""
if not self.model_tester.is_training:
return
SCREAMING_SNAKE_CASE_, SCREAMING_SNAKE_CASE_ : Union[str, Any] = self.model_tester.prepare_config_and_inputs_for_common()
SCREAMING_SNAKE_CASE_ : List[str] = True
for model_class in self.all_model_classes:
# DeiTForImageClassificationWithTeacher supports inference-only
if (
model_class in get_values(lowercase__ )
or model_class.__name__ == "DeiTForImageClassificationWithTeacher"
):
continue
SCREAMING_SNAKE_CASE_ : Optional[Any] = model_class(lowercase__ )
model.to(lowercase__ )
model.train()
SCREAMING_SNAKE_CASE_ : Optional[int] = self._prepare_for_class(lowercase__ , lowercase__ , return_labels=lowercase__ )
SCREAMING_SNAKE_CASE_ : List[Any] = model(**lowercase__ ).loss
loss.backward()
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_, SCREAMING_SNAKE_CASE_ : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
if not self.model_tester.is_training:
return
SCREAMING_SNAKE_CASE_ : Union[str, Any] = False
SCREAMING_SNAKE_CASE_ : Dict = True
for model_class in self.all_model_classes:
if model_class in get_values(lowercase__ ) or not model_class.supports_gradient_checkpointing:
continue
# DeiTForImageClassificationWithTeacher supports inference-only
if model_class.__name__ == "DeiTForImageClassificationWithTeacher":
continue
SCREAMING_SNAKE_CASE_ : Any = model_class(lowercase__ )
model.gradient_checkpointing_enable()
model.to(lowercase__ )
model.train()
SCREAMING_SNAKE_CASE_ : List[Any] = self._prepare_for_class(lowercase__ , lowercase__ , return_labels=lowercase__ )
SCREAMING_SNAKE_CASE_ : Union[str, Any] = model(**lowercase__ ).loss
loss.backward()
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_, SCREAMING_SNAKE_CASE_ : str = self.model_tester.prepare_config_and_inputs_for_common()
SCREAMING_SNAKE_CASE_ : List[Any] = [
{"title": "multi_label_classification", "num_labels": 2, "dtype": torch.float},
{"title": "single_label_classification", "num_labels": 1, "dtype": torch.long},
{"title": "regression", "num_labels": 1, "dtype": torch.float},
]
for model_class in self.all_model_classes:
if (
model_class
not in [
*get_values(lowercase__ ),
*get_values(lowercase__ ),
]
or model_class.__name__ == "DeiTForImageClassificationWithTeacher"
):
continue
for problem_type in problem_types:
with self.subTest(msg=F"Testing {model_class} with {problem_type['title']}" ):
SCREAMING_SNAKE_CASE_ : int = problem_type["title"]
SCREAMING_SNAKE_CASE_ : Union[str, Any] = problem_type["num_labels"]
SCREAMING_SNAKE_CASE_ : List[Any] = model_class(lowercase__ )
model.to(lowercase__ )
model.train()
SCREAMING_SNAKE_CASE_ : int = self._prepare_for_class(lowercase__ , lowercase__ , return_labels=lowercase__ )
if problem_type["num_labels"] > 1:
SCREAMING_SNAKE_CASE_ : Optional[Any] = inputs["labels"].unsqueeze(1 ).repeat(1 , problem_type["num_labels"] )
SCREAMING_SNAKE_CASE_ : Optional[int] = inputs["labels"].to(problem_type["dtype"] )
# This tests that we do not trigger the warning form PyTorch "Using a target size that is different
# to the input size. This will likely lead to incorrect results due to broadcasting. Please ensure
# they have the same size." which is a symptom something in wrong for the regression problem.
# See https://github.com/huggingface/transformers/issues/11780
with warnings.catch_warnings(record=lowercase__ ) as warning_list:
SCREAMING_SNAKE_CASE_ : Any = model(**lowercase__ ).loss
for w in warning_list:
if "Using a target size that is different to the input size" in str(w.message ):
raise ValueError(
F"Something is going wrong in the regression problem: intercepted {w.message}" )
loss.backward()
@slow
def __lowerCamelCase ( self ):
"""simple docstring"""
for model_name in DEIT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
SCREAMING_SNAKE_CASE_ : Tuple = DeiTModel.from_pretrained(lowercase__ )
self.assertIsNotNone(lowercase__ )
def __lowerCamelCase ( ) -> Optional[Any]:
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[Any] = Image.open("./tests/fixtures/tests_samples/COCO/000000039769.png" )
return image
@require_torch
@require_vision
class SCREAMING_SNAKE_CASE__ ( unittest.TestCase ):
@cached_property
def __lowerCamelCase ( self ):
"""simple docstring"""
return (
DeiTImageProcessor.from_pretrained("facebook/deit-base-distilled-patch16-224" )
if is_vision_available()
else None
)
@slow
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : str = DeiTForImageClassificationWithTeacher.from_pretrained("facebook/deit-base-distilled-patch16-224" ).to(
lowercase__ )
SCREAMING_SNAKE_CASE_ : Tuple = self.default_image_processor
SCREAMING_SNAKE_CASE_ : Dict = prepare_img()
SCREAMING_SNAKE_CASE_ : Dict = image_processor(images=lowercase__ , return_tensors="pt" ).to(lowercase__ )
# forward pass
with torch.no_grad():
SCREAMING_SNAKE_CASE_ : List[str] = model(**lowercase__ )
# verify the logits
SCREAMING_SNAKE_CASE_ : Optional[Any] = torch.Size((1, 1000) )
self.assertEqual(outputs.logits.shape , lowercase__ )
SCREAMING_SNAKE_CASE_ : Dict = torch.tensor([-1.0266, 0.1912, -1.2861] ).to(lowercase__ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , lowercase__ , atol=1e-4 ) )
@slow
@require_accelerate
@require_torch_gpu
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : str = DeiTModel.from_pretrained(
"facebook/deit-base-distilled-patch16-224" , torch_dtype=torch.floataa , device_map="auto" )
SCREAMING_SNAKE_CASE_ : List[str] = self.default_image_processor
SCREAMING_SNAKE_CASE_ : str = prepare_img()
SCREAMING_SNAKE_CASE_ : str = image_processor(images=lowercase__ , return_tensors="pt" )
SCREAMING_SNAKE_CASE_ : Union[str, Any] = inputs.pixel_values.to(lowercase__ )
# forward pass to make sure inference works in fp16
with torch.no_grad():
SCREAMING_SNAKE_CASE_ : List[str] = model(lowercase__ )
| 68
|
'''simple docstring'''
import unittest
import numpy as np
from transformers.testing_utils import require_torch, require_vision
from transformers.utils import is_torch_available, is_vision_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin
if is_torch_available():
import torch
if is_vision_available():
from PIL import Image
from transformers import ChineseCLIPImageProcessor
class SCREAMING_SNAKE_CASE__ ( unittest.TestCase ):
def __init__( self , lowercase__ , lowercase__=7 , lowercase__=3 , lowercase__=18 , lowercase__=30 , lowercase__=400 , lowercase__=True , lowercase__=None , lowercase__=True , lowercase__=None , lowercase__=True , lowercase__=[0.48145466, 0.4578275, 0.40821073] , lowercase__=[0.26862954, 0.26130258, 0.27577711] , lowercase__=True , ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[Any] = size if size is not None else {"height": 224, "width": 224}
SCREAMING_SNAKE_CASE_ : Optional[Any] = crop_size if crop_size is not None else {"height": 18, "width": 18}
SCREAMING_SNAKE_CASE_ : str = parent
SCREAMING_SNAKE_CASE_ : List[Any] = batch_size
SCREAMING_SNAKE_CASE_ : Dict = num_channels
SCREAMING_SNAKE_CASE_ : Any = image_size
SCREAMING_SNAKE_CASE_ : Tuple = min_resolution
SCREAMING_SNAKE_CASE_ : Optional[Any] = max_resolution
SCREAMING_SNAKE_CASE_ : Tuple = do_resize
SCREAMING_SNAKE_CASE_ : List[str] = size
SCREAMING_SNAKE_CASE_ : str = do_center_crop
SCREAMING_SNAKE_CASE_ : List[str] = crop_size
SCREAMING_SNAKE_CASE_ : int = do_normalize
SCREAMING_SNAKE_CASE_ : Optional[int] = image_mean
SCREAMING_SNAKE_CASE_ : Dict = image_std
SCREAMING_SNAKE_CASE_ : List[Any] = do_convert_rgb
def __lowerCamelCase ( self ):
"""simple docstring"""
return {
"do_resize": self.do_resize,
"size": self.size,
"do_center_crop": self.do_center_crop,
"crop_size": self.crop_size,
"do_normalize": self.do_normalize,
"image_mean": self.image_mean,
"image_std": self.image_std,
"do_convert_rgb": self.do_convert_rgb,
}
def __lowerCamelCase ( self , lowercase__=False , lowercase__=False , lowercase__=False ):
"""simple docstring"""
assert not (numpify and torchify), "You cannot specify both numpy and PyTorch tensors at the same time"
if equal_resolution:
SCREAMING_SNAKE_CASE_ : Optional[int] = []
for i in range(self.batch_size ):
image_inputs.append(
np.random.randint(
255 , size=(self.num_channels, self.max_resolution, self.max_resolution) , dtype=np.uinta ) )
else:
SCREAMING_SNAKE_CASE_ : str = []
for i in range(self.batch_size ):
SCREAMING_SNAKE_CASE_, SCREAMING_SNAKE_CASE_ : Optional[Any] = np.random.choice(np.arange(self.min_resolution , self.max_resolution ) , 2 )
image_inputs.append(np.random.randint(255 , size=(self.num_channels, width, height) , dtype=np.uinta ) )
if not numpify and not torchify:
# PIL expects the channel dimension as last dimension
SCREAMING_SNAKE_CASE_ : str = [Image.fromarray(np.moveaxis(lowercase__ , 0 , -1 ) ) for x in image_inputs]
if torchify:
SCREAMING_SNAKE_CASE_ : List[str] = [torch.from_numpy(lowercase__ ) for x in image_inputs]
return image_inputs
@require_torch
@require_vision
class SCREAMING_SNAKE_CASE__ ( _UpperCAmelCase,unittest.TestCase ):
_A = ChineseCLIPImageProcessor if is_vision_available() else None
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : List[Any] = ChineseCLIPImageProcessingTester(self , do_center_crop=lowercase__ )
@property
def __lowerCamelCase ( self ):
"""simple docstring"""
return self.image_processor_tester.prepare_image_processor_dict()
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : List[str] = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(lowercase__ , "do_resize" ) )
self.assertTrue(hasattr(lowercase__ , "size" ) )
self.assertTrue(hasattr(lowercase__ , "do_center_crop" ) )
self.assertTrue(hasattr(lowercase__ , "center_crop" ) )
self.assertTrue(hasattr(lowercase__ , "do_normalize" ) )
self.assertTrue(hasattr(lowercase__ , "image_mean" ) )
self.assertTrue(hasattr(lowercase__ , "image_std" ) )
self.assertTrue(hasattr(lowercase__ , "do_convert_rgb" ) )
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : List[Any] = self.image_processing_class.from_dict(self.image_processor_dict )
self.assertEqual(image_processor.size , {"height": 224, "width": 224} )
self.assertEqual(image_processor.crop_size , {"height": 18, "width": 18} )
SCREAMING_SNAKE_CASE_ : str = self.image_processing_class.from_dict(self.image_processor_dict , size=42 , crop_size=84 )
self.assertEqual(image_processor.size , {"shortest_edge": 42} )
self.assertEqual(image_processor.crop_size , {"height": 84, "width": 84} )
def __lowerCamelCase ( self ):
"""simple docstring"""
pass
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Union[str, Any] = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
SCREAMING_SNAKE_CASE_ : Tuple = self.image_processor_tester.prepare_inputs(equal_resolution=lowercase__ )
for image in image_inputs:
self.assertIsInstance(lowercase__ , Image.Image )
# Test not batched input
SCREAMING_SNAKE_CASE_ : Dict = image_processing(image_inputs[0] , return_tensors="pt" ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size["height"],
self.image_processor_tester.crop_size["width"],
) , )
# Test batched
SCREAMING_SNAKE_CASE_ : Optional[int] = image_processing(lowercase__ , return_tensors="pt" ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size["height"],
self.image_processor_tester.crop_size["width"],
) , )
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : List[str] = self.image_processing_class(**self.image_processor_dict )
# create random numpy tensors
SCREAMING_SNAKE_CASE_ : Union[str, Any] = self.image_processor_tester.prepare_inputs(equal_resolution=lowercase__ , numpify=lowercase__ )
for image in image_inputs:
self.assertIsInstance(lowercase__ , np.ndarray )
# Test not batched input
SCREAMING_SNAKE_CASE_ : Optional[int] = image_processing(image_inputs[0] , return_tensors="pt" ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size["height"],
self.image_processor_tester.crop_size["width"],
) , )
# Test batched
SCREAMING_SNAKE_CASE_ : List[str] = image_processing(lowercase__ , return_tensors="pt" ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size["height"],
self.image_processor_tester.crop_size["width"],
) , )
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : str = self.image_processing_class(**self.image_processor_dict )
# create random PyTorch tensors
SCREAMING_SNAKE_CASE_ : List[str] = self.image_processor_tester.prepare_inputs(equal_resolution=lowercase__ , torchify=lowercase__ )
for image in image_inputs:
self.assertIsInstance(lowercase__ , torch.Tensor )
# Test not batched input
SCREAMING_SNAKE_CASE_ : Union[str, Any] = image_processing(image_inputs[0] , return_tensors="pt" ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size["height"],
self.image_processor_tester.crop_size["width"],
) , )
# Test batched
SCREAMING_SNAKE_CASE_ : int = image_processing(lowercase__ , return_tensors="pt" ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.crop_size["height"],
self.image_processor_tester.crop_size["width"],
) , )
@require_torch
@require_vision
class SCREAMING_SNAKE_CASE__ ( _UpperCAmelCase,unittest.TestCase ):
_A = ChineseCLIPImageProcessor if is_vision_available() else None
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : int = ChineseCLIPImageProcessingTester(self , num_channels=4 , do_center_crop=lowercase__ )
SCREAMING_SNAKE_CASE_ : Union[str, Any] = 3
@property
def __lowerCamelCase ( self ):
"""simple docstring"""
return self.image_processor_tester.prepare_image_processor_dict()
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[Any] = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(lowercase__ , "do_resize" ) )
self.assertTrue(hasattr(lowercase__ , "size" ) )
self.assertTrue(hasattr(lowercase__ , "do_center_crop" ) )
self.assertTrue(hasattr(lowercase__ , "center_crop" ) )
self.assertTrue(hasattr(lowercase__ , "do_normalize" ) )
self.assertTrue(hasattr(lowercase__ , "image_mean" ) )
self.assertTrue(hasattr(lowercase__ , "image_std" ) )
self.assertTrue(hasattr(lowercase__ , "do_convert_rgb" ) )
def __lowerCamelCase ( self ):
"""simple docstring"""
pass
def __lowerCamelCase ( self ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : List[str] = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
SCREAMING_SNAKE_CASE_ : Optional[int] = self.image_processor_tester.prepare_inputs(equal_resolution=lowercase__ )
for image in image_inputs:
self.assertIsInstance(lowercase__ , Image.Image )
# Test not batched input
SCREAMING_SNAKE_CASE_ : Optional[Any] = image_processing(image_inputs[0] , return_tensors="pt" ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.expected_encoded_image_num_channels,
self.image_processor_tester.crop_size["height"],
self.image_processor_tester.crop_size["width"],
) , )
# Test batched
SCREAMING_SNAKE_CASE_ : List[str] = image_processing(lowercase__ , return_tensors="pt" ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.expected_encoded_image_num_channels,
self.image_processor_tester.crop_size["height"],
self.image_processor_tester.crop_size["width"],
) , )
| 68
| 1
|
'''simple docstring'''
from collections import OrderedDict
from typing import Mapping
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...utils import logging
UpperCamelCase_ = logging.get_logger(__name__)
UpperCamelCase_ = {
"junnyu/roformer_chinese_small": "https://huggingface.co/junnyu/roformer_chinese_small/resolve/main/config.json",
"junnyu/roformer_chinese_base": "https://huggingface.co/junnyu/roformer_chinese_base/resolve/main/config.json",
"junnyu/roformer_chinese_char_small": (
"https://huggingface.co/junnyu/roformer_chinese_char_small/resolve/main/config.json"
),
"junnyu/roformer_chinese_char_base": (
"https://huggingface.co/junnyu/roformer_chinese_char_base/resolve/main/config.json"
),
"junnyu/roformer_small_discriminator": (
"https://huggingface.co/junnyu/roformer_small_discriminator/resolve/main/config.json"
),
"junnyu/roformer_small_generator": (
"https://huggingface.co/junnyu/roformer_small_generator/resolve/main/config.json"
),
# See all RoFormer models at https://huggingface.co/models?filter=roformer
}
class _a ( SCREAMING_SNAKE_CASE ):
'''simple docstring'''
A : List[str] = '''roformer'''
def __init__( self, A=50_000, A=None, A=768, A=12, A=12, A=3_072, A="gelu", A=0.1, A=0.1, A=1_536, A=2, A=0.02, A=1E-12, A=0, A=False, A=True, **A, ):
'''simple docstring'''
super().__init__(pad_token_id=A, **A )
SCREAMING_SNAKE_CASE : List[Any] = vocab_size
SCREAMING_SNAKE_CASE : int = hidden_size if embedding_size is None else embedding_size
SCREAMING_SNAKE_CASE : Union[str, Any] = hidden_size
SCREAMING_SNAKE_CASE : Dict = num_hidden_layers
SCREAMING_SNAKE_CASE : List[str] = num_attention_heads
SCREAMING_SNAKE_CASE : Any = hidden_act
SCREAMING_SNAKE_CASE : Optional[int] = intermediate_size
SCREAMING_SNAKE_CASE : str = hidden_dropout_prob
SCREAMING_SNAKE_CASE : int = attention_probs_dropout_prob
SCREAMING_SNAKE_CASE : Optional[int] = max_position_embeddings
SCREAMING_SNAKE_CASE : List[str] = type_vocab_size
SCREAMING_SNAKE_CASE : List[str] = initializer_range
SCREAMING_SNAKE_CASE : Tuple = layer_norm_eps
SCREAMING_SNAKE_CASE : Tuple = rotary_value
SCREAMING_SNAKE_CASE : str = use_cache
class _a ( SCREAMING_SNAKE_CASE ):
'''simple docstring'''
@property
def UpperCamelCase_ ( self ):
'''simple docstring'''
if self.task == "multiple-choice":
SCREAMING_SNAKE_CASE : Optional[int] = {0: 'batch', 1: 'choice', 2: 'sequence'}
else:
SCREAMING_SNAKE_CASE : Optional[int] = {0: 'batch', 1: 'sequence'}
SCREAMING_SNAKE_CASE : int = {0: 'batch', 1: 'sequence'}
return OrderedDict(
[
('input_ids', dynamic_axis),
('attention_mask', dynamic_axis),
('token_type_ids', dynamic_axis),
] )
| 28
|
'''simple docstring'''
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel
from diffusers import DDIMScheduler, LDMPipeline, UNetaDModel, VQModel
from diffusers.utils.testing_utils import enable_full_determinism, require_torch, slow, torch_device
enable_full_determinism()
class __magic_name__ ( unittest.TestCase ):
@property
def __lowercase ( self : int ):
torch.manual_seed(0 )
_a : int = UNetaDModel(
block_out_channels=(32, 64) ,layers_per_block=2 ,sample_size=32 ,in_channels=3 ,out_channels=3 ,down_block_types=('DownBlock2D', 'AttnDownBlock2D') ,up_block_types=('AttnUpBlock2D', 'UpBlock2D') ,)
return model
@property
def __lowercase ( self : Any ):
torch.manual_seed(0 )
_a : int = VQModel(
block_out_channels=[32, 64] ,in_channels=3 ,out_channels=3 ,down_block_types=['DownEncoderBlock2D', 'DownEncoderBlock2D'] ,up_block_types=['UpDecoderBlock2D', 'UpDecoderBlock2D'] ,latent_channels=3 ,)
return model
@property
def __lowercase ( self : Union[str, Any] ):
torch.manual_seed(0 )
_a : Dict = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=32 ,intermediate_size=37 ,layer_norm_eps=1E-05 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1000 ,)
return CLIPTextModel(_UpperCAmelCase )
def __lowercase ( self : Dict ):
_a : Union[str, Any] = self.dummy_uncond_unet
_a : Union[str, Any] = DDIMScheduler()
_a : Union[str, Any] = self.dummy_vq_model
_a : List[Any] = LDMPipeline(unet=_UpperCAmelCase ,vqvae=_UpperCAmelCase ,scheduler=_UpperCAmelCase )
ldm.to(_UpperCAmelCase )
ldm.set_progress_bar_config(disable=_UpperCAmelCase )
_a : Dict = torch.manual_seed(0 )
_a : int = ldm(generator=_UpperCAmelCase ,num_inference_steps=2 ,output_type='numpy' ).images
_a : Tuple = torch.manual_seed(0 )
_a : List[Any] = ldm(generator=_UpperCAmelCase ,num_inference_steps=2 ,output_type='numpy' ,return_dict=_UpperCAmelCase )[0]
_a : Any = image[0, -3:, -3:, -1]
_a : List[Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
_a : Union[str, Any] = np.array([0.85_12, 0.8_18, 0.64_11, 0.68_08, 0.44_65, 0.56_18, 0.46, 0.62_31, 0.51_72] )
_a : Any = 1E-2 if torch_device != 'mps' else 3E-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < tolerance
@slow
@require_torch
class __magic_name__ ( unittest.TestCase ):
def __lowercase ( self : Optional[int] ):
_a : Dict = LDMPipeline.from_pretrained('CompVis/ldm-celebahq-256' )
ldm.to(_UpperCAmelCase )
ldm.set_progress_bar_config(disable=_UpperCAmelCase )
_a : str = torch.manual_seed(0 )
_a : Optional[Any] = ldm(generator=_UpperCAmelCase ,num_inference_steps=5 ,output_type='numpy' ).images
_a : Optional[Any] = image[0, -3:, -3:, -1]
assert image.shape == (1, 256, 256, 3)
_a : Union[str, Any] = np.array([0.43_99, 0.4_49_75, 0.4_68_25, 0.4_74, 0.43_59, 0.45_81, 0.4_50_95, 0.43_41, 0.44_47] )
_a : List[str] = 1E-2 if torch_device != 'mps' else 3E-2
assert np.abs(image_slice.flatten() - expected_slice ).max() < tolerance
| 358
| 0
|
# Copyright 2022 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import os
import subprocess
from packaging.version import Version, parse
from accelerate.commands.config.config_args import default_config_file, load_config_from_file
lowercase = """Run commands across TPU VMs for initial setup before running `accelerate launch`."""
def lowerCamelCase_ ( UpperCamelCase__ : Optional[int]=None ):
'''simple docstring'''
if subparsers is not None:
UpperCamelCase__ = subparsers.add_parser('''tpu-config''', description=_description )
else:
UpperCamelCase__ = argparse.ArgumentParser('''Accelerate tpu-config command''', description=_description )
# Core arguments
UpperCamelCase__ = parser.add_argument_group(
'''Config Arguments''', '''Arguments that can be configured through `accelerate config`.''' )
config_args.add_argument(
'''--config_file''', type=UpperCamelCase__, default=UpperCamelCase__, help='''Path to the config file to use for accelerate.''', )
config_args.add_argument(
'''--tpu_name''', default=UpperCamelCase__, help='''The name of the TPU to use. If not specified, will use the TPU specified in the config file.''', )
config_args.add_argument(
'''--tpu_zone''', default=UpperCamelCase__, help='''The zone of the TPU to use. If not specified, will use the zone specified in the config file.''', )
UpperCamelCase__ = parser.add_argument_group('''TPU Arguments''', '''Arguments for options ran inside the TPU.''' )
pod_args.add_argument(
'''--use_alpha''', action='''store_true''', help='''Whether to use `gcloud alpha` when running the TPU training script instead of `gcloud`.''', )
pod_args.add_argument(
'''--command_file''', default=UpperCamelCase__, help='''The path to the file containing the commands to run on the pod on startup.''', )
pod_args.add_argument(
'''--command''', action='''append''', nargs='''+''', help='''A command to run on the pod. Can be passed multiple times.''', )
pod_args.add_argument(
'''--install_accelerate''', action='''store_true''', help='''Whether to install accelerate on the pod. Defaults to False.''', )
pod_args.add_argument(
'''--accelerate_version''', default='''latest''', help='''The version of accelerate to install on the pod. If not specified, will use the latest pypi version. Specify \'dev\' to install from GitHub.''', )
pod_args.add_argument(
'''--debug''', action='''store_true''', help='''If set, will print the command that would be run instead of running it.''' )
if subparsers is not None:
parser.set_defaults(func=UpperCamelCase__ )
return parser
def lowerCamelCase_ ( UpperCamelCase__ : str ):
'''simple docstring'''
UpperCamelCase__ = None
# Get the default from the config file if it exists.
if args.config_file is not None or os.path.isfile(UpperCamelCase__ ):
UpperCamelCase__ = load_config_from_file(args.config_file )
if not args.command_file and defaults.command_file is not None and not args.command:
UpperCamelCase__ = defaults.command_file
if not args.command and defaults.commands is not None:
UpperCamelCase__ = defaults.commands
if not args.tpu_name:
UpperCamelCase__ = defaults.tpu_name
if not args.tpu_zone:
UpperCamelCase__ = defaults.tpu_zone
if args.accelerate_version == "dev":
UpperCamelCase__ = '''git+https://github.com/huggingface/accelerate.git'''
elif args.accelerate_version == "latest":
UpperCamelCase__ = '''accelerate -U'''
elif isinstance(parse(args.accelerate_version ), UpperCamelCase__ ):
UpperCamelCase__ = F"""accelerate=={args.accelerate_version}"""
if not args.command_file and not args.command:
raise ValueError('''You must specify either a command file or a command to run on the pod.''' )
if args.command_file:
with open(args.command_file, '''r''' ) as f:
UpperCamelCase__ = [f.read().splitlines()]
# To turn list of lists into list of strings
if isinstance(args.command[0], UpperCamelCase__ ):
UpperCamelCase__ = [line for cmd in args.command for line in cmd]
# Default to the shared folder and install accelerate
UpperCamelCase__ = ['''cd /usr/share''']
if args.install_accelerate:
new_cmd += [F"""pip install {args.accelerate_version}"""]
new_cmd += args.command
UpperCamelCase__ = '''; '''.join(UpperCamelCase__ )
# Then send it to gcloud
# Eventually try to use google-api-core to do this instead of subprocess
UpperCamelCase__ = ['''gcloud''']
if args.use_alpha:
cmd += ["alpha"]
cmd += [
"compute",
"tpus",
"tpu-vm",
"ssh",
args.tpu_name,
"--zone",
args.tpu_zone,
"--command",
args.command,
"--worker",
"all",
]
if args.debug:
print(F"""Running {" ".join(UpperCamelCase__ )}""" )
return
subprocess.run(UpperCamelCase__ )
print('''Successfully setup pod.''' )
def lowerCamelCase_ ( ):
'''simple docstring'''
UpperCamelCase__ = tpu_command_parser()
UpperCamelCase__ = parser.parse_args()
tpu_command_launcher(UpperCamelCase__ )
| 591
|
import os
import textwrap
import pyarrow as pa
import pytest
from datasets import ClassLabel, Features, Image
from datasets.packaged_modules.csv.csv import Csv
from ..utils import require_pil
@pytest.fixture
def lowerCamelCase_ ( UpperCamelCase__ : Any ):
'''simple docstring'''
UpperCamelCase__ = tmp_path / '''file.csv'''
UpperCamelCase__ = textwrap.dedent(
'''\
header1,header2
1,2
10,20
''' )
with open(UpperCamelCase__, '''w''' ) as f:
f.write(UpperCamelCase__ )
return str(UpperCamelCase__ )
@pytest.fixture
def lowerCamelCase_ ( UpperCamelCase__ : int ):
'''simple docstring'''
UpperCamelCase__ = tmp_path / '''malformed_file.csv'''
UpperCamelCase__ = textwrap.dedent(
'''\
header1,header2
1,2
10,20,
''' )
with open(UpperCamelCase__, '''w''' ) as f:
f.write(UpperCamelCase__ )
return str(UpperCamelCase__ )
@pytest.fixture
def lowerCamelCase_ ( UpperCamelCase__ : Union[str, Any], UpperCamelCase__ : str ):
'''simple docstring'''
UpperCamelCase__ = tmp_path / '''csv_with_image.csv'''
UpperCamelCase__ = textwrap.dedent(
F"""\
image
{image_file}
""" )
with open(UpperCamelCase__, '''w''' ) as f:
f.write(UpperCamelCase__ )
return str(UpperCamelCase__ )
@pytest.fixture
def lowerCamelCase_ ( UpperCamelCase__ : List[str] ):
'''simple docstring'''
UpperCamelCase__ = tmp_path / '''csv_with_label.csv'''
UpperCamelCase__ = textwrap.dedent(
'''\
label
good
bad
good
''' )
with open(UpperCamelCase__, '''w''' ) as f:
f.write(UpperCamelCase__ )
return str(UpperCamelCase__ )
@pytest.fixture
def lowerCamelCase_ ( UpperCamelCase__ : Optional[int] ):
'''simple docstring'''
UpperCamelCase__ = tmp_path / '''csv_with_int_list.csv'''
UpperCamelCase__ = textwrap.dedent(
'''\
int_list
1 2 3
4 5 6
7 8 9
''' )
with open(UpperCamelCase__, '''w''' ) as f:
f.write(UpperCamelCase__ )
return str(UpperCamelCase__ )
def lowerCamelCase_ ( UpperCamelCase__ : int, UpperCamelCase__ : Optional[int], UpperCamelCase__ : Tuple ):
'''simple docstring'''
UpperCamelCase__ = Csv()
UpperCamelCase__ = csv._generate_tables([[csv_file, malformed_csv_file]] )
with pytest.raises(UpperCamelCase__, match='''Error tokenizing data''' ):
for _ in generator:
pass
assert any(
record.levelname == '''ERROR'''
and '''Failed to read file''' in record.message
and os.path.basename(UpperCamelCase__ ) in record.message
for record in caplog.records )
@require_pil
def lowerCamelCase_ ( UpperCamelCase__ : List[str] ):
'''simple docstring'''
with open(UpperCamelCase__, encoding='''utf-8''' ) as f:
UpperCamelCase__ = f.read().splitlines()[1]
UpperCamelCase__ = Csv(encoding='''utf-8''', features=Features({'''image''': Image()} ) )
UpperCamelCase__ = csv._generate_tables([[csv_file_with_image]] )
UpperCamelCase__ = pa.concat_tables([table for _, table in generator] )
assert pa_table.schema.field('''image''' ).type == Image()()
UpperCamelCase__ = pa_table.to_pydict()['''image''']
assert generated_content == [{"path": image_file, "bytes": None}]
def lowerCamelCase_ ( UpperCamelCase__ : Tuple ):
'''simple docstring'''
with open(UpperCamelCase__, encoding='''utf-8''' ) as f:
UpperCamelCase__ = f.read().splitlines()[1:]
UpperCamelCase__ = Csv(encoding='''utf-8''', features=Features({'''label''': ClassLabel(names=['''good''', '''bad'''] )} ) )
UpperCamelCase__ = csv._generate_tables([[csv_file_with_label]] )
UpperCamelCase__ = pa.concat_tables([table for _, table in generator] )
assert pa_table.schema.field('''label''' ).type == ClassLabel(names=['''good''', '''bad'''] )()
UpperCamelCase__ = pa_table.to_pydict()['''label''']
assert generated_content == [ClassLabel(names=['''good''', '''bad'''] ).straint(UpperCamelCase__ ) for label in labels]
def lowerCamelCase_ ( UpperCamelCase__ : Optional[int] ):
'''simple docstring'''
UpperCamelCase__ = Csv(encoding='''utf-8''', sep=''',''', converters={'''int_list''': lambda UpperCamelCase__ : [int(UpperCamelCase__ ) for i in x.split()]} )
UpperCamelCase__ = csv._generate_tables([[csv_file_with_int_list]] )
UpperCamelCase__ = pa.concat_tables([table for _, table in generator] )
assert pa.types.is_list(pa_table.schema.field('''int_list''' ).type )
UpperCamelCase__ = pa_table.to_pydict()['''int_list''']
assert generated_content == [[1, 2, 3], [4, 5, 6], [7, 8, 9]]
| 591
| 1
|
import argparse
import os
import re
import tensorflow as tf
import torch
from transformers import BertConfig, BertModel
from transformers.utils import logging
logging.set_verbosity_info()
UpperCamelCase__ = logging.get_logger(__name__)
def UpperCamelCase__ ( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) -> Optional[int]:
'''simple docstring'''
_lowercase : str = os.path.abspath(UpperCAmelCase_ )
logger.info(F'Converting TensorFlow checkpoint from {tf_path}' )
# Load weights from TF model
_lowercase : List[Any] = tf.train.list_variables(UpperCAmelCase_ )
_lowercase : Union[str, Any] = []
_lowercase : str = []
_lowercase : int = []
for full_name, shape in init_vars:
# logger.info(f"Loading TF weight {name} with shape {shape}")
_lowercase : Optional[int] = full_name.split('''/''' )
if full_name == "_CHECKPOINTABLE_OBJECT_GRAPH" or name[0] in ["global_step", "save_counter"]:
logger.info(F'Skipping non-model layer {full_name}' )
continue
if "optimizer" in full_name:
logger.info(F'Skipping optimization layer {full_name}' )
continue
if name[0] == "model":
# ignore initial 'model'
_lowercase : Optional[Any] = name[1:]
# figure out how many levels deep the name is
_lowercase : List[str] = 0
for _name in name:
if _name.startswith('''layer_with_weights''' ):
depth += 1
else:
break
layer_depth.append(UpperCAmelCase_ )
# read data
_lowercase : List[Any] = tf.train.load_variable(UpperCAmelCase_ , UpperCAmelCase_ )
names.append('''/'''.join(UpperCAmelCase_ ) )
arrays.append(UpperCAmelCase_ )
logger.info(F'Read a total of {len(UpperCAmelCase_ ):,} layers' )
# Sanity check
if len(set(UpperCAmelCase_ ) ) != 1:
raise ValueError(F'Found layer names with different depths (layer depth {list(set(UpperCAmelCase_ ) )})' )
_lowercase : Optional[int] = list(set(UpperCAmelCase_ ) )[0]
if layer_depth != 1:
raise ValueError(
'''The model contains more than just the embedding/encoder layers. This script does not handle MLM/NSP'''
''' heads.''' )
# convert layers
logger.info('''Converting weights...''' )
for full_name, array in zip(UpperCAmelCase_ , UpperCAmelCase_ ):
_lowercase : Dict = full_name.split('''/''' )
_lowercase : List[Any] = model
_lowercase : List[str] = []
for i, m_name in enumerate(UpperCAmelCase_ ):
if m_name == ".ATTRIBUTES":
# variable names end with .ATTRIBUTES/VARIABLE_VALUE
break
if m_name.startswith('''layer_with_weights''' ):
_lowercase : List[Any] = int(m_name.split('''-''' )[-1] )
if layer_num <= 2:
# embedding layers
# layer_num 0: word_embeddings
# layer_num 1: position_embeddings
# layer_num 2: token_type_embeddings
continue
elif layer_num == 3:
# embedding LayerNorm
trace.extend(['''embeddings''', '''LayerNorm'''] )
_lowercase : Any = getattr(UpperCAmelCase_ , '''embeddings''' )
_lowercase : str = getattr(UpperCAmelCase_ , '''LayerNorm''' )
elif layer_num > 3 and layer_num < config.num_hidden_layers + 4:
# encoder layers
trace.extend(['''encoder''', '''layer''', str(layer_num - 4 )] )
_lowercase : List[Any] = getattr(UpperCAmelCase_ , '''encoder''' )
_lowercase : Any = getattr(UpperCAmelCase_ , '''layer''' )
_lowercase : Union[str, Any] = pointer[layer_num - 4]
elif layer_num == config.num_hidden_layers + 4:
# pooler layer
trace.extend(['''pooler''', '''dense'''] )
_lowercase : Tuple = getattr(UpperCAmelCase_ , '''pooler''' )
_lowercase : Dict = getattr(UpperCAmelCase_ , '''dense''' )
elif m_name == "embeddings":
trace.append('''embeddings''' )
_lowercase : int = getattr(UpperCAmelCase_ , '''embeddings''' )
if layer_num == 0:
trace.append('''word_embeddings''' )
_lowercase : str = getattr(UpperCAmelCase_ , '''word_embeddings''' )
elif layer_num == 1:
trace.append('''position_embeddings''' )
_lowercase : List[Any] = getattr(UpperCAmelCase_ , '''position_embeddings''' )
elif layer_num == 2:
trace.append('''token_type_embeddings''' )
_lowercase : Any = getattr(UpperCAmelCase_ , '''token_type_embeddings''' )
else:
raise ValueError(F'Unknown embedding layer with name {full_name}' )
trace.append('''weight''' )
_lowercase : str = getattr(UpperCAmelCase_ , '''weight''' )
elif m_name == "_attention_layer":
# self-attention layer
trace.extend(['''attention''', '''self'''] )
_lowercase : Optional[int] = getattr(UpperCAmelCase_ , '''attention''' )
_lowercase : Dict = getattr(UpperCAmelCase_ , '''self''' )
elif m_name == "_attention_layer_norm":
# output attention norm
trace.extend(['''attention''', '''output''', '''LayerNorm'''] )
_lowercase : Any = getattr(UpperCAmelCase_ , '''attention''' )
_lowercase : Union[str, Any] = getattr(UpperCAmelCase_ , '''output''' )
_lowercase : str = getattr(UpperCAmelCase_ , '''LayerNorm''' )
elif m_name == "_attention_output_dense":
# output attention dense
trace.extend(['''attention''', '''output''', '''dense'''] )
_lowercase : Any = getattr(UpperCAmelCase_ , '''attention''' )
_lowercase : int = getattr(UpperCAmelCase_ , '''output''' )
_lowercase : Optional[Any] = getattr(UpperCAmelCase_ , '''dense''' )
elif m_name == "_output_dense":
# output dense
trace.extend(['''output''', '''dense'''] )
_lowercase : List[str] = getattr(UpperCAmelCase_ , '''output''' )
_lowercase : Optional[Any] = getattr(UpperCAmelCase_ , '''dense''' )
elif m_name == "_output_layer_norm":
# output dense
trace.extend(['''output''', '''LayerNorm'''] )
_lowercase : Union[str, Any] = getattr(UpperCAmelCase_ , '''output''' )
_lowercase : Optional[Any] = getattr(UpperCAmelCase_ , '''LayerNorm''' )
elif m_name == "_key_dense":
# attention key
trace.append('''key''' )
_lowercase : List[Any] = getattr(UpperCAmelCase_ , '''key''' )
elif m_name == "_query_dense":
# attention query
trace.append('''query''' )
_lowercase : Tuple = getattr(UpperCAmelCase_ , '''query''' )
elif m_name == "_value_dense":
# attention value
trace.append('''value''' )
_lowercase : int = getattr(UpperCAmelCase_ , '''value''' )
elif m_name == "_intermediate_dense":
# attention intermediate dense
trace.extend(['''intermediate''', '''dense'''] )
_lowercase : Tuple = getattr(UpperCAmelCase_ , '''intermediate''' )
_lowercase : List[Any] = getattr(UpperCAmelCase_ , '''dense''' )
elif m_name == "_output_layer_norm":
# output layer norm
trace.append('''output''' )
_lowercase : List[Any] = getattr(UpperCAmelCase_ , '''output''' )
# weights & biases
elif m_name in ["bias", "beta"]:
trace.append('''bias''' )
_lowercase : Optional[Any] = getattr(UpperCAmelCase_ , '''bias''' )
elif m_name in ["kernel", "gamma"]:
trace.append('''weight''' )
_lowercase : str = getattr(UpperCAmelCase_ , '''weight''' )
else:
logger.warning(F'Ignored {m_name}' )
# for certain layers reshape is necessary
_lowercase : Optional[int] = '''.'''.join(UpperCAmelCase_ )
if re.match(r'''(\S+)\.attention\.self\.(key|value|query)\.(bias|weight)''' , UpperCAmelCase_ ) or re.match(
r'''(\S+)\.attention\.output\.dense\.weight''' , UpperCAmelCase_ ):
_lowercase : str = array.reshape(pointer.data.shape )
if "kernel" in full_name:
_lowercase : str = array.transpose()
if pointer.shape == array.shape:
_lowercase : str = torch.from_numpy(UpperCAmelCase_ )
else:
raise ValueError(
F'Shape mismatch in layer {full_name}: Model expects shape {pointer.shape} but layer contains shape:'
F' {array.shape}' )
logger.info(F'Successfully set variable {full_name} to PyTorch layer {trace}' )
return model
def UpperCamelCase__ ( UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ ) -> List[str]:
'''simple docstring'''
logger.info(F'Loading model based on config from {config_path}...' )
_lowercase : Any = BertConfig.from_json_file(UpperCAmelCase_ )
_lowercase : int = BertModel(UpperCAmelCase_ )
# Load weights from checkpoint
logger.info(F'Loading weights from checkpoint {tf_checkpoint_path}...' )
load_tfa_weights_in_bert(UpperCAmelCase_ , UpperCAmelCase_ , UpperCAmelCase_ )
# Save pytorch-model
logger.info(F'Saving PyTorch model to {pytorch_dump_path}...' )
torch.save(model.state_dict() , UpperCAmelCase_ )
if __name__ == "__main__":
UpperCamelCase__ = argparse.ArgumentParser()
parser.add_argument(
'--tf_checkpoint_path', type=str, required=True, help='Path to the TensorFlow 2.x checkpoint path.'
)
parser.add_argument(
'--bert_config_file',
type=str,
required=True,
help='The config json file corresponding to the BERT model. This specifies the model architecture.',
)
parser.add_argument(
'--pytorch_dump_path',
type=str,
required=True,
help='Path to the output PyTorch model (must include filename).',
)
UpperCamelCase__ = parser.parse_args()
convert_tfa_checkpoint_to_pytorch(args.tf_checkpoint_path, args.bert_config_file, args.pytorch_dump_path)
| 322
|
import json
import logging
import os
import re
import sys
from dataclasses import dataclass, field
from typing import Any, Dict, List, Optional, Union
import datasets
import numpy as np
import torch
import torchaudio
from packaging import version
from torch import nn
import transformers
from transformers import (
HfArgumentParser,
Trainer,
TrainingArguments,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaForCTC,
WavaVecaProcessor,
is_apex_available,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint, is_main_process
if is_apex_available():
from apex import amp
if version.parse(version.parse(torch.__version__).base_version) >= version.parse('1.6'):
UpperCamelCase__ = True
from torch.cuda.amp import autocast
UpperCamelCase__ = logging.getLogger(__name__)
def UpperCamelCase__ ( UpperCAmelCase_=None , UpperCAmelCase_=None ) -> List[str]:
'''simple docstring'''
return field(default_factory=lambda: default , metadata=UpperCAmelCase_ )
@dataclass
class UpperCAmelCase__ :
'''simple docstring'''
UpperCAmelCase_ = field(
metadata={'''help''': '''Path to pretrained model or model identifier from huggingface.co/models'''} )
UpperCAmelCase_ = field(
default=A_ , metadata={'''help''': '''Where do you want to store the pretrained models downloaded from huggingface.co'''} , )
UpperCAmelCase_ = field(
default=A_ , metadata={'''help''': '''Whether to freeze the feature extractor layers of the model.'''} )
UpperCAmelCase_ = field(
default=0.1 , metadata={'''help''': '''The dropout ratio for the attention probabilities.'''} )
UpperCAmelCase_ = field(
default=0.1 , metadata={'''help''': '''The dropout ratio for activations inside the fully connected layer.'''} )
UpperCAmelCase_ = field(
default=0.1 , metadata={
'''help''': '''The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.'''
} , )
UpperCAmelCase_ = field(
default=0.1 , metadata={'''help''': '''The dropout probabilitiy for all 1D convolutional layers in feature extractor.'''} , )
UpperCAmelCase_ = field(
default=0.05 , metadata={
'''help''': (
'''Propability of each feature vector along the time axis to be chosen as the start of the vector'''
'''span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature'''
'''vectors will be masked along the time axis. This is only relevant if ``apply_spec_augment is True``.'''
)
} , )
UpperCAmelCase_ = field(default=0.0 , metadata={'''help''': '''The LayerDrop probability.'''} )
@dataclass
class UpperCAmelCase__ :
'''simple docstring'''
UpperCAmelCase_ = field(
default=A_ , metadata={'''help''': '''The configuration name of the dataset to use (via the datasets library).'''} )
UpperCAmelCase_ = field(
default='''train+validation''' , metadata={
'''help''': '''The name of the training data set split to use (via the datasets library). Defaults to \'train\''''
} , )
UpperCAmelCase_ = field(
default=A_ , metadata={'''help''': '''Overwrite the cached preprocessed datasets or not.'''} )
UpperCAmelCase_ = field(
default=A_ , metadata={'''help''': '''The number of processes to use for the preprocessing.'''} , )
UpperCAmelCase_ = field(
default=A_ , metadata={
'''help''': (
'''For debugging purposes or quicker training, truncate the number of training examples to this '''
'''value if set.'''
)
} , )
UpperCAmelCase_ = field(
default=A_ , metadata={
'''help''': (
'''For debugging purposes or quicker training, truncate the number of validation examples to this '''
'''value if set.'''
)
} , )
UpperCAmelCase_ = list_field(
default=[''',''', '''?''', '''.''', '''!''', '''-''', ''';''', ''':''', '''""''', '''%''', '''\'''', '''"''', '''�'''] , metadata={'''help''': '''A list of characters to remove from the transcripts.'''} , )
@dataclass
class UpperCAmelCase__ :
'''simple docstring'''
UpperCAmelCase_ = 42
UpperCAmelCase_ = True
UpperCAmelCase_ = None
UpperCAmelCase_ = None
UpperCAmelCase_ = None
UpperCAmelCase_ = None
def __call__( self : List[Any] , UpperCamelCase : List[Dict[str, Union[List[int], torch.Tensor]]] ):
"""simple docstring"""
_lowercase : int = [{'''input_values''': feature['''input_values''']} for feature in features]
_lowercase : Dict = [{'''input_ids''': feature['''labels''']} for feature in features]
_lowercase : int = self.processor.pad(
UpperCamelCase , padding=self.padding , max_length=self.max_length , pad_to_multiple_of=self.pad_to_multiple_of , return_tensors='''pt''' , )
_lowercase : Union[str, Any] = self.processor.pad(
labels=UpperCamelCase , padding=self.padding , max_length=self.max_length_labels , pad_to_multiple_of=self.pad_to_multiple_of_labels , return_tensors='''pt''' , )
# replace padding with -100 to ignore loss correctly
_lowercase : Optional[Any] = labels_batch['''input_ids'''].masked_fill(labels_batch.attention_mask.ne(1 ) , -1_00 )
_lowercase : Optional[Any] = labels
return batch
class UpperCAmelCase__ ( A_ ):
'''simple docstring'''
def lowerCAmelCase_ ( self : List[str] , UpperCamelCase : nn.Module , UpperCamelCase : Dict[str, Union[torch.Tensor, Any]] ):
"""simple docstring"""
model.train()
_lowercase : Tuple = self._prepare_inputs(UpperCamelCase )
if self.use_amp:
with autocast():
_lowercase : Union[str, Any] = self.compute_loss(UpperCamelCase , UpperCamelCase )
else:
_lowercase : List[str] = self.compute_loss(UpperCamelCase , UpperCamelCase )
if self.args.n_gpu > 1:
if model.module.config.ctc_loss_reduction == "mean":
_lowercase : str = loss.mean()
elif model.module.config.ctc_loss_reduction == "sum":
_lowercase : Optional[Any] = loss.sum() / (inputs['''labels'''] >= 0).sum()
else:
raise ValueError(F'{model.config.ctc_loss_reduction} is not valid. Choose one of [\'mean\', \'sum\']' )
if self.args.gradient_accumulation_steps > 1:
_lowercase : Optional[int] = loss / self.args.gradient_accumulation_steps
if self.use_amp:
self.scaler.scale(UpperCamelCase ).backward()
elif self.use_apex:
with amp.scale_loss(UpperCamelCase , self.optimizer ) as scaled_loss:
scaled_loss.backward()
elif self.deepspeed:
self.deepspeed.backward(UpperCamelCase )
else:
loss.backward()
return loss.detach()
def UpperCamelCase__ ( ) -> Optional[Any]:
'''simple docstring'''
_lowercase : List[str] = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments) )
if len(sys.argv ) == 2 and sys.argv[1].endswith('''.json''' ):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
_lowercase , _lowercase , _lowercase : Any = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
_lowercase , _lowercase , _lowercase : int = parser.parse_args_into_dataclasses()
# Detecting last checkpoint.
_lowercase : Dict = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
_lowercase : Optional[int] = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
F'Output directory ({training_args.output_dir}) already exists and is not empty. '
'''Use --overwrite_output_dir to overcome.''' )
elif last_checkpoint is not None:
logger.info(
F'Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change '
'''the `--output_dir` or add `--overwrite_output_dir` to train from scratch.''' )
# Setup logging
logging.basicConfig(
format='''%(asctime)s - %(levelname)s - %(name)s - %(message)s''' , datefmt='''%m/%d/%Y %H:%M:%S''' , handlers=[logging.StreamHandler(sys.stdout )] , )
logger.setLevel(logging.INFO if is_main_process(training_args.local_rank ) else logging.WARN )
# Log on each process the small summary:
logger.warning(
F'Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}'
+ F'distributed training: {bool(training_args.local_rank != -1 )}, 16-bits training: {training_args.fpaa}' )
# Set the verbosity to info of the Transformers logger (on main process only):
if is_main_process(training_args.local_rank ):
transformers.utils.logging.set_verbosity_info()
logger.info('''Training/evaluation parameters %s''' , UpperCAmelCase_ )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets:
_lowercase : Tuple = datasets.load_dataset(
'''common_voice''' , data_args.dataset_config_name , split=data_args.train_split_name )
_lowercase : Dict = datasets.load_dataset('''common_voice''' , data_args.dataset_config_name , split='''test''' )
# Create and save tokenizer
_lowercase : Tuple = F'[{"".join(data_args.chars_to_ignore )}]'
def remove_special_characters(UpperCAmelCase_ ):
_lowercase : List[Any] = re.sub(UpperCAmelCase_ , '''''' , batch['''sentence'''] ).lower() + ''' '''
return batch
_lowercase : Tuple = train_dataset.map(UpperCAmelCase_ , remove_columns=['''sentence'''] )
_lowercase : int = eval_dataset.map(UpperCAmelCase_ , remove_columns=['''sentence'''] )
def extract_all_chars(UpperCAmelCase_ ):
_lowercase : int = ''' '''.join(batch['''text'''] )
_lowercase : int = list(set(UpperCAmelCase_ ) )
return {"vocab": [vocab], "all_text": [all_text]}
_lowercase : List[Any] = train_dataset.map(
UpperCAmelCase_ , batched=UpperCAmelCase_ , batch_size=-1 , keep_in_memory=UpperCAmelCase_ , remove_columns=train_dataset.column_names , )
_lowercase : Any = train_dataset.map(
UpperCAmelCase_ , batched=UpperCAmelCase_ , batch_size=-1 , keep_in_memory=UpperCAmelCase_ , remove_columns=eval_dataset.column_names , )
_lowercase : Optional[int] = list(set(vocab_train['''vocab'''][0] ) | set(vocab_test['''vocab'''][0] ) )
_lowercase : str = {v: k for k, v in enumerate(UpperCAmelCase_ )}
_lowercase : Dict = vocab_dict[''' ''']
del vocab_dict[" "]
_lowercase : Any = len(UpperCAmelCase_ )
_lowercase : str = len(UpperCAmelCase_ )
with open('''vocab.json''' , '''w''' ) as vocab_file:
json.dump(UpperCAmelCase_ , UpperCAmelCase_ )
# Load pretrained model and tokenizer
#
# Distributed training:
# The .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
_lowercase : List[str] = WavaVecaCTCTokenizer(
'''vocab.json''' , unk_token='''[UNK]''' , pad_token='''[PAD]''' , word_delimiter_token='''|''' , )
_lowercase : str = WavaVecaFeatureExtractor(
feature_size=1 , sampling_rate=16000 , padding_value=0.0 , do_normalize=UpperCAmelCase_ , return_attention_mask=UpperCAmelCase_ )
_lowercase : int = WavaVecaProcessor(feature_extractor=UpperCAmelCase_ , tokenizer=UpperCAmelCase_ )
_lowercase : str = WavaVecaForCTC.from_pretrained(
model_args.model_name_or_path , cache_dir=model_args.cache_dir , activation_dropout=model_args.activation_dropout , attention_dropout=model_args.attention_dropout , hidden_dropout=model_args.hidden_dropout , feat_proj_dropout=model_args.feat_proj_dropout , mask_time_prob=model_args.mask_time_prob , gradient_checkpointing=training_args.gradient_checkpointing , layerdrop=model_args.layerdrop , ctc_loss_reduction='''mean''' , pad_token_id=processor.tokenizer.pad_token_id , vocab_size=len(processor.tokenizer ) , )
if data_args.max_train_samples is not None:
_lowercase : List[str] = min(len(UpperCAmelCase_ ) , data_args.max_train_samples )
_lowercase : Tuple = train_dataset.select(range(UpperCAmelCase_ ) )
if data_args.max_val_samples is not None:
_lowercase : List[str] = eval_dataset.select(range(data_args.max_val_samples ) )
_lowercase : Tuple = torchaudio.transforms.Resample(48000 , 16000 )
# Preprocessing the datasets.
# We need to read the aduio files as arrays and tokenize the targets.
def speech_file_to_array_fn(UpperCAmelCase_ ):
_lowercase , _lowercase : List[Any] = torchaudio.load(batch['''path'''] )
_lowercase : Optional[int] = resampler(UpperCAmelCase_ ).squeeze().numpy()
_lowercase : Any = 16000
_lowercase : List[str] = batch['''text''']
return batch
_lowercase : Union[str, Any] = train_dataset.map(
UpperCAmelCase_ , remove_columns=train_dataset.column_names , num_proc=data_args.preprocessing_num_workers , )
_lowercase : Union[str, Any] = eval_dataset.map(
UpperCAmelCase_ , remove_columns=eval_dataset.column_names , num_proc=data_args.preprocessing_num_workers , )
def prepare_dataset(UpperCAmelCase_ ):
# check that all files have the correct sampling rate
assert (
len(set(batch['''sampling_rate'''] ) ) == 1
), F'Make sure all inputs have the same sampling rate of {processor.feature_extractor.sampling_rate}.'
_lowercase : Dict = processor(
audio=batch['''speech'''] , text=batch['''target_text'''] , sampling_rate=batch['''sampling_rate'''][0] )
batch.update(UpperCAmelCase_ )
return batch
_lowercase : Any = train_dataset.map(
UpperCAmelCase_ , remove_columns=train_dataset.column_names , batch_size=training_args.per_device_train_batch_size , batched=UpperCAmelCase_ , num_proc=data_args.preprocessing_num_workers , )
_lowercase : Optional[Any] = eval_dataset.map(
UpperCAmelCase_ , remove_columns=eval_dataset.column_names , batch_size=training_args.per_device_train_batch_size , batched=UpperCAmelCase_ , num_proc=data_args.preprocessing_num_workers , )
# Metric
_lowercase : Any = datasets.load_metric('''wer''' )
def compute_metrics(UpperCAmelCase_ ):
_lowercase : Optional[Any] = pred.predictions
_lowercase : Dict = np.argmax(UpperCAmelCase_ , axis=-1 )
_lowercase : Optional[int] = processor.tokenizer.pad_token_id
_lowercase : List[Any] = processor.batch_decode(UpperCAmelCase_ )
# we do not want to group tokens when computing the metrics
_lowercase : str = processor.batch_decode(pred.label_ids , group_tokens=UpperCAmelCase_ )
_lowercase : Union[str, Any] = wer_metric.compute(predictions=UpperCAmelCase_ , references=UpperCAmelCase_ )
return {"wer": wer}
if model_args.freeze_feature_extractor:
model.freeze_feature_extractor()
# Data collator
_lowercase : List[str] = DataCollatorCTCWithPadding(processor=UpperCAmelCase_ , padding=UpperCAmelCase_ )
# Initialize our Trainer
_lowercase : Dict = CTCTrainer(
model=UpperCAmelCase_ , data_collator=UpperCAmelCase_ , args=UpperCAmelCase_ , compute_metrics=UpperCAmelCase_ , train_dataset=train_dataset if training_args.do_train else None , eval_dataset=eval_dataset if training_args.do_eval else None , tokenizer=processor.feature_extractor , )
# Training
if training_args.do_train:
if last_checkpoint is not None:
_lowercase : Optional[Any] = last_checkpoint
elif os.path.isdir(model_args.model_name_or_path ):
_lowercase : Tuple = model_args.model_name_or_path
else:
_lowercase : Tuple = None
# Save the feature_extractor and the tokenizer
if is_main_process(training_args.local_rank ):
processor.save_pretrained(training_args.output_dir )
_lowercase : Union[str, Any] = trainer.train(resume_from_checkpoint=UpperCAmelCase_ )
trainer.save_model()
_lowercase : Any = train_result.metrics
_lowercase : str = (
data_args.max_train_samples if data_args.max_train_samples is not None else len(UpperCAmelCase_ )
)
_lowercase : Dict = min(UpperCAmelCase_ , len(UpperCAmelCase_ ) )
trainer.log_metrics('''train''' , UpperCAmelCase_ )
trainer.save_metrics('''train''' , UpperCAmelCase_ )
trainer.save_state()
# Evaluation
_lowercase : Optional[Any] = {}
if training_args.do_eval:
logger.info('''*** Evaluate ***''' )
_lowercase : Any = trainer.evaluate()
_lowercase : Union[str, Any] = data_args.max_val_samples if data_args.max_val_samples is not None else len(UpperCAmelCase_ )
_lowercase : str = min(UpperCAmelCase_ , len(UpperCAmelCase_ ) )
trainer.log_metrics('''eval''' , UpperCAmelCase_ )
trainer.save_metrics('''eval''' , UpperCAmelCase_ )
return results
if __name__ == "__main__":
main()
| 322
| 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
UpperCamelCase = logging.get_logger(__name__)
# General docstring
UpperCamelCase = """RegNetConfig"""
# Base docstring
UpperCamelCase = """facebook/regnet-y-040"""
UpperCamelCase = [1, 1088, 7, 7]
# Image classification docstring
UpperCamelCase = """facebook/regnet-y-040"""
UpperCamelCase = """tabby, tabby cat"""
UpperCamelCase = [
"""facebook/regnet-y-040""",
# See all regnet models at https://huggingface.co/models?filter=regnet
]
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = 3 , _SCREAMING_SNAKE_CASE = 1 , _SCREAMING_SNAKE_CASE = 1 , _SCREAMING_SNAKE_CASE = "relu" , **_SCREAMING_SNAKE_CASE , )->str:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
# The padding and conv has been verified in
# https://colab.research.google.com/gist/sayakpaul/854bc10eeaf21c9ee2119e0b9f3841a7/scratchpad.ipynb
A_ : Dict = tf.keras.layers.ZeroPaddingaD(padding=kernel_size // 2 )
A_ : Optional[Any] = tf.keras.layers.ConvaD(
filters=_lowerCAmelCase , kernel_size=_lowerCAmelCase , strides=_lowerCAmelCase , padding='''VALID''' , groups=_lowerCAmelCase , use_bias=_lowerCAmelCase , name='''convolution''' , )
A_ : Any = tf.keras.layers.BatchNormalization(epsilon=1e-5 , momentum=0.9 , name='''normalization''' )
A_ : Optional[int] = ACTaFN[activation] if activation is not None else tf.identity
def _snake_case ( self , _SCREAMING_SNAKE_CASE )->str:
'''simple docstring'''
A_ : Tuple = self.convolution(self.padding(_lowerCAmelCase ) )
A_ : Union[str, Any] = self.normalization(_lowerCAmelCase )
A_ : List[Any] = self.activation(_lowerCAmelCase )
return hidden_state
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )->Union[str, Any]:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
A_ : List[Any] = config.num_channels
A_ : Dict = TFRegNetConvLayer(
out_channels=config.embedding_size , kernel_size=3 , stride=2 , activation=config.hidden_act , name='''embedder''' , )
def _snake_case ( self , _SCREAMING_SNAKE_CASE )->Optional[int]:
'''simple docstring'''
A_ : Any = shape_list(_lowerCAmelCase )[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)
A_ : Union[str, Any] = tf.transpose(_lowerCAmelCase , perm=(0, 2, 3, 1) )
A_ : List[Any] = self.embedder(_lowerCAmelCase )
return hidden_state
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = 2 , **_SCREAMING_SNAKE_CASE )->List[str]:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
A_ : Optional[int] = tf.keras.layers.ConvaD(
filters=_lowerCAmelCase , kernel_size=1 , strides=_lowerCAmelCase , use_bias=_lowerCAmelCase , name='''convolution''' )
A_ : Union[str, Any] = tf.keras.layers.BatchNormalization(epsilon=1e-5 , momentum=0.9 , name='''normalization''' )
def _snake_case ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = False )->tf.Tensor:
'''simple docstring'''
return self.normalization(self.convolution(_lowerCAmelCase ) , training=_lowerCAmelCase )
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )->Dict:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
A_ : List[str] = tf.keras.layers.GlobalAveragePoolingaD(keepdims=_lowerCAmelCase , name='''pooler''' )
A_ : Optional[int] = [
tf.keras.layers.ConvaD(filters=_lowerCAmelCase , kernel_size=1 , activation='''relu''' , name='''attention.0''' ),
tf.keras.layers.ConvaD(filters=_lowerCAmelCase , kernel_size=1 , activation='''sigmoid''' , name='''attention.2''' ),
]
def _snake_case ( self , _SCREAMING_SNAKE_CASE )->Optional[Any]:
'''simple docstring'''
A_ : List[Any] = self.pooler(_lowerCAmelCase )
for layer_module in self.attention:
A_ : Union[str, Any] = layer_module(_lowerCAmelCase )
A_ : Union[str, Any] = hidden_state * pooled
return hidden_state
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = 1 , **_SCREAMING_SNAKE_CASE )->List[str]:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
A_ : int = in_channels != out_channels or stride != 1
A_ : Dict = max(1 , out_channels // config.groups_width )
A_ : Dict = (
TFRegNetShortCut(_lowerCAmelCase , stride=_lowerCAmelCase , 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.
A_ : str = [
TFRegNetConvLayer(_lowerCAmelCase , kernel_size=1 , activation=config.hidden_act , name='''layer.0''' ),
TFRegNetConvLayer(
_lowerCAmelCase , stride=_lowerCAmelCase , groups=_lowerCAmelCase , activation=config.hidden_act , name='''layer.1''' ),
TFRegNetConvLayer(_lowerCAmelCase , kernel_size=1 , activation=_lowerCAmelCase , name='''layer.2''' ),
]
A_ : Tuple = ACTaFN[config.hidden_act]
def _snake_case ( self , _SCREAMING_SNAKE_CASE )->int:
'''simple docstring'''
A_ : str = hidden_state
for layer_module in self.layers:
A_ : Optional[Any] = layer_module(_lowerCAmelCase )
A_ : Optional[int] = self.shortcut(_lowerCAmelCase )
hidden_state += residual
A_ : Optional[Any] = self.activation(_lowerCAmelCase )
return hidden_state
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = 1 , **_SCREAMING_SNAKE_CASE )->Tuple:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
A_ : Dict = in_channels != out_channels or stride != 1
A_ : List[Any] = max(1 , out_channels // config.groups_width )
A_ : Optional[Any] = (
TFRegNetShortCut(_lowerCAmelCase , stride=_lowerCAmelCase , name='''shortcut''' )
if should_apply_shortcut
else tf.keras.layers.Activation('''linear''' , name='''shortcut''' )
)
A_ : Union[str, Any] = [
TFRegNetConvLayer(_lowerCAmelCase , kernel_size=1 , activation=config.hidden_act , name='''layer.0''' ),
TFRegNetConvLayer(
_lowerCAmelCase , stride=_lowerCAmelCase , groups=_lowerCAmelCase , activation=config.hidden_act , name='''layer.1''' ),
TFRegNetSELayer(_lowerCAmelCase , reduced_channels=int(round(in_channels / 4 ) ) , name='''layer.2''' ),
TFRegNetConvLayer(_lowerCAmelCase , kernel_size=1 , activation=_lowerCAmelCase , name='''layer.3''' ),
]
A_ : Optional[int] = ACTaFN[config.hidden_act]
def _snake_case ( self , _SCREAMING_SNAKE_CASE )->Dict:
'''simple docstring'''
A_ : Optional[Any] = hidden_state
for layer_module in self.layers:
A_ : int = layer_module(_lowerCAmelCase )
A_ : Dict = self.shortcut(_lowerCAmelCase )
hidden_state += residual
A_ : Any = self.activation(_lowerCAmelCase )
return hidden_state
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = 2 , _SCREAMING_SNAKE_CASE = 2 , **_SCREAMING_SNAKE_CASE )->Optional[Any]:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
A_ : Dict = TFRegNetXLayer if config.layer_type == '''x''' else TFRegNetYLayer
A_ : Any = [
# downsampling is done in the first layer with stride of 2
layer(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , stride=_lowerCAmelCase , name='''layers.0''' ),
*[layer(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , name=F'''layers.{i+1}''' ) for i in range(depth - 1 )],
]
def _snake_case ( self , _SCREAMING_SNAKE_CASE )->Optional[int]:
'''simple docstring'''
for layer_module in self.layers:
A_ : Any = layer_module(_lowerCAmelCase )
return hidden_state
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )->Union[str, Any]:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
A_ : List[Any] = []
# based on `downsample_in_first_stage`, the first layer of the first stage may or may not downsample the input
self.stages.append(
TFRegNetStage(
_lowerCAmelCase , config.embedding_size , config.hidden_sizes[0] , stride=2 if config.downsample_in_first_stage else 1 , depth=config.depths[0] , name='''stages.0''' , ) )
A_ : Optional[int] = zip(config.hidden_sizes , config.hidden_sizes[1:] )
for i, ((in_channels, out_channels), depth) in enumerate(zip(_lowerCAmelCase , config.depths[1:] ) ):
self.stages.append(TFRegNetStage(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , depth=_lowerCAmelCase , name=F'''stages.{i+1}''' ) )
def _snake_case ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = False , _SCREAMING_SNAKE_CASE = True )->TFBaseModelOutputWithNoAttention:
'''simple docstring'''
A_ : List[Any] = () if output_hidden_states else None
for stage_module in self.stages:
if output_hidden_states:
A_ : List[str] = hidden_states + (hidden_state,)
A_ : int = stage_module(_lowerCAmelCase )
if output_hidden_states:
A_ : Union[str, Any] = 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=_lowerCAmelCase , hidden_states=_lowerCAmelCase )
@keras_serializable
class _lowerCamelCase ( tf.keras.layers.Layer ):
"""simple docstring"""
snake_case = RegNetConfig
def __init__( self , _SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )->Union[str, Any]:
'''simple docstring'''
super().__init__(**_lowerCAmelCase )
A_ : Any = config
A_ : Optional[int] = TFRegNetEmbeddings(_lowerCAmelCase , name='''embedder''' )
A_ : Tuple = TFRegNetEncoder(_lowerCAmelCase , name='''encoder''' )
A_ : Dict = tf.keras.layers.GlobalAveragePoolingaD(keepdims=_lowerCAmelCase , name='''pooler''' )
@unpack_inputs
def _snake_case ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = None , _SCREAMING_SNAKE_CASE = None , _SCREAMING_SNAKE_CASE = False , )->TFBaseModelOutputWithPoolingAndNoAttention:
'''simple docstring'''
A_ : List[str] = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
A_ : List[Any] = return_dict if return_dict is not None else self.config.use_return_dict
A_ : Dict = self.embedder(_lowerCAmelCase , training=_lowerCAmelCase )
A_ : int = self.encoder(
_lowerCAmelCase , output_hidden_states=_lowerCAmelCase , return_dict=_lowerCAmelCase , training=_lowerCAmelCase )
A_ : Any = encoder_outputs[0]
A_ : Tuple = self.pooler(_lowerCAmelCase )
# Change to NCHW output format have uniformity in the modules
A_ : Tuple = tf.transpose(_lowerCAmelCase , perm=(0, 3, 1, 2) )
A_ : Optional[Any] = tf.transpose(_lowerCAmelCase , perm=(0, 3, 1, 2) )
# Change the other hidden state outputs to NCHW as well
if output_hidden_states:
A_ : Tuple = tuple([tf.transpose(_lowerCAmelCase , 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=_lowerCAmelCase , pooler_output=_lowerCAmelCase , hidden_states=hidden_states if output_hidden_states else encoder_outputs.hidden_states , )
class _lowerCamelCase ( _lowerCAmelCase ):
"""simple docstring"""
snake_case = RegNetConfig
snake_case = 'regnet'
snake_case = 'pixel_values'
@property
def _snake_case ( self )->Any:
'''simple docstring'''
return {"pixel_values": tf.TensorSpec(shape=(None, self.config.num_channels, 224, 224) , dtype=tf.floataa )}
UpperCamelCase = 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.
"""
UpperCamelCase = 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 ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , *_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )->List[Any]:
'''simple docstring'''
super().__init__(_lowerCAmelCase , *_lowerCAmelCase , **_lowerCAmelCase )
A_ : Optional[Any] = TFRegNetMainLayer(_lowerCAmelCase , name='''regnet''' )
@unpack_inputs
@add_start_docstrings_to_model_forward(_lowerCAmelCase )
@add_code_sample_docstrings(
checkpoint=_CHECKPOINT_FOR_DOC , output_type=_lowerCAmelCase , config_class=_CONFIG_FOR_DOC , modality='''vision''' , expected_output=_EXPECTED_OUTPUT_SHAPE , )
def _snake_case ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE = None , _SCREAMING_SNAKE_CASE = None , _SCREAMING_SNAKE_CASE=False , )->Union[TFBaseModelOutputWithPoolingAndNoAttention, Tuple[tf.Tensor]]:
'''simple docstring'''
A_ : Optional[Any] = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
A_ : int = return_dict if return_dict is not None else self.config.use_return_dict
A_ : Dict = self.regnet(
pixel_values=_lowerCAmelCase , output_hidden_states=_lowerCAmelCase , return_dict=_lowerCAmelCase , training=_lowerCAmelCase , )
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(
"\n RegNet Model with an image classification head on top (a linear layer on top of the pooled features), e.g. for\n ImageNet.\n " , _lowerCAmelCase , )
class _lowerCamelCase ( _lowerCAmelCase , _lowerCAmelCase ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE , *_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )->int:
'''simple docstring'''
super().__init__(_lowerCAmelCase , *_lowerCAmelCase , **_lowerCAmelCase )
A_ : Optional[Any] = config.num_labels
A_ : Optional[int] = TFRegNetMainLayer(_lowerCAmelCase , name='''regnet''' )
# classification head
A_ : str = [
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(_lowerCAmelCase )
@add_code_sample_docstrings(
checkpoint=_IMAGE_CLASS_CHECKPOINT , output_type=_lowerCAmelCase , config_class=_CONFIG_FOR_DOC , expected_output=_IMAGE_CLASS_EXPECTED_OUTPUT , )
def _snake_case ( self , _SCREAMING_SNAKE_CASE = None , _SCREAMING_SNAKE_CASE = None , _SCREAMING_SNAKE_CASE = None , _SCREAMING_SNAKE_CASE = None , _SCREAMING_SNAKE_CASE=False , )->Union[TFSequenceClassifierOutput, Tuple[tf.Tensor]]:
'''simple docstring'''
A_ : List[Any] = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
A_ : List[Any] = return_dict if return_dict is not None else self.config.use_return_dict
A_ : Dict = self.regnet(
_lowerCAmelCase , output_hidden_states=_lowerCAmelCase , return_dict=_lowerCAmelCase , training=_lowerCAmelCase )
A_ : Dict = outputs.pooler_output if return_dict else outputs[1]
A_ : List[str] = self.classifier[0](_lowerCAmelCase )
A_ : Optional[Any] = self.classifier[1](_lowerCAmelCase )
A_ : Optional[Any] = None if labels is None else self.hf_compute_loss(labels=_lowerCAmelCase , logits=_lowerCAmelCase )
if not return_dict:
A_ : Union[str, Any] = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return TFSequenceClassifierOutput(loss=_lowerCAmelCase , logits=_lowerCAmelCase , hidden_states=outputs.hidden_states )
| 709
|
from typing import Dict
from .base import GenericTensor, Pipeline
class _lowerCamelCase ( UpperCamelCase ):
"""simple docstring"""
def _snake_case ( self , _SCREAMING_SNAKE_CASE=None , _SCREAMING_SNAKE_CASE=None , _SCREAMING_SNAKE_CASE=None , **_SCREAMING_SNAKE_CASE )->Optional[int]:
'''simple docstring'''
if tokenize_kwargs is None:
A_ : Optional[int] = {}
if truncation is not None:
if "truncation" in tokenize_kwargs:
raise ValueError(
'''truncation parameter defined twice (given as keyword argument as well as in tokenize_kwargs)''' )
A_ : Optional[Any] = truncation
A_ : Dict = tokenize_kwargs
A_ : Union[str, Any] = {}
if return_tensors is not None:
A_ : Union[str, Any] = return_tensors
return preprocess_params, {}, postprocess_params
def _snake_case ( self , _SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )->Dict[str, GenericTensor]:
'''simple docstring'''
A_ : Optional[Any] = self.framework
A_ : Tuple = self.tokenizer(_SCREAMING_SNAKE_CASE , return_tensors=_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )
return model_inputs
def _snake_case ( self , _SCREAMING_SNAKE_CASE )->Optional[int]:
'''simple docstring'''
A_ : str = self.model(**_SCREAMING_SNAKE_CASE )
return model_outputs
def _snake_case ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE=False )->Any:
'''simple docstring'''
if return_tensors:
return model_outputs[0]
if self.framework == "pt":
return model_outputs[0].tolist()
elif self.framework == "tf":
return model_outputs[0].numpy().tolist()
def __call__( self , *_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )->List[str]:
'''simple docstring'''
return super().__call__(*_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )
| 152
| 0
|
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_torch_available,
)
lowerCamelCase = {
"""configuration_encodec""": [
"""ENCODEC_PRETRAINED_CONFIG_ARCHIVE_MAP""",
"""EncodecConfig""",
],
"""feature_extraction_encodec""": ["""EncodecFeatureExtractor"""],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
lowerCamelCase = [
"""ENCODEC_PRETRAINED_MODEL_ARCHIVE_LIST""",
"""EncodecModel""",
"""EncodecPreTrainedModel""",
]
if TYPE_CHECKING:
from .configuration_encodec import (
ENCODEC_PRETRAINED_CONFIG_ARCHIVE_MAP,
EncodecConfig,
)
from .feature_extraction_encodec import EncodecFeatureExtractor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_encodec import (
ENCODEC_PRETRAINED_MODEL_ARCHIVE_LIST,
EncodecModel,
EncodecPreTrainedModel,
)
else:
import sys
lowerCamelCase = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 191
|
import unittest
import numpy as np
import requests
from transformers.testing_utils import require_torch, require_vision
from transformers.utils import is_torch_available, is_vision_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin, prepare_image_inputs
if is_torch_available():
import torch
from transformers.pytorch_utils import is_torch_greater_or_equal_than_1_11
else:
lowerCamelCase = False
if is_vision_available():
from PIL import Image
from transformers import PixaStructImageProcessor
class _a ( unittest.TestCase ):
'''simple docstring'''
def __init__( self , __UpperCAmelCase , __UpperCAmelCase=7 , __UpperCAmelCase=3 , __UpperCAmelCase=18 , __UpperCAmelCase=30 , __UpperCAmelCase=400 , __UpperCAmelCase=None , __UpperCAmelCase=True , __UpperCAmelCase=True , __UpperCAmelCase=None , ):
"""simple docstring"""
a__ : str = size if size is not None else {"height": 20, "width": 20}
a__ : Optional[int] = parent
a__ : Union[str, Any] = batch_size
a__ : List[Any] = num_channels
a__ : Union[str, Any] = image_size
a__ : List[str] = min_resolution
a__ : int = max_resolution
a__ : int = size
a__ : List[str] = do_normalize
a__ : Any = do_convert_rgb
a__ : Optional[Any] = [512, 1024, 2048, 4096]
a__ : Any = patch_size if patch_size is not None else {"height": 16, "width": 16}
def _A ( self ):
"""simple docstring"""
return {"do_normalize": self.do_normalize, "do_convert_rgb": self.do_convert_rgb}
def _A ( self ):
"""simple docstring"""
a__ : str = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/australia.jpg"
a__ : Union[str, Any] = Image.open(requests.get(__UpperCAmelCase , stream=__UpperCAmelCase ).raw ).convert("RGB" )
return raw_image
@unittest.skipIf(
not is_torch_greater_or_equal_than_1_11 , reason="`Pix2StructImageProcessor` requires `torch>=1.11.0`." , )
@require_torch
@require_vision
class _a ( SCREAMING_SNAKE_CASE , unittest.TestCase ):
'''simple docstring'''
A :List[str] = PixaStructImageProcessor if is_vision_available() else None
def _A ( self ):
"""simple docstring"""
a__ : Any = PixaStructImageProcessingTester(self )
@property
def _A ( self ):
"""simple docstring"""
return self.image_processor_tester.prepare_image_processor_dict()
def _A ( self ):
"""simple docstring"""
a__ : Union[str, Any] = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(__UpperCAmelCase , "do_normalize" ) )
self.assertTrue(hasattr(__UpperCAmelCase , "do_convert_rgb" ) )
def _A ( self ):
"""simple docstring"""
a__ : Tuple = self.image_processor_tester.prepare_dummy_image()
a__ : Any = self.image_processing_class(**self.image_processor_dict )
a__ : List[str] = 2048
a__ : Union[str, Any] = image_processor(__UpperCAmelCase , return_tensors="pt" , max_patches=__UpperCAmelCase )
self.assertTrue(torch.allclose(inputs.flattened_patches.mean() , torch.tensor(0.0_6_0_6 ) , atol=1E-3 , rtol=1E-3 ) )
def _A ( self ):
"""simple docstring"""
a__ : int = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
a__ : str = prepare_image_inputs(self.image_processor_tester , equal_resolution=__UpperCAmelCase )
for image in image_inputs:
self.assertIsInstance(__UpperCAmelCase , Image.Image )
# Test not batched input
a__ : List[Any] = (
(self.image_processor_tester.patch_size["height"] * self.image_processor_tester.patch_size["width"])
* self.image_processor_tester.num_channels
) + 2
for max_patch in self.image_processor_tester.max_patches:
# Test not batched input
a__ : Optional[Any] = image_processor(
image_inputs[0] , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (1, max_patch, expected_hidden_dim) , )
# Test batched
a__ : Optional[Any] = image_processor(
__UpperCAmelCase , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (self.image_processor_tester.batch_size, max_patch, expected_hidden_dim) , )
def _A ( self ):
"""simple docstring"""
a__ : Union[str, Any] = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
a__ : Any = prepare_image_inputs(self.image_processor_tester , equal_resolution=__UpperCAmelCase )
for image in image_inputs:
self.assertIsInstance(__UpperCAmelCase , Image.Image )
# Test not batched input
a__ : Optional[int] = (
(self.image_processor_tester.patch_size["height"] * self.image_processor_tester.patch_size["width"])
* self.image_processor_tester.num_channels
) + 2
a__ : int = True
for max_patch in self.image_processor_tester.max_patches:
# Test not batched input
with self.assertRaises(__UpperCAmelCase ):
a__ : List[str] = image_processor(
image_inputs[0] , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
a__ : Tuple = "Hello"
a__ : Union[str, Any] = image_processor(
image_inputs[0] , return_tensors="pt" , max_patches=__UpperCAmelCase , header_text=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (1, max_patch, expected_hidden_dim) , )
# Test batched
a__ : Optional[int] = image_processor(
__UpperCAmelCase , return_tensors="pt" , max_patches=__UpperCAmelCase , header_text=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (self.image_processor_tester.batch_size, max_patch, expected_hidden_dim) , )
def _A ( self ):
"""simple docstring"""
a__ : str = self.image_processing_class(**self.image_processor_dict )
# create random numpy tensors
a__ : Tuple = prepare_image_inputs(self.image_processor_tester , equal_resolution=__UpperCAmelCase , numpify=__UpperCAmelCase )
for image in image_inputs:
self.assertIsInstance(__UpperCAmelCase , np.ndarray )
a__ : List[str] = (
(self.image_processor_tester.patch_size["height"] * self.image_processor_tester.patch_size["width"])
* self.image_processor_tester.num_channels
) + 2
for max_patch in self.image_processor_tester.max_patches:
# Test not batched input
a__ : Any = image_processor(
image_inputs[0] , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (1, max_patch, expected_hidden_dim) , )
# Test batched
a__ : str = image_processor(
__UpperCAmelCase , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (self.image_processor_tester.batch_size, max_patch, expected_hidden_dim) , )
def _A ( self ):
"""simple docstring"""
a__ : List[Any] = self.image_processing_class(**self.image_processor_dict )
# create random PyTorch tensors
a__ : Any = prepare_image_inputs(self.image_processor_tester , equal_resolution=__UpperCAmelCase , torchify=__UpperCAmelCase )
for image in image_inputs:
self.assertIsInstance(__UpperCAmelCase , torch.Tensor )
# Test not batched input
a__ : Optional[Any] = (
(self.image_processor_tester.patch_size["height"] * self.image_processor_tester.patch_size["width"])
* self.image_processor_tester.num_channels
) + 2
for max_patch in self.image_processor_tester.max_patches:
# Test not batched input
a__ : str = image_processor(
image_inputs[0] , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (1, max_patch, expected_hidden_dim) , )
# Test batched
a__ : Optional[int] = image_processor(
__UpperCAmelCase , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (self.image_processor_tester.batch_size, max_patch, expected_hidden_dim) , )
@unittest.skipIf(
not is_torch_greater_or_equal_than_1_11 , reason="`Pix2StructImageProcessor` requires `torch>=1.11.0`." , )
@require_torch
@require_vision
class _a ( SCREAMING_SNAKE_CASE , unittest.TestCase ):
'''simple docstring'''
A :List[str] = PixaStructImageProcessor if is_vision_available() else None
def _A ( self ):
"""simple docstring"""
a__ : Dict = PixaStructImageProcessingTester(self , num_channels=4 )
a__ : Tuple = 3
@property
def _A ( self ):
"""simple docstring"""
return self.image_processor_tester.prepare_image_processor_dict()
def _A ( self ):
"""simple docstring"""
a__ : Optional[int] = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(__UpperCAmelCase , "do_normalize" ) )
self.assertTrue(hasattr(__UpperCAmelCase , "do_convert_rgb" ) )
def _A ( self ):
"""simple docstring"""
a__ : Dict = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
a__ : Union[str, Any] = prepare_image_inputs(self.image_processor_tester , equal_resolution=__UpperCAmelCase )
for image in image_inputs:
self.assertIsInstance(__UpperCAmelCase , Image.Image )
# Test not batched input
a__ : str = (
(self.image_processor_tester.patch_size["height"] * self.image_processor_tester.patch_size["width"])
* (self.image_processor_tester.num_channels - 1)
) + 2
for max_patch in self.image_processor_tester.max_patches:
# Test not batched input
a__ : List[str] = image_processor(
image_inputs[0] , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (1, max_patch, expected_hidden_dim) , )
# Test batched
a__ : int = image_processor(
__UpperCAmelCase , return_tensors="pt" , max_patches=__UpperCAmelCase ).flattened_patches
self.assertEqual(
encoded_images.shape , (self.image_processor_tester.batch_size, max_patch, expected_hidden_dim) , )
| 191
| 1
|
"""simple docstring"""
import itertools
import json
import linecache
import os
import pickle
import re
import socket
import string
from collections import Counter
from logging import getLogger
from pathlib import Path
from typing import Callable, Dict, Iterable, List
import git
import torch
from torch.utils.data import Dataset
from transformers import BartTokenizer, RagTokenizer, TaTokenizer
def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : int , lowerCAmelCase__ : int , lowerCAmelCase__ : int , lowerCAmelCase__ : List[str]=True , lowerCAmelCase__ : Union[str, Any]="pt" ) -> int:
__a = {'''add_prefix_space''': True} if isinstance(lowerCAmelCase__ , lowerCAmelCase__ ) and not line.startswith(''' ''' ) else {}
__a = padding_side
return tokenizer(
[line] , max_length=lowerCAmelCase__ , padding='''max_length''' if pad_to_max_length else None , truncation=lowerCAmelCase__ , return_tensors=lowerCAmelCase__ , add_special_tokens=lowerCAmelCase__ , **lowerCAmelCase__ , )
def lowercase ( lowerCAmelCase__ : Dict , lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Any=None , ) -> int:
__a = input_ids.ne(lowerCAmelCase__ ).any(dim=0 )
if attention_mask is None:
return input_ids[:, keep_column_mask]
else:
return (input_ids[:, keep_column_mask], attention_mask[:, keep_column_mask])
class __lowerCAmelCase ( __SCREAMING_SNAKE_CASE ):
'''simple docstring'''
def __init__( self , _a , _a , _a , _a , _a="train" , _a=None , _a=None , _a=None , _a="" , ):
super().__init__()
__a = Path(_a ).joinpath(type_path + '''.source''' )
__a = Path(_a ).joinpath(type_path + '''.target''' )
__a = self.get_char_lens(self.src_file )
__a = max_source_length
__a = max_target_length
assert min(self.src_lens ) > 0, f'''found empty line in {self.src_file}'''
__a = tokenizer
__a = prefix
if n_obs is not None:
__a = self.src_lens[:n_obs]
__a = src_lang
__a = tgt_lang
def __len__( self ):
return len(self.src_lens )
def __getitem__( self , _a ):
__a = index + 1 # linecache starts at 1
__a = self.prefix + linecache.getline(str(self.src_file ) , _a ).rstrip('''\n''' )
__a = linecache.getline(str(self.tgt_file ) , _a ).rstrip('''\n''' )
assert source_line, f'''empty source line for index {index}'''
assert tgt_line, f'''empty tgt line for index {index}'''
# Need to add eos token manually for T5
if isinstance(self.tokenizer , _a ):
source_line += self.tokenizer.eos_token
tgt_line += self.tokenizer.eos_token
# Pad source and target to the right
__a = (
self.tokenizer.question_encoder if isinstance(self.tokenizer , _a ) else self.tokenizer
)
__a = self.tokenizer.generator if isinstance(self.tokenizer , _a ) else self.tokenizer
__a = encode_line(_a , _a , self.max_source_length , '''right''' )
__a = encode_line(_a , _a , self.max_target_length , '''right''' )
__a = source_inputs['''input_ids'''].squeeze()
__a = target_inputs['''input_ids'''].squeeze()
__a = source_inputs['''attention_mask'''].squeeze()
return {
"input_ids": source_ids,
"attention_mask": src_mask,
"decoder_input_ids": target_ids,
}
@staticmethod
def __UpperCAmelCase ( _a ):
return [len(_a ) for x in Path(_a ).open().readlines()]
def __UpperCAmelCase ( self , _a ):
__a = torch.stack([x['''input_ids'''] for x in batch] )
__a = torch.stack([x['''attention_mask'''] for x in batch] )
__a = torch.stack([x['''decoder_input_ids'''] for x in batch] )
__a = (
self.tokenizer.generator.pad_token_id
if isinstance(self.tokenizer , _a )
else self.tokenizer.pad_token_id
)
__a = (
self.tokenizer.question_encoder.pad_token_id
if isinstance(self.tokenizer , _a )
else self.tokenizer.pad_token_id
)
__a = trim_batch(_a , _a )
__a , __a = trim_batch(_a , _a , attention_mask=_a )
__a = {
'''input_ids''': source_ids,
'''attention_mask''': source_mask,
'''decoder_input_ids''': y,
}
return batch
lowercase_ : List[str] = getLogger(__name__)
def lowercase ( lowerCAmelCase__ : List[List] ) -> int:
return list(itertools.chain.from_iterable(lowerCAmelCase__ ) )
def lowercase ( lowerCAmelCase__ : str ) -> None:
__a = get_git_info()
save_json(lowerCAmelCase__ , os.path.join(lowerCAmelCase__ , '''git_log.json''' ) )
def lowercase ( lowerCAmelCase__ : List[Any] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : int=4 , **lowerCAmelCase__ : Optional[int] ) -> str:
with open(lowerCAmelCase__ , '''w''' ) as f:
json.dump(lowerCAmelCase__ , lowerCAmelCase__ , indent=lowerCAmelCase__ , **lowerCAmelCase__ )
def lowercase ( lowerCAmelCase__ : Optional[Any] ) -> Dict:
with open(lowerCAmelCase__ ) as f:
return json.load(lowerCAmelCase__ )
def lowercase ( ) -> Optional[Any]:
__a = git.Repo(search_parent_directories=lowerCAmelCase__ )
__a = {
'''repo_id''': str(lowerCAmelCase__ ),
'''repo_sha''': str(repo.head.object.hexsha ),
'''repo_branch''': str(repo.active_branch ),
'''hostname''': str(socket.gethostname() ),
}
return repo_infos
def lowercase ( lowerCAmelCase__ : Callable , lowerCAmelCase__ : Iterable ) -> List:
return list(map(lowerCAmelCase__ , lowerCAmelCase__ ) )
def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> Optional[int]:
with open(lowerCAmelCase__ , '''wb''' ) as f:
return pickle.dump(lowerCAmelCase__ , lowerCAmelCase__ )
def lowercase ( lowerCAmelCase__ : List[str] ) -> str:
def remove_articles(lowerCAmelCase__ : Any ):
return re.sub(r'''\b(a|an|the)\b''' , ''' ''' , lowerCAmelCase__ )
def white_space_fix(lowerCAmelCase__ : List[Any] ):
return " ".join(text.split() )
def remove_punc(lowerCAmelCase__ : int ):
__a = set(string.punctuation )
return "".join(ch for ch in text if ch not in exclude )
def lower(lowerCAmelCase__ : int ):
return text.lower()
return white_space_fix(remove_articles(remove_punc(lower(lowerCAmelCase__ ) ) ) )
def lowercase ( lowerCAmelCase__ : Dict , lowerCAmelCase__ : str ) -> Union[str, Any]:
__a = normalize_answer(lowerCAmelCase__ ).split()
__a = normalize_answer(lowerCAmelCase__ ).split()
__a = Counter(lowerCAmelCase__ ) & Counter(lowerCAmelCase__ )
__a = sum(common.values() )
if num_same == 0:
return 0
__a = 1.0 * num_same / len(lowerCAmelCase__ )
__a = 1.0 * num_same / len(lowerCAmelCase__ )
__a = (2 * precision * recall) / (precision + recall)
return fa
def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : int ) -> int:
return normalize_answer(lowerCAmelCase__ ) == normalize_answer(lowerCAmelCase__ )
def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : List[str] ) -> Dict:
assert len(lowerCAmelCase__ ) == len(lowerCAmelCase__ )
__a = 0
for hypo, pred in zip(lowerCAmelCase__ , lowerCAmelCase__ ):
em += exact_match_score(lowerCAmelCase__ , lowerCAmelCase__ )
if len(lowerCAmelCase__ ) > 0:
em /= len(lowerCAmelCase__ )
return {"em": em}
def lowercase ( lowerCAmelCase__ : Optional[int] ) -> Tuple:
return model_prefix.startswith('''rag''' )
def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Optional[Any]:
__a = {p: p for p in extra_params}
# T5 models don't have `dropout` param, they have `dropout_rate` instead
__a = '''dropout_rate'''
for p in extra_params:
if getattr(lowerCAmelCase__ , lowerCAmelCase__ , lowerCAmelCase__ ):
if not hasattr(lowerCAmelCase__ , lowerCAmelCase__ ) and not hasattr(lowerCAmelCase__ , equivalent_param[p] ):
logger.info('''config doesn\'t have a `{}` attribute'''.format(lowerCAmelCase__ ) )
delattr(lowerCAmelCase__ , lowerCAmelCase__ )
continue
__a = p if hasattr(lowerCAmelCase__ , lowerCAmelCase__ ) else equivalent_param[p]
setattr(lowerCAmelCase__ , lowerCAmelCase__ , getattr(lowerCAmelCase__ , lowerCAmelCase__ ) )
delattr(lowerCAmelCase__ , lowerCAmelCase__ )
return hparams, config
| 702
|
"""simple docstring"""
import pytest
from datasets import Dataset, DatasetDict, Features, NamedSplit, Value
from datasets.io.text import TextDatasetReader
from ..utils import assert_arrow_memory_doesnt_increase, assert_arrow_memory_increases
def lowercase ( lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : List[Any] ) -> str:
assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ )
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('''keep_in_memory''' , [False, True] )
def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : int ) -> Tuple:
__a = tmp_path / '''cache'''
__a = {'''text''': '''string'''}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read()
_check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ )
@pytest.mark.parametrize(
'''features''' , [
None,
{'''text''': '''string'''},
{'''text''': '''int32'''},
{'''text''': '''float32'''},
] , )
def lowercase ( lowerCAmelCase__ : Any , lowerCAmelCase__ : Tuple , lowerCAmelCase__ : Union[str, Any] ) -> Optional[int]:
__a = tmp_path / '''cache'''
__a = {'''text''': '''string'''}
__a = features.copy() if features else default_expected_features
__a = (
Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None
)
__a = TextDatasetReader(lowerCAmelCase__ , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read()
_check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ )
@pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] )
def lowercase ( lowerCAmelCase__ : int , lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict ) -> Optional[Any]:
__a = tmp_path / '''cache'''
__a = {'''text''': '''string'''}
__a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ , split=lowerCAmelCase__ ).read()
_check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ )
assert dataset.split == split if split else "train"
@pytest.mark.parametrize('''path_type''' , [str, list] )
def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[Any] , lowerCAmelCase__ : Optional[Any] ) -> Dict:
if issubclass(lowerCAmelCase__ , lowerCAmelCase__ ):
__a = text_path
elif issubclass(lowerCAmelCase__ , lowerCAmelCase__ ):
__a = [text_path]
__a = tmp_path / '''cache'''
__a = {'''text''': '''string'''}
__a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read()
_check_text_dataset(lowerCAmelCase__ , lowerCAmelCase__ )
def lowercase ( lowerCAmelCase__ : Union[str, Any] , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Union[str, Any]=("train",) ) -> Optional[Any]:
assert isinstance(lowerCAmelCase__ , lowerCAmelCase__ )
for split in splits:
__a = dataset_dict[split]
assert dataset.num_rows == 4
assert dataset.num_columns == 1
assert dataset.column_names == ["text"]
for feature, expected_dtype in expected_features.items():
assert dataset.features[feature].dtype == expected_dtype
@pytest.mark.parametrize('''keep_in_memory''' , [False, True] )
def lowercase ( lowerCAmelCase__ : Tuple , lowerCAmelCase__ : str , lowerCAmelCase__ : List[Any] ) -> Union[str, Any]:
__a = tmp_path / '''cache'''
__a = {'''text''': '''string'''}
with assert_arrow_memory_increases() if keep_in_memory else assert_arrow_memory_doesnt_increase():
__a = TextDatasetReader({'''train''': text_path} , cache_dir=lowerCAmelCase__ , keep_in_memory=lowerCAmelCase__ ).read()
_check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ )
@pytest.mark.parametrize(
'''features''' , [
None,
{'''text''': '''string'''},
{'''text''': '''int32'''},
{'''text''': '''float32'''},
] , )
def lowercase ( lowerCAmelCase__ : str , lowerCAmelCase__ : Optional[int] , lowerCAmelCase__ : Optional[Any] ) -> str:
__a = tmp_path / '''cache'''
# CSV file loses col_1 string dtype information: default now is "int64" instead of "string"
__a = {'''text''': '''string'''}
__a = features.copy() if features else default_expected_features
__a = (
Features({feature: Value(lowerCAmelCase__ ) for feature, dtype in features.items()} ) if features is not None else None
)
__a = TextDatasetReader({'''train''': text_path} , features=lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read()
_check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ )
@pytest.mark.parametrize('''split''' , [None, NamedSplit('''train''' ), '''train''', '''test'''] )
def lowercase ( lowerCAmelCase__ : List[str] , lowerCAmelCase__ : Dict , lowerCAmelCase__ : Tuple ) -> Dict:
if split:
__a = {split: text_path}
else:
__a = '''train'''
__a = {'''train''': text_path, '''test''': text_path}
__a = tmp_path / '''cache'''
__a = {'''text''': '''string'''}
__a = TextDatasetReader(lowerCAmelCase__ , cache_dir=lowerCAmelCase__ ).read()
_check_text_datasetdict(lowerCAmelCase__ , lowerCAmelCase__ , splits=list(path.keys() ) )
assert all(dataset[split].split == split for split in path.keys() )
| 65
| 0
|
"""simple docstring"""
import unittest
from datasets import load_dataset
from transformers.pipelines import pipeline
from transformers.testing_utils import is_pipeline_test, nested_simplify, require_torch, slow
@is_pipeline_test
@require_torch
class __snake_case ( unittest.TestCase ):
"""simple docstring"""
@require_torch
def UpperCamelCase__( self ):
'''simple docstring'''
__A : int = pipeline(
task='''zero-shot-audio-classification''' , model='''hf-internal-testing/tiny-clap-htsat-unfused''' )
__A : int = load_dataset('''ashraq/esc50''' )
__A : Dict = dataset["train"]["audio"][-1]["array"]
__A : Optional[int] = audio_classifier(snake_case__ , candidate_labels=['''Sound of a dog''', '''Sound of vaccum cleaner'''] )
self.assertEqual(
nested_simplify(snake_case__ ) , [{'''score''': 0.5_0_1, '''label''': '''Sound of a dog'''}, {'''score''': 0.4_9_9, '''label''': '''Sound of vaccum cleaner'''}] , )
@unittest.skip('''No models are available in TF''' )
def UpperCamelCase__( self ):
'''simple docstring'''
pass
@slow
@require_torch
def UpperCamelCase__( self ):
'''simple docstring'''
__A : Union[str, Any] = pipeline(
task='''zero-shot-audio-classification''' , model='''laion/clap-htsat-unfused''' , )
# This is an audio of a dog
__A : str = load_dataset('''ashraq/esc50''' )
__A : int = dataset["train"]["audio"][-1]["array"]
__A : List[str] = audio_classifier(snake_case__ , candidate_labels=['''Sound of a dog''', '''Sound of vaccum cleaner'''] )
self.assertEqual(
nested_simplify(snake_case__ ) , [
{'''score''': 0.9_9_9, '''label''': '''Sound of a dog'''},
{'''score''': 0.0_0_1, '''label''': '''Sound of vaccum cleaner'''},
] , )
__A : Any = audio_classifier([audio] * 5 , candidate_labels=['''Sound of a dog''', '''Sound of vaccum cleaner'''] )
self.assertEqual(
nested_simplify(snake_case__ ) , [
[
{'''score''': 0.9_9_9, '''label''': '''Sound of a dog'''},
{'''score''': 0.0_0_1, '''label''': '''Sound of vaccum cleaner'''},
],
]
* 5 , )
__A : Tuple = audio_classifier(
[audio] * 5 , candidate_labels=['''Sound of a dog''', '''Sound of vaccum cleaner'''] , batch_size=5 )
self.assertEqual(
nested_simplify(snake_case__ ) , [
[
{'''score''': 0.9_9_9, '''label''': '''Sound of a dog'''},
{'''score''': 0.0_0_1, '''label''': '''Sound of vaccum cleaner'''},
],
]
* 5 , )
@unittest.skip('''No models are available in TF''' )
def UpperCamelCase__( self ):
'''simple docstring'''
pass
| 177
|
"""simple docstring"""
import unittest
from transformers import is_flax_available
from transformers.testing_utils import require_flax, require_sentencepiece, require_tokenizers, require_torch, slow
if is_flax_available():
import optax
from flax.training.common_utils import onehot
from transformers import AutoTokenizer, FlaxMTaForConditionalGeneration
from transformers.models.ta.modeling_flax_ta import shift_tokens_right
@require_torch
@require_sentencepiece
@require_tokenizers
@require_flax
class lowerCAmelCase__ ( unittest.TestCase ):
@slow
def __a ( self : List[str] ):
'''simple docstring'''
UpperCAmelCase__ : Union[str, Any] = FlaxMTaForConditionalGeneration.from_pretrained("google/mt5-small" )
UpperCAmelCase__ : str = AutoTokenizer.from_pretrained("google/mt5-small" )
UpperCAmelCase__ : str = tokenizer("Hello there" , return_tensors="np" ).input_ids
UpperCAmelCase__ : Optional[Any] = tokenizer("Hi I am" , return_tensors="np" ).input_ids
UpperCAmelCase__ : Any = shift_tokens_right(snake_case__ , model.config.pad_token_id , model.config.decoder_start_token_id )
UpperCAmelCase__ : Any = model(snake_case__ , decoder_input_ids=snake_case__ ).logits
UpperCAmelCase__ : Any = optax.softmax_cross_entropy(snake_case__ , onehot(snake_case__ , logits.shape[-1] ) ).mean()
UpperCAmelCase__ : List[str] = -(labels.shape[-1] * loss.item())
UpperCAmelCase__ : Dict = -84.9127
self.assertTrue(abs(mtf_score - EXPECTED_SCORE ) < 1e-4 )
| 438
| 0
|
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
UpperCAmelCase_ : Union[str, Any] = logging.get_logger(__name__)
class SCREAMING_SNAKE_CASE__ ( lowercase__ ):
snake_case__ : List[Any] = ['''input_values''', '''padding_mask''']
def __init__( self : Union[str, Any] , SCREAMING_SNAKE_CASE__ : int = 1 , SCREAMING_SNAKE_CASE__ : int = 2_4_0_0_0 , SCREAMING_SNAKE_CASE__ : float = 0.0 , SCREAMING_SNAKE_CASE__ : float = None , SCREAMING_SNAKE_CASE__ : float = None , **SCREAMING_SNAKE_CASE__ : Any , ) -> Union[str, Any]:
super().__init__(feature_size=SCREAMING_SNAKE_CASE__ , sampling_rate=SCREAMING_SNAKE_CASE__ , padding_value=SCREAMING_SNAKE_CASE__ , **SCREAMING_SNAKE_CASE__ )
a_ : Optional[int] = chunk_length_s
a_ : str = overlap
@property
def SCREAMING_SNAKE_CASE ( self : List[str] ) -> Optional[int]:
if self.chunk_length_s is None:
return None
else:
return int(self.chunk_length_s * self.sampling_rate )
@property
def SCREAMING_SNAKE_CASE ( self : str ) -> Optional[int]:
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 : Optional[int] , SCREAMING_SNAKE_CASE__ : Union[np.ndarray, List[float], List[np.ndarray], List[List[float]]] , SCREAMING_SNAKE_CASE__ : Optional[Union[bool, str, PaddingStrategy]] = None , SCREAMING_SNAKE_CASE__ : Optional[bool] = False , SCREAMING_SNAKE_CASE__ : Optional[int] = None , SCREAMING_SNAKE_CASE__ : Optional[Union[str, TensorType]] = None , SCREAMING_SNAKE_CASE__ : Optional[int] = None , ) -> BatchFeature:
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
a_ : int = True
a_ : Any = bool(
isinstance(SCREAMING_SNAKE_CASE__ , (list, tuple) ) and (isinstance(raw_audio[0] , (np.ndarray, tuple, list) )) )
if is_batched:
a_ : Union[str, Any] = [np.asarray(SCREAMING_SNAKE_CASE__ , dtype=np.floataa ).T for audio in raw_audio]
elif not is_batched and not isinstance(SCREAMING_SNAKE_CASE__ , np.ndarray ):
a_ : Union[str, Any] = np.asarray(SCREAMING_SNAKE_CASE__ , dtype=np.floataa )
elif isinstance(SCREAMING_SNAKE_CASE__ , np.ndarray ) and raw_audio.dtype is np.dtype(np.floataa ):
a_ : Any = raw_audio.astype(np.floataa )
# always return batch
if not is_batched:
a_ : Dict = [np.asarray(SCREAMING_SNAKE_CASE__ ).T]
# verify inputs are valid
for idx, example in enumerate(SCREAMING_SNAKE_CASE__ ):
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""" )
a_ : Dict = None
a_ : List[Any] = 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:
a_ : Dict = min(array.shape[0] for array in raw_audio )
a_ : Union[str, Any] = int(np.floor(max_length / self.chunk_stride ) )
a_ : Any = (nb_step - 1) * self.chunk_stride + self.chunk_length
elif padding:
a_ : Union[str, Any] = max(array.shape[0] for array in raw_audio )
a_ : Optional[int] = int(np.ceil(max_length / self.chunk_stride ) )
a_ : int = (nb_step - 1) * self.chunk_stride + self.chunk_length
a_ : Optional[Any] = 'max_length'
else:
a_ : Union[str, Any] = input_values
# normal padding on batch
if padded_inputs is None:
a_ : int = self.pad(
SCREAMING_SNAKE_CASE__ , max_length=SCREAMING_SNAKE_CASE__ , truncation=SCREAMING_SNAKE_CASE__ , padding=SCREAMING_SNAKE_CASE__ , return_attention_mask=SCREAMING_SNAKE_CASE__ , )
if padding:
a_ : Optional[int] = padded_inputs.pop('attention_mask' )
a_ : Union[str, Any] = []
for example in padded_inputs.pop('input_values' ):
if self.feature_size == 1:
a_ : Any = example[..., None]
input_values.append(example.T )
a_ : List[Any] = input_values
if return_tensors is not None:
a_ : Optional[Any] = padded_inputs.convert_to_tensors(SCREAMING_SNAKE_CASE__ )
return padded_inputs
| 716
|
def SCREAMING_SNAKE_CASE_ ( __A : str ) -> list:
"""simple docstring"""
return [
txt[:a] + txt[a].upper() + txt[a + 1 :]
for a in range(len(__A ) )
if txt[a].isalpha()
]
if __name__ == "__main__":
__import__('doctest').testmod()
| 443
| 0
|
import argparse
import json
import os
import fairseq
import torch
from fairseq.data import Dictionary
# Register SEW's fairseq modules
from sew_asapp import tasks # noqa: F401
from transformers import (
SEWConfig,
SEWForCTC,
SEWModel,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaProcessor,
logging,
)
logging.set_verbosity_info()
lowerCAmelCase = logging.get_logger(__name__)
lowerCAmelCase = {
'post_extract_proj': 'feature_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.upsample.0': 'encoder.upsample.projection',
'encoder.layer_norm': 'encoder.layer_norm',
'w2v_model.layer_norm': 'layer_norm',
'w2v_encoder.proj': 'lm_head',
'mask_emb': 'masked_spec_embed',
}
def _a ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ):
"""simple docstring"""
for attribute in key.split('''.''' ):
lowercase__ = getattr(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE )
if weight_type is not None:
lowercase__ = getattr(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ).shape
else:
lowercase__ = hf_pointer.shape
assert hf_shape == value.shape, (
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":
lowercase__ = value
elif weight_type == "weight_g":
lowercase__ = value
elif weight_type == "weight_v":
lowercase__ = value
elif weight_type == "bias":
lowercase__ = value
else:
lowercase__ = value
logger.info(f'{key + "." + weight_type if weight_type is not None else ""} was initialized from {full_name}.' )
def _a ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ):
"""simple docstring"""
lowercase__ = []
lowercase__ = fairseq_model.state_dict()
lowercase__ = hf_model.sew.feature_extractor if is_finetuned else hf_model.feature_extractor
for name, value in fairseq_dict.items():
lowercase__ = False
if "conv_layers" in name:
load_conv_layer(
SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , hf_model.config.feat_extract_norm == '''group''' , )
lowercase__ = True
else:
for key, mapped_key in MAPPING.items():
lowercase__ = '''sew.''' + mapped_key if (is_finetuned and mapped_key != '''lm_head''') else mapped_key
if key in name or key.split('''w2v_model.''' )[-1] == name.split('''.''' )[0]:
lowercase__ = True
if "*" in mapped_key:
lowercase__ = name.split(SCREAMING_SNAKE_CASE )[0].split('''.''' )[-2]
lowercase__ = mapped_key.replace('''*''' , SCREAMING_SNAKE_CASE )
if "weight_g" in name:
lowercase__ = '''weight_g'''
elif "weight_v" in name:
lowercase__ = '''weight_v'''
elif "weight" in name:
lowercase__ = '''weight'''
elif "bias" in name:
lowercase__ = '''bias'''
else:
lowercase__ = None
set_recursively(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE )
continue
if not is_used:
unused_weights.append(SCREAMING_SNAKE_CASE )
logger.warning(f'Unused weights: {unused_weights}' )
def _a ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ):
"""simple docstring"""
lowercase__ = full_name.split('''conv_layers.''' )[-1]
lowercase__ = name.split('''.''' )
lowercase__ = int(items[0] )
lowercase__ = int(items[1] )
if type_id == 0:
if "bias" in name:
assert value.shape == feature_extractor.conv_layers[layer_id].conv.bias.data.shape, (
f'{full_name} has size {value.shape}, but'
f' {feature_extractor.conv_layers[layer_id].conv.bias.data.shape} was found.'
)
lowercase__ = value
logger.info(f'Feat extract conv layer {layer_id} was initialized from {full_name}.' )
elif "weight" in name:
assert value.shape == feature_extractor.conv_layers[layer_id].conv.weight.data.shape, (
f'{full_name} has size {value.shape}, but'
f' {feature_extractor.conv_layers[layer_id].conv.weight.data.shape} was found.'
)
lowercase__ = 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:
assert value.shape == feature_extractor.conv_layers[layer_id].layer_norm.bias.data.shape, (
f'{full_name} has size {value.shape}, but {feature_extractor[layer_id].layer_norm.bias.data.shape} was'
" found."
)
lowercase__ = value
logger.info(f'Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.' )
elif "weight" in name:
assert value.shape == feature_extractor.conv_layers[layer_id].layer_norm.weight.data.shape, (
f'{full_name} has size {value.shape}, but'
f' {feature_extractor[layer_id].layer_norm.weight.data.shape} was found.'
)
lowercase__ = value
logger.info(f'Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.' )
else:
unused_weights.append(SCREAMING_SNAKE_CASE )
def _a ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ):
"""simple docstring"""
lowercase__ = SEWConfig()
if is_finetuned:
lowercase__ = model.wav_encoder.wav_model.cfg
else:
lowercase__ = model.cfg
lowercase__ = fs_config.conv_bias
lowercase__ = eval(fs_config.conv_feature_layers )
lowercase__ = [x[0] for x in conv_layers]
lowercase__ = [x[1] for x in conv_layers]
lowercase__ = [x[2] for x in conv_layers]
lowercase__ = '''gelu'''
lowercase__ = '''layer''' if fs_config.extractor_mode == '''layer_norm''' else '''group'''
lowercase__ = 0.0
lowercase__ = fs_config.activation_fn.name
lowercase__ = fs_config.encoder_embed_dim
lowercase__ = 0.02
lowercase__ = fs_config.encoder_ffn_embed_dim
lowercase__ = 1E-5
lowercase__ = fs_config.encoder_layerdrop
lowercase__ = fs_config.encoder_attention_heads
lowercase__ = fs_config.conv_pos_groups
lowercase__ = fs_config.conv_pos
lowercase__ = len(SCREAMING_SNAKE_CASE )
lowercase__ = fs_config.encoder_layers
lowercase__ = fs_config.squeeze_factor
# take care of any params that are overridden by the Wav2VecCtc model
if is_finetuned:
lowercase__ = model.cfg
lowercase__ = fs_config.final_dropout
lowercase__ = fs_config.layerdrop
lowercase__ = fs_config.activation_dropout
lowercase__ = fs_config.mask_prob > 0 or fs_config.mask_channel_prob > 0
lowercase__ = fs_config.attention_dropout
lowercase__ = fs_config.dropout_input
lowercase__ = fs_config.dropout
lowercase__ = fs_config.mask_channel_length
lowercase__ = fs_config.mask_channel_prob
lowercase__ = fs_config.mask_length
lowercase__ = fs_config.mask_prob
lowercase__ = '''Wav2Vec2FeatureExtractor'''
lowercase__ = '''Wav2Vec2CTCTokenizer'''
return config
@torch.no_grad()
def _a ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE=None , SCREAMING_SNAKE_CASE=None , SCREAMING_SNAKE_CASE=True ):
"""simple docstring"""
if is_finetuned:
lowercase__ , lowercase__ , lowercase__ = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] , arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
lowercase__ , lowercase__ , lowercase__ = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] )
if config_path is not None:
lowercase__ = SEWConfig.from_pretrained(SCREAMING_SNAKE_CASE )
else:
lowercase__ = convert_config(model[0] , SCREAMING_SNAKE_CASE )
lowercase__ = model[0].eval()
lowercase__ = True if config.feat_extract_norm == '''layer''' else False
lowercase__ = WavaVecaFeatureExtractor(
feature_size=1 , sampling_rate=1_60_00 , padding_value=0 , do_normalize=SCREAMING_SNAKE_CASE , return_attention_mask=SCREAMING_SNAKE_CASE , )
if is_finetuned:
if dict_path:
lowercase__ = Dictionary.load(SCREAMING_SNAKE_CASE )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
lowercase__ = target_dict.pad_index
lowercase__ = target_dict.bos_index
lowercase__ = target_dict.pad_index
lowercase__ = target_dict.bos_index
lowercase__ = target_dict.eos_index
lowercase__ = len(target_dict.symbols )
lowercase__ = os.path.join(SCREAMING_SNAKE_CASE , '''vocab.json''' )
if not os.path.isdir(SCREAMING_SNAKE_CASE ):
logger.error('''--pytorch_dump_folder_path ({}) should be a directory'''.format(SCREAMING_SNAKE_CASE ) )
return
os.makedirs(SCREAMING_SNAKE_CASE , exist_ok=SCREAMING_SNAKE_CASE )
with open(SCREAMING_SNAKE_CASE , '''w''' , encoding='''utf-8''' ) as vocab_handle:
json.dump(target_dict.indices , SCREAMING_SNAKE_CASE )
lowercase__ = WavaVecaCTCTokenizer(
SCREAMING_SNAKE_CASE , 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=SCREAMING_SNAKE_CASE , )
lowercase__ = WavaVecaProcessor(feature_extractor=SCREAMING_SNAKE_CASE , tokenizer=SCREAMING_SNAKE_CASE )
processor.save_pretrained(SCREAMING_SNAKE_CASE )
lowercase__ = SEWForCTC(SCREAMING_SNAKE_CASE )
else:
lowercase__ = SEWModel(SCREAMING_SNAKE_CASE )
feature_extractor.save_pretrained(SCREAMING_SNAKE_CASE )
recursively_load_weights(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE )
hf_model.save_pretrained(SCREAMING_SNAKE_CASE )
if __name__ == "__main__":
lowerCAmelCase = 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(
'--is_finetuned', action='store_true', help='Whether the model to convert is a fine-tuned model or not'
)
lowerCAmelCase = parser.parse_args()
convert_sew_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, args.is_finetuned
)
| 43
|
import argparse
import json
from collections import OrderedDict
from pathlib import Path
import requests
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import PoolFormerConfig, PoolFormerForImageClassification, PoolFormerImageProcessor
from transformers.utils import logging
logging.set_verbosity_info()
lowerCAmelCase = logging.get_logger(__name__)
def _a ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ):
"""simple docstring"""
lowercase__ = original_name.split('''.''' )[0]
lowercase__ = key.split('''.''' )
lowercase__ = int(key_list[key_list.index(SCREAMING_SNAKE_CASE ) - 2] )
lowercase__ = int(key_list[key_list.index(SCREAMING_SNAKE_CASE ) - 1] )
lowercase__ = orig_block_num - offset
lowercase__ = key.replace(f'{orig_block_num}.{layer_num}.{original_name}' , f'block.{new_block_num}.{layer_num}.{new_name}' )
return key
def _a ( SCREAMING_SNAKE_CASE ):
"""simple docstring"""
lowercase__ = OrderedDict()
lowercase__ , lowercase__ = 0, 0
for key, value in state_dict.items():
if key.startswith('''network''' ):
lowercase__ = key.replace('''network''' , '''poolformer.encoder''' )
if "proj" in key:
# Works for the first embedding as well as the internal embedding layers
if key.endswith('''bias''' ) and "patch_embed" not in key:
patch_emb_offset += 1
lowercase__ = key[: key.find('''proj''' )]
lowercase__ = key.replace(SCREAMING_SNAKE_CASE , f'patch_embeddings.{total_embed_found}.' )
lowercase__ = key.replace('''proj''' , '''projection''' )
if key.endswith('''bias''' ):
total_embed_found += 1
if "patch_embeddings" in key:
lowercase__ = '''poolformer.encoder.''' + key
if "mlp.fc1" in key:
lowercase__ = replace_key_with_offset(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , '''mlp.fc1''' , '''output.conv1''' )
if "mlp.fc2" in key:
lowercase__ = replace_key_with_offset(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , '''mlp.fc2''' , '''output.conv2''' )
if "norm1" in key:
lowercase__ = replace_key_with_offset(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , '''norm1''' , '''before_norm''' )
if "norm2" in key:
lowercase__ = replace_key_with_offset(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , '''norm2''' , '''after_norm''' )
if "layer_scale_1" in key:
lowercase__ = replace_key_with_offset(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , '''layer_scale_1''' , '''layer_scale_1''' )
if "layer_scale_2" in key:
lowercase__ = replace_key_with_offset(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , '''layer_scale_2''' , '''layer_scale_2''' )
if "head" in key:
lowercase__ = key.replace('''head''' , '''classifier''' )
lowercase__ = value
return new_state_dict
def _a ( ):
"""simple docstring"""
lowercase__ = '''http://images.cocodataset.org/val2017/000000039769.jpg'''
lowercase__ = Image.open(requests.get(SCREAMING_SNAKE_CASE , stream=SCREAMING_SNAKE_CASE ).raw )
return image
@torch.no_grad()
def _a ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ):
"""simple docstring"""
lowercase__ = PoolFormerConfig()
# set attributes based on model_name
lowercase__ = '''huggingface/label-files'''
lowercase__ = model_name[-3:]
lowercase__ = 10_00
lowercase__ = '''imagenet-1k-id2label.json'''
lowercase__ = (1, 10_00)
# set config attributes
lowercase__ = json.load(open(hf_hub_download(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , repo_type='''dataset''' ) , '''r''' ) )
lowercase__ = {int(SCREAMING_SNAKE_CASE ): v for k, v in idalabel.items()}
lowercase__ = idalabel
lowercase__ = {v: k for k, v in idalabel.items()}
if size == "s12":
lowercase__ = [2, 2, 6, 2]
lowercase__ = [64, 1_28, 3_20, 5_12]
lowercase__ = 4.0
lowercase__ = 0.9
elif size == "s24":
lowercase__ = [4, 4, 12, 4]
lowercase__ = [64, 1_28, 3_20, 5_12]
lowercase__ = 4.0
lowercase__ = 0.9
elif size == "s36":
lowercase__ = [6, 6, 18, 6]
lowercase__ = [64, 1_28, 3_20, 5_12]
lowercase__ = 4.0
lowercase__ = 1E-6
lowercase__ = 0.9
elif size == "m36":
lowercase__ = [6, 6, 18, 6]
lowercase__ = [96, 1_92, 3_84, 7_68]
lowercase__ = 4.0
lowercase__ = 1E-6
lowercase__ = 0.95
elif size == "m48":
lowercase__ = [8, 8, 24, 8]
lowercase__ = [96, 1_92, 3_84, 7_68]
lowercase__ = 4.0
lowercase__ = 1E-6
lowercase__ = 0.95
else:
raise ValueError(f'Size {size} not supported' )
# load image processor
lowercase__ = PoolFormerImageProcessor(crop_pct=SCREAMING_SNAKE_CASE )
# Prepare image
lowercase__ = prepare_img()
lowercase__ = image_processor(images=SCREAMING_SNAKE_CASE , return_tensors='''pt''' ).pixel_values
logger.info(f'Converting model {model_name}...' )
# load original state dict
lowercase__ = torch.load(SCREAMING_SNAKE_CASE , map_location=torch.device('''cpu''' ) )
# rename keys
lowercase__ = rename_keys(SCREAMING_SNAKE_CASE )
# create HuggingFace model and load state dict
lowercase__ = PoolFormerForImageClassification(SCREAMING_SNAKE_CASE )
model.load_state_dict(SCREAMING_SNAKE_CASE )
model.eval()
# Define image processor
lowercase__ = PoolFormerImageProcessor(crop_pct=SCREAMING_SNAKE_CASE )
lowercase__ = image_processor(images=prepare_img() , return_tensors='''pt''' ).pixel_values
# forward pass
lowercase__ = model(SCREAMING_SNAKE_CASE )
lowercase__ = outputs.logits
# define expected logit slices for different models
if size == "s12":
lowercase__ = torch.tensor([-0.3_045, -0.6_758, -0.4_869] )
elif size == "s24":
lowercase__ = torch.tensor([0.4_402, -0.1_374, -0.8_045] )
elif size == "s36":
lowercase__ = torch.tensor([-0.6_080, -0.5_133, -0.5_898] )
elif size == "m36":
lowercase__ = torch.tensor([0.3_952, 0.2_263, -1.2_668] )
elif size == "m48":
lowercase__ = torch.tensor([0.1_167, -0.0_656, -0.3_423] )
else:
raise ValueError(f'Size {size} not supported' )
# verify logits
assert logits.shape == expected_shape
assert torch.allclose(logits[0, :3] , SCREAMING_SNAKE_CASE , atol=1E-2 )
# finally, save model and image processor
logger.info(f'Saving PyTorch model and image processor to {pytorch_dump_folder_path}...' )
Path(SCREAMING_SNAKE_CASE ).mkdir(exist_ok=SCREAMING_SNAKE_CASE )
model.save_pretrained(SCREAMING_SNAKE_CASE )
print(f'Saving image processor to {pytorch_dump_folder_path}' )
image_processor.save_pretrained(SCREAMING_SNAKE_CASE )
if __name__ == "__main__":
lowerCAmelCase = argparse.ArgumentParser()
parser.add_argument(
'--model_name',
default='poolformer_s12',
type=str,
help='Name of the model you\'d like to convert.',
)
parser.add_argument(
'--checkpoint_path', default=None, type=str, help='Path to the original PyTorch checkpoint (.pth file).'
)
parser.add_argument(
'--pytorch_dump_folder_path', default=None, type=str, help='Path to the folder to output PyTorch model.'
)
lowerCAmelCase = parser.parse_args()
convert_poolformer_checkpoint(args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path)
| 43
| 1
|
"""simple docstring"""
from __future__ import annotations
import math
def __UpperCAmelCase ( __UpperCamelCase , __UpperCamelCase , __UpperCamelCase , __UpperCamelCase , __UpperCamelCase ):
if depth < 0:
raise ValueError('''Depth cannot be less than 0''' )
if not scores:
raise ValueError('''Scores cannot be empty''' )
if depth == height:
return scores[node_index]
return (
max(
minimax(depth + 1 , node_index * 2 , __UpperCamelCase , __UpperCamelCase , __UpperCamelCase ) , minimax(depth + 1 , node_index * 2 + 1 , __UpperCamelCase , __UpperCamelCase , __UpperCamelCase ) , )
if is_max
else min(
minimax(depth + 1 , node_index * 2 , __UpperCamelCase , __UpperCamelCase , __UpperCamelCase ) , minimax(depth + 1 , node_index * 2 + 1 , __UpperCamelCase , __UpperCamelCase , __UpperCamelCase ) , )
)
def __UpperCAmelCase ( ):
__lowercase : Any = [90, 23, 6, 33, 21, 65, 1_23, 3_44_23]
__lowercase : str = math.log(len(__UpperCamelCase ) , 2 )
print(f"""Optimal value : {minimax(0 , 0 , __UpperCamelCase , __UpperCamelCase , __UpperCamelCase )}""" )
if __name__ == "__main__":
import doctest
doctest.testmod()
main()
| 701
|
"""simple docstring"""
import argparse
import json
from tqdm import tqdm
def __UpperCAmelCase ( ):
__lowercase : List[Any] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--src_path''' , type=__UpperCamelCase , default='''biencoder-nq-dev.json''' , help='''Path to raw DPR training data''' , )
parser.add_argument(
'''--evaluation_set''' , type=__UpperCamelCase , help='''where to store parsed evaluation_set file''' , )
parser.add_argument(
'''--gold_data_path''' , type=__UpperCamelCase , help='''where to store parsed gold_data_path file''' , )
__lowercase : Dict = parser.parse_args()
with open(args.src_path , '''r''' ) as src_file, open(args.evaluation_set , '''w''' ) as eval_file, open(
args.gold_data_path , '''w''' ) as gold_file:
__lowercase : str = json.load(__UpperCamelCase )
for dpr_record in tqdm(__UpperCamelCase ):
__lowercase : Optional[int] = dpr_record['''question''']
__lowercase : Optional[int] = [context['''title'''] for context in dpr_record['''positive_ctxs''']]
eval_file.write(question + '''\n''' )
gold_file.write('''\t'''.join(__UpperCamelCase ) + '''\n''' )
if __name__ == "__main__":
main()
| 523
| 0
|
import contextlib
from multiprocessing import Pool, RLock
from tqdm.auto import tqdm
from ..utils import experimental, logging
A__ = logging.get_logger(__name__)
class a :
__lowerCAmelCase : Optional[Any] = None
@experimental
def _lowerCAmelCase ( __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase ) -> Union[str, Any]:
"""simple docstring"""
if ParallelBackendConfig.backend_name is None:
return _map_with_multiprocessing_pool(
__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase )
return _map_with_joblib(__lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase )
def _lowerCAmelCase ( __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase ) -> Optional[Any]:
"""simple docstring"""
snake_case__ : List[str] = num_proc if num_proc <= len(__lowerCAmelCase ) else len(__lowerCAmelCase )
snake_case__ : int = [] # We organize the splits ourselve (contiguous splits)
for index in range(__lowerCAmelCase ):
snake_case__ : List[Any] = len(__lowerCAmelCase ) // num_proc
snake_case__ : Tuple = len(__lowerCAmelCase ) % num_proc
snake_case__ : Any = div * index + min(__lowerCAmelCase , __lowerCAmelCase )
snake_case__ : int = start + div + (1 if index < mod else 0)
split_kwds.append((function, iterable[start:end], types, index, disable_tqdm, desc) )
if len(__lowerCAmelCase ) != sum(len(i[1] ) for i in split_kwds ):
raise ValueError(
f"""Error dividing inputs iterable among processes. """
f"""Total number of objects {len(__lowerCAmelCase )}, """
f"""length: {sum(len(i[1] ) for i in split_kwds )}""" )
logger.info(
f"""Spawning {num_proc} processes for {len(__lowerCAmelCase )} objects in slices of {[len(i[1] ) for i in split_kwds]}""" )
snake_case__ , snake_case__ : List[str] = None, None
if not disable_tqdm:
snake_case__ , snake_case__ : Any = (RLock(),), tqdm.set_lock
with Pool(__lowerCAmelCase , initargs=__lowerCAmelCase , initializer=__lowerCAmelCase ) as pool:
snake_case__ : Optional[int] = pool.map(__lowerCAmelCase , __lowerCAmelCase )
logger.info(f"""Finished {num_proc} processes""" )
snake_case__ : List[str] = [obj for proc_res in mapped for obj in proc_res]
logger.info(f"""Unpacked {len(__lowerCAmelCase )} objects""" )
return mapped
def _lowerCAmelCase ( __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase , __lowerCAmelCase ) -> Optional[int]:
"""simple docstring"""
import joblib
with joblib.parallel_backend(ParallelBackendConfig.backend_name , n_jobs=__lowerCAmelCase ):
return joblib.Parallel()(
joblib.delayed(__lowerCAmelCase )((function, obj, types, None, True, None) ) for obj in iterable )
@experimental
@contextlib.contextmanager
def _lowerCAmelCase ( __lowerCAmelCase ) -> Tuple:
"""simple docstring"""
snake_case__ : Tuple = backend_name
if backend_name == "spark":
from joblibspark import register_spark
register_spark()
# TODO: call create_cache_and_write_probe if "download" in steps
# TODO: raise NotImplementedError when Dataset.map etc is called
try:
yield
finally:
snake_case__ : int = None
| 252
|
import gc
import random
import unittest
import numpy as np
import torch
from PIL import Image
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, DDPMScheduler, StableDiffusionUpscalePipeline, UNetaDConditionModel
from diffusers.utils import floats_tensor, load_image, load_numpy, slow, torch_device
from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu
enable_full_determinism()
class a ( unittest.TestCase ):
def __lowerCamelCase ( self :Union[str, Any] ):
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def __lowerCamelCase ( self :Dict ):
snake_case__ : Optional[Any] = 1
snake_case__ : int = 3
snake_case__ : Optional[int] = (3_2, 3_2)
snake_case__ : Any = floats_tensor((batch_size, num_channels) + sizes ,rng=random.Random(0 ) ).to(__lowercase )
return image
@property
def __lowerCamelCase ( self :int ):
torch.manual_seed(0 )
snake_case__ : List[Any] = UNetaDConditionModel(
block_out_channels=(3_2, 3_2, 6_4) ,layers_per_block=2 ,sample_size=3_2 ,in_channels=7 ,out_channels=4 ,down_block_types=('''DownBlock2D''', '''CrossAttnDownBlock2D''', '''CrossAttnDownBlock2D''') ,up_block_types=('''CrossAttnUpBlock2D''', '''CrossAttnUpBlock2D''', '''UpBlock2D''') ,cross_attention_dim=3_2 ,attention_head_dim=8 ,use_linear_projection=__lowercase ,only_cross_attention=(True, True, False) ,num_class_embeds=1_0_0 ,)
return model
@property
def __lowerCamelCase ( self :List[Any] ):
torch.manual_seed(0 )
snake_case__ : Tuple = AutoencoderKL(
block_out_channels=[3_2, 3_2, 6_4] ,in_channels=3 ,out_channels=3 ,down_block_types=['''DownEncoderBlock2D''', '''DownEncoderBlock2D''', '''DownEncoderBlock2D'''] ,up_block_types=['''UpDecoderBlock2D''', '''UpDecoderBlock2D''', '''UpDecoderBlock2D'''] ,latent_channels=4 ,)
return model
@property
def __lowerCamelCase ( self :str ):
torch.manual_seed(0 )
snake_case__ : int = CLIPTextConfig(
bos_token_id=0 ,eos_token_id=2 ,hidden_size=3_2 ,intermediate_size=3_7 ,layer_norm_eps=1e-0_5 ,num_attention_heads=4 ,num_hidden_layers=5 ,pad_token_id=1 ,vocab_size=1_0_0_0 ,hidden_act='''gelu''' ,projection_dim=5_1_2 ,)
return CLIPTextModel(__lowercase )
def __lowerCamelCase ( self :List[str] ):
snake_case__ : str = '''cpu''' # ensure determinism for the device-dependent torch.Generator
snake_case__ : str = self.dummy_cond_unet_upscale
snake_case__ : Optional[int] = DDPMScheduler()
snake_case__ : Tuple = DDIMScheduler(prediction_type='''v_prediction''' )
snake_case__ : List[Any] = self.dummy_vae
snake_case__ : Optional[int] = self.dummy_text_encoder
snake_case__ : Union[str, Any] = CLIPTokenizer.from_pretrained('''hf-internal-testing/tiny-random-clip''' )
snake_case__ : Any = self.dummy_image.cpu().permute(0 ,2 ,3 ,1 )[0]
snake_case__ : List[str] = Image.fromarray(np.uinta(__lowercase ) ).convert('''RGB''' ).resize((6_4, 6_4) )
# make sure here that pndm scheduler skips prk
snake_case__ : Union[str, Any] = StableDiffusionUpscalePipeline(
unet=__lowercase ,low_res_scheduler=__lowercase ,scheduler=__lowercase ,vae=__lowercase ,text_encoder=__lowercase ,tokenizer=__lowercase ,max_noise_level=3_5_0 ,)
snake_case__ : List[str] = sd_pipe.to(__lowercase )
sd_pipe.set_progress_bar_config(disable=__lowercase )
snake_case__ : Tuple = '''A painting of a squirrel eating a burger'''
snake_case__ : int = torch.Generator(device=__lowercase ).manual_seed(0 )
snake_case__ : Optional[Any] = sd_pipe(
[prompt] ,image=__lowercase ,generator=__lowercase ,guidance_scale=6.0 ,noise_level=2_0 ,num_inference_steps=2 ,output_type='''np''' ,)
snake_case__ : Optional[int] = output.images
snake_case__ : Dict = torch.Generator(device=__lowercase ).manual_seed(0 )
snake_case__ : Tuple = sd_pipe(
[prompt] ,image=__lowercase ,generator=__lowercase ,guidance_scale=6.0 ,noise_level=2_0 ,num_inference_steps=2 ,output_type='''np''' ,return_dict=__lowercase ,)[0]
snake_case__ : List[str] = image[0, -3:, -3:, -1]
snake_case__ : List[Any] = image_from_tuple[0, -3:, -3:, -1]
snake_case__ : Any = low_res_image.size[0] * 4
assert image.shape == (1, expected_height_width, expected_height_width, 3)
snake_case__ : List[Any] = np.array([0.3113, 0.3910, 0.4272, 0.4859, 0.5061, 0.4652, 0.5362, 0.5715, 0.5661] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1e-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
def __lowerCamelCase ( self :int ):
snake_case__ : Tuple = '''cpu''' # ensure determinism for the device-dependent torch.Generator
snake_case__ : int = self.dummy_cond_unet_upscale
snake_case__ : Optional[int] = DDPMScheduler()
snake_case__ : str = DDIMScheduler(prediction_type='''v_prediction''' )
snake_case__ : Any = self.dummy_vae
snake_case__ : Any = self.dummy_text_encoder
snake_case__ : Tuple = CLIPTokenizer.from_pretrained('''hf-internal-testing/tiny-random-clip''' )
snake_case__ : int = self.dummy_image.cpu().permute(0 ,2 ,3 ,1 )[0]
snake_case__ : Union[str, Any] = Image.fromarray(np.uinta(__lowercase ) ).convert('''RGB''' ).resize((6_4, 6_4) )
# make sure here that pndm scheduler skips prk
snake_case__ : Union[str, Any] = StableDiffusionUpscalePipeline(
unet=__lowercase ,low_res_scheduler=__lowercase ,scheduler=__lowercase ,vae=__lowercase ,text_encoder=__lowercase ,tokenizer=__lowercase ,max_noise_level=3_5_0 ,)
snake_case__ : Union[str, Any] = sd_pipe.to(__lowercase )
sd_pipe.set_progress_bar_config(disable=__lowercase )
snake_case__ : str = '''A painting of a squirrel eating a burger'''
snake_case__ : Tuple = sd_pipe(
2 * [prompt] ,image=2 * [low_res_image] ,guidance_scale=6.0 ,noise_level=2_0 ,num_inference_steps=2 ,output_type='''np''' ,)
snake_case__ : Tuple = output.images
assert image.shape[0] == 2
snake_case__ : Optional[Any] = torch.Generator(device=__lowercase ).manual_seed(0 )
snake_case__ : Dict = sd_pipe(
[prompt] ,image=__lowercase ,generator=__lowercase ,num_images_per_prompt=2 ,guidance_scale=6.0 ,noise_level=2_0 ,num_inference_steps=2 ,output_type='''np''' ,)
snake_case__ : Union[str, Any] = output.images
assert image.shape[0] == 2
@unittest.skipIf(torch_device != '''cuda''' ,'''This test requires a GPU''' )
def __lowerCamelCase ( self :Tuple ):
snake_case__ : Tuple = self.dummy_cond_unet_upscale
snake_case__ : Tuple = DDPMScheduler()
snake_case__ : Dict = DDIMScheduler(prediction_type='''v_prediction''' )
snake_case__ : int = self.dummy_vae
snake_case__ : List[Any] = self.dummy_text_encoder
snake_case__ : List[Any] = CLIPTokenizer.from_pretrained('''hf-internal-testing/tiny-random-clip''' )
snake_case__ : Tuple = self.dummy_image.cpu().permute(0 ,2 ,3 ,1 )[0]
snake_case__ : Tuple = Image.fromarray(np.uinta(__lowercase ) ).convert('''RGB''' ).resize((6_4, 6_4) )
# put models in fp16, except vae as it overflows in fp16
snake_case__ : Optional[Any] = unet.half()
snake_case__ : Any = text_encoder.half()
# make sure here that pndm scheduler skips prk
snake_case__ : Tuple = StableDiffusionUpscalePipeline(
unet=__lowercase ,low_res_scheduler=__lowercase ,scheduler=__lowercase ,vae=__lowercase ,text_encoder=__lowercase ,tokenizer=__lowercase ,max_noise_level=3_5_0 ,)
snake_case__ : str = sd_pipe.to(__lowercase )
sd_pipe.set_progress_bar_config(disable=__lowercase )
snake_case__ : List[Any] = '''A painting of a squirrel eating a burger'''
snake_case__ : Optional[int] = torch.manual_seed(0 )
snake_case__ : str = sd_pipe(
[prompt] ,image=__lowercase ,generator=__lowercase ,num_inference_steps=2 ,output_type='''np''' ,).images
snake_case__ : Optional[Any] = low_res_image.size[0] * 4
assert image.shape == (1, expected_height_width, expected_height_width, 3)
@slow
@require_torch_gpu
class a ( unittest.TestCase ):
def __lowerCamelCase ( self :Tuple ):
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def __lowerCamelCase ( self :Optional[Any] ):
snake_case__ : List[Any] = load_image(
'''https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main'''
'''/sd2-upscale/low_res_cat.png''' )
snake_case__ : Optional[int] = load_numpy(
'''https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd2-upscale'''
'''/upsampled_cat.npy''' )
snake_case__ : int = '''stabilityai/stable-diffusion-x4-upscaler'''
snake_case__ : Optional[Any] = StableDiffusionUpscalePipeline.from_pretrained(__lowercase )
pipe.to(__lowercase )
pipe.set_progress_bar_config(disable=__lowercase )
pipe.enable_attention_slicing()
snake_case__ : List[str] = '''a cat sitting on a park bench'''
snake_case__ : List[Any] = torch.manual_seed(0 )
snake_case__ : Any = pipe(
prompt=__lowercase ,image=__lowercase ,generator=__lowercase ,output_type='''np''' ,)
snake_case__ : Any = output.images[0]
assert image.shape == (5_1_2, 5_1_2, 3)
assert np.abs(expected_image - image ).max() < 1e-3
def __lowerCamelCase ( self :int ):
snake_case__ : Union[str, Any] = load_image(
'''https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main'''
'''/sd2-upscale/low_res_cat.png''' )
snake_case__ : Tuple = load_numpy(
'''https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd2-upscale'''
'''/upsampled_cat_fp16.npy''' )
snake_case__ : Tuple = '''stabilityai/stable-diffusion-x4-upscaler'''
snake_case__ : Optional[Any] = StableDiffusionUpscalePipeline.from_pretrained(
__lowercase ,torch_dtype=torch.floataa ,)
pipe.to(__lowercase )
pipe.set_progress_bar_config(disable=__lowercase )
pipe.enable_attention_slicing()
snake_case__ : Union[str, Any] = '''a cat sitting on a park bench'''
snake_case__ : Optional[int] = torch.manual_seed(0 )
snake_case__ : List[Any] = pipe(
prompt=__lowercase ,image=__lowercase ,generator=__lowercase ,output_type='''np''' ,)
snake_case__ : Tuple = output.images[0]
assert image.shape == (5_1_2, 5_1_2, 3)
assert np.abs(expected_image - image ).max() < 5e-1
def __lowerCamelCase ( self :Union[str, Any] ):
torch.cuda.empty_cache()
torch.cuda.reset_max_memory_allocated()
torch.cuda.reset_peak_memory_stats()
snake_case__ : Optional[int] = load_image(
'''https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main'''
'''/sd2-upscale/low_res_cat.png''' )
snake_case__ : Optional[Any] = '''stabilityai/stable-diffusion-x4-upscaler'''
snake_case__ : List[Any] = StableDiffusionUpscalePipeline.from_pretrained(
__lowercase ,torch_dtype=torch.floataa ,)
pipe.to(__lowercase )
pipe.set_progress_bar_config(disable=__lowercase )
pipe.enable_attention_slicing(1 )
pipe.enable_sequential_cpu_offload()
snake_case__ : List[Any] = '''a cat sitting on a park bench'''
snake_case__ : List[Any] = torch.manual_seed(0 )
snake_case__ : Tuple = pipe(
prompt=__lowercase ,image=__lowercase ,generator=__lowercase ,num_inference_steps=5 ,output_type='''np''' ,)
snake_case__ : str = torch.cuda.max_memory_allocated()
# make sure that less than 2.9 GB is allocated
assert mem_bytes < 2.9 * 1_0**9
| 252
| 1
|
'''simple docstring'''
__lowerCamelCase : Union[str, Any] = frozenset(
[
"""prompt""",
"""height""",
"""width""",
"""guidance_scale""",
"""negative_prompt""",
"""prompt_embeds""",
"""negative_prompt_embeds""",
"""cross_attention_kwargs""",
]
)
__lowerCamelCase : Union[str, Any] = frozenset(["""prompt""", """negative_prompt"""])
__lowerCamelCase : List[Any] = frozenset([])
__lowerCamelCase : Tuple = frozenset(["""image"""])
__lowerCamelCase : Optional[Any] = frozenset(
[
"""image""",
"""height""",
"""width""",
"""guidance_scale""",
]
)
__lowerCamelCase : List[str] = frozenset(["""image"""])
__lowerCamelCase : str = frozenset(
[
"""prompt""",
"""image""",
"""height""",
"""width""",
"""guidance_scale""",
"""negative_prompt""",
"""prompt_embeds""",
"""negative_prompt_embeds""",
]
)
__lowerCamelCase : str = frozenset(["""prompt""", """image""", """negative_prompt"""])
__lowerCamelCase : Optional[Any] = frozenset(
[
# Text guided image variation with an image mask
"""prompt""",
"""image""",
"""mask_image""",
"""height""",
"""width""",
"""guidance_scale""",
"""negative_prompt""",
"""prompt_embeds""",
"""negative_prompt_embeds""",
]
)
__lowerCamelCase : List[str] = frozenset(["""prompt""", """image""", """mask_image""", """negative_prompt"""])
__lowerCamelCase : Tuple = frozenset(
[
# image variation with an image mask
"""image""",
"""mask_image""",
"""height""",
"""width""",
"""guidance_scale""",
]
)
__lowerCamelCase : Union[str, Any] = frozenset(["""image""", """mask_image"""])
__lowerCamelCase : str = frozenset(
[
"""example_image""",
"""image""",
"""mask_image""",
"""height""",
"""width""",
"""guidance_scale""",
]
)
__lowerCamelCase : Tuple = frozenset(["""example_image""", """image""", """mask_image"""])
__lowerCamelCase : int = frozenset(["""class_labels"""])
__lowerCamelCase : Optional[Any] = frozenset(["""class_labels"""])
__lowerCamelCase : Optional[int] = frozenset(["""batch_size"""])
__lowerCamelCase : List[Any] = frozenset([])
__lowerCamelCase : List[Any] = frozenset(["""batch_size"""])
__lowerCamelCase : int = frozenset([])
__lowerCamelCase : Optional[int] = frozenset(
[
"""prompt""",
"""audio_length_in_s""",
"""guidance_scale""",
"""negative_prompt""",
"""prompt_embeds""",
"""negative_prompt_embeds""",
"""cross_attention_kwargs""",
]
)
__lowerCamelCase : List[str] = frozenset(["""prompt""", """negative_prompt"""])
__lowerCamelCase : int = frozenset(["""input_tokens"""])
__lowerCamelCase : Tuple = frozenset(["""input_tokens"""])
| 715
|
'''simple docstring'''
from typing import Any, Dict, Optional
import torch
import torch.nn.functional as F
from torch import nn
from ..utils import maybe_allow_in_graph
from .activations import get_activation
from .attention_processor import Attention
from .embeddings import CombinedTimestepLabelEmbeddings
@maybe_allow_in_graph
class lowerCAmelCase__ ( nn.Module ):
def __init__( self : int , UpperCamelCase_ : int , UpperCamelCase_ : int , UpperCamelCase_ : int , UpperCamelCase_ : Tuple=0.0 , UpperCamelCase_ : Optional[int] = None , UpperCamelCase_ : str = "geglu" , UpperCamelCase_ : Optional[int] = None , UpperCamelCase_ : bool = False , UpperCamelCase_ : bool = False , UpperCamelCase_ : bool = False , UpperCamelCase_ : bool = False , UpperCamelCase_ : bool = True , UpperCamelCase_ : str = "layer_norm" , UpperCamelCase_ : bool = False , ) -> Tuple:
"""simple docstring"""
super().__init__()
lowerCamelCase_ : int = only_cross_attention
lowerCamelCase_ : Dict = (num_embeds_ada_norm is not None) and norm_type == '''ada_norm_zero'''
lowerCamelCase_ : Optional[int] = (num_embeds_ada_norm is not None) and norm_type == '''ada_norm'''
if norm_type in ("ada_norm", "ada_norm_zero") and num_embeds_ada_norm is None:
raise ValueError(
F"""`norm_type` is set to {norm_type}, but `num_embeds_ada_norm` is not defined. Please make sure to"""
F""" define `num_embeds_ada_norm` if setting `norm_type` to {norm_type}.""" )
# Define 3 blocks. Each block has its own normalization layer.
# 1. Self-Attn
if self.use_ada_layer_norm:
lowerCamelCase_ : Optional[int] = AdaLayerNorm(UpperCamelCase_ , UpperCamelCase_ )
elif self.use_ada_layer_norm_zero:
lowerCamelCase_ : Tuple = AdaLayerNormZero(UpperCamelCase_ , UpperCamelCase_ )
else:
lowerCamelCase_ : Any = nn.LayerNorm(UpperCamelCase_ , elementwise_affine=UpperCamelCase_ )
lowerCamelCase_ : Tuple = Attention(
query_dim=UpperCamelCase_ , heads=UpperCamelCase_ , dim_head=UpperCamelCase_ , dropout=UpperCamelCase_ , bias=UpperCamelCase_ , cross_attention_dim=cross_attention_dim if only_cross_attention else None , upcast_attention=UpperCamelCase_ , )
# 2. Cross-Attn
if cross_attention_dim is not None or double_self_attention:
# We currently only use AdaLayerNormZero for self attention where there will only be one attention block.
# I.e. the number of returned modulation chunks from AdaLayerZero would not make sense if returned during
# the second cross attention block.
lowerCamelCase_ : List[str] = (
AdaLayerNorm(UpperCamelCase_ , UpperCamelCase_ )
if self.use_ada_layer_norm
else nn.LayerNorm(UpperCamelCase_ , elementwise_affine=UpperCamelCase_ )
)
lowerCamelCase_ : List[str] = Attention(
query_dim=UpperCamelCase_ , cross_attention_dim=cross_attention_dim if not double_self_attention else None , heads=UpperCamelCase_ , dim_head=UpperCamelCase_ , dropout=UpperCamelCase_ , bias=UpperCamelCase_ , upcast_attention=UpperCamelCase_ , ) # is self-attn if encoder_hidden_states is none
else:
lowerCamelCase_ : Optional[int] = None
lowerCamelCase_ : List[str] = None
# 3. Feed-forward
lowerCamelCase_ : Union[str, Any] = nn.LayerNorm(UpperCamelCase_ , elementwise_affine=UpperCamelCase_ )
lowerCamelCase_ : Optional[int] = FeedForward(UpperCamelCase_ , dropout=UpperCamelCase_ , activation_fn=UpperCamelCase_ , final_dropout=UpperCamelCase_ )
# let chunk size default to None
lowerCamelCase_ : int = None
lowerCamelCase_ : str = 0
def __UpperCamelCase ( self : Dict , UpperCamelCase_ : Optional[int] , UpperCamelCase_ : int ) -> str:
"""simple docstring"""
lowerCamelCase_ : int = chunk_size
lowerCamelCase_ : Dict = dim
def __UpperCamelCase ( self : List[str] , UpperCamelCase_ : torch.FloatTensor , UpperCamelCase_ : Optional[torch.FloatTensor] = None , UpperCamelCase_ : Optional[torch.FloatTensor] = None , UpperCamelCase_ : Optional[torch.FloatTensor] = None , UpperCamelCase_ : Optional[torch.LongTensor] = None , UpperCamelCase_ : Dict[str, Any] = None , UpperCamelCase_ : Optional[torch.LongTensor] = None , ) -> Dict:
"""simple docstring"""
if self.use_ada_layer_norm:
lowerCamelCase_ : int = self.norma(UpperCamelCase_ , UpperCamelCase_ )
elif self.use_ada_layer_norm_zero:
lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ : Any = self.norma(
UpperCamelCase_ , UpperCamelCase_ , UpperCamelCase_ , hidden_dtype=hidden_states.dtype )
else:
lowerCamelCase_ : Optional[Any] = self.norma(UpperCamelCase_ )
lowerCamelCase_ : str = cross_attention_kwargs if cross_attention_kwargs is not None else {}
lowerCamelCase_ : int = self.attna(
UpperCamelCase_ , encoder_hidden_states=encoder_hidden_states if self.only_cross_attention else None , attention_mask=UpperCamelCase_ , **UpperCamelCase_ , )
if self.use_ada_layer_norm_zero:
lowerCamelCase_ : str = gate_msa.unsqueeze(1 ) * attn_output
lowerCamelCase_ : Tuple = attn_output + hidden_states
# 2. Cross-Attention
if self.attna is not None:
lowerCamelCase_ : List[Any] = (
self.norma(UpperCamelCase_ , UpperCamelCase_ ) if self.use_ada_layer_norm else self.norma(UpperCamelCase_ )
)
lowerCamelCase_ : Tuple = self.attna(
UpperCamelCase_ , encoder_hidden_states=UpperCamelCase_ , attention_mask=UpperCamelCase_ , **UpperCamelCase_ , )
lowerCamelCase_ : str = attn_output + hidden_states
# 3. Feed-forward
lowerCamelCase_ : Tuple = self.norma(UpperCamelCase_ )
if self.use_ada_layer_norm_zero:
lowerCamelCase_ : str = norm_hidden_states * (1 + scale_mlp[:, None]) + shift_mlp[:, None]
if self._chunk_size is not None:
# "feed_forward_chunk_size" can be used to save memory
if norm_hidden_states.shape[self._chunk_dim] % self._chunk_size != 0:
raise ValueError(
F"""`hidden_states` dimension to be chunked: {norm_hidden_states.shape[self._chunk_dim]} has to be divisible by chunk size: {self._chunk_size}. Make sure to set an appropriate `chunk_size` when calling `unet.enable_forward_chunking`.""" )
lowerCamelCase_ : Optional[Any] = norm_hidden_states.shape[self._chunk_dim] // self._chunk_size
lowerCamelCase_ : Optional[int] = torch.cat(
[self.ff(UpperCamelCase_ ) for hid_slice in norm_hidden_states.chunk(UpperCamelCase_ , dim=self._chunk_dim )] , dim=self._chunk_dim , )
else:
lowerCamelCase_ : Optional[Any] = self.ff(UpperCamelCase_ )
if self.use_ada_layer_norm_zero:
lowerCamelCase_ : List[str] = gate_mlp.unsqueeze(1 ) * ff_output
lowerCamelCase_ : Optional[int] = ff_output + hidden_states
return hidden_states
class lowerCAmelCase__ ( nn.Module ):
def __init__( self : List[Any] , UpperCamelCase_ : int , UpperCamelCase_ : Optional[int] = None , UpperCamelCase_ : int = 4 , UpperCamelCase_ : float = 0.0 , UpperCamelCase_ : str = "geglu" , UpperCamelCase_ : bool = False , ) -> Dict:
"""simple docstring"""
super().__init__()
lowerCamelCase_ : Tuple = int(dim * mult )
lowerCamelCase_ : List[Any] = dim_out if dim_out is not None else dim
if activation_fn == "gelu":
lowerCamelCase_ : Optional[int] = GELU(UpperCamelCase_ , UpperCamelCase_ )
if activation_fn == "gelu-approximate":
lowerCamelCase_ : Any = GELU(UpperCamelCase_ , UpperCamelCase_ , approximate='''tanh''' )
elif activation_fn == "geglu":
lowerCamelCase_ : Tuple = GEGLU(UpperCamelCase_ , UpperCamelCase_ )
elif activation_fn == "geglu-approximate":
lowerCamelCase_ : Union[str, Any] = ApproximateGELU(UpperCamelCase_ , UpperCamelCase_ )
lowerCamelCase_ : Any = nn.ModuleList([] )
# project in
self.net.append(UpperCamelCase_ )
# project dropout
self.net.append(nn.Dropout(UpperCamelCase_ ) )
# project out
self.net.append(nn.Linear(UpperCamelCase_ , UpperCamelCase_ ) )
# FF as used in Vision Transformer, MLP-Mixer, etc. have a final dropout
if final_dropout:
self.net.append(nn.Dropout(UpperCamelCase_ ) )
def __UpperCamelCase ( self : List[Any] , UpperCamelCase_ : str ) -> Dict:
"""simple docstring"""
for module in self.net:
lowerCamelCase_ : Optional[int] = module(UpperCamelCase_ )
return hidden_states
class lowerCAmelCase__ ( nn.Module ):
def __init__( self : Optional[int] , UpperCamelCase_ : int , UpperCamelCase_ : int , UpperCamelCase_ : str = "none" ) -> int:
"""simple docstring"""
super().__init__()
lowerCamelCase_ : List[str] = nn.Linear(UpperCamelCase_ , UpperCamelCase_ )
lowerCamelCase_ : int = approximate
def __UpperCamelCase ( self : Tuple , UpperCamelCase_ : Optional[int] ) -> Optional[Any]:
"""simple docstring"""
if gate.device.type != "mps":
return F.gelu(UpperCamelCase_ , approximate=self.approximate )
# mps: gelu is not implemented for float16
return F.gelu(gate.to(dtype=torch.floataa ) , approximate=self.approximate ).to(dtype=gate.dtype )
def __UpperCamelCase ( self : Optional[Any] , UpperCamelCase_ : Any ) -> Tuple:
"""simple docstring"""
lowerCamelCase_ : List[str] = self.proj(UpperCamelCase_ )
lowerCamelCase_ : int = self.gelu(UpperCamelCase_ )
return hidden_states
class lowerCAmelCase__ ( nn.Module ):
def __init__( self : Dict , UpperCamelCase_ : int , UpperCamelCase_ : int ) -> Any:
"""simple docstring"""
super().__init__()
lowerCamelCase_ : Optional[Any] = nn.Linear(UpperCamelCase_ , dim_out * 2 )
def __UpperCamelCase ( self : Any , UpperCamelCase_ : Optional[int] ) -> List[str]:
"""simple docstring"""
if gate.device.type != "mps":
return F.gelu(UpperCamelCase_ )
# mps: gelu is not implemented for float16
return F.gelu(gate.to(dtype=torch.floataa ) ).to(dtype=gate.dtype )
def __UpperCamelCase ( self : Dict , UpperCamelCase_ : List[Any] ) -> Any:
"""simple docstring"""
lowerCamelCase_ , lowerCamelCase_ : int = self.proj(UpperCamelCase_ ).chunk(2 , dim=-1 )
return hidden_states * self.gelu(UpperCamelCase_ )
class lowerCAmelCase__ ( nn.Module ):
def __init__( self : Dict , UpperCamelCase_ : int , UpperCamelCase_ : int ) -> List[str]:
"""simple docstring"""
super().__init__()
lowerCamelCase_ : List[Any] = nn.Linear(UpperCamelCase_ , UpperCamelCase_ )
def __UpperCamelCase ( self : List[Any] , UpperCamelCase_ : Union[str, Any] ) -> Optional[int]:
"""simple docstring"""
lowerCamelCase_ : List[Any] = self.proj(UpperCamelCase_ )
return x * torch.sigmoid(1.702 * x )
class lowerCAmelCase__ ( nn.Module ):
def __init__( self : Union[str, Any] , UpperCamelCase_ : Optional[int] , UpperCamelCase_ : List[str] ) -> str:
"""simple docstring"""
super().__init__()
lowerCamelCase_ : Tuple = nn.Embedding(UpperCamelCase_ , UpperCamelCase_ )
lowerCamelCase_ : Tuple = nn.SiLU()
lowerCamelCase_ : List[str] = nn.Linear(UpperCamelCase_ , embedding_dim * 2 )
lowerCamelCase_ : List[Any] = nn.LayerNorm(UpperCamelCase_ , elementwise_affine=UpperCamelCase_ )
def __UpperCamelCase ( self : Tuple , UpperCamelCase_ : Dict , UpperCamelCase_ : Union[str, Any] ) -> Optional[Any]:
"""simple docstring"""
lowerCamelCase_ : Tuple = self.linear(self.silu(self.emb(UpperCamelCase_ ) ) )
lowerCamelCase_ , lowerCamelCase_ : Optional[int] = torch.chunk(UpperCamelCase_ , 2 )
lowerCamelCase_ : List[Any] = self.norm(UpperCamelCase_ ) * (1 + scale) + shift
return x
class lowerCAmelCase__ ( nn.Module ):
def __init__( self : Tuple , UpperCamelCase_ : List[Any] , UpperCamelCase_ : Optional[Any] ) -> Optional[Any]:
"""simple docstring"""
super().__init__()
lowerCamelCase_ : Tuple = CombinedTimestepLabelEmbeddings(UpperCamelCase_ , UpperCamelCase_ )
lowerCamelCase_ : List[Any] = nn.SiLU()
lowerCamelCase_ : str = nn.Linear(UpperCamelCase_ , 6 * embedding_dim , bias=UpperCamelCase_ )
lowerCamelCase_ : Dict = nn.LayerNorm(UpperCamelCase_ , elementwise_affine=UpperCamelCase_ , eps=1e-6 )
def __UpperCamelCase ( self : Tuple , UpperCamelCase_ : Any , UpperCamelCase_ : Optional[Any] , UpperCamelCase_ : Any , UpperCamelCase_ : int=None ) -> Any:
"""simple docstring"""
lowerCamelCase_ : Optional[Any] = self.linear(self.silu(self.emb(UpperCamelCase_ , UpperCamelCase_ , hidden_dtype=UpperCamelCase_ ) ) )
lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ : Union[str, Any] = emb.chunk(6 , dim=1 )
lowerCamelCase_ : Tuple = self.norm(UpperCamelCase_ ) * (1 + scale_msa[:, None]) + shift_msa[:, None]
return x, gate_msa, shift_mlp, scale_mlp, gate_mlp
class lowerCAmelCase__ ( nn.Module ):
def __init__( self : Optional[int] , UpperCamelCase_ : int , UpperCamelCase_ : int , UpperCamelCase_ : int , UpperCamelCase_ : Optional[str] = None , UpperCamelCase_ : float = 1e-5 ) -> Tuple:
"""simple docstring"""
super().__init__()
lowerCamelCase_ : str = num_groups
lowerCamelCase_ : List[Any] = eps
if act_fn is None:
lowerCamelCase_ : Any = None
else:
lowerCamelCase_ : List[str] = get_activation(UpperCamelCase_ )
lowerCamelCase_ : Optional[Any] = nn.Linear(UpperCamelCase_ , out_dim * 2 )
def __UpperCamelCase ( self : Optional[Any] , UpperCamelCase_ : Optional[Any] , UpperCamelCase_ : Optional[Any] ) -> Optional[int]:
"""simple docstring"""
if self.act:
lowerCamelCase_ : Optional[int] = self.act(UpperCamelCase_ )
lowerCamelCase_ : Optional[int] = self.linear(UpperCamelCase_ )
lowerCamelCase_ : List[str] = emb[:, :, None, None]
lowerCamelCase_ , lowerCamelCase_ : int = emb.chunk(2 , dim=1 )
lowerCamelCase_ : List[str] = F.group_norm(UpperCamelCase_ , self.num_groups , eps=self.eps )
lowerCamelCase_ : Optional[Any] = x * (1 + scale) + shift
return x
| 418
| 0
|
from __future__ import annotations
import unittest
from transformers import is_tf_available
from transformers.testing_utils import require_sentencepiece, require_tf, require_tokenizers, slow
if is_tf_available():
import numpy as np
import tensorflow as tf
from transformers import TFCamembertModel
@require_tf
@require_sentencepiece
@require_tokenizers
class _A ( unittest.TestCase ):
"""simple docstring"""
@slow
def _a ( self : List[str] ) -> Tuple:
__UpperCAmelCase =TFCamembertModel.from_pretrained("""jplu/tf-camembert-base""" )
__UpperCAmelCase =tf.convert_to_tensor(
[[5, 121, 11, 660, 16, 730, 25543, 110, 83, 6]] , dtype=tf.intaa , ) # J'aime le camembert !"
__UpperCAmelCase =model(__SCREAMING_SNAKE_CASE )["""last_hidden_state"""]
__UpperCAmelCase =tf.TensorShape((1, 10, 768) )
self.assertEqual(output.shape , __SCREAMING_SNAKE_CASE )
# compare the actual values for a slice.
__UpperCAmelCase =tf.convert_to_tensor(
[[[-0.0_254, 0.0_235, 0.1_027], [0.0_606, -0.1_811, -0.0_418], [-0.1_561, -0.1_127, 0.2_687]]] , dtype=tf.floataa , )
# camembert = torch.hub.load('pytorch/fairseq', 'camembert.v0')
# camembert.eval()
# expected_slice = roberta.model.forward(input_ids)[0][:, :3, :3].detach()
self.assertTrue(np.allclose(output[:, :3, :3].numpy() , expected_slice.numpy() , atol=1e-4 ) )
| 68
|
'''simple docstring'''
# Lint as: python3
# pylint: enable=line-too-long
# pylint: disable=g-import-not-at-top,g-bad-import-order,wrong-import-position
__lowerCAmelCase = '2.13.1'
import platform
import pyarrow
from packaging import version
if version.parse(platform.python_version()) < version.parse('3.7'):
raise ImportWarning(
'To use `datasets`, Python>=3.7 is required, and the current version of Python doesn\'t match this condition.'
)
if version.parse(pyarrow.__version__).major < 8:
raise ImportWarning(
'To use `datasets`, the module `pyarrow>=8.0.0` is required, and the current version of `pyarrow` doesn\'t match this condition.\n'
'If you are running this in a Google Colab, you should probably just restart the runtime to use the right version of `pyarrow`.'
)
del platform
del pyarrow
del version
from .arrow_dataset import Dataset
from .arrow_reader import ReadInstruction
from .builder import ArrowBasedBuilder, BeamBasedBuilder, BuilderConfig, DatasetBuilder, GeneratorBasedBuilder
from .combine import concatenate_datasets, interleave_datasets
from .dataset_dict import DatasetDict, IterableDatasetDict
from .download import *
from .features import *
from .fingerprint import disable_caching, enable_caching, is_caching_enabled, set_caching_enabled
from .info import DatasetInfo, MetricInfo
from .inspect import (
get_dataset_config_info,
get_dataset_config_names,
get_dataset_infos,
get_dataset_split_names,
inspect_dataset,
inspect_metric,
list_datasets,
list_metrics,
)
from .iterable_dataset import IterableDataset
from .load import load_dataset, load_dataset_builder, load_from_disk, load_metric
from .metric import Metric
from .splits import (
NamedSplit,
NamedSplitAll,
Split,
SplitBase,
SplitDict,
SplitGenerator,
SplitInfo,
SubSplitInfo,
percent,
)
from .tasks import *
from .utils import *
from .utils import logging
# deprecated modules
from datasets import arrow_dataset as _arrow_dataset # isort:skip
from datasets import utils as _utils # isort:skip
from datasets.utils import download_manager as _deprecated_download_manager # isort:skip
__lowerCAmelCase = concatenate_datasets
__lowerCAmelCase = DownloadConfig
__lowerCAmelCase = DownloadManager
__lowerCAmelCase = DownloadMode
__lowerCAmelCase = DownloadConfig
__lowerCAmelCase = DownloadMode
__lowerCAmelCase = DownloadManager
del _arrow_dataset, _utils, _deprecated_download_manager
| 585
| 0
|
import importlib
import os
import fsspec
import pytest
from fsspec import register_implementation
from fsspec.registry import _registry as _fsspec_registry
from datasets.filesystems import COMPRESSION_FILESYSTEMS, HfFileSystem, extract_path_from_uri, is_remote_filesystem
from .utils import require_lza, require_zstandard
def lowerCamelCase_ ( UpperCAmelCase_ : Optional[int] ) -> int:
'''simple docstring'''
assert "mock" in _fsspec_registry
assert "bz2" in _fsspec_registry
def lowerCamelCase_ ( ) -> Optional[int]:
'''simple docstring'''
assert "mock" not in _fsspec_registry
assert "bz2" in _fsspec_registry
def lowerCamelCase_ ( ) -> int:
'''simple docstring'''
_UpperCamelCase : Dict = 'mock-s3-bucket'
_UpperCamelCase : List[Any] = F'''s3://{mock_bucket}'''
_UpperCamelCase : Dict = extract_path_from_uri(UpperCAmelCase_ )
assert dataset_path.startswith('s3://' ) is False
_UpperCamelCase : Optional[Any] = './local/path'
_UpperCamelCase : List[str] = extract_path_from_uri(UpperCAmelCase_ )
assert dataset_path == new_dataset_path
def lowerCamelCase_ ( UpperCAmelCase_ : Tuple ) -> Optional[int]:
'''simple docstring'''
_UpperCamelCase : Tuple = is_remote_filesystem(UpperCAmelCase_ )
assert is_remote is True
_UpperCamelCase : Tuple = fsspec.filesystem('file' )
_UpperCamelCase : Tuple = is_remote_filesystem(UpperCAmelCase_ )
assert is_remote is False
@pytest.mark.parametrize('compression_fs_class' , UpperCAmelCase_ )
def lowerCamelCase_ ( UpperCAmelCase_ : int , UpperCAmelCase_ : Dict , UpperCAmelCase_ : int , UpperCAmelCase_ : List[str] , UpperCAmelCase_ : str , UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : Optional[Any] ) -> str:
'''simple docstring'''
_UpperCamelCase : Dict = {'gzip': gz_file, 'xz': xz_file, 'zstd': zstd_file, 'bz2': bza_file, 'lz4': lza_file}
_UpperCamelCase : int = input_paths[compression_fs_class.protocol]
if input_path is None:
_UpperCamelCase : List[str] = F'''for \'{compression_fs_class.protocol}\' compression protocol, '''
if compression_fs_class.protocol == "lz4":
reason += require_lza.kwargs["reason"]
elif compression_fs_class.protocol == "zstd":
reason += require_zstandard.kwargs["reason"]
pytest.skip(UpperCAmelCase_ )
_UpperCamelCase : Union[str, Any] = fsspec.filesystem(compression_fs_class.protocol , fo=UpperCAmelCase_ )
assert isinstance(UpperCAmelCase_ , UpperCAmelCase_ )
_UpperCamelCase : List[str] = os.path.basename(UpperCAmelCase_ )
_UpperCamelCase : Union[str, Any] = expected_filename[: expected_filename.rindex('.' )]
assert fs.glob('*' ) == [expected_filename]
with fs.open(UpperCAmelCase_ , 'r' , encoding='utf-8' ) as f, open(UpperCAmelCase_ , encoding='utf-8' ) as expected_file:
assert f.read() == expected_file.read()
@pytest.mark.parametrize('protocol' , ['zip', 'gzip'] )
def lowerCamelCase_ ( UpperCAmelCase_ : int , UpperCAmelCase_ : Dict , UpperCAmelCase_ : Optional[Any] ) -> Optional[Any]:
'''simple docstring'''
_UpperCamelCase : int = {'zip': zip_jsonl_path, 'gzip': jsonl_gz_path}
_UpperCamelCase : Optional[Any] = compressed_file_paths[protocol]
_UpperCamelCase : Union[str, Any] = 'dataset.jsonl'
_UpperCamelCase : Optional[int] = F'''{protocol}://{member_file_path}::{compressed_file_path}'''
_UpperCamelCase : str = fsspec.get_fs_token_paths(UpperCAmelCase_ )
assert fs.isfile(UpperCAmelCase_ )
assert not fs.isfile('non_existing_' + member_file_path )
@pytest.mark.integration
def lowerCamelCase_ ( UpperCAmelCase_ : List[Any] , UpperCAmelCase_ : int , UpperCAmelCase_ : Dict , UpperCAmelCase_ : List[str] ) -> Tuple:
'''simple docstring'''
_UpperCamelCase : Optional[int] = hf_api.dataset_info(UpperCAmelCase_ , token=UpperCAmelCase_ )
_UpperCamelCase : Any = HfFileSystem(repo_info=UpperCAmelCase_ , token=UpperCAmelCase_ )
assert sorted(hffs.glob('*' ) ) == [".gitattributes", "data"]
assert hffs.isdir('data' )
assert hffs.isfile('.gitattributes' ) and hffs.isfile('data/text_data.txt' )
with open(UpperCAmelCase_ ) as f:
assert hffs.open('data/text_data.txt' , 'r' ).read() == f.read()
def lowerCamelCase_ ( ) -> Dict:
'''simple docstring'''
_UpperCamelCase : List[Any] = 'bz2'
# Import module
import datasets.filesystems
# Overwrite protocol and reload
register_implementation(UpperCAmelCase_ , UpperCAmelCase_ , clobber=UpperCAmelCase_ )
with pytest.warns(UpperCAmelCase_ ) as warning_info:
importlib.reload(datasets.filesystems )
assert len(UpperCAmelCase_ ) == 1
assert (
str(warning_info[0].message )
== F'''A filesystem protocol was already set for {protocol} and will be overwritten.'''
)
| 707
|
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_tokenizers_available, is_torch_available
lowerCAmelCase__ = {
"""configuration_canine""": ["""CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP""", """CanineConfig"""],
"""tokenization_canine""": ["""CanineTokenizer"""],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
lowerCAmelCase__ = [
"""CANINE_PRETRAINED_MODEL_ARCHIVE_LIST""",
"""CanineForMultipleChoice""",
"""CanineForQuestionAnswering""",
"""CanineForSequenceClassification""",
"""CanineForTokenClassification""",
"""CanineLayer""",
"""CanineModel""",
"""CaninePreTrainedModel""",
"""load_tf_weights_in_canine""",
]
if TYPE_CHECKING:
from .configuration_canine import CANINE_PRETRAINED_CONFIG_ARCHIVE_MAP, CanineConfig
from .tokenization_canine import CanineTokenizer
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_canine import (
CANINE_PRETRAINED_MODEL_ARCHIVE_LIST,
CanineForMultipleChoice,
CanineForQuestionAnswering,
CanineForSequenceClassification,
CanineForTokenClassification,
CanineLayer,
CanineModel,
CaninePreTrainedModel,
load_tf_weights_in_canine,
)
else:
import sys
lowerCAmelCase__ = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 648
| 0
|
from typing import TYPE_CHECKING
from ...file_utils import _LazyModule, is_tokenizers_available, is_torch_available, is_vision_available
from ...utils import OptionalDependencyNotAvailable
UpperCamelCase__ ={'configuration_dpt': ['DPT_PRETRAINED_CONFIG_ARCHIVE_MAP', 'DPTConfig']}
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase__ =['DPTFeatureExtractor']
UpperCamelCase__ =['DPTImageProcessor']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase__ =[
'DPT_PRETRAINED_MODEL_ARCHIVE_LIST',
'DPTForDepthEstimation',
'DPTForSemanticSegmentation',
'DPTModel',
'DPTPreTrainedModel',
]
if TYPE_CHECKING:
from .configuration_dpt import DPT_PRETRAINED_CONFIG_ARCHIVE_MAP, DPTConfig
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .feature_extraction_dpt import DPTFeatureExtractor
from .image_processing_dpt import DPTImageProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_dpt import (
DPT_PRETRAINED_MODEL_ARCHIVE_LIST,
DPTForDepthEstimation,
DPTForSemanticSegmentation,
DPTModel,
DPTPreTrainedModel,
)
else:
import sys
UpperCamelCase__ =_LazyModule(__name__, globals()['__file__'], _import_structure, module_spec=__spec__)
| 249
|
"""simple docstring"""
import sys
lowerCAmelCase__ = (
'''73167176531330624919225119674426574742355349194934'''
'''96983520312774506326239578318016984801869478851843'''
'''85861560789112949495459501737958331952853208805511'''
'''12540698747158523863050715693290963295227443043557'''
'''66896648950445244523161731856403098711121722383113'''
'''62229893423380308135336276614282806444486645238749'''
'''30358907296290491560440772390713810515859307960866'''
'''70172427121883998797908792274921901699720888093776'''
'''65727333001053367881220235421809751254540594752243'''
'''52584907711670556013604839586446706324415722155397'''
'''53697817977846174064955149290862569321978468622482'''
'''83972241375657056057490261407972968652414535100474'''
'''82166370484403199890008895243450658541227588666881'''
'''16427171479924442928230863465674813919123162824586'''
'''17866458359124566529476545682848912883142607690042'''
'''24219022671055626321111109370544217506941658960408'''
'''07198403850962455444362981230987879927244284909188'''
'''84580156166097919133875499200524063689912560717606'''
'''05886116467109405077541002256983155200055935729725'''
'''71636269561882670428252483600823257530420752963450'''
)
def a__ ( SCREAMING_SNAKE_CASE : str ):
'''simple docstring'''
lowerCAmelCase : List[str] = 1
for digit in s:
product *= int(SCREAMING_SNAKE_CASE )
return product
def a__ ( SCREAMING_SNAKE_CASE : str = N ):
'''simple docstring'''
lowerCAmelCase : Union[str, Any] = -sys.maxsize - 1
lowerCAmelCase : str = n[:1_3]
lowerCAmelCase : Tuple = 1_3
while cur_index < len(SCREAMING_SNAKE_CASE ) - 1_3:
if int(n[cur_index] ) >= int(substr[0] ):
lowerCAmelCase : Any = substr[1:] + n[cur_index]
cur_index += 1
else:
lowerCAmelCase : int = max(SCREAMING_SNAKE_CASE , str_eval(SCREAMING_SNAKE_CASE ) )
lowerCAmelCase : Dict = n[cur_index : cur_index + 1_3]
cur_index += 1_3
return largest_product
if __name__ == "__main__":
print(F"{solution() = }")
| 645
| 0
|
import argparse
import torch
from datasets import load_dataset
from donut import DonutModel
from transformers import (
DonutImageProcessor,
DonutProcessor,
DonutSwinConfig,
DonutSwinModel,
MBartConfig,
MBartForCausalLM,
VisionEncoderDecoderModel,
XLMRobertaTokenizerFast,
)
def __lowercase ( _a ):
snake_case_ : Optional[int] = model.config
snake_case_ : Optional[int] = DonutSwinConfig(
image_size=original_config.input_size , patch_size=4 , depths=original_config.encoder_layer , num_heads=[4, 8, 16, 32] , window_size=original_config.window_size , embed_dim=128 , )
snake_case_ : List[Any] = MBartConfig(
is_decoder=_a , is_encoder_decoder=_a , add_cross_attention=_a , decoder_layers=original_config.decoder_layer , max_position_embeddings=original_config.max_position_embeddings , vocab_size=len(
model.decoder.tokenizer ) , scale_embedding=_a , add_final_layer_norm=_a , )
return encoder_config, decoder_config
def __lowercase ( _a ):
if "encoder.model" in name:
snake_case_ : Dict = name.replace('''encoder.model''' , '''encoder''' )
if "decoder.model" in name:
snake_case_ : int = name.replace('''decoder.model''' , '''decoder''' )
if "patch_embed.proj" in name:
snake_case_ : str = name.replace('''patch_embed.proj''' , '''embeddings.patch_embeddings.projection''' )
if "patch_embed.norm" in name:
snake_case_ : Optional[Any] = name.replace('''patch_embed.norm''' , '''embeddings.norm''' )
if name.startswith('''encoder''' ):
if "layers" in name:
snake_case_ : List[str] = '''encoder.''' + name
if "attn.proj" in name:
snake_case_ : List[Any] = name.replace('''attn.proj''' , '''attention.output.dense''' )
if "attn" in name and "mask" not in name:
snake_case_ : Dict = name.replace('''attn''' , '''attention.self''' )
if "norm1" in name:
snake_case_ : Any = name.replace('''norm1''' , '''layernorm_before''' )
if "norm2" in name:
snake_case_ : Union[str, Any] = name.replace('''norm2''' , '''layernorm_after''' )
if "mlp.fc1" in name:
snake_case_ : Optional[int] = name.replace('''mlp.fc1''' , '''intermediate.dense''' )
if "mlp.fc2" in name:
snake_case_ : List[str] = name.replace('''mlp.fc2''' , '''output.dense''' )
if name == "encoder.norm.weight":
snake_case_ : str = '''encoder.layernorm.weight'''
if name == "encoder.norm.bias":
snake_case_ : str = '''encoder.layernorm.bias'''
return name
def __lowercase ( _a , _a ):
for key in orig_state_dict.copy().keys():
snake_case_ : int = orig_state_dict.pop(_a )
if "qkv" in key:
snake_case_ : Optional[Any] = key.split('''.''' )
snake_case_ : List[Any] = int(key_split[3] )
snake_case_ : List[Any] = int(key_split[5] )
snake_case_ : Tuple = model.encoder.encoder.layers[layer_num].blocks[block_num].attention.self.all_head_size
if "weight" in key:
snake_case_ : List[Any] = val[:dim, :]
snake_case_ : Optional[Any] = val[dim : dim * 2, :]
snake_case_ : Optional[int] = val[-dim:, :]
else:
snake_case_ : Any = val[:dim]
snake_case_ : Optional[int] = val[dim : dim * 2]
snake_case_ : str = val[-dim:]
elif "attn_mask" in key or key in ["encoder.model.norm.weight", "encoder.model.norm.bias"]:
# HuggingFace implementation doesn't use attn_mask buffer
# and model doesn't use final LayerNorms for the encoder
pass
else:
snake_case_ : Optional[Any] = val
return orig_state_dict
def __lowercase ( _a , _a=None , _a=False ):
snake_case_ : Tuple = DonutModel.from_pretrained(_a ).eval()
# load HuggingFace model
snake_case_ : Optional[int] = get_configs(_a )
snake_case_ : List[str] = DonutSwinModel(_a )
snake_case_ : str = MBartForCausalLM(_a )
snake_case_ : Union[str, Any] = VisionEncoderDecoderModel(encoder=_a , decoder=_a )
model.eval()
snake_case_ : Union[str, Any] = original_model.state_dict()
snake_case_ : Optional[int] = convert_state_dict(_a , _a )
model.load_state_dict(_a )
# verify results on scanned document
snake_case_ : Union[str, Any] = load_dataset('''hf-internal-testing/example-documents''' )
snake_case_ : Tuple = dataset['''test'''][0]['''image'''].convert('''RGB''' )
snake_case_ : str = XLMRobertaTokenizerFast.from_pretrained(_a , from_slow=_a )
snake_case_ : Tuple = DonutImageProcessor(
do_align_long_axis=original_model.config.align_long_axis , size=original_model.config.input_size[::-1] )
snake_case_ : List[Any] = DonutProcessor(_a , _a )
snake_case_ : List[str] = processor(_a , return_tensors='''pt''' ).pixel_values
if model_name == "naver-clova-ix/donut-base-finetuned-docvqa":
snake_case_ : List[str] = '''<s_docvqa><s_question>{user_input}</s_question><s_answer>'''
snake_case_ : Optional[int] = '''When is the coffee break?'''
snake_case_ : List[str] = task_prompt.replace('''{user_input}''' , _a )
elif model_name == "naver-clova-ix/donut-base-finetuned-rvlcdip":
snake_case_ : List[str] = '''<s_rvlcdip>'''
elif model_name in [
"naver-clova-ix/donut-base-finetuned-cord-v1",
"naver-clova-ix/donut-base-finetuned-cord-v1-2560",
]:
snake_case_ : Union[str, Any] = '''<s_cord>'''
elif model_name == "naver-clova-ix/donut-base-finetuned-cord-v2":
snake_case_ : Union[str, Any] = '''s_cord-v2>'''
elif model_name == "naver-clova-ix/donut-base-finetuned-zhtrainticket":
snake_case_ : Optional[Any] = '''<s_zhtrainticket>'''
elif model_name in ["naver-clova-ix/donut-proto", "naver-clova-ix/donut-base"]:
# use a random prompt
snake_case_ : Any = '''hello world'''
else:
raise ValueError('''Model name not supported''' )
snake_case_ : Optional[int] = original_model.decoder.tokenizer(_a , add_special_tokens=_a , return_tensors='''pt''' )[
'''input_ids'''
]
snake_case_ : Optional[int] = original_model.encoder.model.patch_embed(_a )
snake_case_ : Any = model.encoder.embeddings(_a )
assert torch.allclose(_a , _a , atol=1E-3 )
# verify encoder hidden states
snake_case_ : List[Any] = original_model.encoder(_a )
snake_case_ : Tuple = model.encoder(_a ).last_hidden_state
assert torch.allclose(_a , _a , atol=1E-2 )
# verify decoder hidden states
snake_case_ : Optional[int] = original_model(_a , _a , _a ).logits
snake_case_ : Optional[int] = model(_a , decoder_input_ids=_a ).logits
assert torch.allclose(_a , _a , atol=1E-3 )
print('''Looks ok!''' )
if pytorch_dump_folder_path is not None:
print(f"Saving model and processor to {pytorch_dump_folder_path}" )
model.save_pretrained(_a )
processor.save_pretrained(_a )
if push_to_hub:
model.push_to_hub('''nielsr/''' + model_name.split('''/''' )[-1] , commit_message='''Update model''' )
processor.push_to_hub('''nielsr/''' + model_name.split('''/''' )[-1] , commit_message='''Update model''' )
if __name__ == "__main__":
lowercase__ : List[str] = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--model_name''',
default='''naver-clova-ix/donut-base-finetuned-docvqa''',
required=False,
type=str,
help='''Name of the original model you\'d like to convert.''',
)
parser.add_argument(
'''--pytorch_dump_folder_path''',
default=None,
required=False,
type=str,
help='''Path to the output PyTorch model directory.''',
)
parser.add_argument(
'''--push_to_hub''',
action='''store_true''',
help='''Whether or not to push the converted model and processor to the 🤗 hub.''',
)
lowercase__ : str = parser.parse_args()
convert_donut_checkpoint(args.model_name, args.pytorch_dump_folder_path, args.push_to_hub)
| 718
|
"""simple docstring"""
from collections import OrderedDict
from typing import Mapping
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...utils import logging
lowercase__ : Any = logging.get_logger(__name__)
lowercase__ : Tuple = {
'''roberta-base''': '''https://huggingface.co/roberta-base/resolve/main/config.json''',
'''roberta-large''': '''https://huggingface.co/roberta-large/resolve/main/config.json''',
'''roberta-large-mnli''': '''https://huggingface.co/roberta-large-mnli/resolve/main/config.json''',
'''distilroberta-base''': '''https://huggingface.co/distilroberta-base/resolve/main/config.json''',
'''roberta-base-openai-detector''': '''https://huggingface.co/roberta-base-openai-detector/resolve/main/config.json''',
'''roberta-large-openai-detector''': '''https://huggingface.co/roberta-large-openai-detector/resolve/main/config.json''',
}
class _UpperCAmelCase ( lowerCAmelCase__):
_lowerCAmelCase : Dict = """roberta"""
def __init__( self : int , lowercase_ : Any=50265 , lowercase_ : Union[str, Any]=768 , lowercase_ : Optional[int]=12 , lowercase_ : List[str]=12 , lowercase_ : Dict=3072 , lowercase_ : Tuple="gelu" , lowercase_ : Optional[int]=0.1 , lowercase_ : Tuple=0.1 , lowercase_ : List[str]=512 , lowercase_ : List[Any]=2 , lowercase_ : Union[str, Any]=0.02 , lowercase_ : List[Any]=1E-12 , lowercase_ : Optional[int]=1 , lowercase_ : Tuple=0 , lowercase_ : str=2 , lowercase_ : int="absolute" , lowercase_ : str=True , lowercase_ : Tuple=None , **lowercase_ : str , ):
super().__init__(pad_token_id=lowercase_ , bos_token_id=lowercase_ , eos_token_id=lowercase_ , **lowercase_ )
snake_case_ : str = vocab_size
snake_case_ : Optional[Any] = hidden_size
snake_case_ : Any = num_hidden_layers
snake_case_ : List[str] = num_attention_heads
snake_case_ : Dict = hidden_act
snake_case_ : List[Any] = intermediate_size
snake_case_ : Optional[int] = hidden_dropout_prob
snake_case_ : List[str] = attention_probs_dropout_prob
snake_case_ : Any = max_position_embeddings
snake_case_ : Optional[Any] = type_vocab_size
snake_case_ : Optional[int] = initializer_range
snake_case_ : Dict = layer_norm_eps
snake_case_ : Optional[int] = position_embedding_type
snake_case_ : str = use_cache
snake_case_ : Tuple = classifier_dropout
class _UpperCAmelCase ( lowerCAmelCase__):
@property
def _snake_case ( self : Any ):
if self.task == "multiple-choice":
snake_case_ : List[Any] = {0: '''batch''', 1: '''choice''', 2: '''sequence'''}
else:
snake_case_ : Union[str, Any] = {0: '''batch''', 1: '''sequence'''}
return OrderedDict(
[
('''input_ids''', dynamic_axis),
('''attention_mask''', dynamic_axis),
] )
| 485
| 0
|
'''simple docstring'''
from __future__ import annotations
import unittest
from transformers import is_tf_available
from transformers.testing_utils import require_tf, slow
from ...test_configuration_common import ConfigTester
from ...test_modeling_tf_common import TFModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_tf_available():
import numpy
import tensorflow as tf
from transformers import (
TF_DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
TF_DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST,
TF_DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST,
BertConfig,
DPRConfig,
TFDPRContextEncoder,
TFDPRQuestionEncoder,
TFDPRReader,
)
class __SCREAMING_SNAKE_CASE :
def __init__( self : List[str] , UpperCAmelCase__ : List[str] , UpperCAmelCase__ : Optional[int]=13 , UpperCAmelCase__ : Any=7 , UpperCAmelCase__ : Any=True , UpperCAmelCase__ : List[Any]=True , UpperCAmelCase__ : int=True , UpperCAmelCase__ : str=True , UpperCAmelCase__ : Optional[Any]=99 , UpperCAmelCase__ : Optional[Any]=32 , UpperCAmelCase__ : Optional[Any]=2 , UpperCAmelCase__ : Any=4 , UpperCAmelCase__ : Any=37 , UpperCAmelCase__ : str="gelu" , UpperCAmelCase__ : Union[str, Any]=0.1 , UpperCAmelCase__ : List[Any]=0.1 , UpperCAmelCase__ : Any=512 , UpperCAmelCase__ : str=16 , UpperCAmelCase__ : Dict=2 , UpperCAmelCase__ : Tuple=0.02 , UpperCAmelCase__ : Optional[int]=3 , UpperCAmelCase__ : List[str]=4 , UpperCAmelCase__ : Optional[Any]=None , UpperCAmelCase__ : Tuple=0 , ):
'''simple docstring'''
lowercase : Optional[Any] =parent
lowercase : Dict =batch_size
lowercase : List[Any] =seq_length
lowercase : Optional[Any] =is_training
lowercase : Tuple =use_input_mask
lowercase : Dict =use_token_type_ids
lowercase : Any =use_labels
lowercase : List[Any] =vocab_size
lowercase : int =hidden_size
lowercase : List[Any] =num_hidden_layers
lowercase : Dict =num_attention_heads
lowercase : Optional[Any] =intermediate_size
lowercase : str =hidden_act
lowercase : Optional[Any] =hidden_dropout_prob
lowercase : Any =attention_probs_dropout_prob
lowercase : List[Any] =max_position_embeddings
lowercase : Dict =type_vocab_size
lowercase : List[str] =type_sequence_label_size
lowercase : Union[str, Any] =initializer_range
lowercase : Tuple =num_labels
lowercase : Any =num_choices
lowercase : Dict =scope
lowercase : Any =projection_dim
def lowerCamelCase_ ( self : List[Any] ):
'''simple docstring'''
lowercase : Dict =ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
lowercase : Dict =None
if self.use_input_mask:
# follow test_modeling_tf_ctrl.py
lowercase : Tuple =random_attention_mask([self.batch_size, self.seq_length] )
lowercase : List[Any] =None
if self.use_token_type_ids:
lowercase : int =ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
lowercase : Union[str, Any] =None
lowercase : Optional[Any] =None
lowercase : Optional[int] =None
if self.use_labels:
lowercase : Optional[int] =ids_tensor([self.batch_size] , self.type_sequence_label_size )
lowercase : Union[str, Any] =ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
lowercase : Any =ids_tensor([self.batch_size] , self.num_choices )
lowercase : int =BertConfig(
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 , )
lowercase : List[Any] =DPRConfig(projection_dim=self.projection_dim , **config.to_dict() )
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def lowerCamelCase_ ( self : int , UpperCAmelCase__ : int , UpperCAmelCase__ : Any , UpperCAmelCase__ : Dict , UpperCAmelCase__ : Any , UpperCAmelCase__ : Any , UpperCAmelCase__ : List[Any] , UpperCAmelCase__ : Union[str, Any] ):
'''simple docstring'''
lowercase : List[str] =TFDPRContextEncoder(config=UpperCAmelCase__ )
lowercase : List[Any] =model(UpperCAmelCase__ , attention_mask=UpperCAmelCase__ , token_type_ids=UpperCAmelCase__ )
lowercase : Optional[int] =model(UpperCAmelCase__ , token_type_ids=UpperCAmelCase__ )
lowercase : int =model(UpperCAmelCase__ )
self.parent.assertEqual(result.pooler_output.shape , (self.batch_size, self.projection_dim or self.hidden_size) )
def lowerCamelCase_ ( self : Union[str, Any] , UpperCAmelCase__ : int , UpperCAmelCase__ : int , UpperCAmelCase__ : Union[str, Any] , UpperCAmelCase__ : Optional[Any] , UpperCAmelCase__ : Any , UpperCAmelCase__ : str , UpperCAmelCase__ : Optional[int] ):
'''simple docstring'''
lowercase : str =TFDPRQuestionEncoder(config=UpperCAmelCase__ )
lowercase : Dict =model(UpperCAmelCase__ , attention_mask=UpperCAmelCase__ , token_type_ids=UpperCAmelCase__ )
lowercase : Optional[Any] =model(UpperCAmelCase__ , token_type_ids=UpperCAmelCase__ )
lowercase : Any =model(UpperCAmelCase__ )
self.parent.assertEqual(result.pooler_output.shape , (self.batch_size, self.projection_dim or self.hidden_size) )
def lowerCamelCase_ ( self : List[str] , UpperCAmelCase__ : List[str] , UpperCAmelCase__ : Optional[Any] , UpperCAmelCase__ : Optional[Any] , UpperCAmelCase__ : Optional[Any] , UpperCAmelCase__ : List[str] , UpperCAmelCase__ : Optional[int] , UpperCAmelCase__ : int ):
'''simple docstring'''
lowercase : List[Any] =TFDPRReader(config=UpperCAmelCase__ )
lowercase : Tuple =model(UpperCAmelCase__ , attention_mask=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) )
self.parent.assertEqual(result.relevance_logits.shape , (self.batch_size,) )
def lowerCamelCase_ ( self : Any ):
'''simple docstring'''
lowercase : List[Any] =self.prepare_config_and_inputs()
(
(
lowercase
) , (
lowercase
) , (
lowercase
) , (
lowercase
) , (
lowercase
) , (
lowercase
) , (
lowercase
) ,
) : Dict =config_and_inputs
lowercase : Optional[Any] ={'''input_ids''': input_ids}
return config, inputs_dict
@require_tf
class __SCREAMING_SNAKE_CASE ( lowercase__ , lowercase__ , unittest.TestCase ):
lowerCamelCase_ = (
(
TFDPRContextEncoder,
TFDPRQuestionEncoder,
TFDPRReader,
)
if is_tf_available()
else ()
)
lowerCamelCase_ = {'feature-extraction': TFDPRQuestionEncoder} if is_tf_available() else {}
lowerCamelCase_ = False
lowerCamelCase_ = False
lowerCamelCase_ = False
lowerCamelCase_ = False
lowerCamelCase_ = False
def lowerCamelCase_ ( self : List[str] ):
'''simple docstring'''
lowercase : int =TFDPRModelTester(self )
lowercase : Optional[Any] =ConfigTester(self , config_class=UpperCAmelCase__ , hidden_size=37 )
def lowerCamelCase_ ( self : List[Any] ):
'''simple docstring'''
self.config_tester.run_common_tests()
def lowerCamelCase_ ( self : List[Any] ):
'''simple docstring'''
lowercase : int =self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_dpr_context_encoder(*UpperCAmelCase__ )
def lowerCamelCase_ ( self : Optional[Any] ):
'''simple docstring'''
lowercase : str =self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_dpr_question_encoder(*UpperCAmelCase__ )
def lowerCamelCase_ ( self : List[Any] ):
'''simple docstring'''
lowercase : int =self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_dpr_reader(*UpperCAmelCase__ )
@slow
def lowerCamelCase_ ( self : Optional[Any] ):
'''simple docstring'''
for model_name in TF_DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
lowercase : Optional[int] =TFDPRContextEncoder.from_pretrained(UpperCAmelCase__ )
self.assertIsNotNone(UpperCAmelCase__ )
for model_name in TF_DPR_CONTEXT_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
lowercase : Optional[int] =TFDPRContextEncoder.from_pretrained(UpperCAmelCase__ )
self.assertIsNotNone(UpperCAmelCase__ )
for model_name in TF_DPR_QUESTION_ENCODER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
lowercase : Tuple =TFDPRQuestionEncoder.from_pretrained(UpperCAmelCase__ )
self.assertIsNotNone(UpperCAmelCase__ )
for model_name in TF_DPR_READER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
lowercase : List[str] =TFDPRReader.from_pretrained(UpperCAmelCase__ )
self.assertIsNotNone(UpperCAmelCase__ )
@require_tf
class __SCREAMING_SNAKE_CASE ( unittest.TestCase ):
@slow
def lowerCamelCase_ ( self : Any ):
'''simple docstring'''
lowercase : str =TFDPRQuestionEncoder.from_pretrained('''facebook/dpr-question_encoder-single-nq-base''' )
lowercase : Any =tf.constant(
[[101, 7592, 1010, 2003, 2026, 3899, 10140, 1029, 102]] ) # [CLS] hello, is my dog cute? [SEP]
lowercase : int =model(UpperCAmelCase__ )[0] # embedding shape = (1, 768)
# compare the actual values for a slice.
lowercase : Any =tf.constant(
[
[
0.03_23_62_53,
0.12_75_33_35,
0.16_81_85_09,
0.00_27_97_86,
0.3_89_69_33,
0.24_26_49_45,
0.2_17_89_71,
-0.02_33_52_27,
-0.08_48_19_59,
-0.14_32_41_17,
]
] )
self.assertTrue(numpy.allclose(output[:, :10].numpy() , expected_slice.numpy() , atol=1E-4 ) )
| 92
|
"""simple docstring"""
import re
import string
from collections import Counter
import sacrebleu
import sacremoses
from packaging import version
import datasets
__A : List[str] = "\n@inproceedings{xu-etal-2016-optimizing,\n title = {Optimizing Statistical Machine Translation for Text Simplification},\n authors={Xu, Wei and Napoles, Courtney and Pavlick, Ellie and Chen, Quanze and Callison-Burch, Chris},\n journal = {Transactions of the Association for Computational Linguistics},\n volume = {4},\n year={2016},\n url = {https://www.aclweb.org/anthology/Q16-1029},\n pages = {401--415\n},\n@inproceedings{post-2018-call,\n title = \"A Call for Clarity in Reporting {BLEU} Scores\",\n author = \"Post, Matt\",\n booktitle = \"Proceedings of the Third Conference on Machine Translation: Research Papers\",\n month = oct,\n year = \"2018\",\n address = \"Belgium, Brussels\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://www.aclweb.org/anthology/W18-6319\",\n pages = \"186--191\",\n}\n"
__A : List[Any] = "\\nWIKI_SPLIT is the combination of three metrics SARI, EXACT and SACREBLEU\nIt can be used to evaluate the quality of machine-generated texts.\n"
__A : int = "\nCalculates sari score (between 0 and 100) given a list of source and predicted\nsentences, and a list of lists of reference sentences. It also computes the BLEU score as well as the exact match score.\nArgs:\n sources: list of source sentences where each sentence should be a string.\n predictions: list of predicted sentences where each sentence should be a string.\n references: list of lists of reference sentences where each sentence should be a string.\nReturns:\n sari: sari score\n sacrebleu: sacrebleu score\n exact: exact score\n\nExamples:\n >>> sources=[\"About 95 species are currently accepted .\"]\n >>> predictions=[\"About 95 you now get in .\"]\n >>> references=[[\"About 95 species are currently known .\"]]\n >>> wiki_split = datasets.load_metric(\"wiki_split\")\n >>> results = wiki_split.compute(sources=sources, predictions=predictions, references=references)\n >>> print(results)\n {'sari': 21.805555555555557, 'sacrebleu': 14.535768424205482, 'exact': 0.0}\n"
def lowercase ( _SCREAMING_SNAKE_CASE : Dict ):
'''simple docstring'''
def remove_articles(_SCREAMING_SNAKE_CASE : Optional[int] ):
_UpperCAmelCase = re.compile(r'''\b(a|an|the)\b''' , re.UNICODE )
return re.sub(_SCREAMING_SNAKE_CASE , ''' ''' , _SCREAMING_SNAKE_CASE )
def white_space_fix(_SCREAMING_SNAKE_CASE : Union[str, Any] ):
return " ".join(text.split() )
def remove_punc(_SCREAMING_SNAKE_CASE : Any ):
_UpperCAmelCase = set(string.punctuation )
return "".join(ch for ch in text if ch not in exclude )
def lower(_SCREAMING_SNAKE_CASE : Tuple ):
return text.lower()
return white_space_fix(remove_articles(remove_punc(lower(_SCREAMING_SNAKE_CASE ) ) ) )
def lowercase ( _SCREAMING_SNAKE_CASE : Optional[int] , _SCREAMING_SNAKE_CASE : List[str] ):
'''simple docstring'''
return int(normalize_answer(_SCREAMING_SNAKE_CASE ) == normalize_answer(_SCREAMING_SNAKE_CASE ) )
def lowercase ( _SCREAMING_SNAKE_CASE : Union[str, Any] , _SCREAMING_SNAKE_CASE : Optional[Any] ):
'''simple docstring'''
_UpperCAmelCase = [any(compute_exact(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE ) for ref in refs ) for pred, refs in zip(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )]
return (sum(_SCREAMING_SNAKE_CASE ) / len(_SCREAMING_SNAKE_CASE )) * 100
def lowercase ( _SCREAMING_SNAKE_CASE : str , _SCREAMING_SNAKE_CASE : Dict , _SCREAMING_SNAKE_CASE : List[Any] , _SCREAMING_SNAKE_CASE : Tuple ):
'''simple docstring'''
_UpperCAmelCase = [rgram for rgrams in rgramslist for rgram in rgrams]
_UpperCAmelCase = Counter(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = Counter(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = Counter()
for sgram, scount in sgramcounter.items():
_UpperCAmelCase = scount * numref
_UpperCAmelCase = Counter(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = Counter()
for cgram, ccount in cgramcounter.items():
_UpperCAmelCase = ccount * numref
# KEEP
_UpperCAmelCase = sgramcounter_rep & cgramcounter_rep
_UpperCAmelCase = keepgramcounter_rep & rgramcounter
_UpperCAmelCase = sgramcounter_rep & rgramcounter
_UpperCAmelCase = 0
_UpperCAmelCase = 0
for keepgram in keepgramcountergood_rep:
keeptmpscorea += keepgramcountergood_rep[keepgram] / keepgramcounter_rep[keepgram]
# Fix an alleged bug [2] in the keep score computation.
# keeptmpscore2 += keepgramcountergood_rep[keepgram] / keepgramcounterall_rep[keepgram]
keeptmpscorea += keepgramcountergood_rep[keepgram]
# Define 0/0=1 instead of 0 to give higher scores for predictions that match
# a target exactly.
_UpperCAmelCase = 1
_UpperCAmelCase = 1
if len(_SCREAMING_SNAKE_CASE ) > 0:
_UpperCAmelCase = keeptmpscorea / len(_SCREAMING_SNAKE_CASE )
if len(_SCREAMING_SNAKE_CASE ) > 0:
# Fix an alleged bug [2] in the keep score computation.
# keepscore_recall = keeptmpscore2 / len(keepgramcounterall_rep)
_UpperCAmelCase = keeptmpscorea / sum(keepgramcounterall_rep.values() )
_UpperCAmelCase = 0
if keepscore_precision > 0 or keepscore_recall > 0:
_UpperCAmelCase = 2 * keepscore_precision * keepscore_recall / (keepscore_precision + keepscore_recall)
# DELETION
_UpperCAmelCase = sgramcounter_rep - cgramcounter_rep
_UpperCAmelCase = delgramcounter_rep - rgramcounter
_UpperCAmelCase = sgramcounter_rep - rgramcounter
_UpperCAmelCase = 0
_UpperCAmelCase = 0
for delgram in delgramcountergood_rep:
deltmpscorea += delgramcountergood_rep[delgram] / delgramcounter_rep[delgram]
deltmpscorea += delgramcountergood_rep[delgram] / delgramcounterall_rep[delgram]
# Define 0/0=1 instead of 0 to give higher scores for predictions that match
# a target exactly.
_UpperCAmelCase = 1
if len(_SCREAMING_SNAKE_CASE ) > 0:
_UpperCAmelCase = deltmpscorea / len(_SCREAMING_SNAKE_CASE )
# ADDITION
_UpperCAmelCase = set(_SCREAMING_SNAKE_CASE ) - set(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = set(_SCREAMING_SNAKE_CASE ) & set(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = set(_SCREAMING_SNAKE_CASE ) - set(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = 0
for addgram in addgramcountergood:
addtmpscore += 1
# Define 0/0=1 instead of 0 to give higher scores for predictions that match
# a target exactly.
_UpperCAmelCase = 1
_UpperCAmelCase = 1
if len(_SCREAMING_SNAKE_CASE ) > 0:
_UpperCAmelCase = addtmpscore / len(_SCREAMING_SNAKE_CASE )
if len(_SCREAMING_SNAKE_CASE ) > 0:
_UpperCAmelCase = addtmpscore / len(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = 0
if addscore_precision > 0 or addscore_recall > 0:
_UpperCAmelCase = 2 * addscore_precision * addscore_recall / (addscore_precision + addscore_recall)
return (keepscore, delscore_precision, addscore)
def lowercase ( _SCREAMING_SNAKE_CASE : Optional[int] , _SCREAMING_SNAKE_CASE : Union[str, Any] , _SCREAMING_SNAKE_CASE : List[Any] ):
'''simple docstring'''
_UpperCAmelCase = len(_SCREAMING_SNAKE_CASE )
_UpperCAmelCase = ssent.split(''' ''' )
_UpperCAmelCase = csent.split(''' ''' )
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
for rsent in rsents:
_UpperCAmelCase = rsent.split(''' ''' )
_UpperCAmelCase = []
_UpperCAmelCase = []
_UpperCAmelCase = []
ragramslist.append(_SCREAMING_SNAKE_CASE )
for i in range(0 , len(_SCREAMING_SNAKE_CASE ) - 1 ):
if i < len(_SCREAMING_SNAKE_CASE ) - 1:
_UpperCAmelCase = ragrams[i] + ''' ''' + ragrams[i + 1]
ragrams.append(_SCREAMING_SNAKE_CASE )
if i < len(_SCREAMING_SNAKE_CASE ) - 2:
_UpperCAmelCase = ragrams[i] + ''' ''' + ragrams[i + 1] + ''' ''' + ragrams[i + 2]
ragrams.append(_SCREAMING_SNAKE_CASE )
if i < len(_SCREAMING_SNAKE_CASE ) - 3:
_UpperCAmelCase = ragrams[i] + ''' ''' + ragrams[i + 1] + ''' ''' + ragrams[i + 2] + ''' ''' + ragrams[i + 3]
ragrams.append(_SCREAMING_SNAKE_CASE )
ragramslist.append(_SCREAMING_SNAKE_CASE )
ragramslist.append(_SCREAMING_SNAKE_CASE )
ragramslist.append(_SCREAMING_SNAKE_CASE )
for i in range(0 , len(_SCREAMING_SNAKE_CASE ) - 1 ):
if i < len(_SCREAMING_SNAKE_CASE ) - 1:
_UpperCAmelCase = sagrams[i] + ''' ''' + sagrams[i + 1]
sagrams.append(_SCREAMING_SNAKE_CASE )
if i < len(_SCREAMING_SNAKE_CASE ) - 2:
_UpperCAmelCase = sagrams[i] + ''' ''' + sagrams[i + 1] + ''' ''' + sagrams[i + 2]
sagrams.append(_SCREAMING_SNAKE_CASE )
if i < len(_SCREAMING_SNAKE_CASE ) - 3:
_UpperCAmelCase = sagrams[i] + ''' ''' + sagrams[i + 1] + ''' ''' + sagrams[i + 2] + ''' ''' + sagrams[i + 3]
sagrams.append(_SCREAMING_SNAKE_CASE )
for i in range(0 , len(_SCREAMING_SNAKE_CASE ) - 1 ):
if i < len(_SCREAMING_SNAKE_CASE ) - 1:
_UpperCAmelCase = cagrams[i] + ''' ''' + cagrams[i + 1]
cagrams.append(_SCREAMING_SNAKE_CASE )
if i < len(_SCREAMING_SNAKE_CASE ) - 2:
_UpperCAmelCase = cagrams[i] + ''' ''' + cagrams[i + 1] + ''' ''' + cagrams[i + 2]
cagrams.append(_SCREAMING_SNAKE_CASE )
if i < len(_SCREAMING_SNAKE_CASE ) - 3:
_UpperCAmelCase = cagrams[i] + ''' ''' + cagrams[i + 1] + ''' ''' + cagrams[i + 2] + ''' ''' + cagrams[i + 3]
cagrams.append(_SCREAMING_SNAKE_CASE )
((_UpperCAmelCase) , (_UpperCAmelCase) , (_UpperCAmelCase)) = SARIngram(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
((_UpperCAmelCase) , (_UpperCAmelCase) , (_UpperCAmelCase)) = SARIngram(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
((_UpperCAmelCase) , (_UpperCAmelCase) , (_UpperCAmelCase)) = SARIngram(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
((_UpperCAmelCase) , (_UpperCAmelCase) , (_UpperCAmelCase)) = SARIngram(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
_UpperCAmelCase = sum([keepascore, keepascore, keepascore, keepascore] ) / 4
_UpperCAmelCase = sum([delascore, delascore, delascore, delascore] ) / 4
_UpperCAmelCase = sum([addascore, addascore, addascore, addascore] ) / 4
_UpperCAmelCase = (avgkeepscore + avgdelscore + avgaddscore) / 3
return finalscore
def lowercase ( _SCREAMING_SNAKE_CASE : Optional[int] , _SCREAMING_SNAKE_CASE : bool = True , _SCREAMING_SNAKE_CASE : str = "13a" , _SCREAMING_SNAKE_CASE : bool = True ):
'''simple docstring'''
if lowercase:
_UpperCAmelCase = sentence.lower()
if tokenizer in ["13a", "intl"]:
if version.parse(sacrebleu.__version__ ).major >= 2:
_UpperCAmelCase = sacrebleu.metrics.bleu._get_tokenizer(_SCREAMING_SNAKE_CASE )()(_SCREAMING_SNAKE_CASE )
else:
_UpperCAmelCase = sacrebleu.TOKENIZERS[tokenizer]()(_SCREAMING_SNAKE_CASE )
elif tokenizer == "moses":
_UpperCAmelCase = sacremoses.MosesTokenizer().tokenize(_SCREAMING_SNAKE_CASE , return_str=_SCREAMING_SNAKE_CASE , escape=_SCREAMING_SNAKE_CASE )
elif tokenizer == "penn":
_UpperCAmelCase = sacremoses.MosesTokenizer().penn_tokenize(_SCREAMING_SNAKE_CASE , return_str=_SCREAMING_SNAKE_CASE )
else:
_UpperCAmelCase = sentence
if not return_str:
_UpperCAmelCase = normalized_sent.split()
return normalized_sent
def lowercase ( _SCREAMING_SNAKE_CASE : Any , _SCREAMING_SNAKE_CASE : int , _SCREAMING_SNAKE_CASE : Union[str, Any] ):
'''simple docstring'''
if not (len(_SCREAMING_SNAKE_CASE ) == len(_SCREAMING_SNAKE_CASE ) == len(_SCREAMING_SNAKE_CASE )):
raise ValueError('''Sources length must match predictions and references lengths.''' )
_UpperCAmelCase = 0
for src, pred, refs in zip(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE ):
sari_score += SARIsent(normalize(_SCREAMING_SNAKE_CASE ) , normalize(_SCREAMING_SNAKE_CASE ) , [normalize(_SCREAMING_SNAKE_CASE ) for sent in refs] )
_UpperCAmelCase = sari_score / len(_SCREAMING_SNAKE_CASE )
return 100 * sari_score
def lowercase ( _SCREAMING_SNAKE_CASE : Optional[Any] , _SCREAMING_SNAKE_CASE : int , _SCREAMING_SNAKE_CASE : str="exp" , _SCREAMING_SNAKE_CASE : Tuple=None , _SCREAMING_SNAKE_CASE : List[str]=False , _SCREAMING_SNAKE_CASE : str=False , _SCREAMING_SNAKE_CASE : int=False , ):
'''simple docstring'''
_UpperCAmelCase = len(references[0] )
if any(len(_SCREAMING_SNAKE_CASE ) != references_per_prediction for refs in references ):
raise ValueError('''Sacrebleu requires the same number of references for each prediction''' )
_UpperCAmelCase = [[refs[i] for refs in references] for i in range(_SCREAMING_SNAKE_CASE )]
_UpperCAmelCase = sacrebleu.corpus_bleu(
_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , smooth_method=_SCREAMING_SNAKE_CASE , smooth_value=_SCREAMING_SNAKE_CASE , force=_SCREAMING_SNAKE_CASE , lowercase=_SCREAMING_SNAKE_CASE , use_effective_order=_SCREAMING_SNAKE_CASE , )
return output.score
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION)
class _a ( datasets.Metric):
"""simple docstring"""
def lowercase__ ( self : Dict )->Any:
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(
{
'''predictions''': datasets.Value('''string''' , id='''sequence''' ),
'''references''': datasets.Sequence(datasets.Value('''string''' , id='''sequence''' ) , id='''references''' ),
} ) , codebase_urls=[
'''https://github.com/huggingface/transformers/blob/master/src/transformers/data/metrics/squad_metrics.py''',
'''https://github.com/cocoxu/simplification/blob/master/SARI.py''',
'''https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/utils/sari_hook.py''',
'''https://github.com/mjpost/sacreBLEU''',
] , reference_urls=[
'''https://www.aclweb.org/anthology/Q16-1029.pdf''',
'''https://github.com/mjpost/sacreBLEU''',
'''https://en.wikipedia.org/wiki/BLEU''',
'''https://towardsdatascience.com/evaluating-text-output-in-nlp-bleu-at-your-own-risk-e8609665a213''',
] , )
def lowercase__ ( self : Dict , __UpperCamelCase : List[str] , __UpperCamelCase : Dict , __UpperCamelCase : List[Any] )->Any:
_UpperCAmelCase = {}
result.update({'''sari''': compute_sari(sources=__UpperCamelCase , predictions=__UpperCamelCase , references=__UpperCamelCase )} )
result.update({'''sacrebleu''': compute_sacrebleu(predictions=__UpperCamelCase , references=__UpperCamelCase )} )
result.update({'''exact''': compute_em(predictions=__UpperCamelCase , references=__UpperCamelCase )} )
return result
| 602
| 0
|
UpperCAmelCase = """
# Transformers 설치 방법
! pip install transformers datasets
# 마지막 릴리스 대신 소스에서 설치하려면, 위 명령을 주석으로 바꾸고 아래 명령을 해제하세요.
# ! pip install git+https://github.com/huggingface/transformers.git
"""
UpperCAmelCase = [{"""type""": """code""", """content""": INSTALL_CONTENT}]
UpperCAmelCase = {
"""{processor_class}""": """FakeProcessorClass""",
"""{model_class}""": """FakeModelClass""",
"""{object_class}""": """FakeObjectClass""",
}
| 531
|
# Lint as: python3
import sys
from collections.abc import Mapping
from typing import TYPE_CHECKING, Dict, Optional
import numpy as np
import pyarrow as pa
from .. import config
from ..utils.logging import get_logger
from ..utils.py_utils import map_nested
from .formatting import TensorFormatter
if TYPE_CHECKING:
import jax
import jaxlib
UpperCAmelCase = get_logger()
UpperCAmelCase = None
class lowerCAmelCase_ ( TensorFormatter[Mapping, "jax.Array", Mapping] ):
'''simple docstring'''
def __init__( self , _UpperCAmelCase=None , _UpperCAmelCase=None , **_UpperCAmelCase ):
super().__init__(features=_UpperCAmelCase )
import jax
from jaxlib.xla_client import Device
if isinstance(_UpperCAmelCase , _UpperCAmelCase ):
raise ValueError(
F'''Expected {device} to be a `str` not {type(_UpperCAmelCase )}, as `jaxlib.xla_extension.Device` '''
'''is not serializable neither with `pickle` nor with `dill`. Instead you can surround '''
'''the device with `str()` to get its string identifier that will be internally mapped '''
'''to the actual `jaxlib.xla_extension.Device`.''' )
snake_case_ = device if isinstance(_UpperCAmelCase , _UpperCAmelCase ) else str(jax.devices()[0] )
# using global variable since `jaxlib.xla_extension.Device` is not serializable neither
# with `pickle` nor with `dill`, so we need to use a global variable instead
global DEVICE_MAPPING
if DEVICE_MAPPING is None:
snake_case_ = self._map_devices_to_str()
if self.device not in list(DEVICE_MAPPING.keys() ):
logger.warning(
F'''Device with string identifier {self.device} not listed among the available '''
F'''devices: {list(DEVICE_MAPPING.keys() )}, so falling back to the default '''
F'''device: {str(jax.devices()[0] )}.''' )
snake_case_ = str(jax.devices()[0] )
snake_case_ = jnp_array_kwargs
@staticmethod
def UpperCamelCase__ ( ):
import jax
return {str(_UpperCAmelCase ): device for device in jax.devices()}
def UpperCamelCase__ ( self , _UpperCAmelCase ):
import jax
import jax.numpy as jnp
if isinstance(_UpperCAmelCase , _UpperCAmelCase ) and column:
if all(
isinstance(_UpperCAmelCase , jax.Array ) and x.shape == column[0].shape and x.dtype == column[0].dtype for x in column ):
return jnp.stack(_UpperCAmelCase , axis=0 )
return column
def UpperCamelCase__ ( self , _UpperCAmelCase ):
import jax
import jax.numpy as jnp
if isinstance(_UpperCAmelCase , (str, bytes, type(_UpperCAmelCase )) ):
return value
elif isinstance(_UpperCAmelCase , (np.character, np.ndarray) ) and np.issubdtype(value.dtype , np.character ):
return value.tolist()
snake_case_ = {}
if isinstance(_UpperCAmelCase , (np.number, np.ndarray) ) and np.issubdtype(value.dtype , np.integer ):
# the default int precision depends on the jax config
# see https://jax.readthedocs.io/en/latest/notebooks/Common_Gotchas_in_JAX.html#double-64bit-precision
if jax.config.jax_enable_xaa:
snake_case_ = {'''dtype''': jnp.intaa}
else:
snake_case_ = {'''dtype''': jnp.intaa}
elif isinstance(_UpperCAmelCase , (np.number, np.ndarray) ) and np.issubdtype(value.dtype , np.floating ):
snake_case_ = {'''dtype''': jnp.floataa}
elif config.PIL_AVAILABLE and "PIL" in sys.modules:
import PIL.Image
if isinstance(_UpperCAmelCase , PIL.Image.Image ):
snake_case_ = np.asarray(_UpperCAmelCase )
# using global variable since `jaxlib.xla_extension.Device` is not serializable neither
# with `pickle` nor with `dill`, so we need to use a global variable instead
global DEVICE_MAPPING
if DEVICE_MAPPING is None:
snake_case_ = self._map_devices_to_str()
with jax.default_device(DEVICE_MAPPING[self.device] ):
# calling jnp.array on a np.ndarray does copy the data
# see https://github.com/google/jax/issues/4486
return jnp.array(_UpperCAmelCase , **{**default_dtype, **self.jnp_array_kwargs} )
def UpperCamelCase__ ( self , _UpperCAmelCase ):
import jax
# support for torch, tf, jax etc.
if config.TORCH_AVAILABLE and "torch" in sys.modules:
import torch
if isinstance(_UpperCAmelCase , torch.Tensor ):
return self._tensorize(data_struct.detach().cpu().numpy()[()] )
if hasattr(_UpperCAmelCase , '''__array__''' ) and not isinstance(_UpperCAmelCase , jax.Array ):
snake_case_ = data_struct.__array__()
# support for nested types like struct of list of struct
if isinstance(_UpperCAmelCase , np.ndarray ):
if data_struct.dtype == object: # jax arrays cannot be instantied from an array of objects
return self._consolidate([self.recursive_tensorize(_UpperCAmelCase ) for substruct in data_struct] )
elif isinstance(_UpperCAmelCase , (list, tuple) ):
return self._consolidate([self.recursive_tensorize(_UpperCAmelCase ) for substruct in data_struct] )
return self._tensorize(_UpperCAmelCase )
def UpperCamelCase__ ( self , _UpperCAmelCase ):
return map_nested(self._recursive_tensorize , _UpperCAmelCase , map_list=_UpperCAmelCase )
def UpperCamelCase__ ( self , _UpperCAmelCase ):
snake_case_ = self.numpy_arrow_extractor().extract_row(_UpperCAmelCase )
snake_case_ = self.python_features_decoder.decode_row(_UpperCAmelCase )
return self.recursive_tensorize(_UpperCAmelCase )
def UpperCamelCase__ ( self , _UpperCAmelCase ):
snake_case_ = self.numpy_arrow_extractor().extract_column(_UpperCAmelCase )
snake_case_ = self.python_features_decoder.decode_column(_UpperCAmelCase , pa_table.column_names[0] )
snake_case_ = self.recursive_tensorize(_UpperCAmelCase )
snake_case_ = self._consolidate(_UpperCAmelCase )
return column
def UpperCamelCase__ ( self , _UpperCAmelCase ):
snake_case_ = self.numpy_arrow_extractor().extract_batch(_UpperCAmelCase )
snake_case_ = self.python_features_decoder.decode_batch(_UpperCAmelCase )
snake_case_ = self.recursive_tensorize(_UpperCAmelCase )
for column_name in batch:
snake_case_ = self._consolidate(batch[column_name] )
return batch
| 531
| 1
|
'''simple docstring'''
import logging
from dataclasses import dataclass, field
from pathlib import Path
from typing import Optional, Union
from .generation.configuration_utils import GenerationConfig
from .training_args import TrainingArguments
from .utils import add_start_docstrings
A_ = logging.getLogger(__name__)
@dataclass
@add_start_docstrings(TrainingArguments.__doc__ )
class lowercase_ ( lowerCAmelCase_ ):
A_ = field(default=lowerCAmelCase_ , metadata={"help": "Whether to use SortishSampler or not."} )
A_ = field(
default=lowerCAmelCase_ , metadata={"help": "Whether to use generate to calculate generative metrics (ROUGE, BLEU)."} )
A_ = field(
default=lowerCAmelCase_ , metadata={
"help": (
"The `max_length` to use on each evaluation loop when `predict_with_generate=True`. Will default "
"to the `max_length` value of the model configuration."
)
} , )
A_ = field(
default=lowerCAmelCase_ , metadata={
"help": (
"The `num_beams` to use on each evaluation loop when `predict_with_generate=True`. Will default "
"to the `num_beams` value of the model configuration."
)
} , )
A_ = field(
default=lowerCAmelCase_ , metadata={
"help": "Model id, file path or url pointing to a GenerationConfig json file, to use during prediction."
} , )
def _lowerCAmelCase ( self : Optional[Any] ):
snake_case__ : str = super().to_dict()
for k, v in d.items():
if isinstance(__lowerCamelCase , __lowerCamelCase ):
snake_case__ : Optional[Any] = v.to_dict()
return d
| 270
|
'''simple docstring'''
import os
import re
import shutil
import sys
import tempfile
import unittest
import black
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_copies # noqa: E402
# This is the reference code that will be used in the tests.
# If BertLMPredictionHead is changed in modeling_bert.py, this code needs to be manually updated.
A_ = " def __init__(self, config):\n super().__init__()\n self.transform = BertPredictionHeadTransform(config)\n\n # The output weights are the same as the input embeddings, but there is\n # an output-only bias for each token.\n self.decoder = nn.Linear(config.hidden_size, config.vocab_size, bias=False)\n\n self.bias = nn.Parameter(torch.zeros(config.vocab_size))\n\n # Need a link between the two variables so that the bias is correctly resized with `resize_token_embeddings`\n self.decoder.bias = self.bias\n\n def forward(self, hidden_states):\n hidden_states = self.transform(hidden_states)\n hidden_states = self.decoder(hidden_states)\n return hidden_states\n"
class lowercase_ ( unittest.TestCase ):
def _lowerCAmelCase ( self : Union[str, Any] ):
snake_case__ : str = tempfile.mkdtemp()
os.makedirs(os.path.join(self.transformer_dir , 'models/bert/' ) )
snake_case__ : Optional[Any] = self.transformer_dir
shutil.copy(
os.path.join(__lowerCamelCase , 'src/transformers/models/bert/modeling_bert.py' ) , os.path.join(self.transformer_dir , 'models/bert/modeling_bert.py' ) , )
def _lowerCAmelCase ( self : int ):
snake_case__ : Union[str, Any] = 'src/transformers'
shutil.rmtree(self.transformer_dir )
def _lowerCAmelCase ( self : Tuple , __lowerCamelCase : Tuple , __lowerCamelCase : Dict , __lowerCamelCase : Tuple , __lowerCamelCase : List[Any]=None ):
snake_case__ : Optional[int] = comment + F"\nclass {class_name}(nn.Module):\n" + class_code
if overwrite_result is not None:
snake_case__ : List[Any] = comment + F"\nclass {class_name}(nn.Module):\n" + overwrite_result
snake_case__ : Optional[int] = black.Mode(target_versions={black.TargetVersion.PYaa} , line_length=119 )
snake_case__ : Tuple = black.format_str(__lowerCamelCase , mode=__lowerCamelCase )
snake_case__ : Union[str, Any] = os.path.join(self.transformer_dir , 'new_code.py' )
with open(__lowerCamelCase , 'w' , newline='\n' ) as f:
f.write(__lowerCamelCase )
if overwrite_result is None:
self.assertTrue(len(check_copies.is_copy_consistent(__lowerCamelCase ) ) == 0 )
else:
check_copies.is_copy_consistent(f.name , overwrite=__lowerCamelCase )
with open(__lowerCamelCase , 'r' ) as f:
self.assertTrue(f.read() , __lowerCamelCase )
def _lowerCAmelCase ( self : Tuple ):
snake_case__ : List[str] = check_copies.find_code_in_transformers('models.bert.modeling_bert.BertLMPredictionHead' )
self.assertEqual(__lowerCamelCase , __lowerCamelCase )
def _lowerCAmelCase ( self : Any ):
# Base copy consistency
self.check_copy_consistency(
'# Copied from transformers.models.bert.modeling_bert.BertLMPredictionHead' , 'BertLMPredictionHead' , REFERENCE_CODE + '\n' , )
# With no empty line at the end
self.check_copy_consistency(
'# Copied from transformers.models.bert.modeling_bert.BertLMPredictionHead' , 'BertLMPredictionHead' , __lowerCamelCase , )
# Copy consistency with rename
self.check_copy_consistency(
'# Copied from transformers.models.bert.modeling_bert.BertLMPredictionHead with Bert->TestModel' , 'TestModelLMPredictionHead' , re.sub('Bert' , 'TestModel' , __lowerCamelCase ) , )
# Copy consistency with a really long name
snake_case__ : Union[str, Any] = 'TestModelWithAReallyLongNameBecauseSomePeopleLikeThatForSomeReason'
self.check_copy_consistency(
F"# Copied from transformers.models.bert.modeling_bert.BertLMPredictionHead with Bert->{long_class_name}" , F"{long_class_name}LMPredictionHead" , re.sub('Bert' , __lowerCamelCase , __lowerCamelCase ) , )
# Copy consistency with overwrite
self.check_copy_consistency(
'# Copied from transformers.models.bert.modeling_bert.BertLMPredictionHead with Bert->TestModel' , 'TestModelLMPredictionHead' , __lowerCamelCase , overwrite_result=re.sub('Bert' , 'TestModel' , __lowerCamelCase ) , )
def _lowerCAmelCase ( self : Union[str, Any] ):
snake_case__ : List[str] = check_copies.LOCALIZED_READMES['README_zh-hans.md']
snake_case__ : List[str] = (
'1. **[ALBERT](https://huggingface.co/transformers/model_doc/albert.html)** (from Google Research and the'
' Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for'
' Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong'
' Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.\n1.'
' **[DistilBERT](https://huggingface.co/transformers/model_doc/distilbert.html)** (from HuggingFace),'
' released together with the paper [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and'
' lighter](https://arxiv.org/abs/1910.01108) by Victor Sanh, Lysandre Debut and Thomas Wolf. The same'
' method has been applied to compress GPT2 into'
' [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa into'
' [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation),'
' Multilingual BERT into'
' [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) and a German'
' version of DistilBERT.\n1. **[ELECTRA](https://huggingface.co/transformers/model_doc/electra.html)**'
' (from Google Research/Stanford University) released with the paper [ELECTRA: Pre-training text encoders'
' as discriminators rather than generators](https://arxiv.org/abs/2003.10555) by Kevin Clark, Minh-Thang'
' Luong, Quoc V. Le, Christopher D. Manning.'
)
snake_case__ : Union[str, Any] = (
'1. **[ALBERT](https://huggingface.co/transformers/model_doc/albert.html)** (来自 Google Research and the'
' Toyota Technological Institute at Chicago) 伴随论文 [ALBERT: A Lite BERT for Self-supervised Learning of'
' Language Representations](https://arxiv.org/abs/1909.11942), 由 Zhenzhong Lan, Mingda Chen, Sebastian'
' Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut 发布。\n'
)
snake_case__ : Tuple = (
'1. **[ALBERT](https://huggingface.co/transformers/model_doc/albert.html)** (来自 Google Research and the'
' Toyota Technological Institute at Chicago) 伴随论文 [ALBERT: A Lite BERT for Self-supervised Learning of'
' Language Representations](https://arxiv.org/abs/1909.11942), 由 Zhenzhong Lan, Mingda Chen, Sebastian'
' Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut 发布。\n1.'
' **[DistilBERT](https://huggingface.co/transformers/model_doc/distilbert.html)** (来自 HuggingFace) 伴随论文'
' [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and'
' lighter](https://arxiv.org/abs/1910.01108) 由 Victor Sanh, Lysandre Debut and Thomas Wolf 发布。 The same'
' method has been applied to compress GPT2 into'
' [DistilGPT2](https://github.com/huggingface/transformers/tree/main/examples/distillation), RoBERTa into'
' [DistilRoBERTa](https://github.com/huggingface/transformers/tree/main/examples/distillation),'
' Multilingual BERT into'
' [DistilmBERT](https://github.com/huggingface/transformers/tree/main/examples/distillation) and a German'
' version of DistilBERT.\n1. **[ELECTRA](https://huggingface.co/transformers/model_doc/electra.html)** (来自'
' Google Research/Stanford University) 伴随论文 [ELECTRA: Pre-training text encoders as discriminators rather'
' than generators](https://arxiv.org/abs/2003.10555) 由 Kevin Clark, Minh-Thang Luong, Quoc V. Le,'
' Christopher D. Manning 发布。\n'
)
snake_case__ , snake_case__ : Optional[Any] = check_copies.convert_to_localized_md(
__lowerCamelCase , __lowerCamelCase , localized_readme['format_model_list'] )
self.assertFalse(__lowerCamelCase )
self.assertEqual(__lowerCamelCase , __lowerCamelCase )
snake_case__ , snake_case__ : Any = check_copies.convert_to_localized_md(
__lowerCamelCase , __lowerCamelCase , localized_readme['format_model_list'] )
# Check whether the number of models is equal to README.md after conversion.
self.assertTrue(__lowerCamelCase )
snake_case__ : Optional[Any] = (
'1. **[ALBERT](https://huggingface.co/transformers/model_doc/albert.html)** (from Google Research and the'
' Toyota Technological Institute at Chicago) released with the paper [ALBERT: A Lite BERT for'
' Self-supervised Learning of Language Representations](https://arxiv.org/abs/1909.11942), by Zhenzhong'
' Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut.'
)
snake_case__ : int = (
'1. **[ALBERT](https://huggingface.co/transformers/main/model_doc/albert.html)** (来自 Google Research and'
' the Toyota Technological Institute at Chicago) 伴随论文 [ALBERT: A Lite BERT for Self-supervised Learning of'
' Language Representations](https://arxiv.org/abs/1909.11942), 由 Zhenzhong Lan, Mingda Chen, Sebastian'
' Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut 发布。\n'
)
snake_case__ : List[str] = (
'1. **[ALBERT](https://huggingface.co/transformers/model_doc/albert.html)** (来自 Google Research and the'
' Toyota Technological Institute at Chicago) 伴随论文 [ALBERT: A Lite BERT for Self-supervised Learning of'
' Language Representations](https://arxiv.org/abs/1909.11942), 由 Zhenzhong Lan, Mingda Chen, Sebastian'
' Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut 发布。\n'
)
snake_case__ , snake_case__ : Any = check_copies.convert_to_localized_md(
__lowerCamelCase , __lowerCamelCase , localized_readme['format_model_list'] )
# Check if the model link is synchronized.
self.assertEqual(__lowerCamelCase , __lowerCamelCase )
| 270
| 1
|
'''simple docstring'''
import secrets
from random import shuffle
from string import ascii_letters, ascii_lowercase, ascii_uppercase, digits, punctuation
def __lowerCAmelCase ( snake_case__ = 8 ):
__UpperCamelCase : Optional[Any] = ascii_letters + digits + punctuation
return "".join(secrets.choice(snake_case__ ) for _ in range(snake_case__ ) )
def __lowerCAmelCase ( snake_case__ , snake_case__ ):
# Password Generator = full boot with random_number, random_letters, and
# random_character FUNCTIONS
# Put your code here...
i -= len(snake_case__ )
__UpperCamelCase : List[Any] = i // 3
__UpperCamelCase : Optional[int] = i % 3
# chars = chars_incl + random_letters(ascii_letters, i / 3 + remainder) +
# random_number(digits, i / 3) + random_characters(punctuation, i / 3)
__UpperCamelCase : Tuple = (
chars_incl
+ random(snake_case__ , quotient + remainder )
+ random(snake_case__ , snake_case__ )
+ random(snake_case__ , snake_case__ )
)
__UpperCamelCase : List[str] = list(snake_case__ )
shuffle(snake_case__ )
return "".join(snake_case__ )
# random is a generalised function for letters, characters and numbers
def __lowerCAmelCase ( snake_case__ , snake_case__ ):
return "".join(secrets.choice(snake_case__ ) for _ in range(snake_case__ ) )
def __lowerCAmelCase ( snake_case__ , snake_case__ ):
pass # Put your code here...
def __lowerCAmelCase ( snake_case__ , snake_case__ ):
pass # Put your code here...
def __lowerCAmelCase ( snake_case__ , snake_case__ ):
pass # Put your code here...
def __lowerCAmelCase ( snake_case__ , snake_case__ = 8 ):
if len(snake_case__ ) < min_length:
# Your Password must be at least 8 characters long
return False
__UpperCamelCase : Optional[Any] = any(char in ascii_uppercase for char in password )
__UpperCamelCase : Tuple = any(char in ascii_lowercase for char in password )
__UpperCamelCase : Optional[int] = any(char in digits for char in password )
__UpperCamelCase : Dict = any(char in punctuation for char in password )
return upper and lower and num and spec_char
# Passwords should contain UPPERCASE, lowerase
# numbers, and special characters
def __lowerCAmelCase ( ):
__UpperCamelCase : Dict = int(input("Please indicate the max length of your password: " ).strip() )
__UpperCamelCase : Any = input(
"Please indicate the characters that must be in your password: " ).strip()
print("Password generated:" , password_generator(snake_case__ ) )
print(
"Alternative Password generated:" , alternative_password_generator(snake_case__ , snake_case__ ) , )
print("[If you are thinking of using this passsword, You better save it.]" )
if __name__ == "__main__":
main()
| 399
|
'''simple docstring'''
import json
import logging
import os
import sys
from time import time
from unittest.mock import patch
from transformers.testing_utils import TestCasePlus, require_torch_tpu
logging.basicConfig(level=logging.DEBUG)
_lowerCAmelCase = logging.getLogger()
def __lowerCAmelCase ( snake_case__ ):
__UpperCamelCase : int = {}
__UpperCamelCase : int = os.path.join(snake_case__ , "all_results.json" )
if os.path.exists(snake_case__ ):
with open(snake_case__ , "r" ) as f:
__UpperCamelCase : Union[str, Any] = json.load(snake_case__ )
else:
raise ValueError(F"can't find {path}" )
return results
_lowerCAmelCase = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
@require_torch_tpu
class A ( SCREAMING_SNAKE_CASE__ ):
'''simple docstring'''
def a_ (self ) -> Optional[Any]:
import xla_spawn
__UpperCamelCase : Optional[int] = self.get_auto_remove_tmp_dir()
__UpperCamelCase : List[str] = f"\n ./examples/pytorch/text-classification/run_glue.py\n --num_cores=8\n ./examples/pytorch/text-classification/run_glue.py\n --model_name_or_path distilbert-base-uncased\n --output_dir {tmp_dir}\n --overwrite_output_dir\n --train_file ./tests/fixtures/tests_samples/MRPC/train.csv\n --validation_file ./tests/fixtures/tests_samples/MRPC/dev.csv\n --do_train\n --do_eval\n --debug tpu_metrics_debug\n --per_device_train_batch_size=2\n --per_device_eval_batch_size=1\n --learning_rate=1e-4\n --max_steps=10\n --warmup_steps=2\n --seed=42\n --max_seq_length=128\n ".split()
with patch.object(_UpperCAmelCase , "argv" , _UpperCAmelCase ):
__UpperCamelCase : List[Any] = time()
xla_spawn.main()
__UpperCamelCase : Union[str, Any] = time()
__UpperCamelCase : Tuple = get_results(_UpperCAmelCase )
self.assertGreaterEqual(result["eval_accuracy"] , 0.75 )
# Assert that the script takes less than 500 seconds to make sure it doesn't hang.
self.assertLess(end - start , 5_0_0 )
def a_ (self ) -> Optional[Any]:
import xla_spawn
__UpperCamelCase : str = "\n ./tests/test_trainer_tpu.py\n --num_cores=8\n ./tests/test_trainer_tpu.py\n ".split()
with patch.object(_UpperCAmelCase , "argv" , _UpperCAmelCase ):
xla_spawn.main()
| 399
| 1
|
from random import randint
from tempfile import TemporaryFile
import numpy as np
def lowercase__( A , A , A ):
snake_case__ : Any = 0
if start < end:
snake_case__ : int = randint(A , A )
snake_case__ : str = a[end]
snake_case__ : Tuple = a[pivot]
snake_case__ : List[str] = temp
snake_case__ : Tuple = _in_place_partition(A , A , A )
count += _in_place_quick_sort(A , A , p - 1 )
count += _in_place_quick_sort(A , p + 1 , A )
return count
def lowercase__( A , A , A ):
snake_case__ : Optional[int] = 0
snake_case__ : List[Any] = randint(A , A )
snake_case__ : Optional[Any] = a[end]
snake_case__ : Dict = a[pivot]
snake_case__ : str = temp
snake_case__ : Optional[Any] = start - 1
for index in range(A , A ):
count += 1
if a[index] < a[end]: # check if current val is less than pivot value
snake_case__ : str = new_pivot_index + 1
snake_case__ : List[str] = a[new_pivot_index]
snake_case__ : Union[str, Any] = a[index]
snake_case__ : Tuple = temp
snake_case__ : List[str] = a[new_pivot_index + 1]
snake_case__ : Any = a[end]
snake_case__ : int = temp
return new_pivot_index + 1, count
lowerCamelCase : str = TemporaryFile()
lowerCamelCase : Any = 1_0_0 # 1000 elements are to be sorted
lowerCamelCase : List[str] = 0, 1 # mean and standard deviation
lowerCamelCase : int = np.random.normal(mu, sigma, p)
np.save(outfile, X)
print('The array is')
print(X)
outfile.seek(0) # using the same array
lowerCamelCase : List[Any] = np.load(outfile)
lowerCamelCase : List[Any] = len(M) - 1
lowerCamelCase : Dict = _in_place_quick_sort(M, 0, r)
print(
'No of Comparisons for 100 elements selected from a standard normal distribution'
'is :'
)
print(z)
| 170
|
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a : Any = logging.get_logger(__name__)
a : List[str] = {
"""transfo-xl-wt103""": """https://huggingface.co/transfo-xl-wt103/resolve/main/config.json""",
}
class _lowercase ( UpperCAmelCase__ ):
'''simple docstring'''
SCREAMING_SNAKE_CASE: Optional[Any] = 'transfo-xl'
SCREAMING_SNAKE_CASE: List[Any] = ['mems']
SCREAMING_SNAKE_CASE: List[Any] = {
'n_token': 'vocab_size',
'hidden_size': 'd_model',
'num_attention_heads': 'n_head',
'num_hidden_layers': 'n_layer',
}
def __init__( self , lowerCamelCase__=267_735 , lowerCamelCase__=[20_000, 40_000, 200_000] , lowerCamelCase__=1_024 , lowerCamelCase__=1_024 , lowerCamelCase__=16 , lowerCamelCase__=64 , lowerCamelCase__=4_096 , lowerCamelCase__=4 , lowerCamelCase__=False , lowerCamelCase__=18 , lowerCamelCase__=1_600 , lowerCamelCase__=1_000 , lowerCamelCase__=True , lowerCamelCase__=True , lowerCamelCase__=0 , lowerCamelCase__=-1 , lowerCamelCase__=True , lowerCamelCase__=0.1 , lowerCamelCase__=0.0 , lowerCamelCase__=True , lowerCamelCase__="normal" , lowerCamelCase__=0.0_1 , lowerCamelCase__=0.0_1 , lowerCamelCase__=0.0_2 , lowerCamelCase__=1E-5 , lowerCamelCase__=0 , **lowerCamelCase__ , ):
lowerCAmelCase_: Dict = vocab_size
lowerCAmelCase_: Any = []
self.cutoffs.extend(lowerCamelCase__ )
if proj_share_all_but_first:
lowerCAmelCase_: int = [False] + [True] * len(self.cutoffs )
else:
lowerCAmelCase_: Any = [False] + [False] * len(self.cutoffs )
lowerCAmelCase_: Dict = d_model
lowerCAmelCase_: Union[str, Any] = d_embed
lowerCAmelCase_: Dict = d_head
lowerCAmelCase_: Dict = d_inner
lowerCAmelCase_: List[str] = div_val
lowerCAmelCase_: List[str] = pre_lnorm
lowerCAmelCase_: List[str] = n_layer
lowerCAmelCase_: List[str] = n_head
lowerCAmelCase_: str = mem_len
lowerCAmelCase_: Any = same_length
lowerCAmelCase_: Optional[Any] = attn_type
lowerCAmelCase_: int = clamp_len
lowerCAmelCase_: Optional[int] = sample_softmax
lowerCAmelCase_: Optional[Any] = adaptive
lowerCAmelCase_: Optional[int] = dropout
lowerCAmelCase_: Union[str, Any] = dropatt
lowerCAmelCase_: Tuple = untie_r
lowerCAmelCase_: Union[str, Any] = init
lowerCAmelCase_: Optional[Any] = init_range
lowerCAmelCase_: Optional[Any] = proj_init_std
lowerCAmelCase_: Tuple = init_std
lowerCAmelCase_: Tuple = layer_norm_epsilon
super().__init__(eos_token_id=lowerCamelCase__ , **lowerCamelCase__ )
@property
def _a ( self ):
# Message copied from Transformer-XL documentation
logger.info(F'''The model {self.model_type} is one of the few models that has no sequence length limit.''' )
return -1
@max_position_embeddings.setter
def _a ( self , lowerCamelCase__ ):
# Message copied from Transformer-XL documentation
raise NotImplementedError(
F'''The model {self.model_type} is one of the few models that has no sequence length limit.''' )
| 613
| 0
|
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a__ : Optional[Any] = logging.get_logger(__name__)
a__ : List[str] = {
"naver-clova-ix/donut-base": "https://huggingface.co/naver-clova-ix/donut-base/resolve/main/config.json",
# See all Donut models at https://huggingface.co/models?filter=donut-swin
}
class UpperCAmelCase__( _UpperCAmelCase ):
'''simple docstring'''
A : Tuple = "donut-swin"
A : Optional[Any] = {
"num_attention_heads": "num_heads",
"num_hidden_layers": "num_layers",
}
def __init__( self : Union[str, Any] , lowerCAmelCase : int=2_24 , lowerCAmelCase : Dict=4 , lowerCAmelCase : Any=3 , lowerCAmelCase : Any=96 , lowerCAmelCase : Any=[2, 2, 6, 2] , lowerCAmelCase : str=[3, 6, 12, 24] , lowerCAmelCase : Optional[int]=7 , lowerCAmelCase : str=4.0 , lowerCAmelCase : Dict=True , lowerCAmelCase : int=0.0 , lowerCAmelCase : Any=0.0 , lowerCAmelCase : int=0.1 , lowerCAmelCase : Tuple="gelu" , lowerCAmelCase : Optional[Any]=False , lowerCAmelCase : Union[str, Any]=0.02 , lowerCAmelCase : int=1E-5 , **lowerCAmelCase : Any , ) -> Tuple:
"""simple docstring"""
super().__init__(**lowerCamelCase_)
lowercase__ = image_size
lowercase__ = patch_size
lowercase__ = num_channels
lowercase__ = embed_dim
lowercase__ = depths
lowercase__ = len(lowerCamelCase_)
lowercase__ = num_heads
lowercase__ = window_size
lowercase__ = mlp_ratio
lowercase__ = qkv_bias
lowercase__ = hidden_dropout_prob
lowercase__ = attention_probs_dropout_prob
lowercase__ = drop_path_rate
lowercase__ = hidden_act
lowercase__ = use_absolute_embeddings
lowercase__ = layer_norm_eps
lowercase__ = initializer_range
# we set the hidden_size attribute in order to make Swin work with VisionEncoderDecoderModel
# this indicates the channel dimension after the last stage of the model
lowercase__ = int(embed_dim * 2 ** (len(lowerCamelCase_) - 1))
| 718
|
import json
from typing import List, Optional, Tuple
from tokenizers import pre_tokenizers, processors
from ...tokenization_utils_base import AddedToken, BatchEncoding
from ...tokenization_utils_fast import PreTrainedTokenizerFast
from ...utils import logging
from .tokenization_bart import BartTokenizer
a__ : List[Any] = logging.get_logger(__name__)
a__ : Optional[int] = {"vocab_file": "vocab.json", "merges_file": "merges.txt", "tokenizer_file": "tokenizer.json"}
# See all BART models at https://huggingface.co/models?filter=bart
a__ : List[Any] = {
"vocab_file": {
"facebook/bart-base": "https://huggingface.co/facebook/bart-base/resolve/main/vocab.json",
"facebook/bart-large": "https://huggingface.co/facebook/bart-large/resolve/main/vocab.json",
"facebook/bart-large-mnli": "https://huggingface.co/facebook/bart-large-mnli/resolve/main/vocab.json",
"facebook/bart-large-cnn": "https://huggingface.co/facebook/bart-large-cnn/resolve/main/vocab.json",
"facebook/bart-large-xsum": "https://huggingface.co/facebook/bart-large-xsum/resolve/main/vocab.json",
"yjernite/bart_eli5": "https://huggingface.co/yjernite/bart_eli5/resolve/main/vocab.json",
},
"merges_file": {
"facebook/bart-base": "https://huggingface.co/facebook/bart-base/resolve/main/merges.txt",
"facebook/bart-large": "https://huggingface.co/facebook/bart-large/resolve/main/merges.txt",
"facebook/bart-large-mnli": "https://huggingface.co/facebook/bart-large-mnli/resolve/main/merges.txt",
"facebook/bart-large-cnn": "https://huggingface.co/facebook/bart-large-cnn/resolve/main/merges.txt",
"facebook/bart-large-xsum": "https://huggingface.co/facebook/bart-large-xsum/resolve/main/merges.txt",
"yjernite/bart_eli5": "https://huggingface.co/yjernite/bart_eli5/resolve/main/merges.txt",
},
"tokenizer_file": {
"facebook/bart-base": "https://huggingface.co/facebook/bart-base/resolve/main/tokenizer.json",
"facebook/bart-large": "https://huggingface.co/facebook/bart-large/resolve/main/tokenizer.json",
"facebook/bart-large-mnli": "https://huggingface.co/facebook/bart-large-mnli/resolve/main/tokenizer.json",
"facebook/bart-large-cnn": "https://huggingface.co/facebook/bart-large-cnn/resolve/main/tokenizer.json",
"facebook/bart-large-xsum": "https://huggingface.co/facebook/bart-large-xsum/resolve/main/tokenizer.json",
"yjernite/bart_eli5": "https://huggingface.co/yjernite/bart_eli5/resolve/main/tokenizer.json",
},
}
a__ : int = {
"facebook/bart-base": 10_24,
"facebook/bart-large": 10_24,
"facebook/bart-large-mnli": 10_24,
"facebook/bart-large-cnn": 10_24,
"facebook/bart-large-xsum": 10_24,
"yjernite/bart_eli5": 10_24,
}
class UpperCAmelCase__( lowerCamelCase ):
'''simple docstring'''
A : Optional[Any] = VOCAB_FILES_NAMES
A : Dict = PRETRAINED_VOCAB_FILES_MAP
A : str = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
A : int = ["input_ids", "attention_mask"]
A : Any = BartTokenizer
def __init__( self : List[Any] , lowerCAmelCase : Any=None , lowerCAmelCase : Optional[int]=None , lowerCAmelCase : List[Any]=None , lowerCAmelCase : str="replace" , lowerCAmelCase : str="<s>" , lowerCAmelCase : int="</s>" , lowerCAmelCase : Optional[int]="</s>" , lowerCAmelCase : Union[str, Any]="<s>" , lowerCAmelCase : str="<unk>" , lowerCAmelCase : int="<pad>" , lowerCAmelCase : int="<mask>" , lowerCAmelCase : Dict=False , lowerCAmelCase : List[Any]=True , **lowerCAmelCase : Optional[Any] , ) -> Optional[Any]:
"""simple docstring"""
super().__init__(
lowerCAmelCase , lowerCAmelCase , tokenizer_file=lowerCAmelCase , errors=lowerCAmelCase , bos_token=lowerCAmelCase , eos_token=lowerCAmelCase , sep_token=lowerCAmelCase , cls_token=lowerCAmelCase , unk_token=lowerCAmelCase , pad_token=lowerCAmelCase , mask_token=lowerCAmelCase , add_prefix_space=lowerCAmelCase , trim_offsets=lowerCAmelCase , **lowerCAmelCase , )
lowercase__ = json.loads(self.backend_tokenizer.pre_tokenizer.__getstate__())
if pre_tok_state.get('add_prefix_space' , lowerCAmelCase) != add_prefix_space:
lowercase__ = getattr(lowerCAmelCase , pre_tok_state.pop('type'))
lowercase__ = add_prefix_space
lowercase__ = pre_tok_class(**lowerCAmelCase)
lowercase__ = add_prefix_space
# the pre_tokenizer is already updated in the GPT2TokenizerFast `__init__`
lowercase__ = 'post_processor'
lowercase__ = getattr(self.backend_tokenizer , lowerCAmelCase , lowerCAmelCase)
if tokenizer_component_instance:
lowercase__ = json.loads(tokenizer_component_instance.__getstate__())
# The lists 'sep' and 'cls' must be cased in tuples for the object `post_processor_class`
if "sep" in state:
lowercase__ = tuple(state['sep'])
if "cls" in state:
lowercase__ = tuple(state['cls'])
lowercase__ = False
if state.get('add_prefix_space' , lowerCAmelCase) != add_prefix_space:
lowercase__ = add_prefix_space
lowercase__ = True
if state.get('trim_offsets' , lowerCAmelCase) != trim_offsets:
lowercase__ = trim_offsets
lowercase__ = True
if changes_to_apply:
lowercase__ = getattr(lowerCAmelCase , state.pop('type'))
lowercase__ = component_class(**lowerCAmelCase)
setattr(self.backend_tokenizer , lowerCAmelCase , lowerCAmelCase)
@property
def UpperCAmelCase ( self : Union[str, Any]) -> str:
"""simple docstring"""
if self._mask_token is None:
if self.verbose:
logger.error('Using mask_token, but it is not set yet.')
return None
return str(self._mask_token)
@mask_token.setter
def UpperCAmelCase ( self : Tuple , lowerCAmelCase : int) -> Optional[int]:
"""simple docstring"""
lowercase__ = AddedToken(lowerCAmelCase , lstrip=lowerCAmelCase , rstrip=lowerCAmelCase) if isinstance(lowerCAmelCase , lowerCAmelCase) else value
lowercase__ = value
def UpperCAmelCase ( self : List[str] , *lowerCAmelCase : Tuple , **lowerCAmelCase : Optional[int]) -> BatchEncoding:
"""simple docstring"""
lowercase__ = kwargs.get('is_split_into_words' , lowerCAmelCase)
if is_split_into_words and not self.add_prefix_space:
raise ValueError(
f'''You need to instantiate {self.__class__.__name__} with add_prefix_space=True '''
'to use it with pretokenized inputs.')
return super()._batch_encode_plus(*lowerCAmelCase , **lowerCAmelCase)
def UpperCAmelCase ( self : str , *lowerCAmelCase : Tuple , **lowerCAmelCase : str) -> BatchEncoding:
"""simple docstring"""
lowercase__ = kwargs.get('is_split_into_words' , lowerCAmelCase)
if is_split_into_words and not self.add_prefix_space:
raise ValueError(
f'''You need to instantiate {self.__class__.__name__} with add_prefix_space=True '''
'to use it with pretokenized inputs.')
return super()._encode_plus(*lowerCAmelCase , **lowerCAmelCase)
def UpperCAmelCase ( self : Dict , lowerCAmelCase : str , lowerCAmelCase : Optional[str] = None) -> Tuple[str]:
"""simple docstring"""
lowercase__ = self._tokenizer.model.save(lowerCAmelCase , name=lowerCAmelCase)
return tuple(lowerCAmelCase)
def UpperCAmelCase ( self : Any , lowerCAmelCase : str , lowerCAmelCase : Optional[int]=None) -> Tuple:
"""simple docstring"""
lowercase__ = [self.bos_token_id] + token_ids_a + [self.eos_token_id]
if token_ids_a is None:
return output
return output + [self.eos_token_id] + token_ids_a + [self.eos_token_id]
def UpperCAmelCase ( self : Union[str, Any] , lowerCAmelCase : List[int] , lowerCAmelCase : Optional[List[int]] = None) -> List[int]:
"""simple docstring"""
lowercase__ = [self.sep_token_id]
lowercase__ = [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]
| 642
| 0
|
import logging
import os
import sys
from pathlib import Path
from unittest.mock import patch
from parameterized import parameterized
from run_eval import run_generate
from run_eval_search import run_search
from transformers.testing_utils import CaptureStdout, TestCasePlus, slow
from utils import ROUGE_KEYS
logging.basicConfig(level=logging.DEBUG)
_snake_case = logging.getLogger()
def _A ( __magic_name__ , __magic_name__ ):
lowercase__ = "\n".join(__magic_name__ )
Path(__magic_name__ ).open("w" ).writelines(__magic_name__ )
_snake_case = """patrickvonplaten/t5-tiny-random"""
_snake_case = """sshleifer/bart-tiny-random"""
_snake_case = """sshleifer/tiny-mbart"""
_snake_case = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
logging.disable(logging.CRITICAL) # remove noisy download output from tracebacks
class lowerCAmelCase ( lowercase_ ):
def UpperCAmelCase ( self :Any , _lowercase :Optional[int] ):
'''simple docstring'''
lowercase__ = Path(self.get_auto_remove_tmp_dir() ) / "utest_input.source"
lowercase__ = input_file_name.parent / "utest_output.txt"
assert not output_file_name.exists()
lowercase__ = [" New York (CNN)When Liana Barrientos was 23 years old, she got married in Westchester County."]
_dump_articles(_lowercase , _lowercase )
lowercase__ = str(Path(self.get_auto_remove_tmp_dir() ) / "scores.json" )
lowercase__ = "translation_en_to_de" if model == T5_TINY else "summarization"
lowercase__ = f'''
run_eval_search.py
{model}
{input_file_name}
{output_file_name}
--score_path {score_path}
--task {task}
--num_beams 2
--length_penalty 2.0
'''.split()
with patch.object(_lowercase , "argv" , _lowercase ):
run_generate()
assert Path(_lowercase ).exists()
# os.remove(Path(output_file_name))
def UpperCAmelCase ( self :Union[str, Any] ):
'''simple docstring'''
self.run_eval_tester(_lowercase )
@parameterized.expand([BART_TINY, MBART_TINY] )
@slow
def UpperCAmelCase ( self :Union[str, Any] , _lowercase :Optional[Any] ):
'''simple docstring'''
self.run_eval_tester(_lowercase )
@parameterized.expand([T5_TINY, MBART_TINY] )
@slow
def UpperCAmelCase ( self :Union[str, Any] , _lowercase :int ):
'''simple docstring'''
lowercase__ = Path(self.get_auto_remove_tmp_dir() ) / "utest_input.source"
lowercase__ = input_file_name.parent / "utest_output.txt"
assert not output_file_name.exists()
lowercase__ = {
"en": ["Machine learning is great, isn't it?", "I like to eat bananas", "Tomorrow is another great day!"],
"de": [
"Maschinelles Lernen ist großartig, oder?",
"Ich esse gerne Bananen",
"Morgen ist wieder ein toller Tag!",
],
}
lowercase__ = Path(self.get_auto_remove_tmp_dir() )
lowercase__ = str(tmp_dir / "scores.json" )
lowercase__ = str(tmp_dir / "val.target" )
_dump_articles(_lowercase , text["en"] )
_dump_articles(_lowercase , text["de"] )
lowercase__ = "translation_en_to_de" if model == T5_TINY else "summarization"
lowercase__ = f'''
run_eval_search.py
{model}
{str(_lowercase )}
{str(_lowercase )}
--score_path {score_path}
--reference_path {reference_path}
--task {task}
'''.split()
testargs.extend(["--search", "num_beams=1:2 length_penalty=0.9:1.0"] )
with patch.object(_lowercase , "argv" , _lowercase ):
with CaptureStdout() as cs:
run_search()
lowercase__ = [" num_beams | length_penalty", model, "Best score args"]
lowercase__ = ["Info"]
if "translation" in task:
expected_strings.append("bleu" )
else:
expected_strings.extend(_lowercase )
for w in expected_strings:
assert w in cs.out
for w in un_expected_strings:
assert w not in cs.out
assert Path(_lowercase ).exists()
os.remove(Path(_lowercase ) )
| 655
|
import copy
import os
from typing import Union
from ...configuration_utils import PretrainedConfig
from ...utils import logging
_snake_case = logging.get_logger(__name__)
_snake_case = {
"""microsoft/git-base""": """https://huggingface.co/microsoft/git-base/resolve/main/config.json""",
}
class lowerCAmelCase ( lowercase_ ):
__lowerCamelCase = 'git_vision_model'
def __init__( self :Dict , _lowercase :Dict=7_68 , _lowercase :Dict=30_72 , _lowercase :Tuple=12 , _lowercase :List[str]=12 , _lowercase :Tuple=3 , _lowercase :Dict=2_24 , _lowercase :Tuple=16 , _lowercase :Optional[int]="quick_gelu" , _lowercase :Union[str, Any]=1e-5 , _lowercase :Tuple=0.0 , _lowercase :Tuple=0.02 , **_lowercase :Optional[Any] , ):
'''simple docstring'''
super().__init__(**_lowercase )
lowercase__ = hidden_size
lowercase__ = intermediate_size
lowercase__ = num_hidden_layers
lowercase__ = num_attention_heads
lowercase__ = num_channels
lowercase__ = patch_size
lowercase__ = image_size
lowercase__ = initializer_range
lowercase__ = attention_dropout
lowercase__ = layer_norm_eps
lowercase__ = hidden_act
@classmethod
def UpperCAmelCase ( cls :List[str] , _lowercase :Union[str, os.PathLike] , **_lowercase :Optional[int] ):
'''simple docstring'''
cls._set_token_in_kwargs(_lowercase )
lowercase__ , lowercase__ = cls.get_config_dict(_lowercase , **_lowercase )
# get the vision config dict if we are loading from GITConfig
if config_dict.get("model_type" ) == "git":
lowercase__ = config_dict["vision_config"]
if "model_type" in config_dict and hasattr(cls , "model_type" ) and config_dict["model_type"] != cls.model_type:
logger.warning(
f'''You are using a model of type {config_dict['model_type']} to instantiate a model of type '''
f'''{cls.model_type}. This is not supported for all configurations of models and can yield errors.''' )
return cls.from_dict(_lowercase , **_lowercase )
class lowerCAmelCase ( lowercase_ ):
__lowerCamelCase = 'git'
def __init__( self :Union[str, Any] , _lowercase :Dict=None , _lowercase :List[str]=3_05_22 , _lowercase :Tuple=7_68 , _lowercase :Any=6 , _lowercase :Dict=12 , _lowercase :Any=30_72 , _lowercase :List[Any]="gelu" , _lowercase :Tuple=0.1 , _lowercase :Optional[int]=0.1 , _lowercase :Optional[Any]=10_24 , _lowercase :Any=0.02 , _lowercase :int=1e-12 , _lowercase :List[Any]=0 , _lowercase :int="absolute" , _lowercase :List[str]=True , _lowercase :Any=False , _lowercase :int=1_01 , _lowercase :str=1_02 , _lowercase :Dict=None , **_lowercase :List[str] , ):
'''simple docstring'''
super().__init__(bos_token_id=_lowercase , eos_token_id=_lowercase , pad_token_id=_lowercase , **_lowercase )
if vision_config is None:
lowercase__ = {}
logger.info("vision_config is None. initializing the GitVisionConfig with default values." )
lowercase__ = GitVisionConfig(**_lowercase )
lowercase__ = vocab_size
lowercase__ = hidden_size
lowercase__ = num_hidden_layers
lowercase__ = num_attention_heads
lowercase__ = hidden_act
lowercase__ = intermediate_size
lowercase__ = hidden_dropout_prob
lowercase__ = attention_probs_dropout_prob
lowercase__ = max_position_embeddings
lowercase__ = initializer_range
lowercase__ = layer_norm_eps
lowercase__ = position_embedding_type
lowercase__ = use_cache
lowercase__ = tie_word_embeddings
lowercase__ = num_image_with_embedding
lowercase__ = bos_token_id
lowercase__ = eos_token_id
def UpperCAmelCase ( self :Tuple ):
'''simple docstring'''
lowercase__ = copy.deepcopy(self.__dict__ )
lowercase__ = self.vision_config.to_dict()
lowercase__ = self.__class__.model_type
return output
| 655
| 1
|
"""simple docstring"""
import os
from tempfile import TemporaryDirectory
from unittest import TestCase
import pytest
from absl.testing import parameterized
from datasets import config
from datasets.arrow_reader import HF_GCP_BASE_URL
from datasets.builder import DatasetBuilder
from datasets.dataset_dict import IterableDatasetDict
from datasets.iterable_dataset import IterableDataset
from datasets.load import dataset_module_factory, import_main_class
from datasets.utils.file_utils import cached_path
snake_case = [
{'dataset': 'wikipedia', 'config_name': '20220301.de'},
{'dataset': 'wikipedia', 'config_name': '20220301.en'},
{'dataset': 'wikipedia', 'config_name': '20220301.fr'},
{'dataset': 'wikipedia', 'config_name': '20220301.frr'},
{'dataset': 'wikipedia', 'config_name': '20220301.it'},
{'dataset': 'wikipedia', 'config_name': '20220301.simple'},
{'dataset': 'snli', 'config_name': 'plain_text'},
{'dataset': 'eli5', 'config_name': 'LFQA_reddit'},
{'dataset': 'wiki40b', 'config_name': 'en'},
{'dataset': 'wiki_dpr', 'config_name': 'psgs_w100.nq.compressed'},
{'dataset': 'wiki_dpr', 'config_name': 'psgs_w100.nq.no_index'},
{'dataset': 'wiki_dpr', 'config_name': 'psgs_w100.multiset.no_index'},
{'dataset': 'natural_questions', 'config_name': 'default'},
]
def UpperCamelCase_ ( SCREAMING_SNAKE_CASE_=True ):
if with_config:
return [
{
"testcase_name": d["dataset"] + "/" + d["config_name"],
"dataset": d["dataset"],
"config_name": d["config_name"],
}
for d in DATASETS_ON_HF_GCP
]
else:
return [
{"testcase_name": dataset, "dataset": dataset} for dataset in {d["dataset"] for d in DATASETS_ON_HF_GCP}
]
@parameterized.named_parameters(list_datasets_on_hf_gcp_parameters(with_config=__magic_name__ ) )
class UpperCamelCase ( __magic_name__ ):
"""simple docstring"""
UpperCAmelCase_ : Union[str, Any] = None
UpperCAmelCase_ : Dict = None
def A ( self , lowercase__ , lowercase__ ) -> Dict:
"""simple docstring"""
with TemporaryDirectory() as tmp_dir:
SCREAMING_SNAKE_CASE = dataset_module_factory(lowercase__ , cache_dir=lowercase__ )
SCREAMING_SNAKE_CASE = import_main_class(dataset_module.module_path , dataset=lowercase__ )
SCREAMING_SNAKE_CASE = builder_cls(
cache_dir=lowercase__ , config_name=lowercase__ , hash=dataset_module.hash , )
SCREAMING_SNAKE_CASE = '/'.join(
[
HF_GCP_BASE_URL,
builder_instance._relative_data_dir(with_hash=lowercase__ ).replace(os.sep , '/' ),
config.DATASET_INFO_FILENAME,
] )
SCREAMING_SNAKE_CASE = cached_path(lowercase__ , cache_dir=lowercase__ )
self.assertTrue(os.path.exists(lowercase__ ) )
@pytest.mark.integration
def UpperCamelCase_ ( SCREAMING_SNAKE_CASE_ ):
SCREAMING_SNAKE_CASE = tmp_path_factory.mktemp('test_hf_gcp' ) / 'test_wikipedia_simple'
SCREAMING_SNAKE_CASE = dataset_module_factory('wikipedia', cache_dir=SCREAMING_SNAKE_CASE_ )
SCREAMING_SNAKE_CASE = import_main_class(dataset_module.module_path )
SCREAMING_SNAKE_CASE = builder_cls(
cache_dir=SCREAMING_SNAKE_CASE_, config_name='20220301.frr', hash=dataset_module.hash, )
# use the HF cloud storage, not the original download_and_prepare that uses apache-beam
SCREAMING_SNAKE_CASE = None
builder_instance.download_and_prepare()
SCREAMING_SNAKE_CASE = builder_instance.as_dataset()
assert ds
@pytest.mark.integration
def UpperCamelCase_ ( SCREAMING_SNAKE_CASE_ ):
SCREAMING_SNAKE_CASE = dataset_module_factory('wikipedia', cache_dir=SCREAMING_SNAKE_CASE_ )
SCREAMING_SNAKE_CASE = import_main_class(dataset_module.module_path, dataset=SCREAMING_SNAKE_CASE_ )
SCREAMING_SNAKE_CASE = builder_cls(
cache_dir=SCREAMING_SNAKE_CASE_, config_name='20220301.frr', hash=dataset_module.hash, )
SCREAMING_SNAKE_CASE = builder_instance.as_streaming_dataset()
assert ds
assert isinstance(SCREAMING_SNAKE_CASE_, SCREAMING_SNAKE_CASE_ )
assert "train" in ds
assert isinstance(ds['train'], SCREAMING_SNAKE_CASE_ )
assert next(iter(ds['train'] ) )
| 703
|
"""simple docstring"""
from __future__ import annotations
def UpperCamelCase_ ( SCREAMING_SNAKE_CASE_ ):
if len(SCREAMING_SNAKE_CASE_ ) == 0:
return []
SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE = min(SCREAMING_SNAKE_CASE_ ), max(SCREAMING_SNAKE_CASE_ )
SCREAMING_SNAKE_CASE = int(max_value - min_value ) + 1
SCREAMING_SNAKE_CASE = [[] for _ in range(SCREAMING_SNAKE_CASE_ )]
for i in my_list:
buckets[int(i - min_value )].append(SCREAMING_SNAKE_CASE_ )
return [v for bucket in buckets for v in sorted(SCREAMING_SNAKE_CASE_ )]
if __name__ == "__main__":
from doctest import testmod
testmod()
assert bucket_sort([4, 5, 3, 2, 1]) == [1, 2, 3, 4, 5]
assert bucket_sort([0, 1, -1_0, 1_5, 2, -2]) == [-1_0, -2, 0, 1, 2, 1_5]
| 406
| 0
|
import multiprocessing
import time
from arguments import PretokenizationArguments
from datasets import load_dataset
from transformers import AutoTokenizer, HfArgumentParser
def A__ ( SCREAMING_SNAKE_CASE_ : int ) -> List[str]:
"""simple docstring"""
_UpperCAmelCase = {}
_UpperCAmelCase = tokenizer(example['''content'''] , truncation=SCREAMING_SNAKE_CASE_ )['''input_ids''']
_UpperCAmelCase = len(example['''content'''] ) / len(output['''input_ids'''] )
return output
UpperCAmelCase_ = HfArgumentParser(PretokenizationArguments)
UpperCAmelCase_ = parser.parse_args()
if args.num_workers is None:
UpperCAmelCase_ = multiprocessing.cpu_count()
UpperCAmelCase_ = AutoTokenizer.from_pretrained(args.tokenizer_dir)
UpperCAmelCase_ = time.time()
UpperCAmelCase_ = load_dataset(args.dataset_name, split="train")
print(f'''Dataset loaded in {time.time()-t_start:.2f}s''')
UpperCAmelCase_ = time.time()
UpperCAmelCase_ = ds.map(
tokenize,
num_proc=args.num_workers,
remove_columns=[
"repo_name",
"path",
"copies",
"size",
"content",
"license",
"hash",
"line_mean",
"line_max",
"alpha_frac",
"autogenerated",
],
)
print(f'''Dataset tokenized in {time.time()-t_start:.2f}s''')
UpperCAmelCase_ = time.time()
ds.push_to_hub(args.tokenized_data_repo)
print(f'''Data pushed to the hub in {time.time()-t_start:.2f}s''')
| 32
|
import tempfile
import torch
from diffusers import (
DEISMultistepScheduler,
DPMSolverMultistepScheduler,
DPMSolverSinglestepScheduler,
UniPCMultistepScheduler,
)
from .test_schedulers import SchedulerCommonTest
class UpperCAmelCase ( UpperCAmelCase__ ):
'''simple docstring'''
lowerCAmelCase_ = (DPMSolverSinglestepScheduler,)
lowerCAmelCase_ = (('''num_inference_steps''', 25),)
def snake_case__ ( self : str , **__lowercase : Any ):
"""simple docstring"""
snake_case_ = {
"num_train_timesteps": 10_00,
"beta_start": 0.0001,
"beta_end": 0.02,
"beta_schedule": "linear",
"solver_order": 2,
"prediction_type": "epsilon",
"thresholding": False,
"sample_max_value": 1.0,
"algorithm_type": "dpmsolver++",
"solver_type": "midpoint",
"lambda_min_clipped": -float("inf" ),
"variance_type": None,
}
config.update(**__lowercase )
return config
def snake_case__ ( self : str , __lowercase : int=0 , **__lowercase : str ):
"""simple docstring"""
snake_case_ = dict(self.forward_default_kwargs )
snake_case_ = kwargs.pop("num_inference_steps" , __lowercase )
snake_case_ = self.dummy_sample
snake_case_ = 0.1 * sample
snake_case_ = [residual + 0.2, residual + 0.15, residual + 0.10]
for scheduler_class in self.scheduler_classes:
snake_case_ = self.get_scheduler_config(**__lowercase )
snake_case_ = scheduler_class(**__lowercase )
scheduler.set_timesteps(__lowercase )
# copy over dummy past residuals
snake_case_ = dummy_past_residuals[: scheduler.config.solver_order]
with tempfile.TemporaryDirectory() as tmpdirname:
scheduler.save_config(__lowercase )
snake_case_ = scheduler_class.from_pretrained(__lowercase )
new_scheduler.set_timesteps(__lowercase )
# copy over dummy past residuals
snake_case_ = dummy_past_residuals[: new_scheduler.config.solver_order]
snake_case_ , snake_case_ = sample, sample
for t in range(__lowercase , time_step + scheduler.config.solver_order + 1 ):
snake_case_ = scheduler.step(__lowercase , __lowercase , __lowercase , **__lowercase ).prev_sample
snake_case_ = new_scheduler.step(__lowercase , __lowercase , __lowercase , **__lowercase ).prev_sample
assert torch.sum(torch.abs(output - new_output ) ) < 1E-5, "Scheduler outputs are not identical"
def snake_case__ ( self : Union[str, Any] ):
"""simple docstring"""
pass
def snake_case__ ( self : List[Any] , __lowercase : Optional[int]=0 , **__lowercase : int ):
"""simple docstring"""
snake_case_ = dict(self.forward_default_kwargs )
snake_case_ = kwargs.pop("num_inference_steps" , __lowercase )
snake_case_ = self.dummy_sample
snake_case_ = 0.1 * sample
snake_case_ = [residual + 0.2, residual + 0.15, residual + 0.10]
for scheduler_class in self.scheduler_classes:
snake_case_ = self.get_scheduler_config()
snake_case_ = scheduler_class(**__lowercase )
scheduler.set_timesteps(__lowercase )
# copy over dummy past residuals (must be after setting timesteps)
snake_case_ = dummy_past_residuals[: scheduler.config.solver_order]
with tempfile.TemporaryDirectory() as tmpdirname:
scheduler.save_config(__lowercase )
snake_case_ = scheduler_class.from_pretrained(__lowercase )
# copy over dummy past residuals
new_scheduler.set_timesteps(__lowercase )
# copy over dummy past residual (must be after setting timesteps)
snake_case_ = dummy_past_residuals[: new_scheduler.config.solver_order]
snake_case_ = scheduler.step(__lowercase , __lowercase , __lowercase , **__lowercase ).prev_sample
snake_case_ = new_scheduler.step(__lowercase , __lowercase , __lowercase , **__lowercase ).prev_sample
assert torch.sum(torch.abs(output - new_output ) ) < 1E-5, "Scheduler outputs are not identical"
def snake_case__ ( self : List[str] , __lowercase : Any=None , **__lowercase : List[Any] ):
"""simple docstring"""
if scheduler is None:
snake_case_ = self.scheduler_classes[0]
snake_case_ = self.get_scheduler_config(**__lowercase )
snake_case_ = scheduler_class(**__lowercase )
snake_case_ = self.scheduler_classes[0]
snake_case_ = self.get_scheduler_config(**__lowercase )
snake_case_ = scheduler_class(**__lowercase )
snake_case_ = 10
snake_case_ = self.dummy_model()
snake_case_ = self.dummy_sample_deter
scheduler.set_timesteps(__lowercase )
for i, t in enumerate(scheduler.timesteps ):
snake_case_ = model(__lowercase , __lowercase )
snake_case_ = scheduler.step(__lowercase , __lowercase , __lowercase ).prev_sample
return sample
def snake_case__ ( self : Any ):
"""simple docstring"""
snake_case_ = DPMSolverSinglestepScheduler(**self.get_scheduler_config() )
snake_case_ = 50
snake_case_ = self.dummy_model()
snake_case_ = self.dummy_sample_deter
scheduler.set_timesteps(__lowercase )
# make sure that the first t is uneven
for i, t in enumerate(scheduler.timesteps[3:] ):
snake_case_ = model(__lowercase , __lowercase )
snake_case_ = scheduler.step(__lowercase , __lowercase , __lowercase ).prev_sample
snake_case_ = torch.mean(torch.abs(__lowercase ) )
assert abs(result_mean.item() - 0.2574 ) < 1E-3
def snake_case__ ( self : Optional[int] ):
"""simple docstring"""
for timesteps in [25, 50, 1_00, 9_99, 10_00]:
self.check_over_configs(num_train_timesteps=__lowercase )
def snake_case__ ( self : Dict ):
"""simple docstring"""
snake_case_ = DPMSolverSinglestepScheduler(**self.get_scheduler_config() )
snake_case_ = self.full_loop(scheduler=__lowercase )
snake_case_ = torch.mean(torch.abs(__lowercase ) )
assert abs(result_mean.item() - 0.2791 ) < 1E-3
snake_case_ = DEISMultistepScheduler.from_config(scheduler.config )
snake_case_ = DPMSolverMultistepScheduler.from_config(scheduler.config )
snake_case_ = UniPCMultistepScheduler.from_config(scheduler.config )
snake_case_ = DPMSolverSinglestepScheduler.from_config(scheduler.config )
snake_case_ = self.full_loop(scheduler=__lowercase )
snake_case_ = torch.mean(torch.abs(__lowercase ) )
assert abs(result_mean.item() - 0.2791 ) < 1E-3
def snake_case__ ( self : List[str] ):
"""simple docstring"""
self.check_over_configs(thresholding=__lowercase )
for order in [1, 2, 3]:
for solver_type in ["midpoint", "heun"]:
for threshold in [0.5, 1.0, 2.0]:
for prediction_type in ["epsilon", "sample"]:
self.check_over_configs(
thresholding=__lowercase , prediction_type=__lowercase , sample_max_value=__lowercase , algorithm_type="dpmsolver++" , solver_order=__lowercase , solver_type=__lowercase , )
def snake_case__ ( self : Union[str, Any] ):
"""simple docstring"""
for prediction_type in ["epsilon", "v_prediction"]:
self.check_over_configs(prediction_type=__lowercase )
def snake_case__ ( self : Optional[Any] ):
"""simple docstring"""
for algorithm_type in ["dpmsolver", "dpmsolver++"]:
for solver_type in ["midpoint", "heun"]:
for order in [1, 2, 3]:
for prediction_type in ["epsilon", "sample"]:
self.check_over_configs(
solver_order=__lowercase , solver_type=__lowercase , prediction_type=__lowercase , algorithm_type=__lowercase , )
snake_case_ = self.full_loop(
solver_order=__lowercase , solver_type=__lowercase , prediction_type=__lowercase , algorithm_type=__lowercase , )
assert not torch.isnan(__lowercase ).any(), "Samples have nan numbers"
def snake_case__ ( self : List[Any] ):
"""simple docstring"""
self.check_over_configs(lower_order_final=__lowercase )
self.check_over_configs(lower_order_final=__lowercase )
def snake_case__ ( self : Union[str, Any] ):
"""simple docstring"""
self.check_over_configs(lambda_min_clipped=-float("inf" ) )
self.check_over_configs(lambda_min_clipped=-5.1 )
def snake_case__ ( self : List[Any] ):
"""simple docstring"""
self.check_over_configs(variance_type=__lowercase )
self.check_over_configs(variance_type="learned_range" )
def snake_case__ ( self : List[Any] ):
"""simple docstring"""
for num_inference_steps in [1, 2, 3, 5, 10, 50, 1_00, 9_99, 10_00]:
self.check_over_forward(num_inference_steps=__lowercase , time_step=0 )
def snake_case__ ( self : int ):
"""simple docstring"""
snake_case_ = self.full_loop()
snake_case_ = torch.mean(torch.abs(__lowercase ) )
assert abs(result_mean.item() - 0.2791 ) < 1E-3
def snake_case__ ( self : Any ):
"""simple docstring"""
snake_case_ = self.full_loop(use_karras_sigmas=__lowercase )
snake_case_ = torch.mean(torch.abs(__lowercase ) )
assert abs(result_mean.item() - 0.2248 ) < 1E-3
def snake_case__ ( self : str ):
"""simple docstring"""
snake_case_ = self.full_loop(prediction_type="v_prediction" )
snake_case_ = torch.mean(torch.abs(__lowercase ) )
assert abs(result_mean.item() - 0.1453 ) < 1E-3
def snake_case__ ( self : int ):
"""simple docstring"""
snake_case_ = self.full_loop(prediction_type="v_prediction" , use_karras_sigmas=__lowercase )
snake_case_ = torch.mean(torch.abs(__lowercase ) )
assert abs(result_mean.item() - 0.0649 ) < 1E-3
def snake_case__ ( self : Optional[Any] ):
"""simple docstring"""
snake_case_ = self.scheduler_classes[0]
snake_case_ = self.get_scheduler_config(thresholding=__lowercase , dynamic_thresholding_ratio=0 )
snake_case_ = scheduler_class(**__lowercase )
snake_case_ = 10
snake_case_ = self.dummy_model()
snake_case_ = self.dummy_sample_deter.half()
scheduler.set_timesteps(__lowercase )
for i, t in enumerate(scheduler.timesteps ):
snake_case_ = model(__lowercase , __lowercase )
snake_case_ = scheduler.step(__lowercase , __lowercase , __lowercase ).prev_sample
assert sample.dtype == torch.floataa
| 376
| 0
|
'''simple docstring'''
import itertools
import random
import unittest
import numpy as np
from transformers import BatchFeature, SpeechTaFeatureExtractor
from transformers.testing_utils import require_torch
from transformers.utils.import_utils import is_torch_available
from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
if is_torch_available():
import torch
_UpperCamelCase : int =random.Random()
def lowerCamelCase_ ( A_ , A_=1.0 , A_=None , A_=None ):
if rng is None:
__lowerCamelCase = global_rng
__lowerCamelCase = []
for batch_idx in range(shape[0] ):
values.append([] )
for _ in range(shape[1] ):
values[-1].append(rng.random() * scale )
return values
@require_torch
class _SCREAMING_SNAKE_CASE ( unittest.TestCase ):
"""simple docstring"""
def __init__( self , _snake_case , _snake_case=7 , _snake_case=4_00 , _snake_case=20_00 , _snake_case=1 , _snake_case=0.0 , _snake_case=1_60_00 , _snake_case=True , _snake_case=80 , _snake_case=16 , _snake_case=64 , _snake_case="hann_window" , _snake_case=80 , _snake_case=76_00 , _snake_case=1E-10 , _snake_case=True , ):
"""simple docstring"""
__lowerCamelCase = parent
__lowerCamelCase = batch_size
__lowerCamelCase = min_seq_length
__lowerCamelCase = max_seq_length
__lowerCamelCase = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
__lowerCamelCase = feature_size
__lowerCamelCase = padding_value
__lowerCamelCase = sampling_rate
__lowerCamelCase = do_normalize
__lowerCamelCase = num_mel_bins
__lowerCamelCase = hop_length
__lowerCamelCase = win_length
__lowerCamelCase = win_function
__lowerCamelCase = fmin
__lowerCamelCase = fmax
__lowerCamelCase = mel_floor
__lowerCamelCase = return_attention_mask
def _lowerCamelCase ( self ):
"""simple docstring"""
return {
"feature_size": self.feature_size,
"padding_value": self.padding_value,
"sampling_rate": self.sampling_rate,
"do_normalize": self.do_normalize,
"num_mel_bins": self.num_mel_bins,
"hop_length": self.hop_length,
"win_length": self.win_length,
"win_function": self.win_function,
"fmin": self.fmin,
"fmax": self.fmax,
"mel_floor": self.mel_floor,
"return_attention_mask": self.return_attention_mask,
}
def _lowerCamelCase ( self , _snake_case=False , _snake_case=False ):
"""simple docstring"""
def _flatten(_snake_case ):
return list(itertools.chain(*_snake_case ) )
if equal_length:
__lowerCamelCase = floats_list((self.batch_size, self.max_seq_length) )
else:
# make sure that inputs increase in size
__lowerCamelCase = [
_flatten(floats_list((x, self.feature_size) ) )
for x in range(self.min_seq_length , self.max_seq_length , self.seq_length_diff )
]
if numpify:
__lowerCamelCase = [np.asarray(_snake_case ) for x in speech_inputs]
return speech_inputs
def _lowerCamelCase ( self , _snake_case=False , _snake_case=False ):
"""simple docstring"""
if equal_length:
__lowerCamelCase = [floats_list((self.max_seq_length, self.num_mel_bins) ) for _ in range(self.batch_size )]
else:
# make sure that inputs increase in size
__lowerCamelCase = [
floats_list((x, self.num_mel_bins) )
for x in range(self.min_seq_length , self.max_seq_length , self.seq_length_diff )
]
if numpify:
__lowerCamelCase = [np.asarray(_snake_case ) for x in speech_inputs]
return speech_inputs
@require_torch
class _SCREAMING_SNAKE_CASE ( UpperCamelCase , unittest.TestCase ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ = SpeechTaFeatureExtractor
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = SpeechTaFeatureExtractionTester(self )
def _lowerCamelCase ( self , _snake_case ):
"""simple docstring"""
self.assertTrue(np.all(np.mean(_snake_case , axis=0 ) < 1E-3 ) )
self.assertTrue(np.all(np.abs(np.var(_snake_case , axis=0 ) - 1 ) < 1E-3 ) )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
# create three inputs of length 800, 1000, and 1200
__lowerCamelCase = [floats_list((1, x) )[0] for x in range(8_00 , 14_00 , 2_00 )]
__lowerCamelCase = [np.asarray(_snake_case ) for speech_input in speech_inputs]
# Test not batched input
__lowerCamelCase = feat_extract(speech_inputs[0] , return_tensors='''np''' ).input_values
__lowerCamelCase = feat_extract(np_speech_inputs[0] , return_tensors='''np''' ).input_values
self.assertTrue(np.allclose(_snake_case , _snake_case , atol=1E-3 ) )
# Test batched
__lowerCamelCase = feat_extract(_snake_case , return_tensors='''np''' ).input_values
__lowerCamelCase = feat_extract(_snake_case , return_tensors='''np''' ).input_values
for enc_seq_a, enc_seq_a in zip(_snake_case , _snake_case ):
self.assertTrue(np.allclose(_snake_case , _snake_case , atol=1E-3 ) )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__lowerCamelCase = [floats_list((1, x) )[0] for x in range(8_00 , 14_00 , 2_00 )]
__lowerCamelCase = ['''longest''', '''max_length''', '''do_not_pad''']
__lowerCamelCase = [None, 16_00, None]
for max_length, padding in zip(_snake_case , _snake_case ):
__lowerCamelCase = feat_extract(_snake_case , padding=_snake_case , max_length=_snake_case , return_tensors='''np''' )
__lowerCamelCase = processed.input_values
self._check_zero_mean_unit_variance(input_values[0][:8_00] )
self.assertTrue(input_values[0][8_00:].sum() < 1E-6 )
self._check_zero_mean_unit_variance(input_values[1][:10_00] )
self.assertTrue(input_values[0][10_00:].sum() < 1E-6 )
self._check_zero_mean_unit_variance(input_values[2][:12_00] )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__lowerCamelCase = range(8_00 , 14_00 , 2_00 )
__lowerCamelCase = [floats_list((1, x) )[0] for x in lengths]
__lowerCamelCase = ['''longest''', '''max_length''', '''do_not_pad''']
__lowerCamelCase = [None, 16_00, None]
for max_length, padding in zip(_snake_case , _snake_case ):
__lowerCamelCase = feat_extract(_snake_case , max_length=_snake_case , padding=_snake_case )
__lowerCamelCase = processed.input_values
self._check_zero_mean_unit_variance(input_values[0][:8_00] )
self._check_zero_mean_unit_variance(input_values[1][:10_00] )
self._check_zero_mean_unit_variance(input_values[2][:12_00] )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__lowerCamelCase = [floats_list((1, x) )[0] for x in range(8_00 , 14_00 , 2_00 )]
__lowerCamelCase = feat_extract(
_snake_case , truncation=_snake_case , max_length=10_00 , padding='''max_length''' , return_tensors='''np''' )
__lowerCamelCase = processed.input_values
self._check_zero_mean_unit_variance(input_values[0, :8_00] )
self._check_zero_mean_unit_variance(input_values[1] )
self._check_zero_mean_unit_variance(input_values[2] )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__lowerCamelCase = [floats_list((1, x) )[0] for x in range(8_00 , 14_00 , 2_00 )]
__lowerCamelCase = feat_extract(
_snake_case , truncation=_snake_case , max_length=10_00 , padding='''longest''' , return_tensors='''np''' )
__lowerCamelCase = processed.input_values
self._check_zero_mean_unit_variance(input_values[0, :8_00] )
self._check_zero_mean_unit_variance(input_values[1, :10_00] )
self._check_zero_mean_unit_variance(input_values[2] )
# make sure that if max_length < longest -> then pad to max_length
self.assertTrue(input_values.shape == (3, 10_00) )
__lowerCamelCase = [floats_list((1, x) )[0] for x in range(8_00 , 14_00 , 2_00 )]
__lowerCamelCase = feat_extract(
_snake_case , truncation=_snake_case , max_length=20_00 , padding='''longest''' , return_tensors='''np''' )
__lowerCamelCase = processed.input_values
self._check_zero_mean_unit_variance(input_values[0, :8_00] )
self._check_zero_mean_unit_variance(input_values[1, :10_00] )
self._check_zero_mean_unit_variance(input_values[2] )
# make sure that if max_length > longest -> then pad to longest
self.assertTrue(input_values.shape == (3, 12_00) )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
__lowerCamelCase = np.random.rand(1_00 ).astype(np.floataa )
__lowerCamelCase = np_speech_inputs.tolist()
for inputs in [py_speech_inputs, np_speech_inputs]:
__lowerCamelCase = feature_extractor.pad([{'''input_values''': inputs}] , return_tensors='''np''' )
self.assertTrue(np_processed.input_values.dtype == np.floataa )
__lowerCamelCase = feature_extractor.pad([{'''input_values''': inputs}] , return_tensors='''pt''' )
self.assertTrue(pt_processed.input_values.dtype == torch.floataa )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_tester.prepare_feat_extract_dict() )
# create three inputs of length 800, 1000, and 1200
__lowerCamelCase = [floats_list((1, x) )[0] for x in range(8_00 , 14_00 , 2_00 )]
__lowerCamelCase = [np.asarray(_snake_case ) for speech_input in speech_inputs]
# Test feature size
__lowerCamelCase = feature_extractor(audio_target=_snake_case , padding=_snake_case , return_tensors='''np''' ).input_values
self.assertTrue(input_values.ndim == 3 )
self.assertTrue(input_values.shape[-1] == feature_extractor.num_mel_bins )
# Test not batched input
__lowerCamelCase = feature_extractor(speech_inputs[0] , return_tensors='''np''' ).input_values
__lowerCamelCase = feature_extractor(np_speech_inputs[0] , return_tensors='''np''' ).input_values
self.assertTrue(np.allclose(_snake_case , _snake_case , atol=1E-3 ) )
# Test batched
__lowerCamelCase = feature_extractor(_snake_case , return_tensors='''np''' ).input_values
__lowerCamelCase = feature_extractor(_snake_case , return_tensors='''np''' ).input_values
for enc_seq_a, enc_seq_a in zip(_snake_case , _snake_case ):
self.assertTrue(np.allclose(_snake_case , _snake_case , atol=1E-3 ) )
# Test 2-D numpy arrays are batched.
__lowerCamelCase = [floats_list((1, x) )[0] for x in (8_00, 8_00, 8_00)]
__lowerCamelCase = np.asarray(_snake_case )
__lowerCamelCase = feature_extractor(_snake_case , return_tensors='''np''' ).input_values
__lowerCamelCase = feature_extractor(_snake_case , return_tensors='''np''' ).input_values
for enc_seq_a, enc_seq_a in zip(_snake_case , _snake_case ):
self.assertTrue(np.allclose(_snake_case , _snake_case , atol=1E-3 ) )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feat_extract_tester.prepare_inputs_for_target()
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_dict )
__lowerCamelCase = feat_extract.model_input_names[0]
__lowerCamelCase = BatchFeature({input_name: speech_inputs} )
self.assertTrue(all(len(_snake_case ) == len(_snake_case ) for x, y in zip(_snake_case , processed_features[input_name] ) ) )
__lowerCamelCase = self.feat_extract_tester.prepare_inputs_for_target(equal_length=_snake_case )
__lowerCamelCase = BatchFeature({input_name: speech_inputs} , tensor_type='''np''' )
__lowerCamelCase = processed_features[input_name]
if len(batch_features_input.shape ) < 3:
__lowerCamelCase = batch_features_input[:, :, None]
self.assertTrue(
batch_features_input.shape
== (self.feat_extract_tester.batch_size, len(speech_inputs[0] ), self.feat_extract_tester.num_mel_bins) )
@require_torch
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feat_extract_tester.prepare_inputs_for_target(equal_length=_snake_case )
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_dict )
__lowerCamelCase = feat_extract.model_input_names[0]
__lowerCamelCase = BatchFeature({input_name: speech_inputs} , tensor_type='''pt''' )
__lowerCamelCase = processed_features[input_name]
if len(batch_features_input.shape ) < 3:
__lowerCamelCase = batch_features_input[:, :, None]
self.assertTrue(
batch_features_input.shape
== (self.feat_extract_tester.batch_size, len(speech_inputs[0] ), self.feat_extract_tester.num_mel_bins) )
@require_torch
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feature_extraction_class(**self.feat_extract_dict )
__lowerCamelCase = self.feat_extract_tester.prepare_inputs_for_target()
__lowerCamelCase = feat_extract.model_input_names[0]
__lowerCamelCase = BatchFeature({input_name: speech_inputs} )
__lowerCamelCase = feat_extract.num_mel_bins # hack!
__lowerCamelCase = feat_extract.pad(_snake_case , padding='''longest''' , return_tensors='''np''' )[input_name]
__lowerCamelCase = feat_extract.pad(_snake_case , padding='''longest''' , return_tensors='''pt''' )[input_name]
self.assertTrue(abs(input_np.astype(np.floataa ).sum() - input_pt.numpy().astype(np.floataa ).sum() ) < 1E-2 )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feat_extract_dict
__lowerCamelCase = True
__lowerCamelCase = self.feature_extraction_class(**_snake_case )
__lowerCamelCase = self.feat_extract_tester.prepare_inputs_for_target()
__lowerCamelCase = [len(_snake_case ) for x in speech_inputs]
__lowerCamelCase = feat_extract.model_input_names[0]
__lowerCamelCase = BatchFeature({input_name: speech_inputs} )
__lowerCamelCase = feat_extract.num_mel_bins # hack!
__lowerCamelCase = feat_extract.pad(_snake_case , padding='''longest''' , return_tensors='''np''' )
self.assertIn('''attention_mask''' , _snake_case )
self.assertListEqual(list(processed.attention_mask.shape ) , list(processed[input_name].shape[:2] ) )
self.assertListEqual(processed.attention_mask.sum(-1 ).tolist() , _snake_case )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = self.feat_extract_dict
__lowerCamelCase = True
__lowerCamelCase = self.feature_extraction_class(**_snake_case )
__lowerCamelCase = self.feat_extract_tester.prepare_inputs_for_target()
__lowerCamelCase = [len(_snake_case ) for x in speech_inputs]
__lowerCamelCase = feat_extract.model_input_names[0]
__lowerCamelCase = BatchFeature({input_name: speech_inputs} )
__lowerCamelCase = min(_snake_case )
__lowerCamelCase = feat_extract.num_mel_bins # hack!
__lowerCamelCase = feat_extract.pad(
_snake_case , padding='''max_length''' , max_length=_snake_case , truncation=_snake_case , return_tensors='''np''' )
self.assertIn('''attention_mask''' , _snake_case )
self.assertListEqual(
list(processed_pad.attention_mask.shape ) , [processed_pad[input_name].shape[0], max_length] )
self.assertListEqual(
processed_pad.attention_mask[:, :max_length].sum(-1 ).tolist() , [max_length for x in speech_inputs] )
def _lowerCamelCase ( self , _snake_case ):
"""simple docstring"""
from datasets import load_dataset
__lowerCamelCase = load_dataset('''hf-internal-testing/librispeech_asr_dummy''' , '''clean''' , split='''validation''' )
# automatic decoding with librispeech
__lowerCamelCase = ds.sort('''id''' ).select(range(_snake_case ) )[:num_samples]['''audio''']
return [x["array"] for x in speech_samples]
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = torch.tensor(
[2.3804E-03, 2.0752E-03, 1.9836E-03, 2.1057E-03, 1.6174E-03,
3.0518E-04, 9.1553E-05, 3.3569E-04, 9.7656E-04, 1.8311E-03,
2.0142E-03, 2.1057E-03, 1.7395E-03, 4.5776E-04, -3.9673E-04,
4.5776E-04, 1.0071E-03, 9.1553E-05, 4.8828E-04, 1.1597E-03,
7.3242E-04, 9.4604E-04, 1.8005E-03, 1.8311E-03, 8.8501E-04,
4.2725E-04, 4.8828E-04, 7.3242E-04, 1.0986E-03, 2.1057E-03] )
# fmt: on
__lowerCamelCase = self._load_datasamples(1 )
__lowerCamelCase = SpeechTaFeatureExtractor()
__lowerCamelCase = feature_extractor(_snake_case , return_tensors='''pt''' ).input_values
self.assertEquals(input_values.shape , (1, 9_36_80) )
self.assertTrue(torch.allclose(input_values[0, :30] , _snake_case , atol=1E-6 ) )
def _lowerCamelCase ( self ):
"""simple docstring"""
__lowerCamelCase = torch.tensor(
[-2.6_8_7_0, -3.0_1_0_4, -3.1_3_5_6, -3.5_3_5_2, -3.0_0_4_4, -3.0_3_5_3, -3.4_7_1_9, -3.6_7_7_7,
-3.1_5_2_0, -2.9_4_3_5, -2.6_5_5_3, -2.8_7_9_5, -2.9_9_4_4, -2.5_9_2_1, -3.0_2_7_9, -3.0_3_8_6,
-3.0_8_6_4, -3.1_2_9_1, -3.2_3_5_3, -2.7_4_4_4, -2.6_8_3_1, -2.7_2_8_7, -3.1_7_6_1, -3.1_5_7_1,
-3.2_7_2_6, -3.0_5_8_2, -3.1_0_0_7, -3.4_5_3_3, -3.4_6_9_5, -3.0_9_9_8] )
# fmt: on
__lowerCamelCase = self._load_datasamples(1 )
__lowerCamelCase = SpeechTaFeatureExtractor()
__lowerCamelCase = feature_extractor(audio_target=_snake_case , return_tensors='''pt''' ).input_values
self.assertEquals(input_values.shape , (1, 3_66, 80) )
self.assertTrue(torch.allclose(input_values[0, 0, :30] , _snake_case , atol=1E-4 ) )
| 575
|
'''simple docstring'''
from typing import List, Optional, Union
import numpy as np
import torch
import torchaudio.compliance.kaldi as ta_kaldi
from ...feature_extraction_sequence_utils import SequenceFeatureExtractor
from ...feature_extraction_utils import BatchFeature
from ...utils import PaddingStrategy, TensorType, logging
_UpperCamelCase : List[Any] =logging.get_logger(__name__)
class _SCREAMING_SNAKE_CASE ( UpperCamelCase ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ = ['input_features', 'attention_mask']
def __init__( self , _snake_case=80 , _snake_case=1_60_00 , _snake_case=80 , _snake_case=0.0 , _snake_case=True , _snake_case=True , _snake_case=True , **_snake_case , ):
"""simple docstring"""
super().__init__(feature_size=_snake_case , sampling_rate=_snake_case , padding_value=_snake_case , **_snake_case )
__lowerCamelCase = num_mel_bins
__lowerCamelCase = do_ceptral_normalize
__lowerCamelCase = normalize_means
__lowerCamelCase = normalize_vars
__lowerCamelCase = True
def _lowerCamelCase ( self , _snake_case , ):
"""simple docstring"""
__lowerCamelCase = waveform * (2**15) # Kaldi compliance: 16-bit signed integers
__lowerCamelCase = torch.from_numpy(_snake_case ).unsqueeze(0 )
__lowerCamelCase = ta_kaldi.fbank(_snake_case , num_mel_bins=self.num_mel_bins , sample_frequency=self.sampling_rate )
return features.numpy()
@staticmethod
def _lowerCamelCase ( _snake_case , _snake_case , _snake_case = True , _snake_case = True , _snake_case = 0.0 , ):
"""simple docstring"""
if normalize_means:
__lowerCamelCase = x[:input_length].mean(axis=0 )
__lowerCamelCase = np.subtract(_snake_case , _snake_case )
if normalize_vars:
__lowerCamelCase = x[:input_length].std(axis=0 )
__lowerCamelCase = np.divide(_snake_case , _snake_case )
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 , _snake_case , _snake_case = 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.utterance_cmvn(_snake_case , _snake_case , self.normalize_means , self.normalize_vars , self.padding_value )
for x, n in zip(_snake_case , _snake_case )
]
def __call__( self , _snake_case , _snake_case = False , _snake_case = None , _snake_case = False , _snake_case = None , _snake_case = None , _snake_case = None , _snake_case = None , **_snake_case , ):
"""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(_snake_case , 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(_snake_case , (list, tuple) ) and (isinstance(raw_speech[0] , (np.ndarray, tuple, list) ))
)
if is_batched:
__lowerCamelCase = [np.asarray(_snake_case , dtype=np.floataa ) for speech in raw_speech]
elif not is_batched and not isinstance(_snake_case , np.ndarray ):
__lowerCamelCase = np.asarray(_snake_case , dtype=np.floataa )
elif isinstance(_snake_case , 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_fbank_features(_snake_case ) for waveform in raw_speech]
# convert into correct format for padding
__lowerCamelCase = BatchFeature({'''input_features''': features} )
__lowerCamelCase = self.pad(
_snake_case , padding=_snake_case , max_length=_snake_case , truncation=_snake_case , pad_to_multiple_of=_snake_case , return_attention_mask=_snake_case , **_snake_case , )
# make sure list is in array format
__lowerCamelCase = padded_inputs.get('''input_features''' )
if isinstance(input_features[0] , _snake_case ):
__lowerCamelCase = [np.asarray(_snake_case , dtype=np.floataa ) for feature in input_features]
__lowerCamelCase = padded_inputs.get('''attention_mask''' )
if attention_mask is not None:
__lowerCamelCase = [np.asarray(_snake_case , dtype=np.intaa ) for array in attention_mask]
# Utterance-level cepstral mean and variance normalization
if self.do_ceptral_normalize:
__lowerCamelCase = (
np.array(_snake_case , dtype=np.intaa )
if self._get_padding_strategies(_snake_case , max_length=_snake_case ) is not PaddingStrategy.DO_NOT_PAD
else None
)
__lowerCamelCase = self.normalize(
padded_inputs['''input_features'''] , attention_mask=_snake_case )
if return_tensors is not None:
__lowerCamelCase = padded_inputs.convert_to_tensors(_snake_case )
return padded_inputs
| 575
| 1
|
from typing import List, Optional
from tokenizers import ByteLevelBPETokenizer
from ...tokenization_utils_fast import PreTrainedTokenizerFast
from ...utils import logging
from .tokenization_blenderbot_small import BlenderbotSmallTokenizer
lowercase_ = logging.get_logger(__name__)
lowercase_ = {
"""vocab_file""": """vocab.json""",
"""merges_file""": """merges.txt""",
"""tokenizer_config_file""": """tokenizer_config.json""",
}
lowercase_ = {
"""vocab_file""": {
"""facebook/blenderbot_small-90M""": """https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/vocab.json"""
},
"""merges_file""": {
"""facebook/blenderbot_small-90M""": """https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/merges.txt"""
},
"""tokenizer_config_file""": {
"""facebook/blenderbot_small-90M""": (
"""https://huggingface.co/facebook/blenderbot_small-90M/resolve/main/tokenizer_config.json"""
)
},
}
lowercase_ = {
"""facebook/blenderbot_small-90M""": 512,
}
class _snake_case ( __SCREAMING_SNAKE_CASE):
UpperCamelCase__ : str =VOCAB_FILES_NAMES
UpperCamelCase__ : str =PRETRAINED_VOCAB_FILES_MAP
UpperCamelCase__ : str =PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
UpperCamelCase__ : Optional[int] =BlenderbotSmallTokenizer
def __init__( self : Optional[int], __lowercase : Optional[int]=None, __lowercase : Optional[int]=None, __lowercase : List[Any]="<|endoftext|>", __lowercase : Union[str, Any]="<|endoftext|>", __lowercase : str="<|endoftext|>", __lowercase : Dict=False, __lowercase : Union[str, Any]=True, **__lowercase : List[Any], ):
super().__init__(
ByteLevelBPETokenizer(
vocab=A_, merges=A_, add_prefix_space=A_, trim_offsets=A_, ), bos_token=A_, eos_token=A_, unk_token=A_, **A_, )
lowercase__ = add_prefix_space
def A__ ( self : Any, __lowercase : int, __lowercase : Dict=None ):
lowercase__ = [self.bos_token_id] + token_ids_a + [self.eos_token_id]
if token_ids_a is None:
return output
return output + [self.eos_token_id] + token_ids_a + [self.eos_token_id]
def A__ ( self : Any, __lowercase : List[int], __lowercase : Optional[List[int]] = None ):
lowercase__ = [self.sep_token_id]
lowercase__ = [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]
| 413
|
import os
import sys
import tempfile
import torch
from .state import AcceleratorState
from .utils import PrecisionType, PrepareForLaunch, is_mps_available, patch_environment
def UpperCAmelCase_ ( _UpperCAmelCase , _UpperCAmelCase=() , _UpperCAmelCase=None , _UpperCAmelCase="no" , _UpperCAmelCase="29500" ):
lowerCamelCase_: List[str] = False
lowerCamelCase_: Dict = False
if any(key.startswith("""KAGGLE""" ) for key in os.environ.keys() ):
lowerCamelCase_: Dict = True
elif "IPython" in sys.modules:
lowerCamelCase_: Dict = """google.colab""" in str(sys.modules["""IPython"""].get_ipython() )
try:
lowerCamelCase_: str = PrecisionType(mixed_precision.lower() )
except ValueError:
raise ValueError(
f"""Unknown mixed_precision mode: {args.mixed_precision.lower()}. Choose between {PrecisionType.list()}.""" )
if (in_colab or in_kaggle) and (os.environ.get("""TPU_NAME""" , _UpperCAmelCase ) is not None):
# TPU launch
import torch_xla.distributed.xla_multiprocessing as xmp
if len(AcceleratorState._shared_state ) > 0:
raise ValueError(
"""To train on TPU in Colab or Kaggle Kernel, the `Accelerator` should only be initialized inside """
"""your training function. Restart your notebook and make sure no cells initializes an """
"""`Accelerator`.""" )
if num_processes is None:
lowerCamelCase_: Optional[Any] = 8
lowerCamelCase_: List[Any] = PrepareForLaunch(_UpperCAmelCase , distributed_type="""TPU""" )
print(f"""Launching a training on {num_processes} TPU cores.""" )
xmp.spawn(_UpperCAmelCase , args=_UpperCAmelCase , nprocs=_UpperCAmelCase , start_method="""fork""" )
elif in_colab:
# No need for a distributed launch otherwise as it's either CPU or one GPU.
if torch.cuda.is_available():
print("""Launching training on one GPU.""" )
else:
print("""Launching training on one CPU.""" )
function(*_UpperCAmelCase )
else:
if num_processes is None:
raise ValueError(
"""You have to specify the number of GPUs you would like to use, add `num_processes=...` to your call.""" )
if num_processes > 1:
# Multi-GPU launch
from torch.multiprocessing import start_processes
from torch.multiprocessing.spawn import ProcessRaisedException
if len(AcceleratorState._shared_state ) > 0:
raise ValueError(
"""To launch a multi-GPU training from your notebook, the `Accelerator` should only be initialized """
"""inside your training function. Restart your notebook and make sure no cells initializes an """
"""`Accelerator`.""" )
if torch.cuda.is_initialized():
raise ValueError(
"""To launch a multi-GPU training from your notebook, you need to avoid running any instruction """
"""using `torch.cuda` in any cell. Restart your notebook and make sure no cells use any CUDA """
"""function.""" )
# torch.distributed will expect a few environment variable to be here. We set the ones common to each
# process here (the other ones will be set be the launcher).
with patch_environment(
world_size=_UpperCAmelCase , master_addr="""127.0.01""" , master_port=_UpperCAmelCase , mixed_precision=_UpperCAmelCase ):
lowerCamelCase_: Tuple = PrepareForLaunch(_UpperCAmelCase , distributed_type="""MULTI_GPU""" )
print(f"""Launching training on {num_processes} GPUs.""" )
try:
start_processes(_UpperCAmelCase , args=_UpperCAmelCase , nprocs=_UpperCAmelCase , start_method="""fork""" )
except ProcessRaisedException as e:
if "Cannot re-initialize CUDA in forked subprocess" in e.args[0]:
raise RuntimeError(
"""CUDA has been initialized before the `notebook_launcher` could create a forked subprocess. """
"""This likely stems from an outside import causing issues once the `notebook_launcher()` is called. """
"""Please review your imports and test them when running the `notebook_launcher()` to identify """
"""which one is problematic.""" ) from e
else:
# No need for a distributed launch otherwise as it's either CPU, GPU or MPS.
if is_mps_available():
lowerCamelCase_: List[Any] = """1"""
print("""Launching training on MPS.""" )
elif torch.cuda.is_available():
print("""Launching training on one GPU.""" )
else:
print("""Launching training on CPU.""" )
function(*_UpperCAmelCase )
def UpperCAmelCase_ ( _UpperCAmelCase , _UpperCAmelCase=() , _UpperCAmelCase=2 ):
from torch.multiprocessing import start_processes
with tempfile.NamedTemporaryFile() as tmp_file:
# torch.distributed will expect a few environment variable to be here. We set the ones common to each
# process here (the other ones will be set be the launcher).
with patch_environment(
world_size=_UpperCAmelCase , master_addr="""127.0.01""" , master_port="""29500""" , accelerate_mixed_precision="""no""" , accelerate_debug_rdv_file=tmp_file.name , accelerate_use_cpu="""yes""" , ):
lowerCamelCase_: Optional[Any] = PrepareForLaunch(_UpperCAmelCase , debug=_UpperCAmelCase )
start_processes(_UpperCAmelCase , args=_UpperCAmelCase , nprocs=_UpperCAmelCase , start_method="""fork""" )
| 423
| 0
|
import warnings
from ...utils import logging
from .image_processing_flava import FlavaImageProcessor
UpperCamelCase : Any = logging.get_logger(__name__)
class A__ ( A__ ):
"""simple docstring"""
def __init__( self : Optional[int] , *lowerCamelCase__ : Optional[int] , **lowerCamelCase__ : int ):
warnings.warn(
"The class FlavaFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please"
" use FlavaImageProcessor instead." , lowerCamelCase__ , )
super().__init__(*lowerCamelCase__ , **lowerCamelCase__ )
| 151
|
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
UpperCamelCase : List[str] = {
"""configuration_blenderbot""": [
"""BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP""",
"""BlenderbotConfig""",
"""BlenderbotOnnxConfig""",
],
"""tokenization_blenderbot""": ["""BlenderbotTokenizer"""],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase : List[Any] = ["""BlenderbotTokenizerFast"""]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase : Optional[int] = [
"""BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST""",
"""BlenderbotForCausalLM""",
"""BlenderbotForConditionalGeneration""",
"""BlenderbotModel""",
"""BlenderbotPreTrainedModel""",
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase : Any = [
"""TFBlenderbotForConditionalGeneration""",
"""TFBlenderbotModel""",
"""TFBlenderbotPreTrainedModel""",
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase : Any = [
"""FlaxBlenderbotForConditionalGeneration""",
"""FlaxBlenderbotModel""",
"""FlaxBlenderbotPreTrainedModel""",
]
if TYPE_CHECKING:
from .configuration_blenderbot import (
BLENDERBOT_PRETRAINED_CONFIG_ARCHIVE_MAP,
BlenderbotConfig,
BlenderbotOnnxConfig,
)
from .tokenization_blenderbot import BlenderbotTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_blenderbot_fast import BlenderbotTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_blenderbot import (
BLENDERBOT_PRETRAINED_MODEL_ARCHIVE_LIST,
BlenderbotForCausalLM,
BlenderbotForConditionalGeneration,
BlenderbotModel,
BlenderbotPreTrainedModel,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_blenderbot import (
TFBlenderbotForConditionalGeneration,
TFBlenderbotModel,
TFBlenderbotPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_blenderbot import (
FlaxBlenderbotForConditionalGeneration,
FlaxBlenderbotModel,
FlaxBlenderbotPreTrainedModel,
)
else:
import sys
UpperCamelCase : Any = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 151
| 1
|
"""simple docstring"""
import unittest
from transformers import is_torch_available
from transformers.testing_utils import require_torch
if is_torch_available():
import torch
from transformers.activations import gelu_new, gelu_python, get_activation
@require_torch
class lowercase__ ( unittest.TestCase ):
"""simple docstring"""
def _a ( self ):
'''simple docstring'''
UpperCamelCase : Optional[int] = torch.tensor([-1_0_0, -1, -0.1, 0, 0.1, 1.0, 1_0_0] )
UpperCamelCase : Optional[int] = get_activation("""gelu""" )
self.assertTrue(torch.allclose(gelu_python(_A ) , torch_builtin(_A ) ) )
self.assertFalse(torch.allclose(gelu_python(_A ) , gelu_new(_A ) ) )
def _a ( self ):
'''simple docstring'''
UpperCamelCase : Optional[int] = torch.tensor([-1_0_0, -1, -0.1, 0, 0.1, 1.0, 1_0_0] )
UpperCamelCase : int = get_activation("""gelu""" )
UpperCamelCase : str = get_activation("""gelu_10""" )
UpperCamelCase : int = torch_builtin(_A )
UpperCamelCase : Any = geluaa(_A )
UpperCamelCase : List[str] = torch.where(y_gelu_aa < 10.0 , 1 , 0 )
self.assertTrue(torch.max(_A ).item() == 10.0 )
self.assertTrue(torch.allclose(y_gelu * clipped_mask , y_gelu_aa * clipped_mask ) )
def _a ( self ):
'''simple docstring'''
get_activation("""gelu""" )
get_activation("""gelu_10""" )
get_activation("""gelu_fast""" )
get_activation("""gelu_new""" )
get_activation("""gelu_python""" )
get_activation("""gelu_pytorch_tanh""" )
get_activation("""linear""" )
get_activation("""mish""" )
get_activation("""quick_gelu""" )
get_activation("""relu""" )
get_activation("""sigmoid""" )
get_activation("""silu""" )
get_activation("""swish""" )
get_activation("""tanh""" )
with self.assertRaises(_A ):
get_activation("""bogus""" )
with self.assertRaises(_A ):
get_activation(_A )
def _a ( self ):
'''simple docstring'''
UpperCamelCase : Union[str, Any] = get_activation("""gelu""" )
UpperCamelCase : Optional[Any] = 1
UpperCamelCase : Union[str, Any] = get_activation("""gelu""" )
self.assertEqual(acta.a , 1 )
with self.assertRaises(_A ):
UpperCamelCase : List[Any] = acta.a
| 102
|
from argparse import ArgumentParser
from .add_new_model import AddNewModelCommand
from .add_new_model_like import AddNewModelLikeCommand
from .convert import ConvertCommand
from .download import DownloadCommand
from .env import EnvironmentCommand
from .lfs import LfsCommands
from .pt_to_tf import PTtoTFCommand
from .run import RunCommand
from .serving import ServeCommand
from .user import UserCommands
def _lowerCAmelCase ( ):
lowercase__ = ArgumentParser('Transformers CLI tool' , usage='transformers-cli <command> [<args>]' )
lowercase__ = parser.add_subparsers(help='transformers-cli command helpers' )
# Register commands
ConvertCommand.register_subcommand(A__ )
DownloadCommand.register_subcommand(A__ )
EnvironmentCommand.register_subcommand(A__ )
RunCommand.register_subcommand(A__ )
ServeCommand.register_subcommand(A__ )
UserCommands.register_subcommand(A__ )
AddNewModelCommand.register_subcommand(A__ )
AddNewModelLikeCommand.register_subcommand(A__ )
LfsCommands.register_subcommand(A__ )
PTtoTFCommand.register_subcommand(A__ )
# Let's go
lowercase__ = parser.parse_args()
if not hasattr(A__ , 'func' ):
parser.print_help()
exit(1 )
# Run
lowercase__ = args.func(A__ )
service.run()
if __name__ == "__main__":
main()
| 622
| 0
|
'''simple docstring'''
import argparse
import os
import sys
from unittest.mock import patch
import pytorch_lightning as pl
import timeout_decorator
import torch
from distillation import SummarizationDistiller, distill_main
from finetune import SummarizationModule, main
from transformers import MarianMTModel
from transformers.file_utils import cached_path
from transformers.testing_utils import TestCasePlus, require_torch_gpu, slow
from utils import load_json
UpperCAmelCase_ : str = 'sshleifer/mar_enro_6_3_student'
class a ( snake_case__ ):
'''simple docstring'''
def __UpperCamelCase ( self ) -> str:
super().setUp()
_a : List[str] = cached_path(
'https://cdn-datasets.huggingface.co/translation/wmt_en_ro-tr40k-va0.5k-te0.5k.tar.gz' , extract_compressed_file=UpperCAmelCase__ , )
_a : Optional[Any] = F'''{data_cached}/wmt_en_ro-tr40k-va0.5k-te0.5k'''
@slow
@require_torch_gpu
def __UpperCamelCase ( self ) -> Dict:
MarianMTModel.from_pretrained(UpperCAmelCase__ )
@slow
@require_torch_gpu
def __UpperCamelCase ( self ) -> Any:
_a : Union[str, Any] = {
'''$MAX_LEN''': 6_4,
'''$BS''': 6_4,
'''$GAS''': 1,
'''$ENRO_DIR''': self.data_dir,
'''facebook/mbart-large-cc25''': MARIAN_MODEL,
# "val_check_interval=0.25": "val_check_interval=1.0",
'''--learning_rate=3e-5''': '''--learning_rate 3e-4''',
'''--num_train_epochs 6''': '''--num_train_epochs 1''',
}
# Clean up bash script
_a : List[Any] = (self.test_file_dir / '''train_mbart_cc25_enro.sh''').open().read().split('finetune.py' )[1].strip()
_a : Dict = bash_script.replace('\\\n' , '' ).strip().replace('"$@"' , '' )
for k, v in env_vars_to_replace.items():
_a : Tuple = bash_script.replace(UpperCAmelCase__ , str(UpperCAmelCase__ ) )
_a : Tuple = self.get_auto_remove_tmp_dir()
# bash_script = bash_script.replace("--fp16 ", "")
_a : Union[str, Any] = F'''\n --output_dir {output_dir}\n --tokenizer_name Helsinki-NLP/opus-mt-en-ro\n --sortish_sampler\n --do_predict\n --gpus 1\n --freeze_encoder\n --n_train 40000\n --n_val 500\n --n_test 500\n --fp16_opt_level O1\n --num_sanity_val_steps 0\n --eval_beams 2\n '''.split()
# XXX: args.gpus > 1 : handle multi_gpu in the future
_a : str = ['''finetune.py'''] + bash_script.split() + args
with patch.object(UpperCAmelCase__ , 'argv' , UpperCAmelCase__ ):
_a : str = argparse.ArgumentParser()
_a : List[str] = pl.Trainer.add_argparse_args(UpperCAmelCase__ )
_a : Dict = SummarizationModule.add_model_specific_args(UpperCAmelCase__ , os.getcwd() )
_a : List[str] = parser.parse_args()
_a : str = main(UpperCAmelCase__ )
# Check metrics
_a : Tuple = load_json(model.metrics_save_path )
_a : Optional[int] = metrics['''val'''][0]
_a : List[str] = metrics['''val'''][-1]
self.assertEqual(len(metrics['val'] ) , (args.max_epochs / args.val_check_interval) )
assert isinstance(last_step_stats[F'''val_avg_{model.val_metric}'''] , UpperCAmelCase__ )
self.assertGreater(last_step_stats['val_avg_gen_time'] , 0.01 )
# model hanging on generate. Maybe bad config was saved. (XXX: old comment/assert?)
self.assertLessEqual(last_step_stats['val_avg_gen_time'] , 1.0 )
# test learning requirements:
# 1. BLEU improves over the course of training by more than 2 pts
self.assertGreater(last_step_stats['val_avg_bleu'] - first_step_stats['val_avg_bleu'] , 2 )
# 2. BLEU finishes above 17
self.assertGreater(last_step_stats['val_avg_bleu'] , 1_7 )
# 3. test BLEU and val BLEU within ~1.1 pt.
self.assertLess(abs(metrics['val'][-1]['val_avg_bleu'] - metrics['test'][-1]['test_avg_bleu'] ) , 1.1 )
# check lightning ckpt can be loaded and has a reasonable statedict
_a : Dict = os.listdir(UpperCAmelCase__ )
_a : Any = [x for x in contents if x.endswith('.ckpt' )][0]
_a : List[str] = os.path.join(args.output_dir , UpperCAmelCase__ )
_a : str = torch.load(UpperCAmelCase__ , map_location='cpu' )
_a : Optional[int] = '''model.model.decoder.layers.0.encoder_attn_layer_norm.weight'''
assert expected_key in ckpt["state_dict"]
assert ckpt["state_dict"]["model.model.decoder.layers.0.encoder_attn_layer_norm.weight"].dtype == torch.floataa
# TODO: turn on args.do_predict when PL bug fixed.
if args.do_predict:
_a : int = {os.path.basename(UpperCAmelCase__ ) for p in contents}
assert "test_generations.txt" in contents
assert "test_results.txt" in contents
# assert len(metrics["val"]) == desired_n_evals
assert len(metrics['test'] ) == 1
class a ( snake_case__ ):
'''simple docstring'''
@timeout_decorator.timeout(6_0_0 )
@slow
@require_torch_gpu
def __UpperCamelCase ( self ) -> int:
_a : Tuple = F'''{self.test_file_dir_str}/test_data/wmt_en_ro'''
_a : Tuple = {
'''--fp16_opt_level=O1''': '''''',
'''$MAX_LEN''': 1_2_8,
'''$BS''': 1_6,
'''$GAS''': 1,
'''$ENRO_DIR''': data_dir,
'''$m''': '''sshleifer/student_marian_en_ro_6_1''',
'''val_check_interval=0.25''': '''val_check_interval=1.0''',
}
# Clean up bash script
_a : str = (
(self.test_file_dir / '''distil_marian_no_teacher.sh''').open().read().split('distillation.py' )[1].strip()
)
_a : Tuple = bash_script.replace('\\\n' , '' ).strip().replace('"$@"' , '' )
_a : Dict = bash_script.replace('--fp16 ' , ' ' )
for k, v in env_vars_to_replace.items():
_a : Optional[int] = bash_script.replace(UpperCAmelCase__ , str(UpperCAmelCase__ ) )
_a : Any = self.get_auto_remove_tmp_dir()
_a : Optional[Any] = bash_script.replace('--fp16' , '' )
_a : str = 6
_a : Dict = (
['''distillation.py''']
+ bash_script.split()
+ [
F'''--output_dir={output_dir}''',
'''--gpus=1''',
'''--learning_rate=1e-3''',
F'''--num_train_epochs={epochs}''',
'''--warmup_steps=10''',
'''--val_check_interval=1.0''',
'''--do_predict''',
]
)
with patch.object(UpperCAmelCase__ , 'argv' , UpperCAmelCase__ ):
_a : int = argparse.ArgumentParser()
_a : Optional[int] = pl.Trainer.add_argparse_args(UpperCAmelCase__ )
_a : List[Any] = SummarizationDistiller.add_model_specific_args(UpperCAmelCase__ , os.getcwd() )
_a : int = parser.parse_args()
# assert args.gpus == gpus THIS BREAKS for multi_gpu
_a : Any = distill_main(UpperCAmelCase__ )
# Check metrics
_a : Optional[Any] = load_json(model.metrics_save_path )
_a : Any = metrics['''val'''][0]
_a : int = metrics['''val'''][-1]
assert len(metrics['val'] ) >= (args.max_epochs / args.val_check_interval) # +1 accounts for val_sanity_check
assert last_step_stats["val_avg_gen_time"] >= 0.01
assert first_step_stats["val_avg_bleu"] < last_step_stats["val_avg_bleu"] # model learned nothing
assert 1.0 >= last_step_stats["val_avg_gen_time"] # model hanging on generate. Maybe bad config was saved.
assert isinstance(last_step_stats[F'''val_avg_{model.val_metric}'''] , UpperCAmelCase__ )
# check lightning ckpt can be loaded and has a reasonable statedict
_a : List[str] = os.listdir(UpperCAmelCase__ )
_a : int = [x for x in contents if x.endswith('.ckpt' )][0]
_a : str = os.path.join(args.output_dir , UpperCAmelCase__ )
_a : Any = torch.load(UpperCAmelCase__ , map_location='cpu' )
_a : Any = '''model.model.decoder.layers.0.encoder_attn_layer_norm.weight'''
assert expected_key in ckpt["state_dict"]
assert ckpt["state_dict"]["model.model.decoder.layers.0.encoder_attn_layer_norm.weight"].dtype == torch.floataa
# TODO: turn on args.do_predict when PL bug fixed.
if args.do_predict:
_a : List[Any] = {os.path.basename(UpperCAmelCase__ ) for p in contents}
assert "test_generations.txt" in contents
assert "test_results.txt" in contents
# assert len(metrics["val"]) == desired_n_evals
assert len(metrics['test'] ) == 1
| 718
|
'''simple docstring'''
import unittest
from transformers import is_torch_available
from transformers.testing_utils import require_torch, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST,
OpenAIGPTConfig,
OpenAIGPTDoubleHeadsModel,
OpenAIGPTForSequenceClassification,
OpenAIGPTLMHeadModel,
OpenAIGPTModel,
)
class a :
'''simple docstring'''
def __init__( self , lowerCamelCase_ , lowerCamelCase_=1_3 , lowerCamelCase_=7 , lowerCamelCase_=True , lowerCamelCase_=True , lowerCamelCase_=True , lowerCamelCase_=9_9 , lowerCamelCase_=3_2 , lowerCamelCase_=5 , lowerCamelCase_=4 , lowerCamelCase_=3_7 , lowerCamelCase_="gelu" , lowerCamelCase_=0.1 , lowerCamelCase_=0.1 , lowerCamelCase_=5_1_2 , lowerCamelCase_=1_6 , lowerCamelCase_=2 , lowerCamelCase_=0.02 , lowerCamelCase_=3 , lowerCamelCase_=4 , lowerCamelCase_=None , ) -> Optional[Any]:
_a : List[Any] = parent
_a : Optional[int] = batch_size
_a : List[Any] = seq_length
_a : Tuple = is_training
_a : List[str] = use_token_type_ids
_a : Dict = use_labels
_a : Optional[Any] = vocab_size
_a : Tuple = hidden_size
_a : int = num_hidden_layers
_a : Optional[Any] = num_attention_heads
_a : Optional[int] = intermediate_size
_a : Union[str, Any] = hidden_act
_a : Tuple = hidden_dropout_prob
_a : Dict = attention_probs_dropout_prob
_a : Optional[Any] = max_position_embeddings
_a : Union[str, Any] = type_vocab_size
_a : List[Any] = type_sequence_label_size
_a : Dict = initializer_range
_a : Any = num_labels
_a : Dict = num_choices
_a : Optional[int] = scope
_a : Optional[int] = self.vocab_size - 1
def __UpperCamelCase ( self ) -> int:
_a : Optional[Any] = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
_a : int = None
if self.use_token_type_ids:
_a : Optional[int] = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
_a : Union[str, Any] = None
_a : Union[str, Any] = None
_a : int = None
if self.use_labels:
_a : Optional[int] = ids_tensor([self.batch_size] , self.type_sequence_label_size )
_a : List[Any] = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
_a : Tuple = ids_tensor([self.batch_size] , self.num_choices )
_a : Optional[int] = OpenAIGPTConfig(
vocab_size=self.vocab_size , n_embd=self.hidden_size , n_layer=self.num_hidden_layers , n_head=self.num_attention_heads , n_positions=self.max_position_embeddings , pad_token_id=self.pad_token_id , )
_a : Union[str, Any] = ids_tensor([self.num_hidden_layers, self.num_attention_heads] , 2 )
return (
config,
input_ids,
head_mask,
token_type_ids,
sequence_labels,
token_labels,
choice_labels,
)
def __UpperCamelCase ( self , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , *lowerCamelCase_ ) -> int:
_a : Union[str, Any] = OpenAIGPTModel(config=lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
_a : Union[str, Any] = model(lowerCamelCase_ , token_type_ids=lowerCamelCase_ , head_mask=lowerCamelCase_ )
_a : Union[str, Any] = model(lowerCamelCase_ , token_type_ids=lowerCamelCase_ )
_a : Optional[int] = model(lowerCamelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def __UpperCamelCase ( self , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , *lowerCamelCase_ ) -> Optional[Any]:
_a : int = OpenAIGPTLMHeadModel(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
_a : Union[str, Any] = model(lowerCamelCase_ , token_type_ids=lowerCamelCase_ , labels=lowerCamelCase_ )
self.parent.assertEqual(result.loss.shape , () )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def __UpperCamelCase ( self , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , *lowerCamelCase_ ) -> str:
_a : int = OpenAIGPTDoubleHeadsModel(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
_a : Any = model(lowerCamelCase_ , token_type_ids=lowerCamelCase_ , labels=lowerCamelCase_ )
self.parent.assertEqual(result.loss.shape , () )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def __UpperCamelCase ( self , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , *lowerCamelCase_ ) -> List[Any]:
_a : Optional[int] = self.num_labels
_a : int = OpenAIGPTForSequenceClassification(lowerCamelCase_ )
model.to(lowerCamelCase_ )
model.eval()
_a : int = ids_tensor([self.batch_size] , self.type_sequence_label_size )
_a : List[str] = model(lowerCamelCase_ , token_type_ids=lowerCamelCase_ , labels=lowerCamelCase_ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def __UpperCamelCase ( self ) -> Union[str, Any]:
_a : Optional[int] = self.prepare_config_and_inputs()
(
(
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) ,
) : List[str] = config_and_inputs
_a : List[str] = {
'input_ids': input_ids,
'token_type_ids': token_type_ids,
'head_mask': head_mask,
}
return config, inputs_dict
@require_torch
class a ( snake_case__ , snake_case__ , snake_case__ , unittest.TestCase ):
'''simple docstring'''
__lowerCAmelCase : Tuple = (
(OpenAIGPTModel, OpenAIGPTLMHeadModel, OpenAIGPTDoubleHeadsModel, OpenAIGPTForSequenceClassification)
if is_torch_available()
else ()
)
__lowerCAmelCase : List[str] = (
(OpenAIGPTLMHeadModel,) if is_torch_available() else ()
) # TODO (PVP): Add Double HeadsModel when generate() function is changed accordingly
__lowerCAmelCase : Optional[int] = (
{
"""feature-extraction""": OpenAIGPTModel,
"""text-classification""": OpenAIGPTForSequenceClassification,
"""text-generation""": OpenAIGPTLMHeadModel,
"""zero-shot""": OpenAIGPTForSequenceClassification,
}
if is_torch_available()
else {}
)
def __UpperCamelCase ( self , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_ ) -> str:
if pipeline_test_casse_name == "ZeroShotClassificationPipelineTests":
# Get `tokenizer does not have a padding token` error for both fast/slow tokenizers.
# `OpenAIGPTConfig` was never used in pipeline tests, either because of a missing checkpoint or because a
# tiny config could not be created.
return True
return False
def __UpperCamelCase ( self , lowerCamelCase_ , lowerCamelCase_ , lowerCamelCase_=False ) -> Optional[Any]:
_a : Optional[Any] = super()._prepare_for_class(lowerCamelCase_ , lowerCamelCase_ , return_labels=lowerCamelCase_ )
if return_labels:
if model_class.__name__ == "OpenAIGPTDoubleHeadsModel":
_a : str = torch.zeros(
(self.model_tester.batch_size, self.model_tester.num_choices, self.model_tester.seq_length) , dtype=torch.long , device=lowerCamelCase_ , )
_a : Optional[Any] = inputs_dict['labels']
_a : List[Any] = inputs_dict['labels']
_a : Union[str, Any] = torch.zeros(
(self.model_tester.batch_size, self.model_tester.num_choices) , dtype=torch.long , device=lowerCamelCase_ , )
_a : List[Any] = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=lowerCamelCase_ )
return inputs_dict
def __UpperCamelCase ( self ) -> Dict:
_a : Any = OpenAIGPTModelTester(self )
_a : Dict = ConfigTester(self , config_class=lowerCamelCase_ , n_embd=3_7 )
def __UpperCamelCase ( self ) -> Optional[int]:
self.config_tester.run_common_tests()
def __UpperCamelCase ( self ) -> Union[str, Any]:
_a : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_openai_gpt_model(*lowerCamelCase_ )
def __UpperCamelCase ( self ) -> int:
_a : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_lm_head_model(*lowerCamelCase_ )
def __UpperCamelCase ( self ) -> int:
_a : Optional[int] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_double_lm_head_model(*lowerCamelCase_ )
def __UpperCamelCase ( self ) -> Union[str, Any]:
_a : Any = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_openai_gpt_for_sequence_classification(*lowerCamelCase_ )
@slow
def __UpperCamelCase ( self ) -> Optional[Any]:
for model_name in OPENAI_GPT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
_a : Union[str, Any] = OpenAIGPTModel.from_pretrained(lowerCamelCase_ )
self.assertIsNotNone(lowerCamelCase_ )
@require_torch
class a ( unittest.TestCase ):
'''simple docstring'''
@slow
def __UpperCamelCase ( self ) -> str:
_a : int = OpenAIGPTLMHeadModel.from_pretrained('openai-gpt' )
model.to(lowerCamelCase_ )
_a : Tuple = torch.tensor([[4_8_1, 4_7_3_5, 5_4_4]] , dtype=torch.long , device=lowerCamelCase_ ) # the president is
_a : str = [
4_8_1,
4_7_3_5,
5_4_4,
2_4_6,
9_6_3,
8_7_0,
7_6_2,
2_3_9,
2_4_4,
4_0_4_7_7,
2_4_4,
2_4_9,
7_1_9,
8_8_1,
4_8_7,
5_4_4,
2_4_0,
2_4_4,
6_0_3,
4_8_1,
] # the president is a very good man. " \n " i\'m sure he is, " said the
_a : Union[str, Any] = model.generate(lowerCamelCase_ , do_sample=lowerCamelCase_ )
self.assertListEqual(output_ids[0].tolist() , lowerCamelCase_ )
| 424
| 0
|
from tempfile import TemporaryDirectory
from unittest import TestCase
from unittest.mock import MagicMock, patch
from transformers import AutoModel, TFAutoModel
from transformers.onnx import FeaturesManager
from transformers.testing_utils import SMALL_MODEL_IDENTIFIER, require_tf, require_torch
@require_torch
@require_tf
class lowerCamelCase__ ( _UpperCamelCase ):
def SCREAMING_SNAKE_CASE__ ( self : Dict ):
_lowerCAmelCase = SMALL_MODEL_IDENTIFIER
_lowerCAmelCase = """pt"""
_lowerCAmelCase = """tf"""
def SCREAMING_SNAKE_CASE__ ( self : Any , lowercase__ : List[Any] ):
_lowerCAmelCase = AutoModel.from_pretrained(self.test_model )
model_pt.save_pretrained(lowercase__ )
def SCREAMING_SNAKE_CASE__ ( self : List[Any] , lowercase__ : Any ):
_lowerCAmelCase = TFAutoModel.from_pretrained(self.test_model , from_pt=lowercase__ )
model_tf.save_pretrained(lowercase__ )
def SCREAMING_SNAKE_CASE__ ( self : str ):
_lowerCAmelCase = """mock_framework"""
# Framework provided - return whatever the user provides
_lowerCAmelCase = FeaturesManager.determine_framework(self.test_model , lowercase__ )
self.assertEqual(lowercase__ , lowercase__ )
# Local checkpoint and framework provided - return provided framework
# PyTorch checkpoint
with TemporaryDirectory() as local_pt_ckpt:
self._setup_pt_ckpt(lowercase__ )
_lowerCAmelCase = FeaturesManager.determine_framework(lowercase__ , lowercase__ )
self.assertEqual(lowercase__ , lowercase__ )
# TensorFlow checkpoint
with TemporaryDirectory() as local_tf_ckpt:
self._setup_tf_ckpt(lowercase__ )
_lowerCAmelCase = FeaturesManager.determine_framework(lowercase__ , lowercase__ )
self.assertEqual(lowercase__ , lowercase__ )
def SCREAMING_SNAKE_CASE__ ( self : Optional[Any] ):
with TemporaryDirectory() as local_pt_ckpt:
self._setup_pt_ckpt(lowercase__ )
_lowerCAmelCase = FeaturesManager.determine_framework(lowercase__ )
self.assertEqual(lowercase__ , self.framework_pt )
# TensorFlow checkpoint
with TemporaryDirectory() as local_tf_ckpt:
self._setup_tf_ckpt(lowercase__ )
_lowerCAmelCase = FeaturesManager.determine_framework(lowercase__ )
self.assertEqual(lowercase__ , self.framework_tf )
# Invalid local checkpoint
with TemporaryDirectory() as local_invalid_ckpt:
with self.assertRaises(lowercase__ ):
_lowerCAmelCase = FeaturesManager.determine_framework(lowercase__ )
def SCREAMING_SNAKE_CASE__ ( self : Dict ):
_lowerCAmelCase = MagicMock(return_value=lowercase__ )
with patch('transformers.onnx.features.is_tf_available' , lowercase__ ):
_lowerCAmelCase = FeaturesManager.determine_framework(self.test_model )
self.assertEqual(lowercase__ , self.framework_pt )
# PyTorch not in environment -> use TensorFlow
_lowerCAmelCase = MagicMock(return_value=lowercase__ )
with patch('transformers.onnx.features.is_torch_available' , lowercase__ ):
_lowerCAmelCase = FeaturesManager.determine_framework(self.test_model )
self.assertEqual(lowercase__ , self.framework_tf )
# Both in environment -> use PyTorch
_lowerCAmelCase = MagicMock(return_value=lowercase__ )
_lowerCAmelCase = MagicMock(return_value=lowercase__ )
with patch('transformers.onnx.features.is_tf_available' , lowercase__ ), patch(
'transformers.onnx.features.is_torch_available' , lowercase__ ):
_lowerCAmelCase = FeaturesManager.determine_framework(self.test_model )
self.assertEqual(lowercase__ , self.framework_pt )
# Both not in environment -> raise error
_lowerCAmelCase = MagicMock(return_value=lowercase__ )
_lowerCAmelCase = MagicMock(return_value=lowercase__ )
with patch('transformers.onnx.features.is_tf_available' , lowercase__ ), patch(
'transformers.onnx.features.is_torch_available' , lowercase__ ):
with self.assertRaises(lowercase__ ):
_lowerCAmelCase = FeaturesManager.determine_framework(self.test_model )
| 192
|
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 SCREAMING_SNAKE_CASE_ ( _UpperCamelCase ):
"""simple docstring"""
__magic_name__ : Dict = 'linear'
__magic_name__ : Dict = 'cosine'
__magic_name__ : Optional[int] = 'cosine_with_restarts'
__magic_name__ : List[str] = 'polynomial'
__magic_name__ : Any = 'constant'
__magic_name__ : Union[str, Any] = 'constant_with_warmup'
__magic_name__ : str = 'piecewise_constant'
def A__ (snake_case : Optimizer , snake_case : int = -1 ) -> Optional[Any]:
return LambdaLR(snake_case , lambda snake_case : 1 , last_epoch=snake_case )
def A__ (snake_case : Optimizer , snake_case : int , snake_case : int = -1 ) -> List[Any]:
def lr_lambda(snake_case : int ):
if current_step < num_warmup_steps:
return float(snake_case ) / float(max(1.0 , snake_case ) )
return 1.0
return LambdaLR(snake_case , snake_case , last_epoch=snake_case )
def A__ (snake_case : Optimizer , snake_case : str , snake_case : int = -1 ) -> Union[str, Any]:
__UpperCamelCase : Optional[Any] = {}
__UpperCamelCase : int = step_rules.split(""",""" )
for rule_str in rule_list[:-1]:
__UpperCamelCase , __UpperCamelCase : Tuple = rule_str.split(""":""" )
__UpperCamelCase : int = int(snake_case )
__UpperCamelCase : Union[str, Any] = float(snake_case )
__UpperCamelCase : Optional[int] = value
__UpperCamelCase : Dict = float(rule_list[-1] )
def create_rules_function(snake_case : List[str] , snake_case : Any ):
def rule_func(snake_case : int ) -> float:
__UpperCamelCase : Any = sorted(rules_dict.keys() )
for i, sorted_step in enumerate(snake_case ):
if steps < sorted_step:
return rules_dict[sorted_steps[i]]
return last_lr_multiple
return rule_func
__UpperCamelCase : Tuple = create_rules_function(snake_case , snake_case )
return LambdaLR(snake_case , snake_case , last_epoch=snake_case )
def A__ (snake_case : int , snake_case : Optional[Any] , snake_case : Union[str, Any] , snake_case : str=-1 ) -> str:
def lr_lambda(snake_case : int ):
if current_step < num_warmup_steps:
return float(snake_case ) / float(max(1 , snake_case ) )
return max(
0.0 , float(num_training_steps - current_step ) / float(max(1 , num_training_steps - num_warmup_steps ) ) )
return LambdaLR(snake_case , snake_case , snake_case )
def A__ (snake_case : Optimizer , snake_case : int , snake_case : int , snake_case : float = 0.5 , snake_case : int = -1 ) -> List[str]:
def lr_lambda(snake_case : Dict ):
if current_step < num_warmup_steps:
return float(snake_case ) / float(max(1 , snake_case ) )
__UpperCamelCase : Dict = 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(snake_case ) * 2.0 * progress )) )
return LambdaLR(snake_case , snake_case , snake_case )
def A__ (snake_case : Optimizer , snake_case : int , snake_case : int , snake_case : int = 1 , snake_case : int = -1 ) -> Tuple:
def lr_lambda(snake_case : Optional[int] ):
if current_step < num_warmup_steps:
return float(snake_case ) / float(max(1 , snake_case ) )
__UpperCamelCase : List[str] = 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(snake_case ) * progress) % 1.0) )) )
return LambdaLR(snake_case , snake_case , snake_case )
def A__ (snake_case : List[str] , snake_case : Optional[Any] , snake_case : Optional[int] , snake_case : str=1e-7 , snake_case : List[str]=1.0 , snake_case : Dict=-1 ) -> Tuple:
__UpperCamelCase : Tuple = 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(snake_case : int ):
if current_step < num_warmup_steps:
return float(snake_case ) / float(max(1 , snake_case ) )
elif current_step > num_training_steps:
return lr_end / lr_init # as LambdaLR multiplies by lr_init
else:
__UpperCamelCase : List[str] = lr_init - lr_end
__UpperCamelCase : Any = num_training_steps - num_warmup_steps
__UpperCamelCase : List[str] = 1 - (current_step - num_warmup_steps) / decay_steps
__UpperCamelCase : List[Any] = lr_range * pct_remaining**power + lr_end
return decay / lr_init # as LambdaLR multiplies by lr_init
return LambdaLR(snake_case , snake_case , snake_case )
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__ (snake_case : Union[str, SchedulerType] , snake_case : Optimizer , snake_case : Optional[str] = None , snake_case : Optional[int] = None , snake_case : Optional[int] = None , snake_case : int = 1 , snake_case : float = 1.0 , snake_case : int = -1 , ) -> Dict:
__UpperCamelCase : List[str] = SchedulerType(snake_case )
__UpperCamelCase : int = TYPE_TO_SCHEDULER_FUNCTION[name]
if name == SchedulerType.CONSTANT:
return schedule_func(snake_case , last_epoch=snake_case )
if name == SchedulerType.PIECEWISE_CONSTANT:
return schedule_func(snake_case , step_rules=snake_case , last_epoch=snake_case )
# 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(snake_case , num_warmup_steps=snake_case , last_epoch=snake_case )
# 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(
snake_case , num_warmup_steps=snake_case , num_training_steps=snake_case , num_cycles=snake_case , last_epoch=snake_case , )
if name == SchedulerType.POLYNOMIAL:
return schedule_func(
snake_case , num_warmup_steps=snake_case , num_training_steps=snake_case , power=snake_case , last_epoch=snake_case , )
return schedule_func(
snake_case , num_warmup_steps=snake_case , num_training_steps=snake_case , last_epoch=snake_case )
| 279
| 0
|
"""simple docstring"""
import unittest
from transformers import MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING, AutoTokenizer, is_vision_available
from transformers.pipelines import pipeline
from transformers.pipelines.document_question_answering import apply_tesseract
from transformers.testing_utils import (
is_pipeline_test,
nested_simplify,
require_detectrona,
require_pytesseract,
require_tf,
require_torch,
require_vision,
slow,
)
from .test_pipelines_common import ANY
if is_vision_available():
from PIL import Image
from transformers.image_utils import load_image
else:
class __UpperCAmelCase :
"""simple docstring"""
@staticmethod
def A ( *A_ : str , **A_ : int )-> Tuple:
pass
def lowercase (_snake_case ) -> Dict:
'''simple docstring'''
return None
# This is a pinned image from a specific revision of a document question answering space, hosted by HuggingFace,
# so we can expect it to be available.
_A = (
"https://huggingface.co/spaces/impira/docquery/resolve/2f6c96314dc84dfda62d40de9da55f2f5165d403/invoice.png"
)
@is_pipeline_test
@require_torch
@require_vision
class __UpperCAmelCase ( unittest.TestCase ):
"""simple docstring"""
_snake_case : Optional[int] = MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING
@require_pytesseract
@require_vision
def A ( self : List[str] , A_ : List[str] , A_ : Union[str, Any] , A_ : str )-> Optional[Any]:
__UpperCamelCase = pipeline(
"document-question-answering" , model=A_ , tokenizer=A_ , image_processor=A_ )
__UpperCamelCase = INVOICE_URL
__UpperCamelCase = list(zip(*apply_tesseract(load_image(A_ ) , A_ , "" ) ) )
__UpperCamelCase = "What is the placebo?"
__UpperCamelCase = [
{
"image": load_image(A_ ),
"question": question,
},
{
"image": image,
"question": question,
},
{
"image": image,
"question": question,
"word_boxes": word_boxes,
},
]
return dqa_pipeline, examples
def A ( self : Optional[Any] , A_ : int , A_ : Optional[Any] )-> Tuple:
__UpperCamelCase = dqa_pipeline(A_ , top_k=2 )
self.assertEqual(
A_ , [
[
{"score": ANY(A_ ), "answer": ANY(A_ ), "start": ANY(A_ ), "end": ANY(A_ )},
{"score": ANY(A_ ), "answer": ANY(A_ ), "start": ANY(A_ ), "end": ANY(A_ )},
]
]
* 3 , )
@require_torch
@require_detectrona
@require_pytesseract
def A ( self : List[Any] )-> Union[str, Any]:
__UpperCamelCase = pipeline("document-question-answering" , model="hf-internal-testing/tiny-random-layoutlmv2" )
__UpperCamelCase = INVOICE_URL
__UpperCamelCase = "How many cats are there?"
__UpperCamelCase = [
{"score": 0.0_001, "answer": "oy 2312/2019", "start": 38, "end": 39},
{"score": 0.0_001, "answer": "oy 2312/2019 DUE", "start": 38, "end": 40},
]
__UpperCamelCase = dqa_pipeline(image=A_ , question=A_ , top_k=2 )
self.assertEqual(nested_simplify(A_ , decimals=4 ) , A_ )
__UpperCamelCase = dqa_pipeline({"image": image, "question": question} , top_k=2 )
self.assertEqual(nested_simplify(A_ , decimals=4 ) , A_ )
# This image does not detect ANY text in it, meaning layoutlmv2 should fail.
# Empty answer probably
__UpperCamelCase = "./tests/fixtures/tests_samples/COCO/000000039769.png"
__UpperCamelCase = dqa_pipeline(image=A_ , question=A_ , top_k=2 )
self.assertEqual(A_ , [] )
# We can optionnally pass directly the words and bounding boxes
__UpperCamelCase = "./tests/fixtures/tests_samples/COCO/000000039769.png"
__UpperCamelCase = []
__UpperCamelCase = []
__UpperCamelCase = dqa_pipeline(image=A_ , question=A_ , words=A_ , boxes=A_ , top_k=2 )
self.assertEqual(A_ , [] )
@slow
@require_torch
@require_detectrona
@require_pytesseract
def A ( self : Dict )-> Union[str, Any]:
__UpperCamelCase = pipeline(
"document-question-answering" , model="tiennvcs/layoutlmv2-base-uncased-finetuned-docvqa" , revision="9977165" , )
__UpperCamelCase = INVOICE_URL
__UpperCamelCase = "What is the invoice number?"
__UpperCamelCase = dqa_pipeline(image=A_ , question=A_ , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.9_944, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.0_009, "answer": "us-001", "start": 16, "end": 16},
] , )
__UpperCamelCase = dqa_pipeline({"image": image, "question": question} , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.9_944, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.0_009, "answer": "us-001", "start": 16, "end": 16},
] , )
__UpperCamelCase = dqa_pipeline(
[{"image": image, "question": question}, {"image": image, "question": question}] , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
[
{"score": 0.9_944, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.0_009, "answer": "us-001", "start": 16, "end": 16},
],
]
* 2 , )
@slow
@require_torch
@require_detectrona
@require_pytesseract
def A ( self : Optional[int] )-> Optional[int]:
__UpperCamelCase = pipeline(
"document-question-answering" , model="tiennvcs/layoutlmv2-base-uncased-finetuned-docvqa" , revision="9977165" , max_seq_len=50 , )
__UpperCamelCase = INVOICE_URL
__UpperCamelCase = "What is the invoice number?"
__UpperCamelCase = dqa_pipeline(image=A_ , question=A_ , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.9_974, "answer": "1110212019", "start": 23, "end": 23},
{"score": 0.9_948, "answer": "us-001", "start": 16, "end": 16},
] , )
__UpperCamelCase = dqa_pipeline({"image": image, "question": question} , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.9_974, "answer": "1110212019", "start": 23, "end": 23},
{"score": 0.9_948, "answer": "us-001", "start": 16, "end": 16},
] , )
__UpperCamelCase = dqa_pipeline(
[{"image": image, "question": question}, {"image": image, "question": question}] , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
[
{"score": 0.9_974, "answer": "1110212019", "start": 23, "end": 23},
{"score": 0.9_948, "answer": "us-001", "start": 16, "end": 16},
]
]
* 2 , )
@slow
@require_torch
@require_pytesseract
@require_vision
def A ( self : Union[str, Any] )-> Optional[int]:
__UpperCamelCase = AutoTokenizer.from_pretrained(
"impira/layoutlm-document-qa" , revision="3dc6de3" , add_prefix_space=A_ )
__UpperCamelCase = pipeline(
"document-question-answering" , model="impira/layoutlm-document-qa" , tokenizer=A_ , revision="3dc6de3" , )
__UpperCamelCase = INVOICE_URL
__UpperCamelCase = "What is the invoice number?"
__UpperCamelCase = dqa_pipeline(image=A_ , question=A_ , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.4_251, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.0_819, "answer": "1110212019", "start": 23, "end": 23},
] , )
__UpperCamelCase = dqa_pipeline({"image": image, "question": question} , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.4_251, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.0_819, "answer": "1110212019", "start": 23, "end": 23},
] , )
__UpperCamelCase = dqa_pipeline(
[{"image": image, "question": question}, {"image": image, "question": question}] , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
[
{"score": 0.4_251, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.0_819, "answer": "1110212019", "start": 23, "end": 23},
]
]
* 2 , )
__UpperCamelCase = list(zip(*apply_tesseract(load_image(A_ ) , A_ , "" ) ) )
# This model should also work if `image` is set to None
__UpperCamelCase = dqa_pipeline({"image": None, "word_boxes": word_boxes, "question": question} , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.4_251, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.0_819, "answer": "1110212019", "start": 23, "end": 23},
] , )
@slow
@require_torch
@require_pytesseract
@require_vision
def A ( self : List[str] )-> int:
__UpperCamelCase = AutoTokenizer.from_pretrained(
"impira/layoutlm-document-qa" , revision="3dc6de3" , add_prefix_space=A_ )
__UpperCamelCase = pipeline(
"document-question-answering" , model="impira/layoutlm-document-qa" , tokenizer=A_ , revision="3dc6de3" , max_seq_len=50 , )
__UpperCamelCase = INVOICE_URL
__UpperCamelCase = "What is the invoice number?"
__UpperCamelCase = dqa_pipeline(image=A_ , question=A_ , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.9_999, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.9_998, "answer": "us-001", "start": 16, "end": 16},
] , )
__UpperCamelCase = dqa_pipeline(
[{"image": image, "question": question}, {"image": image, "question": question}] , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
[
{"score": 0.9_999, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.9_998, "answer": "us-001", "start": 16, "end": 16},
]
]
* 2 , )
__UpperCamelCase = list(zip(*apply_tesseract(load_image(A_ ) , A_ , "" ) ) )
# This model should also work if `image` is set to None
__UpperCamelCase = dqa_pipeline({"image": None, "word_boxes": word_boxes, "question": question} , top_k=2 )
self.assertEqual(
nested_simplify(A_ , decimals=4 ) , [
{"score": 0.9_999, "answer": "us-001", "start": 16, "end": 16},
{"score": 0.9_998, "answer": "us-001", "start": 16, "end": 16},
] , )
@slow
@require_torch
def A ( self : Optional[int] )-> Optional[Any]:
__UpperCamelCase = pipeline(
"document-question-answering" , model="naver-clova-ix/donut-base-finetuned-docvqa" , tokenizer=AutoTokenizer.from_pretrained("naver-clova-ix/donut-base-finetuned-docvqa" ) , feature_extractor="naver-clova-ix/donut-base-finetuned-docvqa" , )
__UpperCamelCase = INVOICE_URL
__UpperCamelCase = "What is the invoice number?"
__UpperCamelCase = dqa_pipeline(image=A_ , question=A_ , top_k=2 )
self.assertEqual(nested_simplify(A_ , decimals=4 ) , [{"answer": "us-001"}] )
@require_tf
@unittest.skip("Document question answering not implemented in TF" )
def A ( self : List[Any] )-> str:
pass
| 228
|
"""simple docstring"""
from dataclasses import dataclass
from typing import Dict, Optional, Union
import torch
import torch.nn.functional as F
from torch import nn
from ..configuration_utils import ConfigMixin, register_to_config
from ..utils import BaseOutput
from .attention import BasicTransformerBlock
from .attention_processor import AttentionProcessor, AttnProcessor
from .embeddings import TimestepEmbedding, Timesteps
from .modeling_utils import ModelMixin
@dataclass
class __UpperCAmelCase ( snake_case__ ):
"""simple docstring"""
_snake_case : torch.FloatTensor
class __UpperCAmelCase ( snake_case__ , snake_case__ ):
"""simple docstring"""
@register_to_config
def __init__( self : Optional[Any] , A_ : int = 32 , A_ : int = 64 , A_ : int = 20 , A_ : int = 7_68 , A_ : Dict=77 , A_ : Union[str, Any]=4 , A_ : float = 0.0 , A_ : str = "silu" , A_ : Optional[str] = None , A_ : Optional[str] = None , A_ : Optional[str] = "linear" , A_ : Optional[str] = "prd" , A_ : Optional[int] = None , A_ : Optional[int] = None , A_ : Optional[int] = None , )-> Optional[int]:
super().__init__()
__UpperCamelCase = num_attention_heads
__UpperCamelCase = attention_head_dim
__UpperCamelCase = num_attention_heads * attention_head_dim
__UpperCamelCase = additional_embeddings
__UpperCamelCase = time_embed_dim or inner_dim
__UpperCamelCase = embedding_proj_dim or embedding_dim
__UpperCamelCase = clip_embed_dim or embedding_dim
__UpperCamelCase = Timesteps(A_ , A_ , 0 )
__UpperCamelCase = TimestepEmbedding(A_ , A_ , out_dim=A_ , act_fn=A_ )
__UpperCamelCase = nn.Linear(A_ , A_ )
if embedding_proj_norm_type is None:
__UpperCamelCase = None
elif embedding_proj_norm_type == "layer":
__UpperCamelCase = nn.LayerNorm(A_ )
else:
raise ValueError(f"""unsupported embedding_proj_norm_type: {embedding_proj_norm_type}""" )
__UpperCamelCase = nn.Linear(A_ , A_ )
if encoder_hid_proj_type is None:
__UpperCamelCase = None
elif encoder_hid_proj_type == "linear":
__UpperCamelCase = nn.Linear(A_ , A_ )
else:
raise ValueError(f"""unsupported encoder_hid_proj_type: {encoder_hid_proj_type}""" )
__UpperCamelCase = nn.Parameter(torch.zeros(1 , num_embeddings + additional_embeddings , A_ ) )
if added_emb_type == "prd":
__UpperCamelCase = nn.Parameter(torch.zeros(1 , 1 , A_ ) )
elif added_emb_type is None:
__UpperCamelCase = None
else:
raise ValueError(
f"""`added_emb_type`: {added_emb_type} is not supported. Make sure to choose one of `'prd'` or `None`.""" )
__UpperCamelCase = nn.ModuleList(
[
BasicTransformerBlock(
A_ , A_ , A_ , dropout=A_ , activation_fn="gelu" , attention_bias=A_ , )
for d in range(A_ )
] )
if norm_in_type == "layer":
__UpperCamelCase = nn.LayerNorm(A_ )
elif norm_in_type is None:
__UpperCamelCase = None
else:
raise ValueError(f"""Unsupported norm_in_type: {norm_in_type}.""" )
__UpperCamelCase = nn.LayerNorm(A_ )
__UpperCamelCase = nn.Linear(A_ , A_ )
__UpperCamelCase = torch.full(
[num_embeddings + additional_embeddings, num_embeddings + additional_embeddings] , -10_000.0 )
causal_attention_mask.triu_(1 )
__UpperCamelCase = causal_attention_mask[None, ...]
self.register_buffer("causal_attention_mask" , A_ , persistent=A_ )
__UpperCamelCase = nn.Parameter(torch.zeros(1 , A_ ) )
__UpperCamelCase = nn.Parameter(torch.zeros(1 , A_ ) )
@property
# Copied from diffusers.models.unet_2d_condition.UNet2DConditionModel.attn_processors
def A ( self : Tuple )-> Dict[str, AttentionProcessor]:
__UpperCamelCase = {}
def fn_recursive_add_processors(A_ : str , A_ : torch.nn.Module , A_ : Dict[str, AttentionProcessor] ):
if hasattr(A_ , "set_processor" ):
__UpperCamelCase = module.processor
for sub_name, child in module.named_children():
fn_recursive_add_processors(f"""{name}.{sub_name}""" , A_ , A_ )
return processors
for name, module in self.named_children():
fn_recursive_add_processors(A_ , A_ , A_ )
return processors
def A ( self : Tuple , A_ : Union[AttentionProcessor, Dict[str, AttentionProcessor]] )-> Optional[int]:
__UpperCamelCase = len(self.attn_processors.keys() )
if isinstance(A_ , A_ ) and len(A_ ) != count:
raise ValueError(
f"""A dict of processors was passed, but the number of processors {len(A_ )} does not match the"""
f""" number of attention layers: {count}. Please make sure to pass {count} processor classes.""" )
def fn_recursive_attn_processor(A_ : str , A_ : torch.nn.Module , A_ : Any ):
if hasattr(A_ , "set_processor" ):
if not isinstance(A_ , A_ ):
module.set_processor(A_ )
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}""" , A_ , A_ )
for name, module in self.named_children():
fn_recursive_attn_processor(A_ , A_ , A_ )
def A ( self : List[str] )-> List[str]:
self.set_attn_processor(AttnProcessor() )
def A ( self : Dict , A_ : str , A_ : Union[torch.Tensor, float, int] , A_ : torch.FloatTensor , A_ : Optional[torch.FloatTensor] = None , A_ : Optional[torch.BoolTensor] = None , A_ : bool = True , )-> Any:
__UpperCamelCase = hidden_states.shape[0]
__UpperCamelCase = timestep
if not torch.is_tensor(A_ ):
__UpperCamelCase = torch.tensor([timesteps] , dtype=torch.long , device=hidden_states.device )
elif torch.is_tensor(A_ ) and len(timesteps.shape ) == 0:
__UpperCamelCase = timesteps[None].to(hidden_states.device )
# broadcast to batch dimension in a way that's compatible with ONNX/Core ML
__UpperCamelCase = timesteps * torch.ones(A_ , dtype=timesteps.dtype , device=timesteps.device )
__UpperCamelCase = self.time_proj(A_ )
# timesteps does not contain any weights and will always return f32 tensors
# but time_embedding might be fp16, so we need to cast here.
__UpperCamelCase = timesteps_projected.to(dtype=self.dtype )
__UpperCamelCase = self.time_embedding(A_ )
if self.embedding_proj_norm is not None:
__UpperCamelCase = self.embedding_proj_norm(A_ )
__UpperCamelCase = self.embedding_proj(A_ )
if self.encoder_hidden_states_proj is not None and encoder_hidden_states is not None:
__UpperCamelCase = self.encoder_hidden_states_proj(A_ )
elif self.encoder_hidden_states_proj is not None and encoder_hidden_states is None:
raise ValueError("`encoder_hidden_states_proj` requires `encoder_hidden_states` to be set" )
__UpperCamelCase = self.proj_in(A_ )
__UpperCamelCase = self.positional_embedding.to(hidden_states.dtype )
__UpperCamelCase = []
__UpperCamelCase = 0
if encoder_hidden_states is not None:
additional_embeds.append(A_ )
additional_embeddings_len += encoder_hidden_states.shape[1]
if len(proj_embeddings.shape ) == 2:
__UpperCamelCase = proj_embeddings[:, None, :]
if len(hidden_states.shape ) == 2:
__UpperCamelCase = hidden_states[:, None, :]
__UpperCamelCase = additional_embeds + [
proj_embeddings,
time_embeddings[:, None, :],
hidden_states,
]
if self.prd_embedding is not None:
__UpperCamelCase = self.prd_embedding.to(hidden_states.dtype ).expand(A_ , -1 , -1 )
additional_embeds.append(A_ )
__UpperCamelCase = torch.cat(
A_ , dim=1 , )
# Allow positional_embedding to not include the `addtional_embeddings` and instead pad it with zeros for these additional tokens
__UpperCamelCase = additional_embeddings_len + proj_embeddings.shape[1] + 1
if positional_embeddings.shape[1] < hidden_states.shape[1]:
__UpperCamelCase = F.pad(
A_ , (
0,
0,
additional_embeddings_len,
self.prd_embedding.shape[1] if self.prd_embedding is not None else 0,
) , value=0.0 , )
__UpperCamelCase = hidden_states + positional_embeddings
if attention_mask is not None:
__UpperCamelCase = (1 - attention_mask.to(hidden_states.dtype )) * -10_000.0
__UpperCamelCase = F.pad(A_ , (0, self.additional_embeddings) , value=0.0 )
__UpperCamelCase = (attention_mask[:, None, :] + self.causal_attention_mask).to(hidden_states.dtype )
__UpperCamelCase = attention_mask.repeat_interleave(self.config.num_attention_heads , dim=0 )
if self.norm_in is not None:
__UpperCamelCase = self.norm_in(A_ )
for block in self.transformer_blocks:
__UpperCamelCase = block(A_ , attention_mask=A_ )
__UpperCamelCase = self.norm_out(A_ )
if self.prd_embedding is not None:
__UpperCamelCase = hidden_states[:, -1]
else:
__UpperCamelCase = hidden_states[:, additional_embeddings_len:]
__UpperCamelCase = self.proj_to_clip_embeddings(A_ )
if not return_dict:
return (predicted_image_embedding,)
return PriorTransformerOutput(predicted_image_embedding=A_ )
def A ( self : Dict , A_ : Tuple )-> Dict:
__UpperCamelCase = (prior_latents * self.clip_std) + self.clip_mean
return prior_latents
| 228
| 1
|
"""simple docstring"""
from __future__ import annotations
import os
import tempfile
import unittest
import numpy as np
from huggingface_hub import hf_hub_download
from transformers import is_tensorflow_text_available, is_tf_available
from transformers.testing_utils import require_tensorflow_text, require_tf, slow
from ..test_modeling_tf_common import floats_tensor
from .test_framework_agnostic import GenerationIntegrationTestsMixin
if is_tf_available():
import tensorflow as tf
from transformers import (
AutoTokenizer,
TFAutoModelForCausalLM,
TFAutoModelForSeqaSeqLM,
TFAutoModelForSpeechSeqaSeq,
TFAutoModelForVisionaSeq,
TFBartForConditionalGeneration,
TFLogitsProcessorList,
TFMinLengthLogitsProcessor,
tf_top_k_top_p_filtering,
)
if is_tensorflow_text_available():
import tensorflow_text as text
@require_tf
class UpperCAmelCase__ ( unittest.TestCase ):
"""simple docstring"""
def A ( self ) -> Union[str, Any]:
a_ : Union[str, Any] = tf.convert_to_tensor(
[
[
8.2_2_2_0_9_9_1, # 3rd highest value; idx. 0
-0.5_6_2_0_0_4_4,
5.2_3_2_2_9_7_5_2,
4.0_3_8_6_3_9_3,
-6.8_7_9_8_3_7_8,
-0.5_4_7_8_5_8_0_2,
-3.2_0_1_2_1_5_3,
2.9_2_7_7_7_1_7_6,
1.8_8_1_7_1_9_5_3,
7.3_5_3_4_1_2_7_6, # 5th highest value; idx. 9
8.4_3_2_0_7_8_3_3, # 2nd highest value; idx. 10
-9.8_5_7_1_1_8_3_6,
-5.9_6_2_0_9_2_3_6,
-1.1_3_0_3_9_1_6_1,
-7.1_1_1_5_2_9_4,
-0.8_3_6_9_6_3_3,
-5.3_1_8_6_4_0_8,
7.0_6_4_2_7_4_0_7,
0.8_1_3_6_9_3_4_4,
-0.8_2_0_2_3_8_1_7,
-5.9_1_7_9_7_9_6,
0.5_8_8_1_3_4_4_3,
-6.9_9_7_7_8_4_3_8,
4.7_1_5_5_1_1_8_9,
-0.1_8_7_7_1_6_3_7,
7.4_4_0_2_0_7_5_9, # 4th highest value; idx. 25
9.3_8_4_5_0_9_8_7, # 1st highest value; idx. 26
2.1_2_6_6_2_9_4_1,
-9.3_2_5_6_2_0_3_8,
2.3_5_6_5_2_5_2_2,
], # cummulative prob of 5 highest values <= 0.6
[
0.5_8_4_2_5_5_1_8,
4.5_3_1_3_9_2_3_8,
-5.5_7_5_1_0_4_6_4,
-6.2_8_0_3_0_6_9_9,
-7.1_9_5_2_9_5_0_3,
-4.0_2_1_2_2_5_5_1,
1.3_9_3_3_7_0_3_7,
-6.0_6_7_0_7_0_5_7,
1.5_9_4_8_0_5_1_7,
-9.6_4_3_1_1_9,
0.0_3_9_0_7_7_9_9,
0.6_7_2_3_1_7_6_2,
-8.8_8_2_0_6_7_2_6,
6.2_7_1_1_5_9_2_2, # 4th highest value; idx. 13
2.2_8_5_2_0_7_2_3,
4.8_2_7_6_7_5_0_6,
4.3_0_4_2_1_3_6_8,
8.8_2_7_5_3_1_3, # 2nd highest value; idx. 17
5.4_4_0_2_9_9_5_8, # 5th highest value; idx. 18
-4.4_7_3_5_7_9_4,
7.3_8_5_7_9_5_3_6, # 3rd highest value; idx. 20
-2.9_1_0_5_1_6_6_3,
2.6_1_9_4_6_0_7_7,
-2.5_6_7_4_7_6_2,
-9.4_8_9_5_9_3_0_2,
-4.0_2_9_2_2_6_4_5,
-1.3_5_4_1_6_9_1_8,
9.6_7_7_0_2_3_2_3, # 1st highest value; idx. 27
-5.8_9_4_7_8_5_5_3,
1.8_5_3_7_0_4_6_7,
], # cummulative prob of 5 highest values <= 0.6
] , dtype=tf.floataa , )
a_ : Union[str, Any] = tf.convert_to_tensor(
[[0, 0], [0, 9], [0, 1_0], [0, 2_5], [0, 2_6], [1, 1_3], [1, 1_7], [1, 1_8], [1, 2_0], [1, 2_7]] , dtype=tf.intaa , ) # expected non filtered idx as noted above
a_ : Optional[Any] = tf.convert_to_tensor(
[8.2_2_2_0_9_9, 7.3_5_3_4_1_2_6, 8.4_3_2_0_7_8, 7.4_4_0_2_0_7_5, 9.3_8_4_5_1, 6.2_7_1_1_5_9, 8.8_2_7_5_3_1, 5.4_4_0_2_9_9_5, 7.3_8_5_7_9_5_6, 9.6_7_7_0_2_3] , dtype=tf.floataa , ) # expected non filtered values as noted above
a_ : Any = tf_top_k_top_p_filtering(_SCREAMING_SNAKE_CASE , top_k=1_0 , top_p=0.6 , min_tokens_to_keep=4 )
a_ : Optional[int] = output[output != -float("inf" )]
a_ : List[str] = tf.cast(
tf.where(tf.not_equal(_SCREAMING_SNAKE_CASE , tf.constant(-float("inf" ) , dtype=tf.floataa ) ) ) , dtype=tf.intaa , )
tf.debugging.assert_near(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , rtol=1E-12 )
tf.debugging.assert_equal(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
@require_tf
class UpperCAmelCase__ ( unittest.TestCase, __lowerCamelCase ):
"""simple docstring"""
# setting framework_dependent_parameters needs to be gated, just like its contents' imports
if is_tf_available():
lowerCAmelCase__ : Dict = {
"""AutoModelForCausalLM""": TFAutoModelForCausalLM,
"""AutoModelForSpeechSeq2Seq""": TFAutoModelForSpeechSeqaSeq,
"""AutoModelForSeq2SeqLM""": TFAutoModelForSeqaSeqLM,
"""AutoModelForVision2Seq""": TFAutoModelForVisionaSeq,
"""LogitsProcessorList""": TFLogitsProcessorList,
"""MinLengthLogitsProcessor""": TFMinLengthLogitsProcessor,
"""create_tensor_fn""": tf.convert_to_tensor,
"""floats_tensor""": floats_tensor,
"""return_tensors""": """tf""",
}
@slow
def A ( self ) -> Any:
# TF-only test: tf.saved_model export
a_ : Tuple = TFAutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-gpt2" )
a_ : Any = 2
a_ : Optional[int] = 2
class UpperCAmelCase__ ( tf.Module ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE ) -> Union[str, Any]:
super(_SCREAMING_SNAKE_CASE , self ).__init__()
a_ : List[str] = model
@tf.function(
input_signature=(
tf.TensorSpec((None, input_length) , tf.intaa , name="input_ids" ),
tf.TensorSpec((None, input_length) , tf.intaa , name="attention_mask" ),
) , jit_compile=_SCREAMING_SNAKE_CASE , )
def A ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE ) -> Dict:
a_ : Dict = self.model.generate(
input_ids=_SCREAMING_SNAKE_CASE , attention_mask=_SCREAMING_SNAKE_CASE , max_new_tokens=_SCREAMING_SNAKE_CASE , return_dict_in_generate=_SCREAMING_SNAKE_CASE , )
return {"sequences": outputs["sequences"]}
a_ : Tuple = [[2, 0], [1_0_2, 1_0_3]]
a_ : List[str] = [[1, 0], [1, 1]]
a_ : Tuple = DummyModel(model=_SCREAMING_SNAKE_CASE )
with tempfile.TemporaryDirectory() as tmp_dir:
tf.saved_model.save(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , signatures={"serving_default": dummy_model.serving} )
a_ : str = tf.saved_model.load(_SCREAMING_SNAKE_CASE ).signatures["serving_default"]
for batch_size in range(1 , len(_SCREAMING_SNAKE_CASE ) + 1 ):
a_ : Optional[Any] = {
"input_ids": tf.constant(dummy_input_ids[:batch_size] ),
"attention_mask": tf.constant(dummy_attention_masks[:batch_size] ),
}
a_ : Union[str, Any] = serving_func(**_SCREAMING_SNAKE_CASE )["sequences"]
a_ : Optional[int] = test_model.generate(**_SCREAMING_SNAKE_CASE , max_new_tokens=_SCREAMING_SNAKE_CASE )
tf.debugging.assert_equal(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
@slow
def A ( self ) -> Optional[int]:
# TF-only test: tf.saved_model export
a_ : Any = TFAutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-gpt2" )
a_ : Union[str, Any] = 1
a_ : List[str] = 2
class UpperCAmelCase__ ( tf.Module ):
"""simple docstring"""
def __init__( self , _SCREAMING_SNAKE_CASE ) -> Any:
super(_SCREAMING_SNAKE_CASE , self ).__init__()
a_ : str = model
@tf.function(
input_signature=(
tf.TensorSpec((batch_size, None) , tf.intaa , name="input_ids" ),
tf.TensorSpec((batch_size, None) , tf.intaa , name="attention_mask" ),
) , jit_compile=_SCREAMING_SNAKE_CASE , )
def A ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE ) -> Union[str, Any]:
a_ : Dict = self.model.generate(
input_ids=_SCREAMING_SNAKE_CASE , attention_mask=_SCREAMING_SNAKE_CASE , max_new_tokens=_SCREAMING_SNAKE_CASE , return_dict_in_generate=_SCREAMING_SNAKE_CASE , )
return {"sequences": outputs["sequences"]}
a_ : int = [[2], [1_0_2, 1_0_3]]
a_ : Tuple = [[1], [1, 1]]
a_ : int = DummyModel(model=_SCREAMING_SNAKE_CASE )
with tempfile.TemporaryDirectory() as tmp_dir:
tf.saved_model.save(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , signatures={"serving_default": dummy_model.serving} )
a_ : Optional[int] = tf.saved_model.load(_SCREAMING_SNAKE_CASE ).signatures["serving_default"]
for input_row in range(len(_SCREAMING_SNAKE_CASE ) ):
a_ : Union[str, Any] = {
"input_ids": tf.constant([dummy_input_ids[input_row]] ),
"attention_mask": tf.constant([dummy_attention_masks[input_row]] ),
}
a_ : Union[str, Any] = serving_func(**_SCREAMING_SNAKE_CASE )["sequences"]
a_ : List[str] = test_model.generate(**_SCREAMING_SNAKE_CASE , max_new_tokens=_SCREAMING_SNAKE_CASE )
tf.debugging.assert_equal(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
@slow
@require_tensorflow_text
def A ( self ) -> Optional[int]:
# TF-only test: tf.saved_model export
with tempfile.TemporaryDirectory() as tmp_dir:
# file needed to load the TF tokenizer
hf_hub_download(repo_id="google/flan-t5-small" , filename="spiece.model" , local_dir=_SCREAMING_SNAKE_CASE )
class UpperCAmelCase__ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self ) -> List[Any]:
super().__init__()
a_ : Optional[Any] = text.SentencepieceTokenizer(
model=tf.io.gfile.GFile(os.path.join(_SCREAMING_SNAKE_CASE , "spiece.model" ) , "rb" ).read() )
a_ : Optional[int] = TFAutoModelForSeqaSeqLM.from_pretrained("hf-internal-testing/tiny-random-t5" )
def A ( self , _SCREAMING_SNAKE_CASE , *_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE ) -> Dict:
a_ : List[str] = self.tokenizer.tokenize(_SCREAMING_SNAKE_CASE )
a_ , a_ : Optional[int] = text.pad_model_inputs(
_SCREAMING_SNAKE_CASE , max_seq_length=6_4 , pad_value=self.model.config.pad_token_id )
a_ : Tuple = self.model.generate(input_ids=_SCREAMING_SNAKE_CASE , attention_mask=_SCREAMING_SNAKE_CASE )
return self.tokenizer.detokenize(_SCREAMING_SNAKE_CASE )
a_ : Optional[Any] = CompleteSentenceTransformer()
a_ : List[Any] = tf.keras.layers.Input(shape=(1,) , dtype=tf.string , name="inputs" )
a_ : List[Any] = complete_model(_SCREAMING_SNAKE_CASE )
a_ : int = tf.keras.Model(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE )
keras_model.save(_SCREAMING_SNAKE_CASE )
def A ( self ) -> int:
# Has PT equivalent: this test relies on random sampling
a_ : List[Any] = {
"do_sample": True,
"num_beams": 1,
"top_p": 0.7,
"top_k": 1_0,
"temperature": 0.7,
}
a_ : Union[str, Any] = 1_4
a_ : Tuple = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-gpt2" )
a_ : Any = "Hello, my dog is cute and"
a_ : Tuple = tokenizer(_SCREAMING_SNAKE_CASE , return_tensors="tf" )
a_ : str = TFAutoModelForCausalLM.from_pretrained("hf-internal-testing/tiny-random-gpt2" )
a_ : List[Any] = 6_3_8
# forces the generation to happen on CPU, to avoid GPU-related quirks
with tf.device(":/CPU:0" ):
tf.random.set_seed(0 )
a_ : Dict = model.generate(**_SCREAMING_SNAKE_CASE , eos_token_id=_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )
self.assertTrue(expectation == len(generated_tokens[0] ) )
a_ : int = [6_3_8, 1_9_8]
with tf.device(":/CPU:0" ):
tf.random.set_seed(0 )
a_ : Dict = model.generate(**_SCREAMING_SNAKE_CASE , eos_token_id=_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )
self.assertTrue(expectation == len(generated_tokens[0] ) )
def A ( self ) -> Any:
# Has PT equivalent: ample use of framework-specific code
a_ : List[str] = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-bart" )
a_ : Optional[int] = "Hugging Face is a technology company based in New York and Paris."
a_ : Optional[Any] = bart_tokenizer(_SCREAMING_SNAKE_CASE , return_tensors="tf" ).input_ids
a_ : Optional[int] = TFBartForConditionalGeneration.from_pretrained("hf-internal-testing/tiny-random-bart" )
a_ : Any = bart_model.generate(_SCREAMING_SNAKE_CASE ).numpy()
class UpperCAmelCase__ ( __lowerCamelCase ):
"""simple docstring"""
def A ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE=None , **_SCREAMING_SNAKE_CASE ) -> str:
return super().call(_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )
a_ : Optional[Any] = FakeBart.from_pretrained("hf-internal-testing/tiny-random-bart" )
a_ : List[str] = bart_model.generate(_SCREAMING_SNAKE_CASE , foo="bar" ).numpy()
self.assertTrue(np.array_equal(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE ) )
class UpperCAmelCase__ ( bart_model.model.encoder.__class__ ):
"""simple docstring"""
def A ( self , _SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE ) -> Dict:
return super().call(_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )
a_ : int = FakeEncoder(bart_model.config , bart_model.model.shared )
a_ : Optional[int] = fake_encoder
# Normal generation still works (the output will be different because the encoder weights are different)
a_ : Any = bart_model.generate(_SCREAMING_SNAKE_CASE ).numpy()
with self.assertRaises(_SCREAMING_SNAKE_CASE ):
# FakeEncoder.call() accepts **kwargs -> no filtering -> value error due to unexpected input "foo"
bart_model.generate(_SCREAMING_SNAKE_CASE , foo="bar" )
| 473
|
"""simple docstring"""
# Copyright 2023 The HuggingFace Inc. team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import numpy as np
import torch
from ..models.clipseg import CLIPSegForImageSegmentation
from ..utils import is_vision_available, requires_backends
from .base import PipelineTool
if is_vision_available():
from PIL import Image
class UpperCAmelCase__ ( __lowerCamelCase ):
"""simple docstring"""
lowerCAmelCase__ : Any = (
"""This is a tool that creates a segmentation mask of an image according to a label. It cannot create an image."""
"""It takes two arguments named `image` which should be the original image, and `label` which should be a text """
"""describing the elements what should be identified in the segmentation mask. The tool returns the mask."""
)
lowerCAmelCase__ : Tuple = """CIDAS/clipseg-rd64-refined"""
lowerCAmelCase__ : Optional[Any] = """image_segmenter"""
lowerCAmelCase__ : Optional[Any] = CLIPSegForImageSegmentation
lowerCAmelCase__ : Any = ["""image""", """text"""]
lowerCAmelCase__ : Optional[Any] = ["""image"""]
def __init__( self , *_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE ) -> Tuple:
requires_backends(self , ["vision"] )
super().__init__(*_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE )
def A ( self , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE ) -> Dict:
return self.pre_processor(text=[label] , images=[image] , padding=_SCREAMING_SNAKE_CASE , return_tensors="pt" )
def A ( self , _SCREAMING_SNAKE_CASE ) -> Optional[Any]:
with torch.no_grad():
a_ : List[Any] = self.model(**_SCREAMING_SNAKE_CASE ).logits
return logits
def A ( self , _SCREAMING_SNAKE_CASE ) -> List[Any]:
a_ : List[Any] = outputs.cpu().detach().numpy()
a_ : Optional[Any] = 0
a_ : Optional[int] = 1
return Image.fromarray((array * 2_5_5).astype(np.uinta ) )
| 473
| 1
|
import unittest
import numpy as np
from transformers.testing_utils import is_flaky, require_torch, require_vision
from transformers.utils import is_torch_available, is_vision_available
from ...test_image_processing_common import ImageProcessingSavingTestMixin, prepare_image_inputs
if is_torch_available():
import torch
if is_vision_available():
from PIL import Image
from transformers import DonutImageProcessor
class snake_case_ ( unittest.TestCase ):
def __init__( self , __lowerCAmelCase , __lowerCAmelCase=7 , __lowerCAmelCase=3 , __lowerCAmelCase=18 , __lowerCAmelCase=30 , __lowerCAmelCase=400 , __lowerCAmelCase=True , __lowerCAmelCase=None , __lowerCAmelCase=True , __lowerCAmelCase=False , __lowerCAmelCase=True , __lowerCAmelCase=True , __lowerCAmelCase=[0.5, 0.5, 0.5] , __lowerCAmelCase=[0.5, 0.5, 0.5] , ):
SCREAMING_SNAKE_CASE_ : Union[str, Any] = parent
SCREAMING_SNAKE_CASE_ : Any = batch_size
SCREAMING_SNAKE_CASE_ : int = num_channels
SCREAMING_SNAKE_CASE_ : int = image_size
SCREAMING_SNAKE_CASE_ : Dict = min_resolution
SCREAMING_SNAKE_CASE_ : Optional[Any] = max_resolution
SCREAMING_SNAKE_CASE_ : List[Any] = do_resize
SCREAMING_SNAKE_CASE_ : List[Any] = size if size is not None else {'height': 18, 'width': 20}
SCREAMING_SNAKE_CASE_ : Dict = do_thumbnail
SCREAMING_SNAKE_CASE_ : Any = do_align_axis
SCREAMING_SNAKE_CASE_ : Dict = do_pad
SCREAMING_SNAKE_CASE_ : Union[str, Any] = do_normalize
SCREAMING_SNAKE_CASE_ : int = image_mean
SCREAMING_SNAKE_CASE_ : List[str] = image_std
def __A ( self ):
return {
"do_resize": self.do_resize,
"size": self.size,
"do_thumbnail": self.do_thumbnail,
"do_align_long_axis": self.do_align_axis,
"do_pad": self.do_pad,
"do_normalize": self.do_normalize,
"image_mean": self.image_mean,
"image_std": self.image_std,
}
@require_torch
@require_vision
class snake_case_ ( lowerCAmelCase , unittest.TestCase ):
__lowerCamelCase : Dict = DonutImageProcessor if is_vision_available() else None
def __A ( self ):
SCREAMING_SNAKE_CASE_ : Dict = DonutImageProcessingTester(self )
@property
def __A ( self ):
return self.image_processor_tester.prepare_image_processor_dict()
def __A ( self ):
SCREAMING_SNAKE_CASE_ : List[str] = self.image_processing_class(**self.image_processor_dict )
self.assertTrue(hasattr(UpperCamelCase__ , 'do_resize' ) )
self.assertTrue(hasattr(UpperCamelCase__ , 'size' ) )
self.assertTrue(hasattr(UpperCamelCase__ , 'do_thumbnail' ) )
self.assertTrue(hasattr(UpperCamelCase__ , 'do_align_long_axis' ) )
self.assertTrue(hasattr(UpperCamelCase__ , 'do_pad' ) )
self.assertTrue(hasattr(UpperCamelCase__ , 'do_normalize' ) )
self.assertTrue(hasattr(UpperCamelCase__ , 'image_mean' ) )
self.assertTrue(hasattr(UpperCamelCase__ , 'image_std' ) )
def __A ( self ):
SCREAMING_SNAKE_CASE_ : Optional[int] = self.image_processing_class.from_dict(self.image_processor_dict )
self.assertEqual(image_processor.size , {'height': 18, 'width': 20} )
SCREAMING_SNAKE_CASE_ : List[str] = self.image_processing_class.from_dict(self.image_processor_dict , size=42 )
self.assertEqual(image_processor.size , {'height': 42, 'width': 42} )
# Previous config had dimensions in (width, height) order
SCREAMING_SNAKE_CASE_ : List[str] = self.image_processing_class.from_dict(self.image_processor_dict , size=(42, 84) )
self.assertEqual(image_processor.size , {'height': 84, 'width': 42} )
def __A ( self ):
pass
@is_flaky()
def __A ( self ):
SCREAMING_SNAKE_CASE_ : Optional[int] = self.image_processing_class(**self.image_processor_dict )
# create random PIL images
SCREAMING_SNAKE_CASE_ : List[Any] = prepare_image_inputs(self.image_processor_tester , equal_resolution=UpperCamelCase__ )
for image in image_inputs:
self.assertIsInstance(UpperCamelCase__ , Image.Image )
# Test not batched input
SCREAMING_SNAKE_CASE_ : Dict = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size['height'],
self.image_processor_tester.size['width'],
) , )
# Test batched
SCREAMING_SNAKE_CASE_ : List[str] = image_processing(UpperCamelCase__ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size['height'],
self.image_processor_tester.size['width'],
) , )
@is_flaky()
def __A ( self ):
SCREAMING_SNAKE_CASE_ : Dict = self.image_processing_class(**self.image_processor_dict )
# create random numpy tensors
SCREAMING_SNAKE_CASE_ : List[Any] = prepare_image_inputs(self.image_processor_tester , equal_resolution=UpperCamelCase__ , numpify=UpperCamelCase__ )
for image in image_inputs:
self.assertIsInstance(UpperCamelCase__ , np.ndarray )
# Test not batched input
SCREAMING_SNAKE_CASE_ : Optional[int] = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size['height'],
self.image_processor_tester.size['width'],
) , )
# Test batched
SCREAMING_SNAKE_CASE_ : List[Any] = image_processing(UpperCamelCase__ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size['height'],
self.image_processor_tester.size['width'],
) , )
@is_flaky()
def __A ( self ):
SCREAMING_SNAKE_CASE_ : List[Any] = self.image_processing_class(**self.image_processor_dict )
# create random PyTorch tensors
SCREAMING_SNAKE_CASE_ : str = prepare_image_inputs(self.image_processor_tester , equal_resolution=UpperCamelCase__ , torchify=UpperCamelCase__ )
for image in image_inputs:
self.assertIsInstance(UpperCamelCase__ , torch.Tensor )
# Test not batched input
SCREAMING_SNAKE_CASE_ : List[Any] = image_processing(image_inputs[0] , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
1,
self.image_processor_tester.num_channels,
self.image_processor_tester.size['height'],
self.image_processor_tester.size['width'],
) , )
# Test batched
SCREAMING_SNAKE_CASE_ : int = image_processing(UpperCamelCase__ , return_tensors='pt' ).pixel_values
self.assertEqual(
encoded_images.shape , (
self.image_processor_tester.batch_size,
self.image_processor_tester.num_channels,
self.image_processor_tester.size['height'],
self.image_processor_tester.size['width'],
) , )
| 700
|
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_flax_available, is_torch_available
lowerCAmelCase__: Dict = {
"configuration_gpt_neo": ["GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP", "GPTNeoConfig", "GPTNeoOnnxConfig"],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
lowerCAmelCase__: List[str] = [
"GPT_NEO_PRETRAINED_MODEL_ARCHIVE_LIST",
"GPTNeoForCausalLM",
"GPTNeoForQuestionAnswering",
"GPTNeoForSequenceClassification",
"GPTNeoForTokenClassification",
"GPTNeoModel",
"GPTNeoPreTrainedModel",
"load_tf_weights_in_gpt_neo",
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
lowerCAmelCase__: Union[str, Any] = [
"FlaxGPTNeoForCausalLM",
"FlaxGPTNeoModel",
"FlaxGPTNeoPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_gpt_neo import GPT_NEO_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTNeoConfig, GPTNeoOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_gpt_neo import (
GPT_NEO_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTNeoForCausalLM,
GPTNeoForQuestionAnswering,
GPTNeoForSequenceClassification,
GPTNeoForTokenClassification,
GPTNeoModel,
GPTNeoPreTrainedModel,
load_tf_weights_in_gpt_neo,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_gpt_neo import FlaxGPTNeoForCausalLM, FlaxGPTNeoModel, FlaxGPTNeoPreTrainedModel
else:
import sys
lowerCAmelCase__: Optional[Any] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| 311
| 0
|
import os
from typing import Optional
import fsspec
from fsspec.archive import AbstractArchiveFileSystem
from fsspec.utils import DEFAULT_BLOCK_SIZE
class UpperCamelCase_ ( UpperCamelCase__ ):
lowerCamelCase_ = ""
lowerCamelCase_ = (
None # protocol passed in prefix to the url. ex: "gzip", for gzip://file.txt::http://foo.bar/file.txt.gz
)
lowerCamelCase_ = None # compression type in fsspec. ex: "gzip"
lowerCamelCase_ = None # extension of the filename to strip. ex: "".gz" to get file.txt from file.txt.gz
def __init__( self :Optional[int] , __A :str = "" , __A :Optional[str] = None , __A :Optional[dict] = None , **__A :List[str] ) -> Any:
"""simple docstring"""
super().__init__(self , **__A )
# always open as "rb" since fsspec can then use the TextIOWrapper to make it work for "r" mode
SCREAMING_SNAKE_CASE__ = fsspec.open(
__A , mode="""rb""" , protocol=__A , compression=self.compression , client_kwargs={
"""requote_redirect_url""": False, # see https://github.com/huggingface/datasets/pull/5459
"""trust_env""": True, # Enable reading proxy env variables.
**(target_options or {}).pop("""client_kwargs""" , {} ), # To avoid issues if it was already passed.
} , **(target_options or {}) , )
SCREAMING_SNAKE_CASE__ = os.path.basename(self.file.path.split("""::""" )[0] )
SCREAMING_SNAKE_CASE__ = (
self.compressed_name[: self.compressed_name.rindex(""".""" )]
if """.""" in self.compressed_name
else self.compressed_name
)
SCREAMING_SNAKE_CASE__ = None
@classmethod
def _snake_case ( cls :Any , __A :Tuple ) -> List[str]:
"""simple docstring"""
return super()._strip_protocol(__A ).lstrip("""/""" )
def _snake_case ( self :Union[str, Any] ) -> Tuple:
"""simple docstring"""
if self.dir_cache is None:
SCREAMING_SNAKE_CASE__ = {**self.file.fs.info(self.file.path ), """name""": self.uncompressed_name}
SCREAMING_SNAKE_CASE__ = {f["""name"""]: f}
def _snake_case ( self :Optional[int] , __A :str ) -> str:
"""simple docstring"""
return self.file.open().read()
def _snake_case ( self :List[str] , __A :str , __A :str = "rb" , __A :int=None , __A :List[str]=True , __A :Optional[Any]=None , **__A :Union[str, Any] , ) -> Any:
"""simple docstring"""
SCREAMING_SNAKE_CASE__ = self._strip_protocol(__A )
if mode != "rb":
raise ValueError(f'''Tried to read with mode {mode} on file {self.file.path} opened with mode \'rb\'''' )
return self.file.open()
class UpperCamelCase_ ( UpperCamelCase__ ):
lowerCamelCase_ = "bz2"
lowerCamelCase_ = "bz2"
lowerCamelCase_ = ".bz2"
class UpperCamelCase_ ( UpperCamelCase__ ):
lowerCamelCase_ = "gzip"
lowerCamelCase_ = "gzip"
lowerCamelCase_ = ".gz"
class UpperCamelCase_ ( UpperCamelCase__ ):
lowerCamelCase_ = "lz4"
lowerCamelCase_ = "lz4"
lowerCamelCase_ = ".lz4"
class UpperCamelCase_ ( UpperCamelCase__ ):
lowerCamelCase_ = "xz"
lowerCamelCase_ = "xz"
lowerCamelCase_ = ".xz"
class UpperCamelCase_ ( UpperCamelCase__ ):
lowerCamelCase_ = "zstd"
lowerCamelCase_ = "zstd"
lowerCamelCase_ = ".zst"
def __init__( self :List[Any] , __A :str , __A :str = "rb" , __A :Optional[str] = None , __A :Optional[dict] = None , __A :int = DEFAULT_BLOCK_SIZE , **__A :Optional[int] , ) -> Union[str, Any]:
"""simple docstring"""
super().__init__(
fo=__A , mode=__A , target_protocol=__A , target_options=__A , block_size=__A , **__A , )
# We need to wrap the zstd decompressor to avoid this error in fsspec==2021.7.0 and zstandard==0.15.2:
#
# File "/Users/user/.virtualenvs/hf-datasets/lib/python3.7/site-packages/fsspec/core.py", line 145, in open
# out.close = close
# AttributeError: 'zstd.ZstdDecompressionReader' object attribute 'close' is read-only
#
# see https://github.com/intake/filesystem_spec/issues/725
SCREAMING_SNAKE_CASE__ = self.file.__enter__
class UpperCamelCase_ :
def __init__( self :int , __A :Tuple ) -> int:
"""simple docstring"""
SCREAMING_SNAKE_CASE__ = file_
def __enter__( self :Optional[Any] ) -> Optional[int]:
"""simple docstring"""
self._file.__enter__()
return self
def __exit__( self :Optional[Any] , *__A :List[Any] , **__A :int ) -> Tuple:
"""simple docstring"""
self._file.__exit__(*__A , **__A )
def __iter__( self :Any ) -> Optional[int]:
"""simple docstring"""
return iter(self._file )
def _snake_case ( self :Dict ) -> Dict:
"""simple docstring"""
return next(self._file )
def __getattr__( self :Union[str, Any] , __A :List[str] ) -> Optional[Any]:
"""simple docstring"""
return getattr(self._file , __A )
def fixed_enter(*__A :List[Any] , **__A :Optional[int] ):
return WrappedFile(_enter(*__A , **__A ) )
SCREAMING_SNAKE_CASE__ = fixed_enter
| 6
|
'''simple docstring'''
import json
import os
import unittest
from transformers import DebertaTokenizer, DebertaTokenizerFast
from transformers.models.deberta.tokenization_deberta import VOCAB_FILES_NAMES
from transformers.testing_utils import slow
from ...test_tokenization_common import TokenizerTesterMixin
class A ( __UpperCAmelCase , unittest.TestCase ):
lowerCamelCase : Tuple = DebertaTokenizer
lowerCamelCase : Any = True
lowerCamelCase : Dict = DebertaTokenizerFast
def A__ ( self ) -> List[str]:
'''simple docstring'''
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
lowercase__ = [
"""l""",
"""o""",
"""w""",
"""e""",
"""r""",
"""s""",
"""t""",
"""i""",
"""d""",
"""n""",
"""\u0120""",
"""\u0120l""",
"""\u0120n""",
"""\u0120lo""",
"""\u0120low""",
"""er""",
"""\u0120lowest""",
"""\u0120newer""",
"""\u0120wider""",
"""[UNK]""",
]
lowercase__ = dict(zip(lowerCamelCase__ , range(len(lowerCamelCase__ ) ) ) )
lowercase__ = ["""#version: 0.2""", """\u0120 l""", """\u0120l o""", """\u0120lo w""", """e r""", """"""]
lowercase__ = {"""unk_token""": """[UNK]"""}
lowercase__ = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["""vocab_file"""] )
lowercase__ = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["""merges_file"""] )
with open(self.vocab_file , """w""" , encoding="""utf-8""" ) as fp:
fp.write(json.dumps(lowerCamelCase__ ) + """\n""" )
with open(self.merges_file , """w""" , encoding="""utf-8""" ) as fp:
fp.write("""\n""".join(lowerCamelCase__ ) )
def A__ ( self , **lowerCamelCase__ ) -> str:
'''simple docstring'''
kwargs.update(self.special_tokens_map )
return self.tokenizer_class.from_pretrained(self.tmpdirname , **lowerCamelCase__ )
def A__ ( self , lowerCamelCase__ ) -> List[Any]:
'''simple docstring'''
lowercase__ = """lower newer"""
lowercase__ = """lower newer"""
return input_text, output_text
def A__ ( self ) -> Optional[int]:
'''simple docstring'''
lowercase__ = self.get_tokenizer()
lowercase__ = """lower newer"""
lowercase__ = ["""l""", """o""", """w""", """er""", """\u0120""", """n""", """e""", """w""", """er"""]
lowercase__ = tokenizer.tokenize(lowerCamelCase__ )
self.assertListEqual(lowerCamelCase__ , lowerCamelCase__ )
lowercase__ = tokens + [tokenizer.unk_token]
lowercase__ = [0, 1, 2, 15, 10, 9, 3, 2, 15, 19]
self.assertListEqual(tokenizer.convert_tokens_to_ids(lowerCamelCase__ ) , lowerCamelCase__ )
def A__ ( self ) -> Optional[Any]:
'''simple docstring'''
lowercase__ = self.get_tokenizer()
lowercase__ = tokenizer("""Hello""" , """World""" )
lowercase__ = [0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1]
self.assertListEqual(tokd["""token_type_ids"""] , lowerCamelCase__ )
@slow
def A__ ( self ) -> Any:
'''simple docstring'''
lowercase__ = self.tokenizer_class.from_pretrained("""microsoft/deberta-base""" )
lowercase__ = tokenizer.encode("""sequence builders""" , add_special_tokens=lowerCamelCase__ )
lowercase__ = tokenizer.encode("""multi-sequence build""" , add_special_tokens=lowerCamelCase__ )
lowercase__ = tokenizer.encode(
"""sequence builders""" , add_special_tokens=lowerCamelCase__ , add_prefix_space=lowerCamelCase__ )
lowercase__ = tokenizer.encode(
"""sequence builders""" , """multi-sequence build""" , add_special_tokens=lowerCamelCase__ , add_prefix_space=lowerCamelCase__ )
lowercase__ = tokenizer.build_inputs_with_special_tokens(lowerCamelCase__ )
lowercase__ = tokenizer.build_inputs_with_special_tokens(lowerCamelCase__ , lowerCamelCase__ )
assert encoded_sentence == encoded_text_from_decode
assert encoded_pair == encoded_pair_from_decode
@slow
def A__ ( self ) -> Tuple:
'''simple docstring'''
lowercase__ = [self.tokenizer_class]
if self.test_rust_tokenizer:
tokenizer_classes.append(self.rust_tokenizer_class )
for tokenizer_class in tokenizer_classes:
lowercase__ = tokenizer_class.from_pretrained("""microsoft/deberta-base""" )
lowercase__ = [
"""ALBERT: A Lite BERT for Self-supervised Learning of Language Representations""",
"""ALBERT incorporates two parameter reduction techniques""",
"""The first one is a factorized embedding parameterization. By decomposing the large vocabulary"""
""" embedding matrix into two small matrices, we separate the size of the hidden layers from the size of"""
""" vocabulary embedding.""",
]
lowercase__ = tokenizer(lowerCamelCase__ , padding=lowerCamelCase__ )
lowercase__ = [tokenizer.decode(lowerCamelCase__ , skip_special_tokens=lowerCamelCase__ ) for seq in encoding["""input_ids"""]]
# fmt: off
lowercase__ = {
"""input_ids""": [
[1, 2_118, 11_126, 565, 35, 83, 25_191, 163, 18_854, 13, 12_156, 12, 16_101, 25_376, 13_807, 9, 22_205, 27_893, 1_635, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 2_118, 11_126, 565, 24_536, 80, 43_797, 4_878, 7_373, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[1, 133, 78, 65, 16, 10, 3_724, 1_538, 33_183, 11_303, 43_797, 1_938, 4, 870, 24_165, 29_105, 5, 739, 32_644, 33_183, 11_303, 36_173, 88, 80, 650, 7_821, 45_940, 6, 52, 2_559, 5, 1_836, 9, 5, 7_397, 13_171, 31, 5, 1_836, 9, 32_644, 33_183, 11_303, 4, 2]
],
"""token_type_ids""": [
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
],
"""attention_mask""": [
[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],
[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],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
]
}
# fmt: on
lowercase__ = [
"""ALBERT: A Lite BERT for Self-supervised Learning of Language Representations""",
"""ALBERT incorporates two parameter reduction techniques""",
"""The first one is a factorized embedding parameterization. By decomposing the large vocabulary"""
""" embedding matrix into two small matrices, we separate the size of the hidden layers from the size of"""
""" vocabulary embedding.""",
]
self.assertDictEqual(encoding.data , lowerCamelCase__ )
for expected, decoded in zip(lowerCamelCase__ , lowerCamelCase__ ):
self.assertEqual(lowerCamelCase__ , lowerCamelCase__ )
| 325
| 0
|
'''simple docstring'''
from typing import TYPE_CHECKING
# rely on isort to merge the imports
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
UpperCAmelCase_ = {
"configuration_informer": [
"INFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"InformerConfig",
],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCAmelCase_ = [
"INFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"InformerForPrediction",
"InformerModel",
"InformerPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_informer import INFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP, InformerConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_informer import (
INFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
InformerForPrediction,
InformerModel,
InformerPreTrainedModel,
)
else:
import sys
UpperCAmelCase_ = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| 490
|
'''simple docstring'''
def SCREAMING_SNAKE_CASE ( a_ : str ):
__a = ''
for ch in key:
if ch == " " or ch not in key_no_dups and ch.isalpha():
key_no_dups += ch
return key_no_dups
def SCREAMING_SNAKE_CASE ( a_ : str ):
__a = [chr(i + 65 ) for i in range(26 )]
# Remove duplicate characters from key
__a = remove_duplicates(key.upper() )
__a = len(a_ )
# First fill cipher with key characters
__a = {alphabet[i]: char for i, char in enumerate(a_ )}
# Then map remaining characters in alphabet to
# the alphabet from the beginning
for i in range(len(a_ ) , 26 ):
__a = alphabet[i - offset]
# Ensure we are not mapping letters to letters previously mapped
while char in key:
offset -= 1
__a = alphabet[i - offset]
__a = char
return cipher_alphabet
def SCREAMING_SNAKE_CASE ( a_ : str , a_ : dict[str, str] ):
return "".join(cipher_map.get(a_ , a_ ) for ch in message.upper() )
def SCREAMING_SNAKE_CASE ( a_ : str , a_ : dict[str, str] ):
__a = {v: k for k, v in cipher_map.items()}
return "".join(rev_cipher_map.get(a_ , a_ ) for ch in message.upper() )
def SCREAMING_SNAKE_CASE ( ):
__a = input('Enter message to encode or decode: ' ).strip()
__a = input('Enter keyword: ' ).strip()
__a = input('Encipher or decipher? E/D:' ).strip()[0].lower()
try:
__a = {'e': encipher, 'd': decipher}[option]
except KeyError:
raise KeyError('invalid input option' )
__a = create_cipher_map(a_ )
print(func(a_ , a_ ) )
if __name__ == "__main__":
import doctest
doctest.testmod()
main()
| 490
| 1
|
from PIL import Image
def A_ ( _UpperCAmelCase , _UpperCAmelCase ):
SCREAMING_SNAKE_CASE_: Optional[int] = (2_59 * (level + 2_55)) / (2_55 * (2_59 - level))
def contrast(_UpperCAmelCase ) -> 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
lowerCAmelCase : int = change_contrast(img, 170)
cont_img.save("""image_data/lena_high_contrast.png""", format="""png""")
| 671
|
import argparse
import os
import evaluate
import torch
from datasets import load_dataset
from torch.optim import AdamW
from torch.utils.data import DataLoader
from transformers import AutoModelForSequenceClassification, AutoTokenizer, get_linear_schedule_with_warmup, set_seed
from accelerate import Accelerator, DistributedType
########################################################################
# This is a fully working simple example to use Accelerate
# and perform gradient accumulation
#
# This example trains a Bert base model on GLUE MRPC
# in any of the following settings (with the same script):
# - single CPU or single GPU
# - multi GPUS (using PyTorch distributed mode)
# - (multi) TPUs
# - fp16 (mixed-precision) or fp32 (normal precision)
#
# To run it in each of these various modes, follow the instructions
# in the readme for examples:
# https://github.com/huggingface/accelerate/tree/main/examples
#
########################################################################
lowerCAmelCase : str = 16
lowerCAmelCase : List[Any] = 32
def A_ ( _UpperCAmelCase , _UpperCAmelCase = 16 ):
SCREAMING_SNAKE_CASE_: List[Any] = AutoTokenizer.from_pretrained("bert-base-cased" )
SCREAMING_SNAKE_CASE_: Tuple = load_dataset("glue" , "mrpc" )
def tokenize_function(_UpperCAmelCase ):
# max_length=None => use the model max length (it's actually the default)
SCREAMING_SNAKE_CASE_: List[Any] = 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():
SCREAMING_SNAKE_CASE_: str = 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
SCREAMING_SNAKE_CASE_: Optional[Any] = tokenized_datasets.rename_column("label" , "labels" )
def collate_fn(_UpperCAmelCase ):
# On TPU it's best to pad everything to the same length or training will be very slow.
SCREAMING_SNAKE_CASE_: List[Any] = 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":
SCREAMING_SNAKE_CASE_: Tuple = 16
elif accelerator.mixed_precision != "no":
SCREAMING_SNAKE_CASE_: int = 8
else:
SCREAMING_SNAKE_CASE_: Any = None
return tokenizer.pad(
_UpperCAmelCase , padding="longest" , max_length=_UpperCAmelCase , pad_to_multiple_of=_UpperCAmelCase , return_tensors="pt" , )
# Instantiate dataloaders.
SCREAMING_SNAKE_CASE_: Optional[Any] = DataLoader(
tokenized_datasets["train"] , shuffle=_UpperCAmelCase , collate_fn=_UpperCAmelCase , batch_size=_UpperCAmelCase )
SCREAMING_SNAKE_CASE_: Tuple = 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
lowerCAmelCase : Optional[int] = mocked_dataloaders # noqa: F811
def A_ ( _UpperCAmelCase , _UpperCAmelCase ):
# For testing only
if os.environ.get("TESTING_MOCKED_DATALOADERS" , _UpperCAmelCase ) == "1":
SCREAMING_SNAKE_CASE_: Tuple = 2
# New Code #
SCREAMING_SNAKE_CASE_: List[str] = int(args.gradient_accumulation_steps )
# Initialize accelerator
SCREAMING_SNAKE_CASE_: int = Accelerator(
cpu=args.cpu , mixed_precision=args.mixed_precision , gradient_accumulation_steps=_UpperCAmelCase )
if accelerator.distributed_type == DistributedType.TPU and gradient_accumulation_steps > 1:
raise NotImplementedError(
"Gradient accumulation on TPUs is currently not supported. Pass `gradient_accumulation_steps=1`" )
# Sample hyper-parameters for learning rate, batch size, seed and a few other HPs
SCREAMING_SNAKE_CASE_: Tuple = config["lr"]
SCREAMING_SNAKE_CASE_: List[str] = int(config["num_epochs"] )
SCREAMING_SNAKE_CASE_: List[str] = int(config["seed"] )
SCREAMING_SNAKE_CASE_: Optional[int] = int(config["batch_size"] )
SCREAMING_SNAKE_CASE_: str = evaluate.load("glue" , "mrpc" )
set_seed(_UpperCAmelCase )
SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_: List[str] = get_dataloaders(_UpperCAmelCase , _UpperCAmelCase )
# Instantiate the model (we build the model here so that the seed also control new weights initialization)
SCREAMING_SNAKE_CASE_: Union[str, Any] = 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).
SCREAMING_SNAKE_CASE_: List[Any] = model.to(accelerator.device )
# Instantiate optimizer
SCREAMING_SNAKE_CASE_: Union[str, Any] = AdamW(params=model.parameters() , lr=_UpperCAmelCase )
# Instantiate scheduler
SCREAMING_SNAKE_CASE_: str = 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.
SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_: Dict = 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 )
# New code #
# We use the new `accumulate` context manager to perform gradient accumulation
# We also currently do not support TPUs nor advise it as bugs were found on the XLA side when running our tests.
with accelerator.accumulate(_UpperCAmelCase ):
SCREAMING_SNAKE_CASE_: List[Any] = model(**_UpperCAmelCase )
SCREAMING_SNAKE_CASE_: List[Any] = output.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():
SCREAMING_SNAKE_CASE_: Optional[Any] = model(**_UpperCAmelCase )
SCREAMING_SNAKE_CASE_: List[Any] = outputs.logits.argmax(dim=-1 )
SCREAMING_SNAKE_CASE_ , SCREAMING_SNAKE_CASE_: List[Any] = accelerator.gather_for_metrics((predictions, batch["labels"]) )
metric.add_batch(
predictions=_UpperCAmelCase , references=_UpperCAmelCase , )
SCREAMING_SNAKE_CASE_: List[str] = metric.compute()
# Use accelerator.print to print only on the main process.
accelerator.print(f"epoch {epoch}:" , _UpperCAmelCase )
def A_ ( ):
SCREAMING_SNAKE_CASE_: str = 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." , )
# New Code #
parser.add_argument(
"--gradient_accumulation_steps" , type=_UpperCAmelCase , default=1 , help="The number of minibatches to be ran before gradients are accumulated." , )
parser.add_argument("--cpu" , action="store_true" , help="If passed, will train on the CPU." )
SCREAMING_SNAKE_CASE_: List[Any] = parser.parse_args()
SCREAMING_SNAKE_CASE_: Tuple = {"lr": 2e-5, "num_epochs": 3, "seed": 42, "batch_size": 16}
training_function(_UpperCAmelCase , _UpperCAmelCase )
if __name__ == "__main__":
main()
| 671
| 1
|
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_sentencepiece_available,
is_speech_available,
is_torch_available,
)
lowerCAmelCase__: Optional[Any] = {
'''configuration_trocr''': ['''TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''TrOCRConfig'''],
'''processing_trocr''': ['''TrOCRProcessor'''],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
lowerCAmelCase__: int = [
'''TROCR_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''TrOCRForCausalLM''',
'''TrOCRPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_trocr import TROCR_PRETRAINED_CONFIG_ARCHIVE_MAP, TrOCRConfig
from .processing_trocr import TrOCRProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_trocr import TROCR_PRETRAINED_MODEL_ARCHIVE_LIST, TrOCRForCausalLM, TrOCRPreTrainedModel
else:
import sys
lowerCAmelCase__: Optional[int] = _LazyModule(__name__, globals()["__file__"], _import_structure, module_spec=__spec__)
| 711
|
from string import ascii_lowercase, ascii_uppercase
def __SCREAMING_SNAKE_CASE ( SCREAMING_SNAKE_CASE ) -> str:
if not sentence:
return ""
SCREAMING_SNAKE_CASE_ : int = dict(zip(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ) )
return lower_to_upper.get(sentence[0] , sentence[0] ) + sentence[1:]
if __name__ == "__main__":
from doctest import testmod
testmod()
| 311
| 0
|
SCREAMING_SNAKE_CASE__ = [
'''DownloadConfig''',
'''DownloadManager''',
'''DownloadMode''',
'''StreamingDownloadManager''',
]
from .download_config import DownloadConfig
from .download_manager import DownloadManager, DownloadMode
from .streaming_download_manager import StreamingDownloadManager
| 9
|
import json
import os
import unittest
from transformers.models.roc_bert.tokenization_roc_bert import (
VOCAB_FILES_NAMES,
RoCBertBasicTokenizer,
RoCBertTokenizer,
RoCBertWordpieceTokenizer,
_is_control,
_is_punctuation,
_is_whitespace,
)
from transformers.testing_utils import require_tokenizers, slow
from ...test_tokenization_common import TokenizerTesterMixin, filter_non_english
@require_tokenizers
class _a (__magic_name__ , unittest.TestCase ):
'''simple docstring'''
UpperCAmelCase__: List[str] = RoCBertTokenizer
UpperCAmelCase__: Dict = None
UpperCAmelCase__: Optional[Any] = False
UpperCAmelCase__: Union[str, Any] = True
UpperCAmelCase__: Union[str, Any] = filter_non_english
def __A ( self ):
super().setUp()
A__ : Tuple = ["""[UNK]""", """[CLS]""", """[SEP]""", """[PAD]""", """[MASK]""", """你""", """好""", """是""", """谁""", """a""", """b""", """c""", """d"""]
A__ : Union[str, Any] = {}
A__ : Dict = {}
for i, value in enumerate(A__ ):
A__ : Optional[int] = i
A__ : Optional[int] = i
A__ : Dict = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["""vocab_file"""] )
A__ : Optional[Any] = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["""word_shape_file"""] )
A__ : int = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES["""word_pronunciation_file"""] )
with open(self.vocab_file , """w""" , encoding="""utf-8""" ) as vocab_writer:
vocab_writer.write("""""".join([x + """\n""" for x in vocab_tokens] ) )
with open(self.word_shape_file , """w""" , encoding="""utf-8""" ) as word_shape_writer:
json.dump(A__ , A__ , ensure_ascii=A__ )
with open(self.word_pronunciation_file , """w""" , encoding="""utf-8""" ) as word_pronunciation_writer:
json.dump(A__ , A__ , ensure_ascii=A__ )
def __A ( self ):
A__ : Dict = self.tokenizer_class(self.vocab_file , self.word_shape_file , self.word_pronunciation_file )
A__ : Optional[Any] = tokenizer.tokenize("""你好[SEP]你是谁""" )
self.assertListEqual(A__ , ["""你""", """好""", """[SEP]""", """你""", """是""", """谁"""] )
self.assertListEqual(tokenizer.convert_tokens_to_ids(A__ ) , [5, 6, 2, 5, 7, 8] )
self.assertListEqual(tokenizer.convert_tokens_to_shape_ids(A__ ) , [5, 6, 2, 5, 7, 8] )
self.assertListEqual(tokenizer.convert_tokens_to_pronunciation_ids(A__ ) , [5, 6, 2, 5, 7, 8] )
def __A ( self ):
A__ : Dict = RoCBertBasicTokenizer()
self.assertListEqual(tokenizer.tokenize("""ah\u535A\u63A8zz""" ) , ["""ah""", """\u535A""", """\u63A8""", """zz"""] )
def __A ( self ):
A__ : List[str] = RoCBertBasicTokenizer(do_lower_case=A__ )
self.assertListEqual(
tokenizer.tokenize(""" \tHeLLo!how \n Are yoU? """ ) , ["""hello""", """!""", """how""", """are""", """you""", """?"""] )
self.assertListEqual(tokenizer.tokenize("""H\u00E9llo""" ) , ["""hello"""] )
def __A ( self ):
A__ : int = RoCBertBasicTokenizer(do_lower_case=A__ , strip_accents=A__ )
self.assertListEqual(
tokenizer.tokenize(""" \tHäLLo!how \n Are yoU? """ ) , ["""hällo""", """!""", """how""", """are""", """you""", """?"""] )
self.assertListEqual(tokenizer.tokenize("""H\u00E9llo""" ) , ["""h\u00E9llo"""] )
def __A ( self ):
A__ : Optional[Any] = RoCBertBasicTokenizer(do_lower_case=A__ , strip_accents=A__ )
self.assertListEqual(
tokenizer.tokenize(""" \tHäLLo!how \n Are yoU? """ ) , ["""hallo""", """!""", """how""", """are""", """you""", """?"""] )
self.assertListEqual(tokenizer.tokenize("""H\u00E9llo""" ) , ["""hello"""] )
def __A ( self ):
A__ : Union[str, Any] = RoCBertBasicTokenizer(do_lower_case=A__ )
self.assertListEqual(
tokenizer.tokenize(""" \tHäLLo!how \n Are yoU? """ ) , ["""hallo""", """!""", """how""", """are""", """you""", """?"""] )
self.assertListEqual(tokenizer.tokenize("""H\u00E9llo""" ) , ["""hello"""] )
def __A ( self ):
A__ : Union[str, Any] = RoCBertBasicTokenizer(do_lower_case=A__ )
self.assertListEqual(
tokenizer.tokenize(""" \tHeLLo!how \n Are yoU? """ ) , ["""HeLLo""", """!""", """how""", """Are""", """yoU""", """?"""] )
def __A ( self ):
A__ : Dict = RoCBertBasicTokenizer(do_lower_case=A__ , strip_accents=A__ )
self.assertListEqual(
tokenizer.tokenize(""" \tHäLLo!how \n Are yoU? """ ) , ["""HäLLo""", """!""", """how""", """Are""", """yoU""", """?"""] )
def __A ( self ):
A__ : Optional[Any] = RoCBertBasicTokenizer(do_lower_case=A__ , strip_accents=A__ )
self.assertListEqual(
tokenizer.tokenize(""" \tHäLLo!how \n Are yoU? """ ) , ["""HaLLo""", """!""", """how""", """Are""", """yoU""", """?"""] )
def __A ( self ):
A__ : Dict = RoCBertBasicTokenizer(do_lower_case=A__ , never_split=["""[UNK]"""] )
self.assertListEqual(
tokenizer.tokenize(""" \tHeLLo!how \n Are yoU? [UNK]""" ) , ["""HeLLo""", """!""", """how""", """Are""", """yoU""", """?""", """[UNK]"""] )
def __A ( self ):
A__ : List[Any] = ["""[UNK]""", """[CLS]""", """[SEP]""", """want""", """##want""", """##ed""", """wa""", """un""", """runn""", """##ing"""]
A__ : Any = {}
for i, token in enumerate(A__ ):
A__ : Optional[int] = i
A__ : Optional[Any] = RoCBertWordpieceTokenizer(vocab=A__ , unk_token="""[UNK]""" )
self.assertListEqual(tokenizer.tokenize("""""" ) , [] )
self.assertListEqual(tokenizer.tokenize("""unwanted running""" ) , ["""un""", """##want""", """##ed""", """runn""", """##ing"""] )
self.assertListEqual(tokenizer.tokenize("""unwantedX running""" ) , ["""[UNK]""", """runn""", """##ing"""] )
def __A ( self ):
self.assertTrue(_is_whitespace(""" """ ) )
self.assertTrue(_is_whitespace("""\t""" ) )
self.assertTrue(_is_whitespace("""\r""" ) )
self.assertTrue(_is_whitespace("""\n""" ) )
self.assertTrue(_is_whitespace("""\u00A0""" ) )
self.assertFalse(_is_whitespace("""A""" ) )
self.assertFalse(_is_whitespace("""-""" ) )
def __A ( self ):
self.assertTrue(_is_control("""\u0005""" ) )
self.assertFalse(_is_control("""A""" ) )
self.assertFalse(_is_control(""" """ ) )
self.assertFalse(_is_control("""\t""" ) )
self.assertFalse(_is_control("""\r""" ) )
def __A ( self ):
self.assertTrue(_is_punctuation("""-""" ) )
self.assertTrue(_is_punctuation("""$""" ) )
self.assertTrue(_is_punctuation("""`""" ) )
self.assertTrue(_is_punctuation(""".""" ) )
self.assertFalse(_is_punctuation("""A""" ) )
self.assertFalse(_is_punctuation(""" """ ) )
def __A ( self ):
A__ : Optional[int] = self.get_tokenizer()
# Example taken from the issue https://github.com/huggingface/tokenizers/issues/340
self.assertListEqual([tokenizer.tokenize(A__ ) for t in ["""Test""", """\xad""", """test"""]] , [["""[UNK]"""], [], ["""[UNK]"""]] )
if self.test_rust_tokenizer:
A__ : Tuple = self.get_rust_tokenizer()
self.assertListEqual(
[rust_tokenizer.tokenize(A__ ) for t in ["""Test""", """\xad""", """test"""]] , [["""[UNK]"""], [], ["""[UNK]"""]] )
def __A ( self ):
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(F"""{tokenizer.__class__.__name__} ({pretrained_name})""" ):
A__ : Dict = self.rust_tokenizer_class.from_pretrained(A__ , **A__ )
A__ : str = F"""A, naïve {tokenizer_r.mask_token} AllenNLP sentence."""
A__ : Union[str, Any] = tokenizer_r.encode_plus(
A__ , return_attention_mask=A__ , return_token_type_ids=A__ , return_offsets_mapping=A__ , add_special_tokens=A__ , )
A__ : Any = tokenizer_r.do_lower_case if hasattr(A__ , """do_lower_case""" ) else False
A__ : List[str] = (
[
((0, 0), tokenizer_r.cls_token),
((0, 1), """A"""),
((1, 2), ""","""),
((3, 5), """na"""),
((5, 6), """##ï"""),
((6, 8), """##ve"""),
((9, 15), tokenizer_r.mask_token),
((16, 21), """Allen"""),
((21, 23), """##NL"""),
((23, 24), """##P"""),
((25, 33), """sentence"""),
((33, 34), """."""),
((0, 0), tokenizer_r.sep_token),
]
if not do_lower_case
else [
((0, 0), tokenizer_r.cls_token),
((0, 1), """a"""),
((1, 2), ""","""),
((3, 8), """naive"""),
((9, 15), tokenizer_r.mask_token),
((16, 21), """allen"""),
((21, 23), """##nl"""),
((23, 24), """##p"""),
((25, 33), """sentence"""),
((33, 34), """."""),
((0, 0), tokenizer_r.sep_token),
]
)
self.assertEqual(
[e[1] for e in expected_results] , tokenizer_r.convert_ids_to_tokens(tokens["""input_ids"""] ) )
self.assertEqual([e[0] for e in expected_results] , tokens["""offset_mapping"""] )
def __A ( self ):
A__ : Union[str, Any] = ["""的""", """人""", """有"""]
A__ : List[str] = """""".join(A__ )
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(F"""{tokenizer.__class__.__name__} ({pretrained_name})""" ):
A__ : Any = True
A__ : int = self.tokenizer_class.from_pretrained(A__ , **A__ )
A__ : Dict = self.rust_tokenizer_class.from_pretrained(A__ , **A__ )
A__ : List[str] = tokenizer_p.encode(A__ , add_special_tokens=A__ )
A__ : List[Any] = tokenizer_r.encode(A__ , add_special_tokens=A__ )
A__ : Tuple = tokenizer_r.convert_ids_to_tokens(A__ )
A__ : int = tokenizer_p.convert_ids_to_tokens(A__ )
# it is expected that each Chinese character is not preceded by "##"
self.assertListEqual(A__ , A__ )
self.assertListEqual(A__ , A__ )
A__ : Optional[int] = False
A__ : Union[str, Any] = self.rust_tokenizer_class.from_pretrained(A__ , **A__ )
A__ : str = self.tokenizer_class.from_pretrained(A__ , **A__ )
A__ : Tuple = tokenizer_r.encode(A__ , add_special_tokens=A__ )
A__ : List[str] = tokenizer_p.encode(A__ , add_special_tokens=A__ )
A__ : Union[str, Any] = tokenizer_r.convert_ids_to_tokens(A__ )
A__ : Tuple = tokenizer_p.convert_ids_to_tokens(A__ )
# it is expected that only the first Chinese character is not preceded by "##".
A__ : str = [
F"""##{token}""" if idx != 0 else token for idx, token in enumerate(A__ )
]
self.assertListEqual(A__ , A__ )
self.assertListEqual(A__ , A__ )
@slow
def __A ( self ):
A__ : str = self.tokenizer_class(self.vocab_file , self.word_shape_file , self.word_pronunciation_file )
A__ : Optional[Any] = tokenizer.encode("""你好""" , add_special_tokens=A__ )
A__ : List[Any] = tokenizer.encode("""你是谁""" , add_special_tokens=A__ )
A__ : Any = tokenizer.build_inputs_with_special_tokens(A__ )
A__ : Any = tokenizer.build_inputs_with_special_tokens(A__ , A__ )
assert encoded_sentence == [1] + text + [2]
assert encoded_pair == [1] + text + [2] + text_a + [2]
def __A ( self ):
A__ : List[str] = self.get_tokenizers(do_lower_case=A__ )
for tokenizer in tokenizers:
with self.subTest(F"""{tokenizer.__class__.__name__}""" ):
A__ : Optional[int] = """你好，你是谁"""
A__ : List[str] = tokenizer.tokenize(A__ )
A__ : Union[str, Any] = tokenizer.convert_tokens_to_ids(A__ )
A__ : str = tokenizer.convert_tokens_to_shape_ids(A__ )
A__ : Optional[int] = tokenizer.convert_tokens_to_pronunciation_ids(A__ )
A__ : Union[str, Any] = tokenizer.prepare_for_model(
A__ , A__ , A__ , add_special_tokens=A__ )
A__ : int = tokenizer.encode_plus(A__ , add_special_tokens=A__ )
self.assertEqual(A__ , A__ )
| 456
| 0
|
import shutil
import tempfile
import unittest
from transformers import ClapFeatureExtractor, ClapProcessor, RobertaTokenizer, RobertaTokenizerFast
from transformers.testing_utils import require_sentencepiece, require_torchaudio
from .test_feature_extraction_clap import floats_list
@require_torchaudio
@require_sentencepiece
class __lowerCAmelCase ( unittest.TestCase ):
"""simple docstring"""
def snake_case_ ( self : Optional[Any] ):
__lowercase : List[str] = '''laion/clap-htsat-unfused'''
__lowercase : Optional[Any] = tempfile.mkdtemp()
def snake_case_ ( self : str , **_snake_case : Any ):
return RobertaTokenizer.from_pretrained(self.checkpoint , **_snake_case )
def snake_case_ ( self : Dict , **_snake_case : Any ):
return ClapFeatureExtractor.from_pretrained(self.checkpoint , **_snake_case )
def snake_case_ ( self : Optional[Any] ):
shutil.rmtree(self.tmpdirname )
def snake_case_ ( self : str ):
__lowercase : int = self.get_tokenizer()
__lowercase : int = self.get_feature_extractor()
__lowercase : Optional[Any] = ClapProcessor(tokenizer=_snake_case , feature_extractor=_snake_case )
processor.save_pretrained(self.tmpdirname )
__lowercase : List[str] = ClapProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() )
self.assertIsInstance(processor.tokenizer , _snake_case )
self.assertEqual(processor.feature_extractor.to_json_string() , feature_extractor.to_json_string() )
self.assertIsInstance(processor.feature_extractor , _snake_case )
def snake_case_ ( self : Tuple ):
__lowercase : Dict = ClapProcessor(tokenizer=self.get_tokenizer() , feature_extractor=self.get_feature_extractor() )
processor.save_pretrained(self.tmpdirname )
__lowercase : Any = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' )
__lowercase : Tuple = self.get_feature_extractor(do_normalize=_snake_case , padding_value=1.0 )
__lowercase : int = ClapProcessor.from_pretrained(
self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=_snake_case , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , _snake_case )
self.assertEqual(processor.feature_extractor.to_json_string() , feature_extractor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.feature_extractor , _snake_case )
def snake_case_ ( self : Union[str, Any] ):
__lowercase : Optional[Any] = self.get_feature_extractor()
__lowercase : Optional[int] = self.get_tokenizer()
__lowercase : Optional[int] = ClapProcessor(tokenizer=_snake_case , feature_extractor=_snake_case )
__lowercase : Tuple = floats_list((3, 1000) )
__lowercase : Any = feature_extractor(_snake_case , return_tensors='''np''' )
__lowercase : Optional[Any] = processor(audios=_snake_case , 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 snake_case_ ( self : Union[str, Any] ):
__lowercase : Dict = self.get_feature_extractor()
__lowercase : Optional[int] = self.get_tokenizer()
__lowercase : int = ClapProcessor(tokenizer=_snake_case , feature_extractor=_snake_case )
__lowercase : Any = '''This is a test string'''
__lowercase : str = processor(text=_snake_case )
__lowercase : Dict = tokenizer(_snake_case )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def snake_case_ ( self : Optional[Any] ):
__lowercase : int = self.get_feature_extractor()
__lowercase : Union[str, Any] = self.get_tokenizer()
__lowercase : Dict = ClapProcessor(tokenizer=_snake_case , feature_extractor=_snake_case )
__lowercase : Dict = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
__lowercase : Dict = processor.batch_decode(_snake_case )
__lowercase : Tuple = tokenizer.batch_decode(_snake_case )
self.assertListEqual(_snake_case , _snake_case )
def snake_case_ ( self : Optional[int] ):
__lowercase : Optional[Any] = self.get_feature_extractor()
__lowercase : Optional[int] = self.get_tokenizer()
__lowercase : Union[str, Any] = ClapProcessor(tokenizer=_snake_case , feature_extractor=_snake_case )
self.assertListEqual(
processor.model_input_names[2:] , feature_extractor.model_input_names , msg='''`processor` and `feature_extractor` model input names do not match''' , )
| 284
|
from math import factorial
__lowerCAmelCase : Dict = {str(d): factorial(d) for d in range(10)}
def UpperCAmelCase_ ( __lowerCAmelCase ) -> int:
return sum(DIGIT_FACTORIAL[d] for d in str(__lowerCAmelCase ) )
def UpperCAmelCase_ ( ) -> int:
__lowercase : int = 7 * factorial(9 ) + 1
return sum(i for i in range(3 , __lowerCAmelCase ) if sum_of_digit_factorial(__lowerCAmelCase ) == i )
if __name__ == "__main__":
print(F'{solution() = }')
| 284
| 1
|
'''simple docstring'''
import argparse
import json
from pathlib import Path
import requests
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import YolosConfig, YolosForObjectDetection, YolosImageProcessor
from transformers.utils import logging
logging.set_verbosity_info()
UpperCAmelCase = logging.get_logger(__name__)
def __UpperCamelCase ( lowercase__ : str ):
'''simple docstring'''
__lowercase =YolosConfig()
# size of the architecture
if "yolos_ti" in yolos_name:
__lowercase =1_92
__lowercase =7_68
__lowercase =12
__lowercase =3
__lowercase =[8_00, 13_33]
__lowercase =False
elif yolos_name == "yolos_s_dWr":
__lowercase =3_30
__lowercase =14
__lowercase =6
__lowercase =13_20
elif "yolos_s" in yolos_name:
__lowercase =3_84
__lowercase =15_36
__lowercase =12
__lowercase =6
elif "yolos_b" in yolos_name:
__lowercase =[8_00, 13_44]
__lowercase =91
__lowercase ='huggingface/label-files'
__lowercase ='coco-detection-id2label.json'
__lowercase =json.load(open(hf_hub_download(lowercase__, lowercase__, repo_type='dataset' ), 'r' ) )
__lowercase ={int(lowercase__ ): v for k, v in idalabel.items()}
__lowercase =idalabel
__lowercase ={v: k for k, v in idalabel.items()}
return config
def __UpperCamelCase ( lowercase__ : dict, lowercase__ : YolosConfig, lowercase__ : bool = False ):
'''simple docstring'''
for i in range(config.num_hidden_layers ):
# read in weights + bias of input projection layer (in timm, this is a single matrix + bias)
__lowercase =state_dict.pop(F'''blocks.{i}.attn.qkv.weight''' )
__lowercase =state_dict.pop(F'''blocks.{i}.attn.qkv.bias''' )
# next, add query, keys and values (in that order) to the state dict
__lowercase =in_proj_weight[: config.hidden_size, :]
__lowercase =in_proj_bias[: config.hidden_size]
__lowercase =in_proj_weight[
config.hidden_size : config.hidden_size * 2, :
]
__lowercase =in_proj_bias[
config.hidden_size : config.hidden_size * 2
]
__lowercase =in_proj_weight[-config.hidden_size :, :]
__lowercase =in_proj_bias[-config.hidden_size :]
def __UpperCamelCase ( lowercase__ : str ):
'''simple docstring'''
if "backbone" in name:
__lowercase =name.replace('backbone', 'vit' )
if "cls_token" in name:
__lowercase =name.replace('cls_token', 'embeddings.cls_token' )
if "det_token" in name:
__lowercase =name.replace('det_token', 'embeddings.detection_tokens' )
if "mid_pos_embed" in name:
__lowercase =name.replace('mid_pos_embed', 'encoder.mid_position_embeddings' )
if "pos_embed" in name:
__lowercase =name.replace('pos_embed', 'embeddings.position_embeddings' )
if "patch_embed.proj" in name:
__lowercase =name.replace('patch_embed.proj', 'embeddings.patch_embeddings.projection' )
if "blocks" in name:
__lowercase =name.replace('blocks', 'encoder.layer' )
if "attn.proj" in name:
__lowercase =name.replace('attn.proj', 'attention.output.dense' )
if "attn" in name:
__lowercase =name.replace('attn', 'attention.self' )
if "norm1" in name:
__lowercase =name.replace('norm1', 'layernorm_before' )
if "norm2" in name:
__lowercase =name.replace('norm2', 'layernorm_after' )
if "mlp.fc1" in name:
__lowercase =name.replace('mlp.fc1', 'intermediate.dense' )
if "mlp.fc2" in name:
__lowercase =name.replace('mlp.fc2', 'output.dense' )
if "class_embed" in name:
__lowercase =name.replace('class_embed', 'class_labels_classifier' )
if "bbox_embed" in name:
__lowercase =name.replace('bbox_embed', 'bbox_predictor' )
if "vit.norm" in name:
__lowercase =name.replace('vit.norm', 'vit.layernorm' )
return name
def __UpperCamelCase ( lowercase__ : dict, lowercase__ : YolosForObjectDetection ):
'''simple docstring'''
for key in orig_state_dict.copy().keys():
__lowercase =orig_state_dict.pop(lowercase__ )
if "qkv" in key:
__lowercase =key.split('.' )
__lowercase =int(key_split[2] )
__lowercase =model.vit.encoder.layer[layer_num].attention.attention.all_head_size
if "weight" in key:
__lowercase =val[:dim, :]
__lowercase =val[
dim : dim * 2, :
]
__lowercase =val[-dim:, :]
else:
__lowercase =val[:dim]
__lowercase =val[dim : dim * 2]
__lowercase =val[-dim:]
else:
__lowercase =val
return orig_state_dict
def __UpperCamelCase ( ):
'''simple docstring'''
__lowercase ='http://images.cocodataset.org/val2017/000000039769.jpg'
__lowercase =Image.open(requests.get(lowercase__, stream=lowercase__ ).raw )
return im
@torch.no_grad()
def __UpperCamelCase ( lowercase__ : str, lowercase__ : str, lowercase__ : str, lowercase__ : bool = False ):
'''simple docstring'''
__lowercase =get_yolos_config(lowercase__ )
# load original state_dict
__lowercase =torch.load(lowercase__, map_location='cpu' )['model']
# load 🤗 model
__lowercase =YolosForObjectDetection(lowercase__ )
model.eval()
__lowercase =convert_state_dict(lowercase__, lowercase__ )
model.load_state_dict(lowercase__ )
# Check outputs on an image, prepared by YolosImageProcessor
__lowercase =8_00 if yolos_name != 'yolos_ti' else 5_12
__lowercase =YolosImageProcessor(format='coco_detection', size=lowercase__ )
__lowercase =image_processor(images=prepare_img(), return_tensors='pt' )
__lowercase =model(**lowercase__ )
__lowercase , __lowercase =outputs.logits, outputs.pred_boxes
__lowercase , __lowercase =None, None
if yolos_name == "yolos_ti":
__lowercase =torch.tensor(
[[-39.5022, -11.9820, -17.6888], [-29.9574, -9.9769, -17.7691], [-42.3281, -20.7200, -30.6294]] )
__lowercase =torch.tensor(
[[0.4021, 0.0836, 0.7979], [0.0184, 0.2609, 0.0364], [0.1781, 0.2004, 0.2095]] )
elif yolos_name == "yolos_s_200_pre":
__lowercase =torch.tensor(
[[-24.0248, -10.3024, -14.8290], [-42.0392, -16.8200, -27.4334], [-27.2743, -11.8154, -18.7148]] )
__lowercase =torch.tensor(
[[0.2559, 0.5455, 0.4706], [0.2989, 0.7279, 0.1875], [0.7732, 0.4017, 0.4462]] )
elif yolos_name == "yolos_s_300_pre":
__lowercase =torch.tensor(
[[-36.2220, -14.4385, -23.5457], [-35.6970, -14.7583, -21.3935], [-31.5939, -13.6042, -16.8049]] )
__lowercase =torch.tensor(
[[0.7614, 0.2316, 0.4728], [0.7168, 0.4495, 0.3855], [0.4996, 0.1466, 0.9996]] )
elif yolos_name == "yolos_s_dWr":
__lowercase =torch.tensor(
[[-42.8668, -24.1049, -41.1690], [-34.7456, -14.1274, -24.9194], [-33.7898, -12.1946, -25.6495]] )
__lowercase =torch.tensor(
[[0.5587, 0.2773, 0.0605], [0.5004, 0.3014, 0.9994], [0.4999, 0.1548, 0.9994]] )
elif yolos_name == "yolos_base":
__lowercase =torch.tensor(
[[-40.6064, -24.3084, -32.6447], [-55.1990, -30.7719, -35.5877], [-51.4311, -33.3507, -35.6462]] )
__lowercase =torch.tensor(
[[0.5555, 0.2794, 0.0655], [0.9049, 0.2664, 0.1894], [0.9183, 0.1984, 0.1635]] )
else:
raise ValueError(F'''Unknown yolos_name: {yolos_name}''' )
assert torch.allclose(logits[0, :3, :3], lowercase__, atol=1E-4 )
assert torch.allclose(pred_boxes[0, :3, :3], lowercase__, atol=1E-4 )
Path(lowercase__ ).mkdir(exist_ok=lowercase__ )
print(F'''Saving model {yolos_name} to {pytorch_dump_folder_path}''' )
model.save_pretrained(lowercase__ )
print(F'''Saving image processor to {pytorch_dump_folder_path}''' )
image_processor.save_pretrained(lowercase__ )
if push_to_hub:
__lowercase ={
'yolos_ti': 'yolos-tiny',
'yolos_s_200_pre': 'yolos-small',
'yolos_s_300_pre': 'yolos-small-300',
'yolos_s_dWr': 'yolos-small-dwr',
'yolos_base': 'yolos-base',
}
print('Pushing to the hub...' )
__lowercase =model_mapping[yolos_name]
image_processor.push_to_hub(lowercase__, organization='hustvl' )
model.push_to_hub(lowercase__, organization='hustvl' )
if __name__ == "__main__":
UpperCAmelCase = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--yolos_name''',
default='''yolos_s_200_pre''',
type=str,
help=(
'''Name of the YOLOS model you\'d like to convert. Should be one of \'yolos_ti\', \'yolos_s_200_pre\','''
''' \'yolos_s_300_pre\', \'yolos_s_dWr\', \'yolos_base\'.'''
),
)
parser.add_argument(
'''--checkpoint_path''', default=None, type=str, help='''Path to the original state dict (.pth file).'''
)
parser.add_argument(
'''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model directory.'''
)
parser.add_argument(
'''--push_to_hub''', action='''store_true''', help='''Whether or not to push the converted model to the 🤗 hub.'''
)
UpperCAmelCase = parser.parse_args()
convert_yolos_checkpoint(args.yolos_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub)
| 119
|
'''simple docstring'''
import os
from shutil import copyfile
from typing import Any, Dict, List, Optional, Tuple
import sentencepiece as spm
from ...tokenization_utils import AddedToken, BatchEncoding, PreTrainedTokenizer
from ...utils import logging
UpperCAmelCase = logging.get_logger(__name__)
UpperCAmelCase = '''▁'''
UpperCAmelCase = {'''vocab_file''': '''sentencepiece.bpe.model'''}
UpperCAmelCase = {
'''vocab_file''': {
'''facebook/nllb-200-distilled-600M''': (
'''https://huggingface.co/facebook/nllb-200-distilled-600M/blob/main/sentencepiece.bpe.model'''
),
}
}
UpperCAmelCase = {
'''facebook/nllb-200-distilled-600M''': 1024,
}
# fmt: off
UpperCAmelCase = ['''ace_Arab''', '''ace_Latn''', '''acm_Arab''', '''acq_Arab''', '''aeb_Arab''', '''afr_Latn''', '''ajp_Arab''', '''aka_Latn''', '''amh_Ethi''', '''apc_Arab''', '''arb_Arab''', '''ars_Arab''', '''ary_Arab''', '''arz_Arab''', '''asm_Beng''', '''ast_Latn''', '''awa_Deva''', '''ayr_Latn''', '''azb_Arab''', '''azj_Latn''', '''bak_Cyrl''', '''bam_Latn''', '''ban_Latn''', '''bel_Cyrl''', '''bem_Latn''', '''ben_Beng''', '''bho_Deva''', '''bjn_Arab''', '''bjn_Latn''', '''bod_Tibt''', '''bos_Latn''', '''bug_Latn''', '''bul_Cyrl''', '''cat_Latn''', '''ceb_Latn''', '''ces_Latn''', '''cjk_Latn''', '''ckb_Arab''', '''crh_Latn''', '''cym_Latn''', '''dan_Latn''', '''deu_Latn''', '''dik_Latn''', '''dyu_Latn''', '''dzo_Tibt''', '''ell_Grek''', '''eng_Latn''', '''epo_Latn''', '''est_Latn''', '''eus_Latn''', '''ewe_Latn''', '''fao_Latn''', '''pes_Arab''', '''fij_Latn''', '''fin_Latn''', '''fon_Latn''', '''fra_Latn''', '''fur_Latn''', '''fuv_Latn''', '''gla_Latn''', '''gle_Latn''', '''glg_Latn''', '''grn_Latn''', '''guj_Gujr''', '''hat_Latn''', '''hau_Latn''', '''heb_Hebr''', '''hin_Deva''', '''hne_Deva''', '''hrv_Latn''', '''hun_Latn''', '''hye_Armn''', '''ibo_Latn''', '''ilo_Latn''', '''ind_Latn''', '''isl_Latn''', '''ita_Latn''', '''jav_Latn''', '''jpn_Jpan''', '''kab_Latn''', '''kac_Latn''', '''kam_Latn''', '''kan_Knda''', '''kas_Arab''', '''kas_Deva''', '''kat_Geor''', '''knc_Arab''', '''knc_Latn''', '''kaz_Cyrl''', '''kbp_Latn''', '''kea_Latn''', '''khm_Khmr''', '''kik_Latn''', '''kin_Latn''', '''kir_Cyrl''', '''kmb_Latn''', '''kon_Latn''', '''kor_Hang''', '''kmr_Latn''', '''lao_Laoo''', '''lvs_Latn''', '''lij_Latn''', '''lim_Latn''', '''lin_Latn''', '''lit_Latn''', '''lmo_Latn''', '''ltg_Latn''', '''ltz_Latn''', '''lua_Latn''', '''lug_Latn''', '''luo_Latn''', '''lus_Latn''', '''mag_Deva''', '''mai_Deva''', '''mal_Mlym''', '''mar_Deva''', '''min_Latn''', '''mkd_Cyrl''', '''plt_Latn''', '''mlt_Latn''', '''mni_Beng''', '''khk_Cyrl''', '''mos_Latn''', '''mri_Latn''', '''zsm_Latn''', '''mya_Mymr''', '''nld_Latn''', '''nno_Latn''', '''nob_Latn''', '''npi_Deva''', '''nso_Latn''', '''nus_Latn''', '''nya_Latn''', '''oci_Latn''', '''gaz_Latn''', '''ory_Orya''', '''pag_Latn''', '''pan_Guru''', '''pap_Latn''', '''pol_Latn''', '''por_Latn''', '''prs_Arab''', '''pbt_Arab''', '''quy_Latn''', '''ron_Latn''', '''run_Latn''', '''rus_Cyrl''', '''sag_Latn''', '''san_Deva''', '''sat_Beng''', '''scn_Latn''', '''shn_Mymr''', '''sin_Sinh''', '''slk_Latn''', '''slv_Latn''', '''smo_Latn''', '''sna_Latn''', '''snd_Arab''', '''som_Latn''', '''sot_Latn''', '''spa_Latn''', '''als_Latn''', '''srd_Latn''', '''srp_Cyrl''', '''ssw_Latn''', '''sun_Latn''', '''swe_Latn''', '''swh_Latn''', '''szl_Latn''', '''tam_Taml''', '''tat_Cyrl''', '''tel_Telu''', '''tgk_Cyrl''', '''tgl_Latn''', '''tha_Thai''', '''tir_Ethi''', '''taq_Latn''', '''taq_Tfng''', '''tpi_Latn''', '''tsn_Latn''', '''tso_Latn''', '''tuk_Latn''', '''tum_Latn''', '''tur_Latn''', '''twi_Latn''', '''tzm_Tfng''', '''uig_Arab''', '''ukr_Cyrl''', '''umb_Latn''', '''urd_Arab''', '''uzn_Latn''', '''vec_Latn''', '''vie_Latn''', '''war_Latn''', '''wol_Latn''', '''xho_Latn''', '''ydd_Hebr''', '''yor_Latn''', '''yue_Hant''', '''zho_Hans''', '''zho_Hant''', '''zul_Latn''']
class lowerCAmelCase ( A ):
lowerCAmelCase_ = VOCAB_FILES_NAMES
lowerCAmelCase_ = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
lowerCAmelCase_ = PRETRAINED_VOCAB_FILES_MAP
lowerCAmelCase_ = ["input_ids", "attention_mask"]
lowerCAmelCase_ = []
lowerCAmelCase_ = []
def __init__( self : Dict , __lowercase : List[str] , __lowercase : int="<s>" , __lowercase : Any="</s>" , __lowercase : Tuple="</s>" , __lowercase : Dict="<s>" , __lowercase : Dict="<unk>" , __lowercase : List[Any]="<pad>" , __lowercase : Dict="<mask>" , __lowercase : Tuple=None , __lowercase : List[str]=None , __lowercase : Union[str, Any]=None , __lowercase : Optional[Dict[str, Any]] = None , __lowercase : List[str]=None , __lowercase : List[str]=False , **__lowercase : Union[str, Any] , ):
"""simple docstring"""
__lowercase =AddedToken(__lowercase , lstrip=__lowercase , rstrip=__lowercase ) if isinstance(__lowercase , __lowercase ) else mask_token
__lowercase ={} if sp_model_kwargs is None else sp_model_kwargs
__lowercase =legacy_behaviour
super().__init__(
bos_token=__lowercase , eos_token=__lowercase , unk_token=__lowercase , sep_token=__lowercase , cls_token=__lowercase , pad_token=__lowercase , mask_token=__lowercase , tokenizer_file=__lowercase , src_lang=__lowercase , tgt_lang=__lowercase , additional_special_tokens=__lowercase , sp_model_kwargs=self.sp_model_kwargs , legacy_behaviour=__lowercase , **__lowercase , )
__lowercase =spm.SentencePieceProcessor(**self.sp_model_kwargs )
self.sp_model.Load(str(__lowercase ) )
__lowercase =vocab_file
# Original fairseq vocab and spm vocab must be "aligned":
# Vocab | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9
# -------- | ------- | ------- | ------ | ------- | ---- | ---- | ---- | ---- | ---- | ----
# fairseq | '<s>' | '<pad>' | '</s>' | '<unk>' | 'an' | '▁n' | '▁m' | '▁t' | '▁k' | '▁a'
# spm | '<unk>' | '<s>' | '</s>' | 'an' | '▁n' | '▁m' | '▁t' | '▁k' | '▁a' | '▁s'
# Mimic fairseq token-to-id alignment for the first 4 token
__lowercase ={'<s>': 0, '<pad>': 1, '</s>': 2, '<unk>': 3}
# The first "real" token "," has position 4 in the original fairseq vocab and position 3 in the spm vocab
__lowercase =1
__lowercase =len(self.sp_model )
__lowercase ={
code: self.sp_model_size + i + self.fairseq_offset for i, code in enumerate(__lowercase )
}
__lowercase ={v: k for k, v in self.lang_code_to_id.items()}
__lowercase =len(self.sp_model ) + len(self.lang_code_to_id ) + self.fairseq_offset
self.fairseq_tokens_to_ids.update(self.lang_code_to_id )
__lowercase ={v: k for k, v in self.fairseq_tokens_to_ids.items()}
__lowercase =list(self.lang_code_to_id.keys() )
if additional_special_tokens is not None:
# Only add those special tokens if they are not already there.
self._additional_special_tokens.extend(
[t for t in additional_special_tokens if t not in self._additional_special_tokens] )
__lowercase =src_lang if src_lang is not None else 'eng_Latn'
__lowercase =self.lang_code_to_id[self._src_lang]
__lowercase =tgt_lang
self.set_src_lang_special_tokens(self._src_lang )
def __getstate__( self : Tuple ):
"""simple docstring"""
__lowercase =self.__dict__.copy()
__lowercase =None
__lowercase =self.sp_model.serialized_model_proto()
return state
def __setstate__( self : List[Any] , __lowercase : List[Any] ):
"""simple docstring"""
__lowercase =d
# for backward compatibility
if not hasattr(self , 'sp_model_kwargs' ):
__lowercase ={}
__lowercase =spm.SentencePieceProcessor(**self.sp_model_kwargs )
self.sp_model.LoadFromSerializedProto(self.sp_model_proto )
@property
def snake_case ( self : Dict ):
"""simple docstring"""
return len(self.sp_model ) + len(self.lang_code_to_id ) + self.fairseq_offset + 1 # Plus 1 for the mask token
@property
def snake_case ( self : Tuple ):
"""simple docstring"""
return self._src_lang
@src_lang.setter
def snake_case ( self : Optional[Any] , __lowercase : str ):
"""simple docstring"""
__lowercase =new_src_lang
self.set_src_lang_special_tokens(self._src_lang )
def snake_case ( self : List[str] , __lowercase : List[int] , __lowercase : Optional[List[int]] = None , __lowercase : bool = False ):
"""simple docstring"""
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=__lowercase , token_ids_a=__lowercase , already_has_special_tokens=__lowercase )
__lowercase =[1] * len(self.prefix_tokens )
__lowercase =[1] * len(self.suffix_tokens )
if token_ids_a is None:
return prefix_ones + ([0] * len(__lowercase )) + suffix_ones
return prefix_ones + ([0] * len(__lowercase )) + ([0] * len(__lowercase )) + suffix_ones
def snake_case ( self : Optional[Any] , __lowercase : List[int] , __lowercase : Optional[List[int]] = None ):
"""simple docstring"""
if token_ids_a is None:
return self.prefix_tokens + token_ids_a + self.suffix_tokens
# We don't expect to process pairs, but leave the pair logic for API consistency
return self.prefix_tokens + token_ids_a + token_ids_a + self.suffix_tokens
def snake_case ( self : int , __lowercase : List[int] , __lowercase : Optional[List[int]] = None ):
"""simple docstring"""
__lowercase =[self.sep_token_id]
__lowercase =[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 snake_case ( self : Optional[Any] , __lowercase : Dict , __lowercase : str , __lowercase : Optional[str] , __lowercase : Optional[str] , **__lowercase : Optional[int] ):
"""simple docstring"""
if src_lang is None or tgt_lang is None:
raise ValueError('Translation requires a `src_lang` and a `tgt_lang` for this model' )
__lowercase =src_lang
__lowercase =self(__lowercase , add_special_tokens=__lowercase , return_tensors=__lowercase , **__lowercase )
__lowercase =self.convert_tokens_to_ids(__lowercase )
__lowercase =tgt_lang_id
return inputs
def snake_case ( self : Any ):
"""simple docstring"""
__lowercase ={self.convert_ids_to_tokens(__lowercase ): i for i in range(self.vocab_size )}
vocab.update(self.added_tokens_encoder )
return vocab
def snake_case ( self : Any , __lowercase : str ):
"""simple docstring"""
return self.sp_model.encode(__lowercase , out_type=__lowercase )
def snake_case ( self : int , __lowercase : int ):
"""simple docstring"""
if token in self.fairseq_tokens_to_ids:
return self.fairseq_tokens_to_ids[token]
__lowercase =self.sp_model.PieceToId(__lowercase )
# Need to return unknown token if the SP model returned 0
return spm_id + self.fairseq_offset if spm_id else self.unk_token_id
def snake_case ( self : int , __lowercase : Any ):
"""simple docstring"""
if index in self.fairseq_ids_to_tokens:
return self.fairseq_ids_to_tokens[index]
return self.sp_model.IdToPiece(index - self.fairseq_offset )
def snake_case ( self : str , __lowercase : List[str] ):
"""simple docstring"""
__lowercase =''.join(__lowercase ).replace(__lowercase , ' ' ).strip()
return out_string
def snake_case ( self : Optional[Any] , __lowercase : str , __lowercase : Optional[str] = None ):
"""simple docstring"""
if not os.path.isdir(__lowercase ):
logger.error(f'''Vocabulary path ({save_directory}) should be a directory''' )
return
__lowercase =os.path.join(
__lowercase , (filename_prefix + '-' if filename_prefix else '') + VOCAB_FILES_NAMES['vocab_file'] )
if os.path.abspath(self.vocab_file ) != os.path.abspath(__lowercase ) and os.path.isfile(self.vocab_file ):
copyfile(self.vocab_file , __lowercase )
elif not os.path.isfile(self.vocab_file ):
with open(__lowercase , 'wb' ) as fi:
__lowercase =self.sp_model.serialized_model_proto()
fi.write(__lowercase )
return (out_vocab_file,)
def snake_case ( self : List[str] , __lowercase : List[str] , __lowercase : str = "eng_Latn" , __lowercase : Optional[List[str]] = None , __lowercase : str = "fra_Latn" , **__lowercase : List[str] , ):
"""simple docstring"""
__lowercase =src_lang
__lowercase =tgt_lang
return super().prepare_seqaseq_batch(__lowercase , __lowercase , **__lowercase )
def snake_case ( self : Optional[int] ):
"""simple docstring"""
return self.set_src_lang_special_tokens(self.src_lang )
def snake_case ( self : List[Any] ):
"""simple docstring"""
return self.set_tgt_lang_special_tokens(self.tgt_lang )
def snake_case ( self : Optional[Any] , __lowercase : List[str] ):
"""simple docstring"""
__lowercase =self.lang_code_to_id[src_lang]
if self.legacy_behaviour:
__lowercase =[]
__lowercase =[self.eos_token_id, self.cur_lang_code]
else:
__lowercase =[self.cur_lang_code]
__lowercase =[self.eos_token_id]
def snake_case ( self : Union[str, Any] , __lowercase : str ):
"""simple docstring"""
__lowercase =self.lang_code_to_id[lang]
if self.legacy_behaviour:
__lowercase =[]
__lowercase =[self.eos_token_id, self.cur_lang_code]
else:
__lowercase =[self.cur_lang_code]
__lowercase =[self.eos_token_id]
| 119
| 1
|
"""simple docstring"""
from ..utils import DummyObject, requires_backends
class __A ( metaclass=A_ ):
'''simple docstring'''
lowerCAmelCase : int = ["torch", "transformers", "onnx"]
def __init__( self : int ,*_snake_case : Any ,**_snake_case : Union[str, Any] ) -> Dict:
"""simple docstring"""
requires_backends(self ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : Dict ,*_snake_case : Optional[int] ,**_snake_case : Dict ) -> Any:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : Optional[int] ,*_snake_case : Optional[int] ,**_snake_case : int ) -> int:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
class __A ( metaclass=A_ ):
'''simple docstring'''
lowerCAmelCase : Dict = ["torch", "transformers", "onnx"]
def __init__( self : Dict ,*_snake_case : Optional[Any] ,**_snake_case : List[Any] ) -> List[str]:
"""simple docstring"""
requires_backends(self ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : List[str] ,*_snake_case : Any ,**_snake_case : str ) -> Union[str, Any]:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : Tuple ,*_snake_case : str ,**_snake_case : Optional[Any] ) -> List[str]:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
class __A ( metaclass=A_ ):
'''simple docstring'''
lowerCAmelCase : Tuple = ["torch", "transformers", "onnx"]
def __init__( self : int ,*_snake_case : int ,**_snake_case : Optional[int] ) -> List[Any]:
"""simple docstring"""
requires_backends(self ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : List[Any] ,*_snake_case : Optional[int] ,**_snake_case : str ) -> Optional[Any]:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : Optional[Any] ,*_snake_case : str ,**_snake_case : int ) -> Optional[int]:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
class __A ( metaclass=A_ ):
'''simple docstring'''
lowerCAmelCase : List[str] = ["torch", "transformers", "onnx"]
def __init__( self : Optional[int] ,*_snake_case : Optional[Any] ,**_snake_case : List[Any] ) -> Optional[Any]:
"""simple docstring"""
requires_backends(self ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : Optional[Any] ,*_snake_case : Optional[int] ,**_snake_case : List[Any] ) -> str:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : str ,*_snake_case : List[str] ,**_snake_case : Any ) -> Tuple:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
class __A ( metaclass=A_ ):
'''simple docstring'''
lowerCAmelCase : Optional[int] = ["torch", "transformers", "onnx"]
def __init__( self : Tuple ,*_snake_case : List[Any] ,**_snake_case : int ) -> Optional[int]:
"""simple docstring"""
requires_backends(self ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : int ,*_snake_case : Optional[int] ,**_snake_case : List[Any] ) -> List[str]:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : int ,*_snake_case : List[Any] ,**_snake_case : Union[str, Any] ) -> List[str]:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
class __A ( metaclass=A_ ):
'''simple docstring'''
lowerCAmelCase : Union[str, Any] = ["torch", "transformers", "onnx"]
def __init__( self : Any ,*_snake_case : Any ,**_snake_case : str ) -> Optional[int]:
"""simple docstring"""
requires_backends(self ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : List[Any] ,*_snake_case : str ,**_snake_case : Optional[int] ) -> str:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
@classmethod
def UpperCAmelCase ( cls : Any ,*_snake_case : int ,**_snake_case : Dict ) -> Optional[int]:
"""simple docstring"""
requires_backends(cls ,['''torch''', '''transformers''', '''onnx'''] )
| 706
|
"""simple docstring"""
import argparse
import os
import pickle
import sys
import torch
from transformers import TransfoXLConfig, TransfoXLLMHeadModel, load_tf_weights_in_transfo_xl
from transformers.models.transfo_xl import tokenization_transfo_xl as data_utils
from transformers.models.transfo_xl.tokenization_transfo_xl import CORPUS_NAME, VOCAB_FILES_NAMES
from transformers.utils import CONFIG_NAME, WEIGHTS_NAME, logging
logging.set_verbosity_info()
# We do this to be able to load python 2 datasets pickles
# See e.g. https://stackoverflow.com/questions/2121874/python-pickling-after-changing-a-modules-directory/2121918#2121918
lowerCAmelCase_ = data_utils.TransfoXLTokenizer
lowerCAmelCase_ = data_utils.TransfoXLCorpus
lowerCAmelCase_ = data_utils
lowerCAmelCase_ = data_utils
def __UpperCAmelCase ( __lowerCamelCase , __lowerCamelCase , __lowerCamelCase , __lowerCamelCase ) -> Dict:
if transfo_xl_dataset_file:
# Convert a pre-processed corpus (see original TensorFlow repo)
with open(__lowerCamelCase , '''rb''' ) as fp:
lowercase__ : Any = pickle.load(__lowerCamelCase , encoding='''latin1''' )
# Save vocabulary and dataset cache as Dictionaries (should be better than pickles for the long-term)
lowercase__ : str = pytorch_dump_folder_path + '''/''' + VOCAB_FILES_NAMES['''pretrained_vocab_file''']
print(f"""Save vocabulary to {pytorch_vocab_dump_path}""" )
lowercase__ : Dict = corpus.vocab.__dict__
torch.save(__lowerCamelCase , __lowerCamelCase )
lowercase__ : Optional[Any] = corpus.__dict__
corpus_dict_no_vocab.pop('''vocab''' , __lowerCamelCase )
lowercase__ : int = pytorch_dump_folder_path + '''/''' + CORPUS_NAME
print(f"""Save dataset to {pytorch_dataset_dump_path}""" )
torch.save(__lowerCamelCase , __lowerCamelCase )
if tf_checkpoint_path:
# Convert a pre-trained TensorFlow model
lowercase__ : int = os.path.abspath(__lowerCamelCase )
lowercase__ : List[Any] = os.path.abspath(__lowerCamelCase )
print(f"""Converting Transformer XL checkpoint from {tf_path} with config at {config_path}.""" )
# Initialise PyTorch model
if transfo_xl_config_file == "":
lowercase__ : Union[str, Any] = TransfoXLConfig()
else:
lowercase__ : str = TransfoXLConfig.from_json_file(__lowerCamelCase )
print(f"""Building PyTorch model from configuration: {config}""" )
lowercase__ : List[str] = TransfoXLLMHeadModel(__lowerCamelCase )
lowercase__ : Tuple = load_tf_weights_in_transfo_xl(__lowerCamelCase , __lowerCamelCase , __lowerCamelCase )
# Save pytorch-model
lowercase__ : int = os.path.join(__lowerCamelCase , __lowerCamelCase )
lowercase__ : Optional[int] = os.path.join(__lowerCamelCase , __lowerCamelCase )
print(f"""Save PyTorch model to {os.path.abspath(__lowerCamelCase )}""" )
torch.save(model.state_dict() , __lowerCamelCase )
print(f"""Save configuration file to {os.path.abspath(__lowerCamelCase )}""" )
with open(__lowerCamelCase , '''w''' , encoding='''utf-8''' ) as f:
f.write(config.to_json_string() )
if __name__ == "__main__":
lowerCAmelCase_ = argparse.ArgumentParser()
parser.add_argument(
'--pytorch_dump_folder_path',
default=None,
type=str,
required=True,
help='Path to the folder to store the PyTorch model or dataset/vocab.',
)
parser.add_argument(
'--tf_checkpoint_path',
default='',
type=str,
help='An optional path to a TensorFlow checkpoint path to be converted.',
)
parser.add_argument(
'--transfo_xl_config_file',
default='',
type=str,
help=(
'An optional config json file corresponding to the pre-trained BERT model. \n'
'This specifies the model architecture.'
),
)
parser.add_argument(
'--transfo_xl_dataset_file',
default='',
type=str,
help='An optional dataset file to be converted in a vocabulary.',
)
lowerCAmelCase_ = parser.parse_args()
convert_transfo_xl_checkpoint_to_pytorch(
args.tf_checkpoint_path,
args.transfo_xl_config_file,
args.pytorch_dump_folder_path,
args.transfo_xl_dataset_file,
)
| 122
| 0
|
"""simple docstring"""
import argparse
import torch
from transformers import MobileBertConfig, MobileBertForPreTraining, load_tf_weights_in_mobilebert
from transformers.utils import logging
logging.set_verbosity_info()
def SCREAMING_SNAKE_CASE__ ( __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ) -> str:
# Initialise PyTorch model
lowercase__: Tuple = MobileBertConfig.from_json_file(__UpperCAmelCase )
print(F"""Building PyTorch model from configuration: {config}""" )
lowercase__: str = MobileBertForPreTraining(__UpperCAmelCase )
# Load weights from tf checkpoint
lowercase__: int = load_tf_weights_in_mobilebert(__UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase )
# Save pytorch-model
print(F"""Save PyTorch model to {pytorch_dump_path}""" )
torch.save(model.state_dict() , __UpperCAmelCase )
if __name__ == "__main__":
__A = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--tf_checkpoint_path", default=None, type=str, required=True, help="Path to the TensorFlow checkpoint path."
)
parser.add_argument(
"--mobilebert_config_file",
default=None,
type=str,
required=True,
help=(
"The config json file corresponding to the pre-trained MobileBERT model. \n"
"This specifies the model architecture."
),
)
parser.add_argument(
"--pytorch_dump_path", default=None, type=str, required=True, help="Path to the output PyTorch model."
)
__A = parser.parse_args()
convert_tf_checkpoint_to_pytorch(args.tf_checkpoint_path, args.mobilebert_config_file, args.pytorch_dump_path)
| 586
|
"""simple docstring"""
import argparse
import requests
import torch
from PIL import Image
from torchvision.transforms import Compose, Normalize, Resize, ToTensor
from transformers import SwinaSRConfig, SwinaSRForImageSuperResolution, SwinaSRImageProcessor
def SCREAMING_SNAKE_CASE__ ( __UpperCAmelCase ) -> Optional[Any]:
lowercase__: Any = SwinaSRConfig()
if "Swin2SR_ClassicalSR_X4_64" in checkpoint_url:
lowercase__: Any = 4
elif "Swin2SR_CompressedSR_X4_48" in checkpoint_url:
lowercase__: int = 4
lowercase__: Tuple = 4_8
lowercase__: str = '''pixelshuffle_aux'''
elif "Swin2SR_Lightweight_X2_64" in checkpoint_url:
lowercase__: List[str] = [6, 6, 6, 6]
lowercase__: Union[str, Any] = 6_0
lowercase__: int = [6, 6, 6, 6]
lowercase__: List[Any] = '''pixelshuffledirect'''
elif "Swin2SR_RealworldSR_X4_64_BSRGAN_PSNR" in checkpoint_url:
lowercase__: Optional[Any] = 4
lowercase__: Union[str, Any] = '''nearest+conv'''
elif "Swin2SR_Jpeg_dynamic" in checkpoint_url:
lowercase__: Tuple = 1
lowercase__: Union[str, Any] = 1
lowercase__: Optional[int] = 1_2_6
lowercase__: Optional[int] = 7
lowercase__: str = 2_5_5.0
lowercase__: int = ''''''
return config
def SCREAMING_SNAKE_CASE__ ( __UpperCAmelCase , __UpperCAmelCase ) -> str:
if "patch_embed.proj" in name and "layers" not in name:
lowercase__: Optional[int] = name.replace('''patch_embed.proj''' , '''embeddings.patch_embeddings.projection''' )
if "patch_embed.norm" in name:
lowercase__: List[str] = name.replace('''patch_embed.norm''' , '''embeddings.patch_embeddings.layernorm''' )
if "layers" in name:
lowercase__: Union[str, Any] = name.replace('''layers''' , '''encoder.stages''' )
if "residual_group.blocks" in name:
lowercase__: int = name.replace('''residual_group.blocks''' , '''layers''' )
if "attn.proj" in name:
lowercase__: Optional[Any] = name.replace('''attn.proj''' , '''attention.output.dense''' )
if "attn" in name:
lowercase__: Any = name.replace('''attn''' , '''attention.self''' )
if "norm1" in name:
lowercase__: Optional[int] = name.replace('''norm1''' , '''layernorm_before''' )
if "norm2" in name:
lowercase__: Optional[Any] = name.replace('''norm2''' , '''layernorm_after''' )
if "mlp.fc1" in name:
lowercase__: Any = name.replace('''mlp.fc1''' , '''intermediate.dense''' )
if "mlp.fc2" in name:
lowercase__: List[Any] = name.replace('''mlp.fc2''' , '''output.dense''' )
if "q_bias" in name:
lowercase__: Dict = name.replace('''q_bias''' , '''query.bias''' )
if "k_bias" in name:
lowercase__: List[Any] = name.replace('''k_bias''' , '''key.bias''' )
if "v_bias" in name:
lowercase__: Any = name.replace('''v_bias''' , '''value.bias''' )
if "cpb_mlp" in name:
lowercase__: int = name.replace('''cpb_mlp''' , '''continuous_position_bias_mlp''' )
if "patch_embed.proj" in name:
lowercase__: Tuple = name.replace('''patch_embed.proj''' , '''patch_embed.projection''' )
if name == "norm.weight":
lowercase__: int = '''layernorm.weight'''
if name == "norm.bias":
lowercase__: Tuple = '''layernorm.bias'''
if "conv_first" in name:
lowercase__: List[str] = name.replace('''conv_first''' , '''first_convolution''' )
if (
"upsample" in name
or "conv_before_upsample" in name
or "conv_bicubic" in name
or "conv_up" in name
or "conv_hr" in name
or "conv_last" in name
or "aux" in name
):
# heads
if "conv_last" in name:
lowercase__: Tuple = name.replace('''conv_last''' , '''final_convolution''' )
if config.upsampler in ["pixelshuffle", "pixelshuffle_aux", "nearest+conv"]:
if "conv_before_upsample.0" in name:
lowercase__: List[Any] = name.replace('''conv_before_upsample.0''' , '''conv_before_upsample''' )
if "upsample.0" in name:
lowercase__: Optional[int] = name.replace('''upsample.0''' , '''upsample.convolution_0''' )
if "upsample.2" in name:
lowercase__: Tuple = name.replace('''upsample.2''' , '''upsample.convolution_1''' )
lowercase__: Union[str, Any] = '''upsample.''' + name
elif config.upsampler == "pixelshuffledirect":
lowercase__: Tuple = name.replace('''upsample.0.weight''' , '''upsample.conv.weight''' )
lowercase__: List[Any] = name.replace('''upsample.0.bias''' , '''upsample.conv.bias''' )
else:
pass
else:
lowercase__: Any = '''swin2sr.''' + name
return name
def SCREAMING_SNAKE_CASE__ ( __UpperCAmelCase , __UpperCAmelCase ) -> Tuple:
for key in orig_state_dict.copy().keys():
lowercase__: List[Any] = orig_state_dict.pop(__UpperCAmelCase )
if "qkv" in key:
lowercase__: Optional[Any] = key.split('''.''' )
lowercase__: str = int(key_split[1] )
lowercase__: Tuple = int(key_split[4] )
lowercase__: Union[str, Any] = config.embed_dim
if "weight" in key:
lowercase__: Tuple = val[:dim, :]
lowercase__: Dict = val[dim : dim * 2, :]
lowercase__: Dict = val[-dim:, :]
else:
lowercase__: Optional[Any] = val[:dim]
lowercase__: Any = val[dim : dim * 2]
lowercase__: str = val[-dim:]
pass
else:
lowercase__: int = val
return orig_state_dict
def SCREAMING_SNAKE_CASE__ ( __UpperCAmelCase , __UpperCAmelCase , __UpperCAmelCase ) -> Tuple:
lowercase__: int = get_config(__UpperCAmelCase )
lowercase__: str = SwinaSRForImageSuperResolution(__UpperCAmelCase )
model.eval()
lowercase__: Optional[Any] = torch.hub.load_state_dict_from_url(__UpperCAmelCase , map_location='''cpu''' )
lowercase__: int = convert_state_dict(__UpperCAmelCase , __UpperCAmelCase )
lowercase__, lowercase__: int = model.load_state_dict(__UpperCAmelCase , strict=__UpperCAmelCase )
if len(__UpperCAmelCase ) > 0:
raise ValueError('''Missing keys when converting: {}'''.format(__UpperCAmelCase ) )
for key in unexpected_keys:
if not ("relative_position_index" in key or "relative_coords_table" in key or "self_mask" in key):
raise ValueError(F"""Unexpected key {key} in state_dict""" )
# verify values
lowercase__: List[Any] = '''https://github.com/mv-lab/swin2sr/blob/main/testsets/real-inputs/shanghai.jpg?raw=true'''
lowercase__: Dict = Image.open(requests.get(__UpperCAmelCase , stream=__UpperCAmelCase ).raw ).convert('''RGB''' )
lowercase__: Tuple = SwinaSRImageProcessor()
# pixel_values = processor(image, return_tensors="pt").pixel_values
lowercase__: Union[str, Any] = 1_2_6 if '''Jpeg''' in checkpoint_url else 2_5_6
lowercase__: List[str] = Compose(
[
Resize((image_size, image_size) ),
ToTensor(),
Normalize(mean=[0.4_8_5, 0.4_5_6, 0.4_0_6] , std=[0.2_2_9, 0.2_2_4, 0.2_2_5] ),
] )
lowercase__: Union[str, Any] = transforms(__UpperCAmelCase ).unsqueeze(0 )
if config.num_channels == 1:
lowercase__: Optional[int] = pixel_values[:, 0, :, :].unsqueeze(1 )
lowercase__: int = model(__UpperCAmelCase )
# assert values
if "Swin2SR_ClassicalSR_X2_64" in checkpoint_url:
lowercase__: List[str] = torch.Size([1, 3, 5_1_2, 5_1_2] )
lowercase__: int = torch.tensor(
[[-0.7_0_8_7, -0.7_1_3_8, -0.6_7_2_1], [-0.8_3_4_0, -0.8_0_9_5, -0.7_2_9_8], [-0.9_1_4_9, -0.8_4_1_4, -0.7_9_4_0]] )
elif "Swin2SR_ClassicalSR_X4_64" in checkpoint_url:
lowercase__: Tuple = torch.Size([1, 3, 1_0_2_4, 1_0_2_4] )
lowercase__: Any = torch.tensor(
[[-0.7_7_7_5, -0.8_1_0_5, -0.8_9_3_3], [-0.7_7_6_4, -0.8_3_5_6, -0.9_2_2_5], [-0.7_9_7_6, -0.8_6_8_6, -0.9_5_7_9]] )
elif "Swin2SR_CompressedSR_X4_48" in checkpoint_url:
# TODO values didn't match exactly here
lowercase__: Tuple = torch.Size([1, 3, 1_0_2_4, 1_0_2_4] )
lowercase__: List[Any] = torch.tensor(
[[-0.8_0_3_5, -0.7_5_0_4, -0.7_4_9_1], [-0.8_5_3_8, -0.8_1_2_4, -0.7_7_8_2], [-0.8_8_0_4, -0.8_6_5_1, -0.8_4_9_3]] )
elif "Swin2SR_Lightweight_X2_64" in checkpoint_url:
lowercase__: str = torch.Size([1, 3, 5_1_2, 5_1_2] )
lowercase__: Any = torch.tensor(
[[-0.7_6_6_9, -0.8_6_6_2, -0.8_7_6_7], [-0.8_8_1_0, -0.9_9_6_2, -0.9_8_2_0], [-0.9_3_4_0, -1.0_3_2_2, -1.1_1_4_9]] )
elif "Swin2SR_RealworldSR_X4_64_BSRGAN_PSNR" in checkpoint_url:
lowercase__: int = torch.Size([1, 3, 1_0_2_4, 1_0_2_4] )
lowercase__: List[str] = torch.tensor(
[[-0.5_2_3_8, -0.5_5_5_7, -0.6_3_2_1], [-0.6_0_1_6, -0.5_9_0_3, -0.6_3_9_1], [-0.6_2_4_4, -0.6_3_3_4, -0.6_8_8_9]] )
assert (
outputs.reconstruction.shape == expected_shape
), F"""Shape of reconstruction should be {expected_shape}, but is {outputs.reconstruction.shape}"""
assert torch.allclose(outputs.reconstruction[0, 0, :3, :3] , __UpperCAmelCase , atol=1e-3 )
print('''Looks ok!''' )
lowercase__: Tuple = {
'''https://github.com/mv-lab/swin2sr/releases/download/v0.0.1/Swin2SR_ClassicalSR_X2_64.pth''': (
'''swin2SR-classical-sr-x2-64'''
),
'''https://github.com/mv-lab/swin2sr/releases/download/v0.0.1/Swin2SR_ClassicalSR_X4_64.pth''': (
'''swin2SR-classical-sr-x4-64'''
),
'''https://github.com/mv-lab/swin2sr/releases/download/v0.0.1/Swin2SR_CompressedSR_X4_48.pth''': (
'''swin2SR-compressed-sr-x4-48'''
),
'''https://github.com/mv-lab/swin2sr/releases/download/v0.0.1/Swin2SR_Lightweight_X2_64.pth''': (
'''swin2SR-lightweight-x2-64'''
),
'''https://github.com/mv-lab/swin2sr/releases/download/v0.0.1/Swin2SR_RealworldSR_X4_64_BSRGAN_PSNR.pth''': (
'''swin2SR-realworld-sr-x4-64-bsrgan-psnr'''
),
}
lowercase__: str = url_to_name[checkpoint_url]
if pytorch_dump_folder_path is not None:
print(F"""Saving model {model_name} to {pytorch_dump_folder_path}""" )
model.save_pretrained(__UpperCAmelCase )
print(F"""Saving image processor to {pytorch_dump_folder_path}""" )
processor.save_pretrained(__UpperCAmelCase )
if push_to_hub:
model.push_to_hub(F"""caidas/{model_name}""" )
processor.push_to_hub(F"""caidas/{model_name}""" )
if __name__ == "__main__":
__A = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--checkpoint_url",
default="https://github.com/mv-lab/swin2sr/releases/download/v0.0.1/Swin2SR_ClassicalSR_X2_64.pth",
type=str,
help="URL of the original Swin2SR checkpoint you'd like to convert.",
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, help="Path to the output PyTorch model directory."
)
parser.add_argument("--push_to_hub", action="store_true", help="Whether to push the converted model to the hub.")
__A = parser.parse_args()
convert_swinasr_checkpoint(args.checkpoint_url, args.pytorch_dump_folder_path, args.push_to_hub)
| 586
| 1
|
from typing import Dict, List, Optional, Union
import numpy as np
from transformers.utils import is_vision_available
from transformers.utils.generic import TensorType
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
is_valid_image,
to_numpy_array,
valid_images,
)
from ...utils import logging
if is_vision_available():
import PIL
SCREAMING_SNAKE_CASE = logging.get_logger(__name__)
def _lowerCamelCase ( __A : List[Any] ) -> List[List[ImageInput]]:
if isinstance(_A , (list, tuple) ) and isinstance(videos[0] , (list, tuple) ) and is_valid_image(videos[0][0] ):
return videos
elif isinstance(_A , (list, tuple) ) and is_valid_image(videos[0] ):
return [videos]
elif is_valid_image(_A ):
return [[videos]]
raise ValueError(f'''Could not make batched video from {videos}''' )
class A_ ( lowerCAmelCase__ ):
'''simple docstring'''
_SCREAMING_SNAKE_CASE : Union[str, Any] = ["pixel_values"]
def __init__( self , _A = True , _A = None , _A = PILImageResampling.BILINEAR , _A = True , _A = None , _A = True , _A = 1 / 255 , _A = True , _A = True , _A = None , _A = None , **_A , ) -> None:
"""simple docstring"""
super().__init__(**_SCREAMING_SNAKE_CASE)
_UpperCAmelCase : Dict = size if size is not None else {'''shortest_edge''': 256}
_UpperCAmelCase : Optional[Any] = get_size_dict(_SCREAMING_SNAKE_CASE , default_to_square=_SCREAMING_SNAKE_CASE)
_UpperCAmelCase : str = crop_size if crop_size is not None else {'''height''': 224, '''width''': 224}
_UpperCAmelCase : int = get_size_dict(_SCREAMING_SNAKE_CASE , param_name='''crop_size''')
_UpperCAmelCase : Optional[int] = do_resize
_UpperCAmelCase : Tuple = size
_UpperCAmelCase : int = do_center_crop
_UpperCAmelCase : Any = crop_size
_UpperCAmelCase : Optional[int] = resample
_UpperCAmelCase : Optional[Any] = do_rescale
_UpperCAmelCase : Optional[Any] = rescale_factor
_UpperCAmelCase : List[Any] = offset
_UpperCAmelCase : List[Any] = do_normalize
_UpperCAmelCase : Dict = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
_UpperCAmelCase : str = image_std if image_std is not None else IMAGENET_STANDARD_STD
def snake_case__ ( self , _A , _A , _A = PILImageResampling.BILINEAR , _A = None , **_A , ) -> np.ndarray:
"""simple docstring"""
_UpperCAmelCase : Dict = get_size_dict(_SCREAMING_SNAKE_CASE , default_to_square=_SCREAMING_SNAKE_CASE)
if "shortest_edge" in size:
_UpperCAmelCase : Tuple = get_resize_output_image_size(_SCREAMING_SNAKE_CASE , size['''shortest_edge'''] , default_to_square=_SCREAMING_SNAKE_CASE)
elif "height" in size and "width" in size:
_UpperCAmelCase : Tuple = (size['''height'''], size['''width'''])
else:
raise ValueError(f'''Size must have \'height\' and \'width\' or \'shortest_edge\' as keys. Got {size.keys()}''')
return resize(_SCREAMING_SNAKE_CASE , size=_SCREAMING_SNAKE_CASE , resample=_SCREAMING_SNAKE_CASE , data_format=_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE)
def snake_case__ ( self , _A , _A , _A = None , **_A , ) -> np.ndarray:
"""simple docstring"""
_UpperCAmelCase : Tuple = get_size_dict(_SCREAMING_SNAKE_CASE)
if "height" not in size or "width" not in size:
raise ValueError(f'''Size must have \'height\' and \'width\' as keys. Got {size.keys()}''')
return center_crop(_SCREAMING_SNAKE_CASE , size=(size['''height'''], size['''width''']) , data_format=_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE)
def snake_case__ ( self , _A , _A , _A = True , _A = None , **_A , ) -> Optional[Any]:
"""simple docstring"""
_UpperCAmelCase : Optional[Any] = image.astype(np.floataa)
if offset:
_UpperCAmelCase : List[str] = image - (scale / 2)
return rescale(_SCREAMING_SNAKE_CASE , scale=_SCREAMING_SNAKE_CASE , data_format=_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE)
def snake_case__ ( self , _A , _A , _A , _A = None , **_A , ) -> np.ndarray:
"""simple docstring"""
return normalize(_SCREAMING_SNAKE_CASE , mean=_SCREAMING_SNAKE_CASE , std=_SCREAMING_SNAKE_CASE , data_format=_SCREAMING_SNAKE_CASE , **_SCREAMING_SNAKE_CASE)
def snake_case__ ( self , _A , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = ChannelDimension.FIRST , ) -> np.ndarray:
"""simple docstring"""
if do_resize and size is None or resample is None:
raise ValueError('''Size and resample must be specified if do_resize is True.''')
if do_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.''')
if offset and not do_rescale:
raise ValueError('''For offset, do_rescale must also be set to True.''')
# All transformations expect numpy arrays.
_UpperCAmelCase : int = to_numpy_array(_SCREAMING_SNAKE_CASE)
if do_resize:
_UpperCAmelCase : List[str] = self.resize(image=_SCREAMING_SNAKE_CASE , size=_SCREAMING_SNAKE_CASE , resample=_SCREAMING_SNAKE_CASE)
if do_center_crop:
_UpperCAmelCase : int = self.center_crop(_SCREAMING_SNAKE_CASE , size=_SCREAMING_SNAKE_CASE)
if do_rescale:
_UpperCAmelCase : Optional[Any] = self.rescale(image=_SCREAMING_SNAKE_CASE , scale=_SCREAMING_SNAKE_CASE , offset=_SCREAMING_SNAKE_CASE)
if do_normalize:
_UpperCAmelCase : List[Any] = self.normalize(image=_SCREAMING_SNAKE_CASE , mean=_SCREAMING_SNAKE_CASE , std=_SCREAMING_SNAKE_CASE)
_UpperCAmelCase : Tuple = to_channel_dimension_format(_SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE)
return image
def snake_case__ ( self , _A , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = None , _A = ChannelDimension.FIRST , **_A , ) -> PIL.Image.Image:
"""simple docstring"""
_UpperCAmelCase : Optional[Any] = do_resize if do_resize is not None else self.do_resize
_UpperCAmelCase : Dict = resample if resample is not None else self.resample
_UpperCAmelCase : str = do_center_crop if do_center_crop is not None else self.do_center_crop
_UpperCAmelCase : Tuple = do_rescale if do_rescale is not None else self.do_rescale
_UpperCAmelCase : List[Any] = rescale_factor if rescale_factor is not None else self.rescale_factor
_UpperCAmelCase : Optional[int] = offset if offset is not None else self.offset
_UpperCAmelCase : Any = do_normalize if do_normalize is not None else self.do_normalize
_UpperCAmelCase : Dict = image_mean if image_mean is not None else self.image_mean
_UpperCAmelCase : str = image_std if image_std is not None else self.image_std
_UpperCAmelCase : Optional[int] = size if size is not None else self.size
_UpperCAmelCase : Optional[int] = get_size_dict(_SCREAMING_SNAKE_CASE , default_to_square=_SCREAMING_SNAKE_CASE)
_UpperCAmelCase : Any = crop_size if crop_size is not None else self.crop_size
_UpperCAmelCase : List[str] = get_size_dict(_SCREAMING_SNAKE_CASE , param_name='''crop_size''')
if not valid_images(_SCREAMING_SNAKE_CASE):
raise ValueError(
'''Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '''
'''torch.Tensor, tf.Tensor or jax.ndarray.''')
_UpperCAmelCase : List[str] = make_batched(_SCREAMING_SNAKE_CASE)
_UpperCAmelCase : Dict = [
[
self._preprocess_image(
image=_SCREAMING_SNAKE_CASE , do_resize=_SCREAMING_SNAKE_CASE , size=_SCREAMING_SNAKE_CASE , resample=_SCREAMING_SNAKE_CASE , do_center_crop=_SCREAMING_SNAKE_CASE , crop_size=_SCREAMING_SNAKE_CASE , do_rescale=_SCREAMING_SNAKE_CASE , rescale_factor=_SCREAMING_SNAKE_CASE , offset=_SCREAMING_SNAKE_CASE , do_normalize=_SCREAMING_SNAKE_CASE , image_mean=_SCREAMING_SNAKE_CASE , image_std=_SCREAMING_SNAKE_CASE , data_format=_SCREAMING_SNAKE_CASE , )
for img in video
]
for video in videos
]
_UpperCAmelCase : List[Any] = {'''pixel_values''': videos}
return BatchFeature(data=_SCREAMING_SNAKE_CASE , tensor_type=_SCREAMING_SNAKE_CASE)
| 709
|
import math_equivalence # From: git+https://github.com/hendrycks/math.git
import datasets
SCREAMING_SNAKE_CASE = '\\n@article{hendrycksmath2021,\n title={Measuring Mathematical Problem Solving With the MATH Dataset},\n author={Dan Hendrycks\n and Collin Burns\n and Saurav Kadavath\n and Akul Arora\n and Steven Basart\n and Eric Tang\n and Dawn Song\n and Jacob Steinhardt},\n journal={arXiv preprint arXiv:2103.03874},\n year={2021}\n}\n'
SCREAMING_SNAKE_CASE = '\\nThis metric is used to assess performance on the Mathematics Aptitude Test of Heuristics (MATH) dataset.\nIt first canonicalizes the inputs (e.g., converting "1/2" to "\\frac{1}{2}") and then computes accuracy.\n'
SCREAMING_SNAKE_CASE = R'\nCalculates accuracy after canonicalizing inputs.\n\nArgs:\n predictions: list of predictions to score. Each prediction\n is a string that contains natural language and LaTex.\n references: list of reference for each prediction. Each\n reference is a string that contains natural language\n and LaTex.\nReturns:\n accuracy: accuracy after canonicalizing inputs\n (e.g., converting "1/2" to "\\frac{1}{2}")\n\nExamples:\n >>> metric = datasets.load_metric("competition_math")\n >>> results = metric.compute(references=["\\frac{1}{2}"], predictions=["1/2"])\n >>> print(results)\n {\'accuracy\': 1.0}\n'
@datasets.utils.file_utils.add_end_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class A_ ( datasets.Metric ):
'''simple docstring'''
def snake_case__ ( self) -> Optional[Any]:
"""simple docstring"""
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(
{
'''predictions''': datasets.Value('''string'''),
'''references''': datasets.Value('''string'''),
}) , homepage='''https://github.com/hendrycks/math''' , codebase_urls=['''https://github.com/hendrycks/math'''] , )
def snake_case__ ( self , _A , _A) -> Tuple:
"""simple docstring"""
_UpperCAmelCase : Tuple = 0.0
for i, j in zip(_A , _A):
n_correct += 1.0 if math_equivalence.is_equiv(_A , _A) else 0.0
_UpperCAmelCase : Tuple = n_correct / len(_A)
return {
"accuracy": accuracy,
}
| 186
| 0
|
'''simple docstring'''
__lowercase : Optional[Any] = [
999,
800,
799,
600,
599,
500,
400,
399,
377,
355,
333,
311,
288,
266,
244,
222,
200,
199,
177,
155,
133,
111,
88,
66,
44,
22,
0,
]
__lowercase : Tuple = [
999,
976,
952,
928,
905,
882,
858,
857,
810,
762,
715,
714,
572,
429,
428,
286,
285,
238,
190,
143,
142,
118,
95,
71,
47,
24,
0,
]
__lowercase : Any = [
999,
988,
977,
966,
955,
944,
933,
922,
911,
900,
899,
879,
859,
840,
820,
800,
799,
766,
733,
700,
699,
650,
600,
599,
500,
499,
400,
399,
350,
300,
299,
266,
233,
200,
199,
179,
159,
140,
120,
100,
99,
88,
77,
66,
55,
44,
33,
22,
11,
0,
]
__lowercase : str = [
999,
995,
992,
989,
985,
981,
978,
975,
971,
967,
964,
961,
957,
956,
951,
947,
942,
937,
933,
928,
923,
919,
914,
913,
908,
903,
897,
892,
887,
881,
876,
871,
870,
864,
858,
852,
846,
840,
834,
828,
827,
820,
813,
806,
799,
792,
785,
784,
777,
770,
763,
756,
749,
742,
741,
733,
724,
716,
707,
699,
698,
688,
677,
666,
656,
655,
645,
634,
623,
613,
612,
598,
584,
570,
569,
555,
541,
527,
526,
505,
484,
483,
462,
440,
439,
396,
395,
352,
351,
308,
307,
264,
263,
220,
219,
176,
132,
88,
44,
0,
]
__lowercase : List[str] = [
999,
997,
995,
992,
990,
988,
986,
984,
981,
979,
977,
975,
972,
970,
968,
966,
964,
961,
959,
957,
956,
954,
951,
949,
946,
944,
941,
939,
936,
934,
931,
929,
926,
924,
921,
919,
916,
914,
913,
910,
907,
905,
902,
899,
896,
893,
891,
888,
885,
882,
879,
877,
874,
871,
870,
867,
864,
861,
858,
855,
852,
849,
846,
843,
840,
837,
834,
831,
828,
827,
824,
821,
817,
814,
811,
808,
804,
801,
798,
795,
791,
788,
785,
784,
780,
777,
774,
770,
766,
763,
760,
756,
752,
749,
746,
742,
741,
737,
733,
730,
726,
722,
718,
714,
710,
707,
703,
699,
698,
694,
690,
685,
681,
677,
673,
669,
664,
660,
656,
655,
650,
646,
641,
636,
632,
627,
622,
618,
613,
612,
607,
602,
596,
591,
586,
580,
575,
570,
569,
563,
557,
551,
545,
539,
533,
527,
526,
519,
512,
505,
498,
491,
484,
483,
474,
466,
457,
449,
440,
439,
428,
418,
407,
396,
395,
381,
366,
352,
351,
330,
308,
307,
286,
264,
263,
242,
220,
219,
176,
175,
132,
131,
88,
44,
0,
]
__lowercase : Tuple = [
999,
991,
982,
974,
966,
958,
950,
941,
933,
925,
916,
908,
900,
899,
874,
850,
825,
800,
799,
700,
600,
500,
400,
300,
200,
100,
0,
]
__lowercase : List[Any] = [
999,
992,
985,
978,
971,
964,
957,
949,
942,
935,
928,
921,
914,
907,
900,
899,
879,
859,
840,
820,
800,
799,
766,
733,
700,
699,
650,
600,
599,
500,
499,
400,
399,
300,
299,
200,
199,
100,
99,
0,
]
__lowercase : Union[str, Any] = [
999,
996,
992,
989,
985,
982,
979,
975,
972,
968,
965,
961,
958,
955,
951,
948,
944,
941,
938,
934,
931,
927,
924,
920,
917,
914,
910,
907,
903,
900,
899,
891,
884,
876,
869,
861,
853,
846,
838,
830,
823,
815,
808,
800,
799,
788,
777,
766,
755,
744,
733,
722,
711,
700,
699,
688,
677,
666,
655,
644,
633,
622,
611,
600,
599,
585,
571,
557,
542,
528,
514,
500,
499,
485,
471,
457,
442,
428,
414,
400,
399,
379,
359,
340,
320,
300,
299,
279,
259,
240,
220,
200,
199,
166,
133,
100,
99,
66,
33,
0,
]
| 422
|
lowercase_ = [4, 1, 7, 4, 2, 6, 4, 1, 5, 3, 7, 5]
lowercase_ = [3, 7, 7, 4, 2, 6, 4, 1, 5, 3, 7, 5]
lowercase_ = {
0: '''Sunday''',
1: '''Monday''',
2: '''Tuesday''',
3: '''Wednesday''',
4: '''Thursday''',
5: '''Friday''',
6: '''Saturday''',
}
def lowerCAmelCase ( UpperCAmelCase, UpperCAmelCase, UpperCAmelCase ) ->str:
"""simple docstring"""
assert len(str(UpperCAmelCase ) ) > 2, "year should be in YYYY format"
assert 1 <= month <= 12, "month should be between 1 to 12"
assert 1 <= day <= 31, "day should be between 1 to 31"
# Doomsday algorithm:
__magic_name__ : Optional[Any] = year // 100
__magic_name__ : Any = (5 * (century % 4) + 2) % 7
__magic_name__ : Any = year % 100
__magic_name__ : str = centurian % 12
__magic_name__ : Union[str, Any] = (
(centurian // 12) + centurian_m + (centurian_m // 4) + century_anchor
) % 7
__magic_name__ : Optional[Any] = (
DOOMSDAY_NOT_LEAP[month - 1]
if (year % 4 != 0) or (centurian == 0 and (year % 400) == 0)
else DOOMSDAY_LEAP[month - 1]
)
__magic_name__ : List[Any] = (dooms_day + day - day_anchor) % 7
return WEEK_DAY_NAMES[week_day]
if __name__ == "__main__":
import doctest
doctest.testmod()
| 154
| 0
|
from operator import delitem, getitem, setitem
import pytest
from data_structures.hashing.hash_map import HashMap
def __SCREAMING_SNAKE_CASE ( UpperCamelCase : Optional[int] ) -> int:
"""simple docstring"""
return getitem, k
def __SCREAMING_SNAKE_CASE ( UpperCamelCase : Dict , UpperCamelCase : Any ) -> Tuple:
"""simple docstring"""
return setitem, k, v
def __SCREAMING_SNAKE_CASE ( UpperCamelCase : Tuple ) -> List[Any]:
"""simple docstring"""
return delitem, k
def __SCREAMING_SNAKE_CASE ( UpperCamelCase : List[str] , UpperCamelCase : Dict , *UpperCamelCase : Optional[int] ) -> str:
"""simple docstring"""
try:
return fun(UpperCamelCase , *UpperCamelCase ), None
except Exception as e:
return None, e
_A = (
_set('key_a', 'val_a'),
_set('key_b', 'val_b'),
)
_A = [
_set('key_a', 'val_a'),
_set('key_a', 'val_b'),
]
_A = [
_set('key_a', 'val_a'),
_set('key_b', 'val_b'),
_del('key_a'),
_del('key_b'),
_set('key_a', 'val_a'),
_del('key_a'),
]
_A = [
_get('key_a'),
_del('key_a'),
_set('key_a', 'val_a'),
_del('key_a'),
_del('key_a'),
_get('key_a'),
]
_A = [
*[_set(x, x) for x in range(5)], # guaranteed upsize
]
_A = [
*[_set(x, x) for x in range(5)], # guaranteed upsize
*[_del(x) for x in range(5)],
_set('key_a', 'val_b'),
]
@pytest.mark.parametrize(
"""operations""" , (
pytest.param(_add_items , id="""add items""" ),
pytest.param(_overwrite_items , id="""overwrite items""" ),
pytest.param(_delete_items , id="""delete items""" ),
pytest.param(_access_absent_items , id="""access absent items""" ),
pytest.param(_add_with_resize_up , id="""add with resize up""" ),
pytest.param(_add_with_resize_down , id="""add with resize down""" ),
) , )
def __SCREAMING_SNAKE_CASE ( UpperCamelCase : Optional[Any] ) -> Tuple:
"""simple docstring"""
a_ = HashMap(initial_block_size=4 )
a_ = {}
for _, (fun, *args) in enumerate(UpperCamelCase ):
a_ , a_ = _run_operation(UpperCamelCase , UpperCamelCase , *UpperCamelCase )
a_ , a_ = _run_operation(UpperCamelCase , UpperCamelCase , *UpperCamelCase )
assert my_res == py_res
assert str(UpperCamelCase ) == str(UpperCamelCase )
assert set(UpperCamelCase ) == set(UpperCamelCase )
assert len(UpperCamelCase ) == len(UpperCamelCase )
assert set(my.items() ) == set(py.items() )
def __SCREAMING_SNAKE_CASE ( ) -> List[Any]:
"""simple docstring"""
def is_public(UpperCamelCase : str ) -> bool:
return not name.startswith("""_""" )
a_ = {name for name in dir({} ) if is_public(UpperCamelCase )}
a_ = {name for name in dir(HashMap() ) if is_public(UpperCamelCase )}
assert dict_public_names > hash_public_names
| 709
|
# 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 .config import config_command_parser
from .config_args import default_config_file, load_config_from_file # noqa: F401
from .default import default_command_parser
from .update import update_command_parser
def __SCREAMING_SNAKE_CASE ( UpperCamelCase : Optional[int]=None ) -> Optional[Any]:
"""simple docstring"""
a_ = argparse.ArgumentParser(add_help=UpperCamelCase , allow_abbrev=UpperCamelCase )
# The main config parser
a_ = config_command_parser(UpperCamelCase )
# The subparser to add commands to
a_ = config_parser.add_subparsers(title="""subcommands""" , dest="""subcommand""" )
# Then add other parsers with the parent parser
default_command_parser(UpperCamelCase , parents=[parent_parser] )
update_command_parser(UpperCamelCase , parents=[parent_parser] )
return config_parser
def __SCREAMING_SNAKE_CASE ( ) -> Any:
"""simple docstring"""
a_ = get_config_parser()
a_ = config_parser.parse_args()
if not hasattr(UpperCamelCase , """func""" ):
config_parser.print_help()
exit(1 )
# Run
args.func(UpperCamelCase )
if __name__ == "__main__":
main()
| 403
| 0
|
import json
import logging
import os
import sys
from time import time
from unittest.mock import patch
from transformers.testing_utils import TestCasePlus, require_torch_tpu
logging.basicConfig(level=logging.DEBUG)
UpperCamelCase__ : Optional[int] = logging.getLogger()
def __UpperCAmelCase ( lowerCamelCase_ : List[str] ) -> Dict:
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : str = {}
SCREAMING_SNAKE_CASE_ : Any = os.path.join(lowerCamelCase_ , 'all_results.json' )
if os.path.exists(lowerCamelCase_ ):
with open(lowerCamelCase_ , 'r' ) as f:
SCREAMING_SNAKE_CASE_ : Optional[int] = json.load(lowerCamelCase_ )
else:
raise ValueError(F'can\'t find {path}' )
return results
UpperCamelCase__ : Dict = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
@require_torch_tpu
class lowerCAmelCase_ ( lowerCamelCase_ ):
def snake_case ( self ):
import xla_spawn
SCREAMING_SNAKE_CASE_ : Dict = self.get_auto_remove_tmp_dir()
SCREAMING_SNAKE_CASE_ : str = F'\n ./examples/pytorch/text-classification/run_glue.py\n --num_cores=8\n ./examples/pytorch/text-classification/run_glue.py\n --model_name_or_path distilbert-base-uncased\n --output_dir {tmp_dir}\n --overwrite_output_dir\n --train_file ./tests/fixtures/tests_samples/MRPC/train.csv\n --validation_file ./tests/fixtures/tests_samples/MRPC/dev.csv\n --do_train\n --do_eval\n --debug tpu_metrics_debug\n --per_device_train_batch_size=2\n --per_device_eval_batch_size=1\n --learning_rate=1e-4\n --max_steps=10\n --warmup_steps=2\n --seed=42\n --max_seq_length=128\n '.split()
with patch.object(snake_case__ ,'argv' ,snake_case__ ):
SCREAMING_SNAKE_CASE_ : Optional[int] = time()
xla_spawn.main()
SCREAMING_SNAKE_CASE_ : Union[str, Any] = time()
SCREAMING_SNAKE_CASE_ : Any = get_results(snake_case__ )
self.assertGreaterEqual(result['eval_accuracy'] ,0.75 )
# Assert that the script takes less than 500 seconds to make sure it doesn't hang.
self.assertLess(end - start ,500 )
def snake_case ( self ):
import xla_spawn
SCREAMING_SNAKE_CASE_ : str = '\n ./tests/test_trainer_tpu.py\n --num_cores=8\n ./tests/test_trainer_tpu.py\n '.split()
with patch.object(snake_case__ ,'argv' ,snake_case__ ):
xla_spawn.main()
| 105
|
from __future__ import annotations
import math
import numpy as np
from numpy.linalg import norm
def __UpperCAmelCase ( lowerCamelCase_ : np.ndarray , lowerCamelCase_ : np.ndarray ) -> float:
"""simple docstring"""
return math.sqrt(sum(pow(a - b , 2 ) for a, b in zip(lowerCamelCase_ , lowerCamelCase_ ) ) )
def __UpperCAmelCase ( lowerCamelCase_ : np.ndarray , lowerCamelCase_ : np.ndarray ) -> list[list[list[float] | float]]:
"""simple docstring"""
if dataset.ndim != value_array.ndim:
SCREAMING_SNAKE_CASE_ : Optional[int] = (
'Wrong input data\'s dimensions... '
F'dataset : {dataset.ndim}, value_array : {value_array.ndim}'
)
raise ValueError(lowerCamelCase_ )
try:
if dataset.shape[1] != value_array.shape[1]:
SCREAMING_SNAKE_CASE_ : List[Any] = (
'Wrong input data\'s shape... '
F'dataset : {dataset.shape[1]}, value_array : {value_array.shape[1]}'
)
raise ValueError(lowerCamelCase_ )
except IndexError:
if dataset.ndim != value_array.ndim:
raise TypeError('Wrong shape' )
if dataset.dtype != value_array.dtype:
SCREAMING_SNAKE_CASE_ : Tuple = (
'Input data have different datatype... '
F'dataset : {dataset.dtype}, value_array : {value_array.dtype}'
)
raise TypeError(lowerCamelCase_ )
SCREAMING_SNAKE_CASE_ : Optional[Any] = []
for value in value_array:
SCREAMING_SNAKE_CASE_ : str = euclidean(lowerCamelCase_ , dataset[0] )
SCREAMING_SNAKE_CASE_ : List[Any] = dataset[0].tolist()
for dataset_value in dataset[1:]:
SCREAMING_SNAKE_CASE_ : Optional[Any] = euclidean(lowerCamelCase_ , lowerCamelCase_ )
if dist > temp_dist:
SCREAMING_SNAKE_CASE_ : Optional[int] = temp_dist
SCREAMING_SNAKE_CASE_ : Union[str, Any] = dataset_value.tolist()
answer.append([vector, dist] )
return answer
def __UpperCAmelCase ( lowerCamelCase_ : np.ndarray , lowerCamelCase_ : np.ndarray ) -> float:
"""simple docstring"""
return np.dot(lowerCamelCase_ , lowerCamelCase_ ) / (norm(lowerCamelCase_ ) * norm(lowerCamelCase_ ))
if __name__ == "__main__":
import doctest
doctest.testmod()
| 105
| 1
|
import argparse
import gc
import json
import os
import shutil
import warnings
import torch
from transformers import LlamaConfig, LlamaForCausalLM, LlamaTokenizer
try:
from transformers import LlamaTokenizerFast
except ImportError as e:
warnings.warn(e)
warnings.warn(
'''The converted tokenizer will be the `slow` tokenizer. To use the fast, update your `tokenizers` library and re-run the tokenizer conversion'''
)
lowerCamelCase__ = None
lowerCamelCase__ = {
'''7B''': 1_10_08,
'''13B''': 1_38_24,
'''30B''': 1_79_20,
'''65B''': 2_20_16,
'''70B''': 2_86_72,
}
lowerCamelCase__ = {
'''7B''': 1,
'''7Bf''': 1,
'''13B''': 2,
'''13Bf''': 2,
'''30B''': 4,
'''65B''': 8,
'''70B''': 8,
'''70Bf''': 8,
}
def A(__a: List[str] , __a: str=1 , __a: Tuple=256 ):
return multiple_of * ((int(ffn_dim_multiplier * int(8 * n / 3 ) ) + multiple_of - 1) // multiple_of)
def A(__a: Union[str, Any] ):
with open(__a , "r" ) as f:
return json.load(__a )
def A(__a: Any , __a: Optional[Any] ):
with open(__a , "w" ) as f:
json.dump(__a , __a )
def A(__a: str , __a: Tuple , __a: List[str] , __a: List[Any]=True ):
os.makedirs(__a , exist_ok=__a )
lowerCAmelCase_ = os.path.join(__a , "tmp" )
os.makedirs(__a , exist_ok=__a )
lowerCAmelCase_ = read_json(os.path.join(__a , "params.json" ) )
lowerCAmelCase_ = NUM_SHARDS[model_size]
lowerCAmelCase_ = params["n_layers"]
lowerCAmelCase_ = params["n_heads"]
lowerCAmelCase_ = n_heads // num_shards
lowerCAmelCase_ = params["dim"]
lowerCAmelCase_ = dim // n_heads
lowerCAmelCase_ = 1_0000.0
lowerCAmelCase_ = 1.0 / (base ** (torch.arange(0 , __a , 2 ).float() / dims_per_head))
if "n_kv_heads" in params:
lowerCAmelCase_ = params["n_kv_heads"] # for GQA / MQA
lowerCAmelCase_ = n_heads_per_shard // num_key_value_heads
lowerCAmelCase_ = dim // num_key_value_heads
else: # compatibility with other checkpoints
lowerCAmelCase_ = n_heads
lowerCAmelCase_ = n_heads_per_shard
lowerCAmelCase_ = dim
# permute for sliced rotary
def permute(__a: List[str] , __a: List[str]=n_heads , __a: Optional[Any]=dim , __a: List[Any]=dim ):
return w.view(__a , dima // n_heads // 2 , 2 , __a ).transpose(1 , 2 ).reshape(__a , __a )
print(F"Fetching all parameters from the checkpoint at {input_base_path}." )
# Load weights
if model_size == "7B":
# Not sharded
# (The sharded implementation would also work, but this is simpler.)
lowerCAmelCase_ = torch.load(os.path.join(__a , "consolidated.00.pth" ) , map_location="cpu" )
else:
# Sharded
lowerCAmelCase_ = [
torch.load(os.path.join(__a , F"consolidated.{i:02d}.pth" ) , map_location="cpu" )
for i in range(__a )
]
lowerCAmelCase_ = 0
lowerCAmelCase_ = {"weight_map": {}}
for layer_i in range(__a ):
lowerCAmelCase_ = F"pytorch_model-{layer_i + 1}-of-{n_layers + 1}.bin"
if model_size == "7B":
# Unsharded
lowerCAmelCase_ = {
F"model.layers.{layer_i}.self_attn.q_proj.weight": permute(
loaded[F"layers.{layer_i}.attention.wq.weight"] ),
F"model.layers.{layer_i}.self_attn.k_proj.weight": permute(
loaded[F"layers.{layer_i}.attention.wk.weight"] ),
F"model.layers.{layer_i}.self_attn.v_proj.weight": loaded[F"layers.{layer_i}.attention.wv.weight"],
F"model.layers.{layer_i}.self_attn.o_proj.weight": loaded[F"layers.{layer_i}.attention.wo.weight"],
F"model.layers.{layer_i}.mlp.gate_proj.weight": loaded[F"layers.{layer_i}.feed_forward.w1.weight"],
F"model.layers.{layer_i}.mlp.down_proj.weight": loaded[F"layers.{layer_i}.feed_forward.w2.weight"],
F"model.layers.{layer_i}.mlp.up_proj.weight": loaded[F"layers.{layer_i}.feed_forward.w3.weight"],
F"model.layers.{layer_i}.input_layernorm.weight": loaded[F"layers.{layer_i}.attention_norm.weight"],
F"model.layers.{layer_i}.post_attention_layernorm.weight": loaded[F"layers.{layer_i}.ffn_norm.weight"],
}
else:
# Sharded
# Note that attention.w{q,k,v,o}, feed_fordward.w[1,2,3], attention_norm.weight and ffn_norm.weight share
# the same storage object, saving attention_norm and ffn_norm will save other weights too, which is
# redundant as other weights will be stitched from multiple shards. To avoid that, they are cloned.
lowerCAmelCase_ = {
F"model.layers.{layer_i}.input_layernorm.weight": loaded[0][
F"layers.{layer_i}.attention_norm.weight"
].clone(),
F"model.layers.{layer_i}.post_attention_layernorm.weight": loaded[0][
F"layers.{layer_i}.ffn_norm.weight"
].clone(),
}
lowerCAmelCase_ = permute(
torch.cat(
[
loaded[i][F"layers.{layer_i}.attention.wq.weight"].view(__a , __a , __a )
for i in range(__a )
] , dim=0 , ).reshape(__a , __a ) )
lowerCAmelCase_ = permute(
torch.cat(
[
loaded[i][F"layers.{layer_i}.attention.wk.weight"].view(
__a , __a , __a )
for i in range(__a )
] , dim=0 , ).reshape(__a , __a ) , __a , __a , __a , )
lowerCAmelCase_ = torch.cat(
[
loaded[i][F"layers.{layer_i}.attention.wv.weight"].view(
__a , __a , __a )
for i in range(__a )
] , dim=0 , ).reshape(__a , __a )
lowerCAmelCase_ = torch.cat(
[loaded[i][F"layers.{layer_i}.attention.wo.weight"] for i in range(__a )] , dim=1 )
lowerCAmelCase_ = torch.cat(
[loaded[i][F"layers.{layer_i}.feed_forward.w1.weight"] for i in range(__a )] , dim=0 )
lowerCAmelCase_ = torch.cat(
[loaded[i][F"layers.{layer_i}.feed_forward.w2.weight"] for i in range(__a )] , dim=1 )
lowerCAmelCase_ = torch.cat(
[loaded[i][F"layers.{layer_i}.feed_forward.w3.weight"] for i in range(__a )] , dim=0 )
lowerCAmelCase_ = inv_freq
for k, v in state_dict.items():
lowerCAmelCase_ = filename
param_count += v.numel()
torch.save(__a , os.path.join(__a , __a ) )
lowerCAmelCase_ = F"pytorch_model-{n_layers + 1}-of-{n_layers + 1}.bin"
if model_size == "7B":
# Unsharded
lowerCAmelCase_ = {
"model.embed_tokens.weight": loaded["tok_embeddings.weight"],
"model.norm.weight": loaded["norm.weight"],
"lm_head.weight": loaded["output.weight"],
}
else:
lowerCAmelCase_ = {
"model.norm.weight": loaded[0]["norm.weight"],
"model.embed_tokens.weight": torch.cat(
[loaded[i]["tok_embeddings.weight"] for i in range(__a )] , dim=1 ),
"lm_head.weight": torch.cat([loaded[i]["output.weight"] for i in range(__a )] , dim=0 ),
}
for k, v in state_dict.items():
lowerCAmelCase_ = filename
param_count += v.numel()
torch.save(__a , os.path.join(__a , __a ) )
# Write configs
lowerCAmelCase_ = {"total_size": param_count * 2}
write_json(__a , os.path.join(__a , "pytorch_model.bin.index.json" ) )
lowerCAmelCase_ = params["ffn_dim_multiplier"] if "ffn_dim_multiplier" in params else 1
lowerCAmelCase_ = params["multiple_of"] if "multiple_of" in params else 256
lowerCAmelCase_ = LlamaConfig(
hidden_size=__a , intermediate_size=compute_intermediate_size(__a , __a , __a ) , num_attention_heads=params["n_heads"] , num_hidden_layers=params["n_layers"] , rms_norm_eps=params["norm_eps"] , num_key_value_heads=__a , )
config.save_pretrained(__a )
# Make space so we can load the model properly now.
del state_dict
del loaded
gc.collect()
print("Loading the checkpoint in a Llama model." )
lowerCAmelCase_ = LlamaForCausalLM.from_pretrained(__a , torch_dtype=torch.floataa , low_cpu_mem_usage=__a )
# Avoid saving this as part of the config.
del model.config._name_or_path
print("Saving in the Transformers format." )
model.save_pretrained(__a , safe_serialization=__a )
shutil.rmtree(__a )
def A(__a: Union[str, Any] , __a: List[str] ):
# Initialize the tokenizer based on the `spm` model
lowerCAmelCase_ = LlamaTokenizer if LlamaTokenizerFast is None else LlamaTokenizerFast
print(F"Saving a {tokenizer_class.__name__} to {tokenizer_path}." )
lowerCAmelCase_ = tokenizer_class(__a )
tokenizer.save_pretrained(__a )
def A():
lowerCAmelCase_ = argparse.ArgumentParser()
parser.add_argument(
"--input_dir" , help="Location of LLaMA weights, which contains tokenizer.model and model folders" , )
parser.add_argument(
"--model_size" , choices=["7B", "7Bf", "13B", "13Bf", "30B", "65B", "70B", "70Bf", "tokenizer_only"] , )
parser.add_argument(
"--output_dir" , help="Location to write HF model and tokenizer" , )
parser.add_argument("--safe_serialization" , type=__a , help="Whether or not to save using `safetensors`." )
lowerCAmelCase_ = parser.parse_args()
if args.model_size != "tokenizer_only":
write_model(
model_path=args.output_dir , input_base_path=os.path.join(args.input_dir , args.model_size ) , model_size=args.model_size , safe_serialization=args.safe_serialization , )
lowerCAmelCase_ = os.path.join(args.input_dir , "tokenizer.model" )
write_tokenizer(args.output_dir , __a )
if __name__ == "__main__":
main()
| 226
|
import gc
import random
import tempfile
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import AutoencoderKL, DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler, UNetaDConditionModel
from diffusers.pipelines.stable_diffusion_safe import StableDiffusionPipelineSafe as StableDiffusionPipeline
from diffusers.utils import floats_tensor, nightly, torch_device
from diffusers.utils.testing_utils import require_torch_gpu
class __magic_name__ (unittest.TestCase ):
def __a ( self ) -> Dict:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def __a ( self ) -> Dict:
lowerCAmelCase_ = 1
lowerCAmelCase_ = 3
lowerCAmelCase_ = (32, 32)
lowerCAmelCase_ = floats_tensor((batch_size, num_channels) + sizes , rng=random.Random(0 ) ).to(_a )
return image
@property
def __a ( self ) -> int:
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") , cross_attention_dim=32 , )
return model
@property
def __a ( self ) -> Union[str, Any]:
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 , )
return model
@property
def __a ( self ) -> int:
torch.manual_seed(0 )
lowerCAmelCase_ = CLIPTextConfig(
bos_token_id=0 , eos_token_id=2 , hidden_size=32 , intermediate_size=37 , layer_norm_eps=1E-05 , num_attention_heads=4 , num_hidden_layers=5 , pad_token_id=1 , vocab_size=1000 , )
return CLIPTextModel(_a )
@property
def __a ( self ) -> List[str]:
def extract(*_a , **_a ):
class __magic_name__ :
def __init__( self ) -> List[str]:
lowerCAmelCase_ = torch.ones([0] )
def __a ( self , _a ) -> int:
self.pixel_values.to(_a )
return self
return Out()
return extract
def __a ( self ) -> Dict:
lowerCAmelCase_ = "cpu" # ensure determinism for the device-dependent torch.Generator
lowerCAmelCase_ = self.dummy_cond_unet
lowerCAmelCase_ = DDIMScheduler(
beta_start=0.0_0_0_8_5 , beta_end=0.0_1_2 , beta_schedule="scaled_linear" , clip_sample=_a , set_alpha_to_one=_a , )
lowerCAmelCase_ = self.dummy_vae
lowerCAmelCase_ = self.dummy_text_encoder
lowerCAmelCase_ = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" )
# make sure here that pndm scheduler skips prk
lowerCAmelCase_ = StableDiffusionPipeline(
unet=_a , scheduler=_a , vae=_a , text_encoder=_a , tokenizer=_a , safety_checker=_a , feature_extractor=self.dummy_extractor , )
lowerCAmelCase_ = sd_pipe.to(_a )
sd_pipe.set_progress_bar_config(disable=_a )
lowerCAmelCase_ = "A painting of a squirrel eating a burger"
lowerCAmelCase_ = torch.Generator(device=_a ).manual_seed(0 )
lowerCAmelCase_ = sd_pipe([prompt] , generator=_a , guidance_scale=6.0 , num_inference_steps=2 , output_type="np" )
lowerCAmelCase_ = output.images
lowerCAmelCase_ = torch.Generator(device=_a ).manual_seed(0 )
lowerCAmelCase_ = sd_pipe(
[prompt] , generator=_a , guidance_scale=6.0 , num_inference_steps=2 , output_type="np" , return_dict=_a , )[0]
lowerCAmelCase_ = image[0, -3:, -3:, -1]
lowerCAmelCase_ = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
lowerCAmelCase_ = np.array([0.5_7_5_6, 0.6_1_1_8, 0.5_0_0_5, 0.5_0_4_1, 0.5_4_7_1, 0.4_7_2_6, 0.4_9_7_6, 0.4_8_6_5, 0.4_8_6_4] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1E-2
def __a ( self ) -> Union[str, Any]:
lowerCAmelCase_ = "cpu" # ensure determinism for the device-dependent torch.Generator
lowerCAmelCase_ = self.dummy_cond_unet
lowerCAmelCase_ = PNDMScheduler(skip_prk_steps=_a )
lowerCAmelCase_ = self.dummy_vae
lowerCAmelCase_ = self.dummy_text_encoder
lowerCAmelCase_ = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" )
# make sure here that pndm scheduler skips prk
lowerCAmelCase_ = StableDiffusionPipeline(
unet=_a , scheduler=_a , vae=_a , text_encoder=_a , tokenizer=_a , safety_checker=_a , feature_extractor=self.dummy_extractor , )
lowerCAmelCase_ = sd_pipe.to(_a )
sd_pipe.set_progress_bar_config(disable=_a )
lowerCAmelCase_ = "A painting of a squirrel eating a burger"
lowerCAmelCase_ = torch.Generator(device=_a ).manual_seed(0 )
lowerCAmelCase_ = sd_pipe([prompt] , generator=_a , guidance_scale=6.0 , num_inference_steps=2 , output_type="np" )
lowerCAmelCase_ = output.images
lowerCAmelCase_ = torch.Generator(device=_a ).manual_seed(0 )
lowerCAmelCase_ = sd_pipe(
[prompt] , generator=_a , guidance_scale=6.0 , num_inference_steps=2 , output_type="np" , return_dict=_a , )[0]
lowerCAmelCase_ = image[0, -3:, -3:, -1]
lowerCAmelCase_ = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 64, 64, 3)
lowerCAmelCase_ = np.array([0.5_1_2_5, 0.5_7_1_6, 0.4_8_2_8, 0.5_0_6_0, 0.5_6_5_0, 0.4_7_6_8, 0.5_1_8_5, 0.4_8_9_5, 0.4_9_9_3] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1E-2
def __a ( self ) -> Any:
lowerCAmelCase_ = StableDiffusionPipeline.from_pretrained(
"hf-internal-testing/tiny-stable-diffusion-lms-pipe" , safety_checker=_a )
assert isinstance(_a , _a )
assert isinstance(pipe.scheduler , _a )
assert pipe.safety_checker is None
lowerCAmelCase_ = pipe("example prompt" , num_inference_steps=2 ).images[0]
assert image is not None
# check that there's no error when saving a pipeline with one of the models being None
with tempfile.TemporaryDirectory() as tmpdirname:
pipe.save_pretrained(_a )
lowerCAmelCase_ = StableDiffusionPipeline.from_pretrained(_a )
# sanity check that the pipeline still works
assert pipe.safety_checker is None
lowerCAmelCase_ = pipe("example prompt" , num_inference_steps=2 ).images[0]
assert image is not None
@unittest.skipIf(torch_device != "cuda" , "This test requires a GPU" )
def __a ( self ) -> Any:
lowerCAmelCase_ = self.dummy_cond_unet
lowerCAmelCase_ = PNDMScheduler(skip_prk_steps=_a )
lowerCAmelCase_ = self.dummy_vae
lowerCAmelCase_ = self.dummy_text_encoder
lowerCAmelCase_ = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip" )
# put models in fp16
lowerCAmelCase_ = unet.half()
lowerCAmelCase_ = vae.half()
lowerCAmelCase_ = bert.half()
# make sure here that pndm scheduler skips prk
lowerCAmelCase_ = StableDiffusionPipeline(
unet=_a , scheduler=_a , vae=_a , text_encoder=_a , tokenizer=_a , safety_checker=_a , feature_extractor=self.dummy_extractor , )
lowerCAmelCase_ = sd_pipe.to(_a )
sd_pipe.set_progress_bar_config(disable=_a )
lowerCAmelCase_ = "A painting of a squirrel eating a burger"
lowerCAmelCase_ = sd_pipe([prompt] , num_inference_steps=2 , output_type="np" ).images
assert image.shape == (1, 64, 64, 3)
@nightly
@require_torch_gpu
class __magic_name__ (unittest.TestCase ):
def __a ( self ) -> Union[str, Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def __a ( self ) -> Any:
lowerCAmelCase_ = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5" , safety_checker=_a )
lowerCAmelCase_ = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
lowerCAmelCase_ = sd_pipe.to(_a )
sd_pipe.set_progress_bar_config(disable=_a )
lowerCAmelCase_ = (
"portrait of girl with smokey eyes makeup in abandoned hotel, grange clothes, redshift, wide high angle"
" coloured polaroid photograph with flash, kodak film, hyper real, stunning moody cinematography, with"
" anamorphic lenses, by maripol, fallen angels by wong kar - wai, style of suspiria and neon demon and"
" children from bahnhof zoo, detailed "
)
lowerCAmelCase_ = 4003660346
lowerCAmelCase_ = 7
# without safety guidance (sld_guidance_scale = 0)
lowerCAmelCase_ = torch.manual_seed(_a )
lowerCAmelCase_ = sd_pipe(
[prompt] , generator=_a , guidance_scale=_a , num_inference_steps=50 , output_type="np" , width=512 , height=512 , sld_guidance_scale=0 , )
lowerCAmelCase_ = output.images
lowerCAmelCase_ = image[0, -3:, -3:, -1]
lowerCAmelCase_ = [0.2_2_7_8, 0.2_2_3_1, 0.2_2_4_9, 0.2_3_3_3, 0.2_3_0_3, 0.1_8_8_5, 0.2_2_7_3, 0.2_1_4_4, 0.2_1_7_6]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
# without safety guidance (strong configuration)
lowerCAmelCase_ = torch.manual_seed(_a )
lowerCAmelCase_ = sd_pipe(
[prompt] , generator=_a , guidance_scale=_a , num_inference_steps=50 , output_type="np" , width=512 , height=512 , sld_guidance_scale=2000 , sld_warmup_steps=7 , sld_threshold=0.0_2_5 , sld_momentum_scale=0.5 , sld_mom_beta=0.7 , )
lowerCAmelCase_ = output.images
lowerCAmelCase_ = image[0, -3:, -3:, -1]
lowerCAmelCase_ = [0.2_3_8_3, 0.2_2_7_6, 0.2_3_6, 0.2_1_9_2, 0.2_1_8_6, 0.2_0_5_3, 0.1_9_7_1, 0.1_9_0_1, 0.1_7_1_9]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
def __a ( self ) -> Optional[Any]:
lowerCAmelCase_ = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5" , safety_checker=_a )
lowerCAmelCase_ = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config )
lowerCAmelCase_ = sd_pipe.to(_a )
sd_pipe.set_progress_bar_config(disable=_a )
lowerCAmelCase_ = "padme amidala taking a bath artwork, safe for work, no nudity"
lowerCAmelCase_ = 2734971755
lowerCAmelCase_ = 7
lowerCAmelCase_ = torch.manual_seed(_a )
lowerCAmelCase_ = sd_pipe(
[prompt] , generator=_a , guidance_scale=_a , num_inference_steps=50 , output_type="np" , width=512 , height=512 , sld_guidance_scale=0 , )
lowerCAmelCase_ = output.images
lowerCAmelCase_ = image[0, -3:, -3:, -1]
lowerCAmelCase_ = [0.3_5_0_2, 0.3_6_2_2, 0.3_3_9_6, 0.3_6_4_2, 0.3_4_7_8, 0.3_3_1_8, 0.3_5, 0.3_3_4_8, 0.3_2_9_7]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
lowerCAmelCase_ = torch.manual_seed(_a )
lowerCAmelCase_ = sd_pipe(
[prompt] , generator=_a , guidance_scale=_a , num_inference_steps=50 , output_type="np" , width=512 , height=512 , sld_guidance_scale=2000 , sld_warmup_steps=7 , sld_threshold=0.0_2_5 , sld_momentum_scale=0.5 , sld_mom_beta=0.7 , )
lowerCAmelCase_ = output.images
lowerCAmelCase_ = image[0, -3:, -3:, -1]
lowerCAmelCase_ = [0.5_5_3_1, 0.5_2_0_6, 0.4_8_9_5, 0.5_1_5_6, 0.5_1_8_2, 0.4_7_5_1, 0.4_8_0_2, 0.4_8_0_3, 0.4_4_4_3]
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
def __a ( self ) -> int:
lowerCAmelCase_ = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5" )
lowerCAmelCase_ = sd_pipe.to(_a )
sd_pipe.set_progress_bar_config(disable=_a )
lowerCAmelCase_ = (
"the four horsewomen of the apocalypse, painting by tom of finland, gaston bussiere, craig mullins, j. c."
" leyendecker"
)
lowerCAmelCase_ = 1044355234
lowerCAmelCase_ = 12
lowerCAmelCase_ = torch.manual_seed(_a )
lowerCAmelCase_ = sd_pipe(
[prompt] , generator=_a , guidance_scale=_a , num_inference_steps=50 , output_type="np" , width=512 , height=512 , sld_guidance_scale=0 , )
lowerCAmelCase_ = output.images
lowerCAmelCase_ = image[0, -3:, -3:, -1]
lowerCAmelCase_ = np.array([0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-7
lowerCAmelCase_ = torch.manual_seed(_a )
lowerCAmelCase_ = sd_pipe(
[prompt] , generator=_a , guidance_scale=_a , num_inference_steps=50 , output_type="np" , width=512 , height=512 , sld_guidance_scale=2000 , sld_warmup_steps=7 , sld_threshold=0.0_2_5 , sld_momentum_scale=0.5 , sld_mom_beta=0.7 , )
lowerCAmelCase_ = output.images
lowerCAmelCase_ = image[0, -3:, -3:, -1]
lowerCAmelCase_ = np.array([0.5_8_1_8, 0.6_2_8_5, 0.6_8_3_5, 0.6_0_1_9, 0.6_2_5, 0.6_7_5_4, 0.6_0_9_6, 0.6_3_3_4, 0.6_5_6_1] )
assert image.shape == (1, 512, 512, 3)
assert np.abs(image_slice.flatten() - expected_slice ).max() < 1E-2
| 226
| 1
|
import logging
import os
import random
import sys
from dataclasses import dataclass, field
from typing import Optional
import datasets
import numpy as np
import pandas as pd
from datasets import load_dataset
import transformers
from transformers import (
AutoConfig,
BartForSequenceClassification,
DataCollatorWithPadding,
EvalPrediction,
HfArgumentParser,
TapexTokenizer,
Trainer,
TrainingArguments,
default_data_collator,
set_seed,
)
from transformers.trainer_utils import get_last_checkpoint
from transformers.utils import check_min_version
from transformers.utils.versions import require_version
# Will error if the minimal version of Transformers is not installed. Remove at your own risks.
check_min_version('4.17.0.dev0')
require_version('datasets>=1.8.0', 'To fix: pip install -r examples/pytorch/text-classification/requirements.txt')
__a = logging.getLogger(__name__)
@dataclass
class __a:
"""simple docstring"""
lowerCAmelCase = field(
default='''tab_fact''' , metadata={'''help''': '''The name of the dataset to use (via the datasets library).'''} )
lowerCAmelCase = field(
default='''tab_fact''' , metadata={'''help''': '''The configuration name of the dataset to use (via the datasets library).'''} , )
lowerCAmelCase = field(
default=1024 , metadata={
'''help''': (
'''The maximum total input sequence length after tokenization. Sequences longer '''
'''than this will be truncated, sequences shorter will be padded.'''
)
} , )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={'''help''': '''Overwrite the cached preprocessed datasets or not.'''} )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={
'''help''': (
'''Whether to pad all samples to `max_seq_length`. '''
'''If False, will pad the samples dynamically when batching to the maximum length in the batch.'''
)
} , )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={
'''help''': (
'''For debugging purposes or quicker training, truncate the number of training examples to this '''
'''value if set.'''
)
} , )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={
'''help''': (
'''For debugging purposes or quicker training, truncate the number of evaluation examples to this '''
'''value if set.'''
)
} , )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={
'''help''': (
'''For debugging purposes or quicker training, truncate the number of prediction examples to this '''
'''value if set.'''
)
} , )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={'''help''': '''A csv or a json file containing the training data.'''} )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={'''help''': '''A csv or a json file containing the validation data.'''} )
lowerCAmelCase = field(default=__lowerCamelCase , metadata={'''help''': '''A csv or a json file containing the test data.'''} )
def a__ ( self ) -> str:
if self.dataset_name is not None:
pass
elif self.train_file is None or self.validation_file is None:
raise ValueError('''Need either a GLUE task, a training/validation file or a dataset name.''' )
else:
UpperCAmelCase_ : List[str] = self.train_file.split('''.''' )[-1]
assert train_extension in ["csv", "json"], "`train_file` should be a csv or a json file."
UpperCAmelCase_ : Union[str, Any] = self.validation_file.split('''.''' )[-1]
assert (
validation_extension == train_extension
), "`validation_file` should have the same extension (csv or json) as `train_file`."
@dataclass
class __a:
"""simple docstring"""
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={'''help''': '''Path to pretrained model or model identifier from huggingface.co/models'''} )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={'''help''': '''Pretrained config name or path if not the same as model_name'''} )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={'''help''': '''Pretrained tokenizer name or path if not the same as model_name'''} )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={'''help''': '''Where do you want to store the pretrained models downloaded from huggingface.co'''} , )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={'''help''': '''Whether to use one of the fast tokenizer (backed by the tokenizers library) or not.'''} , )
lowerCAmelCase = field(
default='''main''' , metadata={'''help''': '''The specific model version to use (can be a branch name, tag name or commit id).'''} , )
lowerCAmelCase = field(
default=__lowerCamelCase , metadata={
'''help''': (
'''Will use the token generated when running `huggingface-cli login` (necessary to use this script '''
'''with private models).'''
)
} , )
def lowerCamelCase__ ( ):
'''simple docstring'''
UpperCAmelCase_ : Dict = HfArgumentParser((ModelArguments, DataTrainingArguments, TrainingArguments) )
if len(sys.argv ) == 2 and sys.argv[1].endswith('''.json''' ):
# If we pass only one argument to the script and it's the path to a json file,
# let's parse it to get our arguments.
UpperCAmelCase_, UpperCAmelCase_, UpperCAmelCase_ : Dict = parser.parse_json_file(json_file=os.path.abspath(sys.argv[1] ) )
else:
UpperCAmelCase_, UpperCAmelCase_, UpperCAmelCase_ : int = parser.parse_args_into_dataclasses()
# Setup logging
logging.basicConfig(
format='''%(asctime)s - %(levelname)s - %(name)s - %(message)s''' , datefmt='''%m/%d/%Y %H:%M:%S''' , handlers=[logging.StreamHandler(sys.stdout )] , )
UpperCAmelCase_ : int = training_args.get_process_log_level()
logger.setLevel(__UpperCamelCase )
datasets.utils.logging.set_verbosity(__UpperCamelCase )
transformers.utils.logging.set_verbosity(__UpperCamelCase )
transformers.utils.logging.enable_default_handler()
transformers.utils.logging.enable_explicit_format()
# Log on each process the small summary:
logger.warning(
f'''Process rank: {training_args.local_rank}, device: {training_args.device}, n_gpu: {training_args.n_gpu}'''
+ f'''distributed training: {bool(training_args.local_rank != -1 )}, 16-bits training: {training_args.fpaa}''' )
logger.info(f'''Training/evaluation parameters {training_args}''' )
# Detecting last checkpoint.
UpperCAmelCase_ : List[Any] = None
if os.path.isdir(training_args.output_dir ) and training_args.do_train and not training_args.overwrite_output_dir:
UpperCAmelCase_ : Any = get_last_checkpoint(training_args.output_dir )
if last_checkpoint is None and len(os.listdir(training_args.output_dir ) ) > 0:
raise ValueError(
f'''Output directory ({training_args.output_dir}) already exists and is not empty. '''
'''Use --overwrite_output_dir to overcome.''' )
elif last_checkpoint is not None and training_args.resume_from_checkpoint is None:
logger.info(
f'''Checkpoint detected, resuming training at {last_checkpoint}. To avoid this behavior, change '''
'''the `--output_dir` or add `--overwrite_output_dir` to train from scratch.''' )
# Set seed before initializing model.
set_seed(training_args.seed )
# Get the datasets: you can either provide your own CSV/JSON training and evaluation files (see below)
# or specify a GLUE benchmark task (the dataset will be downloaded automatically from the datasets Hub).
#
# For JSON files, this script will use the `question` column for the input question and `table` column for the corresponding table.
#
# If the CSVs/JSONs contain only one non-label column, the script does single sentence classification on this
# single column. You can easily tweak this behavior (see below)
#
# In distributed training, the load_dataset function guarantee that only one local process can concurrently
# download the dataset.
if data_args.dataset_name is not None:
# Downloading and loading a dataset from the hub.
UpperCAmelCase_ : Optional[Any] = load_dataset(
data_args.dataset_name , data_args.dataset_config_name , cache_dir=model_args.cache_dir )
else:
# Loading a dataset from your local files.
# CSV/JSON training and evaluation files are needed.
UpperCAmelCase_ : Tuple = {'''train''': data_args.train_file, '''validation''': data_args.validation_file}
# Get the test dataset: you can provide your own CSV/JSON test file (see below)
# when you use `do_predict` without specifying a GLUE benchmark task.
if training_args.do_predict:
if data_args.test_file is not None:
UpperCAmelCase_ : Optional[int] = data_args.train_file.split('''.''' )[-1]
UpperCAmelCase_ : Dict = data_args.test_file.split('''.''' )[-1]
assert (
test_extension == train_extension
), "`test_file` should have the same extension (csv or json) as `train_file`."
UpperCAmelCase_ : Optional[int] = data_args.test_file
else:
raise ValueError('''Need either a GLUE task or a test file for `do_predict`.''' )
for key in data_files.keys():
logger.info(f'''load a local file for {key}: {data_files[key]}''' )
if data_args.train_file.endswith('''.csv''' ):
# Loading a dataset from local csv files
UpperCAmelCase_ : Optional[int] = load_dataset('''csv''' , data_files=__UpperCamelCase , cache_dir=model_args.cache_dir )
else:
# Loading a dataset from local json files
UpperCAmelCase_ : Union[str, Any] = load_dataset('''json''' , data_files=__UpperCamelCase , cache_dir=model_args.cache_dir )
# See more about loading any type of standard or custom dataset at
# https://huggingface.co/docs/datasets/loading_datasets.html.
# Labels
UpperCAmelCase_ : str = raw_datasets['''train'''].features['''label'''].names
UpperCAmelCase_ : int = len(__UpperCamelCase )
# Load pretrained model and tokenizer
#
# In distributed training, the .from_pretrained methods guarantee that only one local process can concurrently
# download model & vocab.
UpperCAmelCase_ : Dict = AutoConfig.from_pretrained(
model_args.config_name if model_args.config_name else model_args.model_name_or_path , num_labels=__UpperCamelCase , cache_dir=model_args.cache_dir , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , )
# load tapex tokenizer
UpperCAmelCase_ : int = TapexTokenizer.from_pretrained(
model_args.tokenizer_name if model_args.tokenizer_name else model_args.model_name_or_path , cache_dir=model_args.cache_dir , use_fast=model_args.use_fast_tokenizer , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , add_prefix_space=__UpperCamelCase , )
UpperCAmelCase_ : Any = BartForSequenceClassification.from_pretrained(
model_args.model_name_or_path , from_tf=bool('''.ckpt''' in model_args.model_name_or_path ) , config=__UpperCamelCase , cache_dir=model_args.cache_dir , revision=model_args.model_revision , use_auth_token=True if model_args.use_auth_token else None , )
# Padding strategy
if data_args.pad_to_max_length:
UpperCAmelCase_ : Union[str, Any] = '''max_length'''
else:
# We will pad later, dynamically at batch creation, to the max sequence length in each batch
UpperCAmelCase_ : Union[str, Any] = False
# Some models have set the order of the labels to use, so let's make sure we do use it.
UpperCAmelCase_ : Optional[int] = {'''Refused''': 0, '''Entailed''': 1}
UpperCAmelCase_ : Union[str, Any] = {0: '''Refused''', 1: '''Entailed'''}
if data_args.max_seq_length > tokenizer.model_max_length:
logger.warning(
f'''The max_seq_length passed ({data_args.max_seq_length}) is larger than the maximum length for the'''
f'''model ({tokenizer.model_max_length}). Using max_seq_length={tokenizer.model_max_length}.''' )
UpperCAmelCase_ : str = min(data_args.max_seq_length , tokenizer.model_max_length )
def preprocess_tabfact_function(_lowercase ):
# Tokenize the texts
def _convert_table_text_to_pandas(_lowercase ):
UpperCAmelCase_ : Union[str, Any] = [_table_row.split('''#''' ) for _table_row in _table_text.strip('''\n''' ).split('''\n''' )]
UpperCAmelCase_ : List[Any] = pd.DataFrame.from_records(_table_content[1:] , columns=_table_content[0] )
return _table_pd
UpperCAmelCase_ : int = examples['''statement''']
UpperCAmelCase_ : Any = list(map(_convert_table_text_to_pandas , examples['''table_text'''] ) )
UpperCAmelCase_ : Union[str, Any] = tokenizer(__UpperCamelCase , __UpperCamelCase , padding=__UpperCamelCase , max_length=__UpperCamelCase , truncation=__UpperCamelCase )
UpperCAmelCase_ : List[Any] = examples['''label''']
return result
with training_args.main_process_first(desc='''dataset map pre-processing''' ):
UpperCAmelCase_ : Dict = raw_datasets.map(
__UpperCamelCase , batched=__UpperCamelCase , load_from_cache_file=not data_args.overwrite_cache , desc='''Running tokenizer on dataset''' , )
if training_args.do_train:
if "train" not in raw_datasets:
raise ValueError('''--do_train requires a train dataset''' )
UpperCAmelCase_ : Union[str, Any] = raw_datasets['''train''']
if data_args.max_train_samples is not None:
UpperCAmelCase_ : Union[str, Any] = train_dataset.select(range(data_args.max_train_samples ) )
if training_args.do_eval:
if "validation" not in raw_datasets and "validation_matched" not in raw_datasets:
raise ValueError('''--do_eval requires a validation dataset''' )
UpperCAmelCase_ : Dict = raw_datasets['''validation''']
if data_args.max_eval_samples is not None:
UpperCAmelCase_ : List[str] = eval_dataset.select(range(data_args.max_eval_samples ) )
if training_args.do_predict or data_args.test_file is not None:
if "test" not in raw_datasets and "test_matched" not in raw_datasets:
raise ValueError('''--do_predict requires a test dataset''' )
UpperCAmelCase_ : Union[str, Any] = raw_datasets['''test''']
if data_args.max_predict_samples is not None:
UpperCAmelCase_ : Optional[int] = predict_dataset.select(range(data_args.max_predict_samples ) )
# Log a few random samples from the training set:
if training_args.do_train:
for index in random.sample(range(len(__UpperCamelCase ) ) , 3 ):
logger.info(f'''Sample {index} of the training set: {train_dataset[index]}.''' )
# You can define your custom compute_metrics function. It takes an `EvalPrediction` object (a namedtuple with a
# predictions and label_ids field) and has to return a dictionary string to float.
def compute_metrics(_lowercase ):
UpperCAmelCase_ : int = p.predictions[0] if isinstance(p.predictions , __UpperCamelCase ) else p.predictions
UpperCAmelCase_ : Any = np.argmax(__UpperCamelCase , axis=1 )
return {"accuracy": (preds == p.label_ids).astype(np.floataa ).mean().item()}
# Data collator will default to DataCollatorWithPadding, so we change it if we already did the padding.
if data_args.pad_to_max_length:
UpperCAmelCase_ : Tuple = default_data_collator
elif training_args.fpaa:
UpperCAmelCase_ : Tuple = DataCollatorWithPadding(__UpperCamelCase , pad_to_multiple_of=8 )
else:
UpperCAmelCase_ : Tuple = None
# Initialize our Trainer
UpperCAmelCase_ : Optional[Any] = Trainer(
model=__UpperCamelCase , args=__UpperCamelCase , train_dataset=train_dataset if training_args.do_train else None , eval_dataset=eval_dataset if training_args.do_eval else None , compute_metrics=__UpperCamelCase , tokenizer=__UpperCamelCase , data_collator=__UpperCamelCase , )
# Training
if training_args.do_train:
UpperCAmelCase_ : List[Any] = None
if training_args.resume_from_checkpoint is not None:
UpperCAmelCase_ : Union[str, Any] = training_args.resume_from_checkpoint
elif last_checkpoint is not None:
UpperCAmelCase_ : List[str] = last_checkpoint
UpperCAmelCase_ : int = trainer.train(resume_from_checkpoint=__UpperCamelCase )
UpperCAmelCase_ : Union[str, Any] = train_result.metrics
UpperCAmelCase_ : Union[str, Any] = (
data_args.max_train_samples if data_args.max_train_samples is not None else len(__UpperCamelCase )
)
UpperCAmelCase_ : Any = min(__UpperCamelCase , len(__UpperCamelCase ) )
trainer.save_model() # Saves the tokenizer too for easy upload
trainer.log_metrics('''train''' , __UpperCamelCase )
trainer.save_metrics('''train''' , __UpperCamelCase )
trainer.save_state()
# Evaluation
if training_args.do_eval:
logger.info('''*** Evaluate ***''' )
UpperCAmelCase_ : Tuple = trainer.evaluate(eval_dataset=__UpperCamelCase )
UpperCAmelCase_ : Dict = data_args.max_eval_samples if data_args.max_eval_samples is not None else len(__UpperCamelCase )
UpperCAmelCase_ : Optional[Any] = min(__UpperCamelCase , len(__UpperCamelCase ) )
trainer.log_metrics('''eval''' , __UpperCamelCase )
trainer.save_metrics('''eval''' , __UpperCamelCase )
if training_args.do_predict:
logger.info('''*** Predict ***''' )
# Removing the `label` columns because it contains -1 and Trainer won't like that.
UpperCAmelCase_ : int = predict_dataset.remove_columns('''label''' )
UpperCAmelCase_ : Dict = trainer.predict(__UpperCamelCase , metric_key_prefix='''predict''' ).predictions
UpperCAmelCase_ : List[str] = np.argmax(__UpperCamelCase , axis=1 )
UpperCAmelCase_ : Optional[Any] = os.path.join(training_args.output_dir , '''predict_results_tabfact.txt''' )
if trainer.is_world_process_zero():
with open(__UpperCamelCase , '''w''' ) as writer:
logger.info('''***** Predict Results *****''' )
writer.write('''index\tprediction\n''' )
for index, item in enumerate(__UpperCamelCase ):
UpperCAmelCase_ : List[Any] = label_list[item]
writer.write(f'''{index}\t{item}\n''' )
UpperCAmelCase_ : Dict = {'''finetuned_from''': model_args.model_name_or_path, '''tasks''': '''text-classification'''}
if training_args.push_to_hub:
trainer.push_to_hub(**__UpperCamelCase )
else:
trainer.create_model_card(**__UpperCamelCase )
def lowerCamelCase__ ( _lowercase ):
'''simple docstring'''
main()
if __name__ == "__main__":
main()
| 30
|
import argparse
import json
import os
import re
import torch
from transformers import BloomConfig, BloomModel
from transformers.file_utils import CONFIG_NAME, WEIGHTS_NAME
from transformers.utils import logging
logging.set_verbosity_info()
lowerCamelCase_ = [
"word_embeddings_layernorm.weight",
"word_embeddings_layernorm.bias",
"input_layernorm.weight",
"input_layernorm.bias",
"post_attention_layernorm.weight",
"post_attention_layernorm.bias",
"self_attention.dense.bias",
"mlp.dense_4h_to_h.bias",
"ln_f.weight",
"ln_f.bias",
]
lowerCamelCase_ = [
"mlp.dense_4h_to_h.weight",
"self_attention.dense.weight",
]
def UpperCAmelCase_ ( __UpperCamelCase, __UpperCamelCase ):
SCREAMING_SNAKE_CASE__ ={
"""word_embeddings.weight""": """word_embeddings.weight""",
"""word_embeddings.norm.weight""": """word_embeddings_layernorm.weight""",
"""word_embeddings.norm.bias""": """word_embeddings_layernorm.bias""",
"""weight""": """ln_f.weight""",
"""bias""": """ln_f.bias""",
}
if key in layer_rename_map:
return layer_rename_map[key]
# Handle transformer blocks
SCREAMING_SNAKE_CASE__ =int(re.match(R""".*layer_(\d*).*""", __UpperCamelCase )[1] )
layer_number -= 3
return f"""h.{layer_number}.""" + key
def UpperCAmelCase_ ( __UpperCamelCase ):
if dtype == torch.bool:
return 1 / 8
SCREAMING_SNAKE_CASE__ =re.search(R"""[^\d](\d+)$""", str(__UpperCamelCase ) )
if bit_search is None:
raise ValueError(f"""`dtype` is not a valid dtype: {dtype}.""" )
SCREAMING_SNAKE_CASE__ =int(bit_search.groups()[0] )
return bit_size // 8
def UpperCAmelCase_ ( __UpperCamelCase, __UpperCamelCase, __UpperCamelCase, __UpperCamelCase, __UpperCamelCase ):
# Construct model
if bloom_config_file == "":
SCREAMING_SNAKE_CASE__ =BloomConfig()
else:
SCREAMING_SNAKE_CASE__ =BloomConfig.from_json_file(__UpperCamelCase )
if shard_model:
SCREAMING_SNAKE_CASE__ =os.listdir(__UpperCamelCase )
SCREAMING_SNAKE_CASE__ =sorted(filter(lambda __UpperCamelCase : s.startswith("""layer""" ) and "model_00" in s, __UpperCamelCase ) )
SCREAMING_SNAKE_CASE__ ={"""weight_map""": {}, """metadata""": {}}
SCREAMING_SNAKE_CASE__ =0
SCREAMING_SNAKE_CASE__ =None
SCREAMING_SNAKE_CASE__ =BloomConfig()
for j, file in enumerate(__UpperCamelCase ):
print("""Processing file: {}""".format(__UpperCamelCase ) )
SCREAMING_SNAKE_CASE__ =None
for i in range(__UpperCamelCase ):
# load all TP files
SCREAMING_SNAKE_CASE__ =file.replace("""model_00""", f"""model_0{i}""" )
SCREAMING_SNAKE_CASE__ =torch.load(os.path.join(__UpperCamelCase, __UpperCamelCase ), map_location="""cpu""" )
# Rename keys in the transformers names
SCREAMING_SNAKE_CASE__ =list(temp.keys() )
for key in keys:
SCREAMING_SNAKE_CASE__ =temp.pop(__UpperCamelCase )
if tensors is None:
SCREAMING_SNAKE_CASE__ =temp
else:
for key in tensors.keys():
if any(key.endswith(__UpperCamelCase ) for end in WEIGHTS_TO_AVERAGE_ENDSWITH ):
# We average (sum and then divide) some weights accross TP ranks (see https://github.com/bigscience-workshop/Megatron-DeepSpeed/blob/olruwase/sync_layer_norms/megatron/training.py#L425)
tensors[key] += temp[key]
else:
# Some weights are RowParallelLinear in Megatron-Deepspeed, others are ColumnParallel
SCREAMING_SNAKE_CASE__ =1 if any(text in key for text in WEIGHTS_WITH_ROW_PARALLELISM_CONTAIN ) else 0
# We concatenate these weights accross TP ranks
SCREAMING_SNAKE_CASE__ =torch.cat([tensors[key], temp[key]], dim=__UpperCamelCase )
# Divide by the number of TP the weights we want to average
for key in tensors.keys():
if any(key.endswith(__UpperCamelCase ) for end in WEIGHTS_TO_AVERAGE_ENDSWITH ):
SCREAMING_SNAKE_CASE__ =tensors[key] / pretraining_tp
torch.save(
__UpperCamelCase, os.path.join(
__UpperCamelCase, """pytorch_model_{}-of-{}.bin""".format(str(j + 1 ).zfill(5 ), str(len(__UpperCamelCase ) ).zfill(5 ) ), ), )
for key in tensors.keys():
SCREAMING_SNAKE_CASE__ =tensors[key]
total_size += value.numel() * get_dtype_size(value.dtype )
if key not in index_dict["weight_map"]:
SCREAMING_SNAKE_CASE__ ="""pytorch_model_{}-of-{}.bin""".format(
str(j + 1 ).zfill(5 ), str(len(__UpperCamelCase ) ).zfill(5 ) )
SCREAMING_SNAKE_CASE__ =BloomConfig()
SCREAMING_SNAKE_CASE__ =pytorch_dump_folder_path + """/""" + CONFIG_NAME
SCREAMING_SNAKE_CASE__ =total_size
with open(__UpperCamelCase, """w""", encoding="""utf-8""" ) as f:
f.write(config.to_json_string() )
with open(os.path.join(__UpperCamelCase, WEIGHTS_NAME + """.index.json""" ), """w""", encoding="""utf-8""" ) as f:
SCREAMING_SNAKE_CASE__ =json.dumps(__UpperCamelCase, indent=2, sort_keys=__UpperCamelCase ) + """\n"""
f.write(__UpperCamelCase )
else:
SCREAMING_SNAKE_CASE__ =BloomModel(__UpperCamelCase )
SCREAMING_SNAKE_CASE__ =os.listdir(__UpperCamelCase )
SCREAMING_SNAKE_CASE__ =sorted(filter(lambda __UpperCamelCase : s.startswith("""layer""" ) and "model_00" in s, __UpperCamelCase ) )
SCREAMING_SNAKE_CASE__ =None
for i, file in enumerate(__UpperCamelCase ):
SCREAMING_SNAKE_CASE__ =None
for i in range(__UpperCamelCase ):
# load all TP files
SCREAMING_SNAKE_CASE__ =file.replace("""model_00""", f"""model_0{i}""" )
SCREAMING_SNAKE_CASE__ =torch.load(os.path.join(__UpperCamelCase, __UpperCamelCase ), map_location="""cpu""" )
# Rename keys in the transformers names
SCREAMING_SNAKE_CASE__ =list(temp.keys() )
for key in keys:
SCREAMING_SNAKE_CASE__ =temp.pop(__UpperCamelCase )
if tensors is None:
SCREAMING_SNAKE_CASE__ =temp
else:
for key in tensors.keys():
# We average (sum and then divide) some weights accross TP ranks (see https://github.com/bigscience-workshop/Megatron-DeepSpeed/blob/olruwase/sync_layer_norms/megatron/training.py#L425)
if any(key.endswith(__UpperCamelCase ) for end in WEIGHTS_TO_AVERAGE_ENDSWITH ):
tensors[key] += temp[key]
else:
# Some weights are RowParallelLinear in Megatron-Deepspeed, others are ColumnParallel
SCREAMING_SNAKE_CASE__ =1 if any(text in key for text in WEIGHTS_WITH_ROW_PARALLELISM_CONTAIN ) else 0
# We concatenate these weights accross TP ranks
SCREAMING_SNAKE_CASE__ =torch.cat([tensors[key], temp[key]], dim=__UpperCamelCase )
# Divide by the number of TP the weights we want to average
for key in tensors.keys():
if any(key.endswith(__UpperCamelCase ) for end in WEIGHTS_TO_AVERAGE_ENDSWITH ):
SCREAMING_SNAKE_CASE__ =tensors[key] / pretraining_tp
SCREAMING_SNAKE_CASE__ =model.load_state_dict(__UpperCamelCase, strict=__UpperCamelCase )
assert not other_keys.unexpected_keys, f"""The keys {other_keys.unexpected_keys} are unexpected"""
if missing_keys is None:
SCREAMING_SNAKE_CASE__ =set(other_keys.missing_keys )
else:
SCREAMING_SNAKE_CASE__ =missing_keys.intersection(set(other_keys.missing_keys ) )
assert not missing_keys, f"""The keys {missing_keys} are missing"""
# Save pytorch-model
os.makedirs(__UpperCamelCase, exist_ok=__UpperCamelCase )
SCREAMING_SNAKE_CASE__ =pytorch_dump_folder_path + """/""" + WEIGHTS_NAME
SCREAMING_SNAKE_CASE__ =pytorch_dump_folder_path + """/""" + CONFIG_NAME
print(f"""Save PyTorch model to {pytorch_weights_dump_path} with dtype {config.torch_dtype}""" )
if config.torch_dtype is not None:
SCREAMING_SNAKE_CASE__ =model.to(config.torch_dtype )
torch.save(model.state_dict(), __UpperCamelCase )
print(f"""Save configuration file to {pytorch_config_dump_path}""" )
with open(__UpperCamelCase, """w""", encoding="""utf-8""" ) as f:
f.write(config.to_json_string() )
if __name__ == "__main__":
lowerCamelCase_ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--bloom_checkpoint_path",
default=None,
type=str,
required=True,
help="Path to the Megatron-LM checkpoint path.",
)
parser.add_argument(
"--pytorch_dump_folder_path", default=None, type=str, required=True, help="Path to the output PyTorch model."
)
parser.add_argument(
"--bloom_config_file",
default="",
type=str,
help=(
"An optional config json file corresponding to the pre-trained model. \n"
"This specifies the model architecture."
),
)
parser.add_argument(
"--shard_model",
action="store_true",
help="An optional setting to shard the output model \nThis enables sharding the converted checkpoint",
)
parser.add_argument(
"--pretraining_tp",
default=4,
type=int,
help="Pretraining TP rank that has been used when training the model in Megatron-LM \n",
)
lowerCamelCase_ = parser.parse_args()
convert_bloom_checkpoint_to_pytorch(
args.bloom_checkpoint_path,
args.bloom_config_file,
args.pytorch_dump_folder_path,
args.shard_model,
args.pretraining_tp,
)
| 151
| 0
|
from __future__ import annotations
from math import gcd
def _A (UpperCamelCase : int , UpperCamelCase : int = 2 , UpperCamelCase : int = 1 , UpperCamelCase : int = 3 , ) ->int | None:
'''simple docstring'''
if num < 2:
raise ValueError("""The input value cannot be less than 2""" )
# Because of the relationship between ``f(f(x))`` and ``f(x)``, this
# algorithm struggles to find factors that are divisible by two.
# As a workaround, we specifically check for two and even inputs.
# See: https://math.stackexchange.com/a/2856214/165820
if num > 2 and num % 2 == 0:
return 2
# Pollard's Rho algorithm requires a function that returns pseudorandom
# values between 0 <= X < ``num``. It doesn't need to be random in the
# sense that the output value is cryptographically secure or difficult
# to calculate, it only needs to be random in the sense that all output
# values should be equally likely to appear.
# For this reason, Pollard suggested using ``f(x) = (x**2 - 1) % num``
# However, the success of Pollard's algorithm isn't guaranteed and is
# determined in part by the initial seed and the chosen random function.
# To make retries easier, we will instead use ``f(x) = (x**2 + C) % num``
# where ``C`` is a value that we can modify between each attempt.
def rand_fn(UpperCamelCase : int , UpperCamelCase : int , UpperCamelCase : int ) -> int:
return (pow(UpperCamelCase , 2 ) + step) % modulus
for _ in range(UpperCamelCase ):
# These track the position within the cycle detection logic.
lowerCamelCase__ : int = seed
lowerCamelCase__ : str = seed
while True:
# At each iteration, the tortoise moves one step and the hare moves two.
lowerCamelCase__ : List[Any] = rand_fn(UpperCamelCase , UpperCamelCase , UpperCamelCase )
lowerCamelCase__ : str = rand_fn(UpperCamelCase , UpperCamelCase , UpperCamelCase )
lowerCamelCase__ : str = rand_fn(UpperCamelCase , UpperCamelCase , UpperCamelCase )
# At some point both the tortoise and the hare will enter a cycle whose
# length ``p`` is a divisor of ``num``. Once in that cycle, at some point
# the tortoise and hare will end up on the same value modulo ``p``.
# We can detect when this happens because the position difference between
# the tortoise and the hare will share a common divisor with ``num``.
lowerCamelCase__ : Any = gcd(hare - tortoise , UpperCamelCase )
if divisor == 1:
# No common divisor yet, just keep searching.
continue
else:
# We found a common divisor!
if divisor == num:
# Unfortunately, the divisor is ``num`` itself and is useless.
break
else:
# The divisor is a nontrivial factor of ``num``!
return divisor
# If we made it here, then this attempt failed.
# We need to pick a new starting seed for the tortoise and hare
# in addition to a new step value for the random function.
# To keep this example implementation deterministic, the
# new values will be generated based on currently available
# values instead of using something like ``random.randint``.
# We can use the hare's position as the new seed.
# This is actually what Richard Brent's the "optimized" variant does.
lowerCamelCase__ : List[Any] = hare
# The new step value for the random function can just be incremented.
# At first the results will be similar to what the old function would
# have produced, but the value will quickly diverge after a bit.
step += 1
# We haven't found a divisor within the requested number of attempts.
# We were unlucky or ``num`` itself is actually prime.
return None
if __name__ == "__main__":
import argparse
_lowercase = argparse.ArgumentParser()
parser.add_argument(
'''num''',
type=int,
help='''The value to find a divisor of''',
)
parser.add_argument(
'''--attempts''',
type=int,
default=3,
help='''The number of attempts before giving up''',
)
_lowercase = parser.parse_args()
_lowercase = pollard_rho(args.num, attempts=args.attempts)
if divisor is None:
print(F'''{args.num} is probably prime''')
else:
_lowercase = args.num // divisor
print(F'''{args.num} = {divisor} * {quotient}''')
| 96
|
def _A (UpperCamelCase : List[Any] , UpperCamelCase : Dict ) ->List[str]:
'''simple docstring'''
lowerCamelCase__ : int = [1]
for i in range(2 , UpperCamelCase ):
factorials.append(factorials[-1] * i )
assert 0 <= k < factorials[-1] * n, "k out of bounds"
lowerCamelCase__ : Union[str, Any] = []
lowerCamelCase__ : Optional[Any] = list(range(UpperCamelCase ) )
# Find permutation
while factorials:
lowerCamelCase__ : Union[str, Any] = factorials.pop()
lowerCamelCase__ ,lowerCamelCase__ : Union[str, Any] = divmod(UpperCamelCase , UpperCamelCase )
permutation.append(elements[number] )
elements.remove(elements[number] )
permutation.append(elements[0] )
return permutation
if __name__ == "__main__":
import doctest
doctest.testmod()
| 96
| 1
|
"""simple docstring"""
import os
import tempfile
import unittest
from transformers import NezhaConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_torch, require_torch_gpu, slow, torch_device
from ...generation.test_utils import GenerationTesterMixin
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
NezhaForMaskedLM,
NezhaForMultipleChoice,
NezhaForNextSentencePrediction,
NezhaForPreTraining,
NezhaForQuestionAnswering,
NezhaForSequenceClassification,
NezhaForTokenClassification,
NezhaModel,
)
from transformers.models.nezha.modeling_nezha import NEZHA_PRETRAINED_MODEL_ARCHIVE_LIST
class __snake_case :
"""simple docstring"""
def __init__( self :str , UpperCamelCase__ :List[str] , UpperCamelCase__ :List[Any]=13 , UpperCamelCase__ :Tuple=7 , UpperCamelCase__ :Tuple=True , UpperCamelCase__ :Union[str, Any]=True , UpperCamelCase__ :Dict=True , UpperCamelCase__ :Tuple=True , UpperCamelCase__ :Dict=99 , UpperCamelCase__ :Union[str, Any]=32 , UpperCamelCase__ :Optional[int]=5 , UpperCamelCase__ :List[str]=4 , UpperCamelCase__ :Any=37 , UpperCamelCase__ :Union[str, Any]="gelu" , UpperCamelCase__ :Tuple=0.1 , UpperCamelCase__ :Union[str, Any]=0.1 , UpperCamelCase__ :Optional[int]=128 , UpperCamelCase__ :Any=32 , UpperCamelCase__ :Union[str, Any]=16 , UpperCamelCase__ :Tuple=2 , UpperCamelCase__ :int=0.02 , UpperCamelCase__ :Tuple=3 , UpperCamelCase__ :Optional[Any]=4 , UpperCamelCase__ :List[Any]=None , ):
_a = parent
_a = batch_size
_a = seq_length
_a = is_training
_a = use_input_mask
_a = use_token_type_ids
_a = use_labels
_a = vocab_size
_a = hidden_size
_a = num_hidden_layers
_a = num_attention_heads
_a = intermediate_size
_a = hidden_act
_a = hidden_dropout_prob
_a = attention_probs_dropout_prob
_a = max_position_embeddings
_a = type_vocab_size
_a = type_sequence_label_size
_a = initializer_range
_a = num_labels
_a = num_choices
_a = scope
def SCREAMING_SNAKE_CASE_ ( self :List[str] ):
_a = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
_a = None
if self.use_input_mask:
_a = random_attention_mask([self.batch_size, self.seq_length] )
_a = None
if self.use_token_type_ids:
_a = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
_a = None
_a = None
_a = None
if self.use_labels:
_a = ids_tensor([self.batch_size] , self.type_sequence_label_size )
_a = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
_a = ids_tensor([self.batch_size] , self.num_choices )
_a = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def SCREAMING_SNAKE_CASE_ ( self :Optional[int] ):
return NezhaConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=UpperCamelCase__ , initializer_range=self.initializer_range , )
def SCREAMING_SNAKE_CASE_ ( self :Optional[int] ):
(
(
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) ,
) = self.prepare_config_and_inputs()
_a = True
_a = floats_tensor([self.batch_size, self.seq_length, self.hidden_size] )
_a = ids_tensor([self.batch_size, self.seq_length] , vocab_size=2 )
return (
config,
input_ids,
token_type_ids,
input_mask,
sequence_labels,
token_labels,
choice_labels,
encoder_hidden_states,
encoder_attention_mask,
)
def SCREAMING_SNAKE_CASE_ ( self :Optional[int] , UpperCamelCase__ :str , UpperCamelCase__ :Union[str, Any] , UpperCamelCase__ :Dict , UpperCamelCase__ :List[Any] , UpperCamelCase__ :Tuple , UpperCamelCase__ :Dict , UpperCamelCase__ :List[str] ):
_a = NezhaModel(config=UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = model(UpperCamelCase__ , attention_mask=UpperCamelCase__ , token_type_ids=UpperCamelCase__ )
_a = model(UpperCamelCase__ , token_type_ids=UpperCamelCase__ )
_a = model(UpperCamelCase__ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
self.parent.assertEqual(result.pooler_output.shape , (self.batch_size, self.hidden_size) )
def SCREAMING_SNAKE_CASE_ ( self :Union[str, Any] , UpperCamelCase__ :List[Any] , UpperCamelCase__ :List[Any] , UpperCamelCase__ :Any , UpperCamelCase__ :int , UpperCamelCase__ :Any , UpperCamelCase__ :Tuple , UpperCamelCase__ :str , UpperCamelCase__ :Optional[int] , UpperCamelCase__ :Union[str, Any] , ):
_a = True
_a = NezhaModel(UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = model(
UpperCamelCase__ , attention_mask=UpperCamelCase__ , token_type_ids=UpperCamelCase__ , encoder_hidden_states=UpperCamelCase__ , encoder_attention_mask=UpperCamelCase__ , )
_a = model(
UpperCamelCase__ , attention_mask=UpperCamelCase__ , token_type_ids=UpperCamelCase__ , encoder_hidden_states=UpperCamelCase__ , )
_a = 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) )
self.parent.assertEqual(result.pooler_output.shape , (self.batch_size, self.hidden_size) )
def SCREAMING_SNAKE_CASE_ ( self :List[str] , UpperCamelCase__ :List[Any] , UpperCamelCase__ :List[str] , UpperCamelCase__ :List[str] , UpperCamelCase__ :str , UpperCamelCase__ :str , UpperCamelCase__ :Tuple , UpperCamelCase__ :Dict ):
_a = NezhaForMaskedLM(config=UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = 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 SCREAMING_SNAKE_CASE_ ( self :str , UpperCamelCase__ :int , UpperCamelCase__ :List[str] , UpperCamelCase__ :Tuple , UpperCamelCase__ :Tuple , UpperCamelCase__ :List[str] , UpperCamelCase__ :Union[str, Any] , UpperCamelCase__ :int ):
_a = NezhaForNextSentencePrediction(config=UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = model(
UpperCamelCase__ , attention_mask=UpperCamelCase__ , token_type_ids=UpperCamelCase__ , labels=UpperCamelCase__ , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, 2) )
def SCREAMING_SNAKE_CASE_ ( self :Optional[int] , UpperCamelCase__ :Union[str, Any] , UpperCamelCase__ :List[str] , UpperCamelCase__ :Any , UpperCamelCase__ :Optional[int] , UpperCamelCase__ :List[Any] , UpperCamelCase__ :Union[str, Any] , UpperCamelCase__ :int ):
_a = NezhaForPreTraining(config=UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = model(
UpperCamelCase__ , attention_mask=UpperCamelCase__ , token_type_ids=UpperCamelCase__ , labels=UpperCamelCase__ , next_sentence_label=UpperCamelCase__ , )
self.parent.assertEqual(result.prediction_logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
self.parent.assertEqual(result.seq_relationship_logits.shape , (self.batch_size, 2) )
def SCREAMING_SNAKE_CASE_ ( self :Dict , UpperCamelCase__ :str , UpperCamelCase__ :Optional[int] , UpperCamelCase__ :Tuple , UpperCamelCase__ :Any , UpperCamelCase__ :Any , UpperCamelCase__ :List[Any] , UpperCamelCase__ :Tuple ):
_a = NezhaForQuestionAnswering(config=UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = 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 SCREAMING_SNAKE_CASE_ ( self :Optional[Any] , UpperCamelCase__ :int , UpperCamelCase__ :Dict , UpperCamelCase__ :Optional[Any] , UpperCamelCase__ :Optional[int] , UpperCamelCase__ :Tuple , UpperCamelCase__ :Optional[Any] , UpperCamelCase__ :Optional[int] ):
_a = self.num_labels
_a = NezhaForSequenceClassification(UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = model(UpperCamelCase__ , attention_mask=UpperCamelCase__ , token_type_ids=UpperCamelCase__ , labels=UpperCamelCase__ )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def SCREAMING_SNAKE_CASE_ ( self :Optional[int] , UpperCamelCase__ :Dict , UpperCamelCase__ :Any , UpperCamelCase__ :str , UpperCamelCase__ :List[Any] , UpperCamelCase__ :Dict , UpperCamelCase__ :Tuple , UpperCamelCase__ :Tuple ):
_a = self.num_labels
_a = NezhaForTokenClassification(config=UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = 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 SCREAMING_SNAKE_CASE_ ( self :List[str] , UpperCamelCase__ :str , UpperCamelCase__ :Optional[int] , UpperCamelCase__ :Union[str, Any] , UpperCamelCase__ :int , UpperCamelCase__ :Optional[int] , UpperCamelCase__ :Any , UpperCamelCase__ :int ):
_a = self.num_choices
_a = NezhaForMultipleChoice(config=UpperCamelCase__ )
model.to(UpperCamelCase__ )
model.eval()
_a = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
_a = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
_a = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
_a = model(
UpperCamelCase__ , attention_mask=UpperCamelCase__ , token_type_ids=UpperCamelCase__ , labels=UpperCamelCase__ , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def SCREAMING_SNAKE_CASE_ ( self :Union[str, Any] ):
_a = self.prepare_config_and_inputs()
(
(
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) ,
) = config_and_inputs
_a = {"input_ids": input_ids, "token_type_ids": token_type_ids, "attention_mask": input_mask}
return config, inputs_dict
@require_torch
class __snake_case ( _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , unittest.TestCase ):
"""simple docstring"""
lowerCAmelCase_ : Union[str, Any] = (
(
NezhaModel,
NezhaForMaskedLM,
NezhaForMultipleChoice,
NezhaForNextSentencePrediction,
NezhaForPreTraining,
NezhaForQuestionAnswering,
NezhaForSequenceClassification,
NezhaForTokenClassification,
)
if is_torch_available()
else ()
)
lowerCAmelCase_ : int = (
{
'feature-extraction': NezhaModel,
'fill-mask': NezhaForMaskedLM,
'question-answering': NezhaForQuestionAnswering,
'text-classification': NezhaForSequenceClassification,
'token-classification': NezhaForTokenClassification,
'zero-shot': NezhaForSequenceClassification,
}
if is_torch_available()
else {}
)
lowerCAmelCase_ : Optional[int] = True
def SCREAMING_SNAKE_CASE_ ( self :Dict , UpperCamelCase__ :Any , UpperCamelCase__ :Optional[Any] , UpperCamelCase__ :str=False ):
_a = super()._prepare_for_class(UpperCamelCase__ , UpperCamelCase__ , return_labels=UpperCamelCase__ )
if return_labels:
if model_class in get_values(UpperCamelCase__ ):
_a = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) , dtype=torch.long , device=UpperCamelCase__ )
_a = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=UpperCamelCase__ )
return inputs_dict
def SCREAMING_SNAKE_CASE_ ( self :Union[str, Any] ):
_a = NezhaModelTester(self )
_a = ConfigTester(self , config_class=UpperCamelCase__ , hidden_size=37 )
def SCREAMING_SNAKE_CASE_ ( self :str ):
self.config_tester.run_common_tests()
def SCREAMING_SNAKE_CASE_ ( self :Any ):
_a = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*UpperCamelCase__ )
def SCREAMING_SNAKE_CASE_ ( self :Union[str, Any] ):
_a = self.model_tester.prepare_config_and_inputs_for_decoder()
self.model_tester.create_and_check_model_as_decoder(*UpperCamelCase__ )
def SCREAMING_SNAKE_CASE_ ( self :Any ):
# This regression test was failing with PyTorch < 1.3
(
(
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) , (
_a
) ,
) = self.model_tester.prepare_config_and_inputs_for_decoder()
_a = None
self.model_tester.create_and_check_model_as_decoder(
UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ , )
def SCREAMING_SNAKE_CASE_ ( self :List[str] ):
_a = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*UpperCamelCase__ )
def SCREAMING_SNAKE_CASE_ ( self :Dict ):
_a = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*UpperCamelCase__ )
def SCREAMING_SNAKE_CASE_ ( self :Tuple ):
_a = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_next_sequence_prediction(*UpperCamelCase__ )
def SCREAMING_SNAKE_CASE_ ( self :List[str] ):
_a = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_pretraining(*UpperCamelCase__ )
def SCREAMING_SNAKE_CASE_ ( self :Dict ):
_a = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*UpperCamelCase__ )
def SCREAMING_SNAKE_CASE_ ( self :Optional[int] ):
_a = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*UpperCamelCase__ )
def SCREAMING_SNAKE_CASE_ ( self :List[str] ):
_a = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_token_classification(*UpperCamelCase__ )
@slow
def SCREAMING_SNAKE_CASE_ ( self :Optional[Any] ):
for model_name in NEZHA_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
_a = NezhaModel.from_pretrained(UpperCamelCase__ )
self.assertIsNotNone(UpperCamelCase__ )
@slow
@require_torch_gpu
def SCREAMING_SNAKE_CASE_ ( self :List[Any] ):
_a , _a = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
# NezhaForMultipleChoice behaves incorrectly in JIT environments.
if model_class == NezhaForMultipleChoice:
return
_a = True
_a = model_class(config=UpperCamelCase__ )
_a = self._prepare_for_class(UpperCamelCase__ , UpperCamelCase__ )
_a = torch.jit.trace(
UpperCamelCase__ , (inputs_dict["input_ids"].to("cpu" ), inputs_dict["attention_mask"].to("cpu" )) )
with tempfile.TemporaryDirectory() as tmp:
torch.jit.save(UpperCamelCase__ , os.path.join(UpperCamelCase__ , "bert.pt" ) )
_a = torch.jit.load(os.path.join(UpperCamelCase__ , "bert.pt" ) , map_location=UpperCamelCase__ )
loaded(inputs_dict["input_ids"].to(UpperCamelCase__ ) , inputs_dict["attention_mask"].to(UpperCamelCase__ ) )
@require_torch
class __snake_case ( unittest.TestCase ):
"""simple docstring"""
@slow
def SCREAMING_SNAKE_CASE_ ( self :Tuple ):
_a = NezhaModel.from_pretrained("sijunhe/nezha-cn-base" )
_a = torch.tensor([[0, 1, 2, 3, 4, 5]] )
_a = torch.tensor([[0, 1, 1, 1, 1, 1]] )
with torch.no_grad():
_a = model(UpperCamelCase__ , attention_mask=UpperCamelCase__ )[0]
_a = torch.Size((1, 6, 768) )
self.assertEqual(output.shape , UpperCamelCase__ )
_a = torch.tensor([[[0.0685, 0.2441, 0.1102], [0.0600, 0.1906, 0.1349], [0.0221, 0.0819, 0.0586]]] )
self.assertTrue(torch.allclose(output[:, 1:4, 1:4] , UpperCamelCase__ , atol=1E-4 ) )
@slow
def SCREAMING_SNAKE_CASE_ ( self :Optional[int] ):
_a = NezhaForMaskedLM.from_pretrained("sijunhe/nezha-cn-base" )
_a = torch.tensor([[0, 1, 2, 3, 4, 5]] )
_a = torch.tensor([[1, 1, 1, 1, 1, 1]] )
with torch.no_grad():
_a = model(UpperCamelCase__ , attention_mask=UpperCamelCase__ )[0]
_a = torch.Size((1, 6, 21_128) )
self.assertEqual(output.shape , UpperCamelCase__ )
_a = torch.tensor(
[[-2.7939, -1.7902, -2.2189], [-2.8585, -1.8908, -2.3723], [-2.6499, -1.7750, -2.2558]] )
self.assertTrue(torch.allclose(output[:, 1:4, 1:4] , UpperCamelCase__ , atol=1E-4 ) )
| 388
|
"""simple docstring"""
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
__SCREAMING_SNAKE_CASE = {
"""configuration_megatron_bert""": ["""MEGATRON_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP""", """MegatronBertConfig"""],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
__SCREAMING_SNAKE_CASE = [
"""MEGATRON_BERT_PRETRAINED_MODEL_ARCHIVE_LIST""",
"""MegatronBertForCausalLM""",
"""MegatronBertForMaskedLM""",
"""MegatronBertForMultipleChoice""",
"""MegatronBertForNextSentencePrediction""",
"""MegatronBertForPreTraining""",
"""MegatronBertForQuestionAnswering""",
"""MegatronBertForSequenceClassification""",
"""MegatronBertForTokenClassification""",
"""MegatronBertModel""",
"""MegatronBertPreTrainedModel""",
]
if TYPE_CHECKING:
from .configuration_megatron_bert import MEGATRON_BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, MegatronBertConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_megatron_bert import (
MEGATRON_BERT_PRETRAINED_MODEL_ARCHIVE_LIST,
MegatronBertForCausalLM,
MegatronBertForMaskedLM,
MegatronBertForMultipleChoice,
MegatronBertForNextSentencePrediction,
MegatronBertForPreTraining,
MegatronBertForQuestionAnswering,
MegatronBertForSequenceClassification,
MegatronBertForTokenClassification,
MegatronBertModel,
MegatronBertPreTrainedModel,
)
else:
import sys
__SCREAMING_SNAKE_CASE = _LazyModule(__name__, globals()["""__file__"""], _import_structure, module_spec=__spec__)
| 388
| 1
|
'''simple docstring'''
from ....configuration_utils import PretrainedConfig
from ....utils import logging
A: int = logging.get_logger(__name__)
A: Optional[int] = {
"CarlCochet/trajectory-transformer-halfcheetah-medium-v2": (
"https://huggingface.co/CarlCochet/trajectory-transformer-halfcheetah-medium-v2/resolve/main/config.json"
),
# See all TrajectoryTransformer models at https://huggingface.co/models?filter=trajectory_transformer
}
class __magic_name__ ( UpperCAmelCase_ ):
"""simple docstring"""
SCREAMING_SNAKE_CASE_ : Optional[int] = 'trajectory_transformer'
SCREAMING_SNAKE_CASE_ : int = ['past_key_values']
SCREAMING_SNAKE_CASE_ : Optional[Any] = {
'hidden_size': 'n_embd',
'num_attention_heads': 'n_head',
'num_hidden_layers': 'n_layer',
}
def __init__( self , _lowercase=100 , _lowercase=5 , _lowercase=1 , _lowercase=1 , _lowercase=249 , _lowercase=6 , _lowercase=17 , _lowercase=25 , _lowercase=4 , _lowercase=4 , _lowercase=128 , _lowercase=0.1 , _lowercase=0.1 , _lowercase=0.1 , _lowercase=0.00_06 , _lowercase=512 , _lowercase=0.02 , _lowercase=1E-1_2 , _lowercase=1 , _lowercase=True , _lowercase=1 , _lowercase=5_0256 , _lowercase=5_0256 , **_lowercase , ) -> Tuple:
lowercase_ : Optional[Any] = vocab_size
lowercase_ : Union[str, Any] = action_weight
lowercase_ : Any = reward_weight
lowercase_ : str = value_weight
lowercase_ : List[str] = max_position_embeddings
lowercase_ : Dict = block_size
lowercase_ : Union[str, Any] = action_dim
lowercase_ : Tuple = observation_dim
lowercase_ : Any = transition_dim
lowercase_ : Optional[int] = learning_rate
lowercase_ : Optional[int] = n_layer
lowercase_ : Tuple = n_head
lowercase_ : int = n_embd
lowercase_ : List[str] = embd_pdrop
lowercase_ : Optional[Any] = attn_pdrop
lowercase_ : Tuple = resid_pdrop
lowercase_ : List[str] = initializer_range
lowercase_ : List[str] = layer_norm_eps
lowercase_ : int = kaiming_initializer_range
lowercase_ : int = use_cache
super().__init__(pad_token_id=_lowercase , bos_token_id=_lowercase , eos_token_id=_lowercase , **_lowercase )
| 7
|
'''simple docstring'''
def _UpperCAmelCase ( a : list[list[float]] ) -> list[list[float]]:
"""simple docstring"""
lowercase_ : list[list[float]] = []
for data in source_data:
for i, el in enumerate(a ):
if len(a ) < i + 1:
data_lists.append([] )
data_lists[i].append(float(a ) )
return data_lists
def _UpperCAmelCase ( a : list[list[float]] , a : list[int] ) -> list[list[float]]:
"""simple docstring"""
lowercase_ : list[list[float]] = []
for dlist, weight in zip(a , a ):
lowercase_ : Tuple = min(a )
lowercase_ : Any = max(a )
lowercase_ : list[float] = []
# for weight 0 score is 1 - actual score
if weight == 0:
for item in dlist:
try:
score.append(1 - ((item - mind) / (maxd - mind)) )
except ZeroDivisionError:
score.append(1 )
elif weight == 1:
for item in dlist:
try:
score.append((item - mind) / (maxd - mind) )
except ZeroDivisionError:
score.append(0 )
# weight not 0 or 1
else:
lowercase_ : str = f"Invalid weight of {weight:f} provided"
raise ValueError(a )
score_lists.append(a )
return score_lists
def _UpperCAmelCase ( a : list[list[float]] ) -> list[float]:
"""simple docstring"""
lowercase_ : list[float] = [0 for i in range(len(score_lists[0] ) )]
for slist in score_lists:
for j, ele in enumerate(a ):
lowercase_ : List[Any] = final_scores[j] + ele
return final_scores
def _UpperCAmelCase ( a : list[list[float]] , a : list[int] ) -> list[list[float]]:
"""simple docstring"""
lowercase_ : int = get_data(a )
lowercase_ : Optional[int] = calculate_each_score(a , a )
lowercase_ : Dict = generate_final_scores(a )
# append scores to source data
for i, ele in enumerate(a ):
source_data[i].append(a )
return source_data
| 7
| 1
|
'''simple docstring'''
A = frozenset(
[
'prompt',
'height',
'width',
'guidance_scale',
'negative_prompt',
'prompt_embeds',
'negative_prompt_embeds',
'cross_attention_kwargs',
]
)
A = frozenset(['prompt', 'negative_prompt'])
A = frozenset([])
A = frozenset(['image'])
A = frozenset(
[
'image',
'height',
'width',
'guidance_scale',
]
)
A = frozenset(['image'])
A = frozenset(
[
'prompt',
'image',
'height',
'width',
'guidance_scale',
'negative_prompt',
'prompt_embeds',
'negative_prompt_embeds',
]
)
A = frozenset(['prompt', 'image', 'negative_prompt'])
A = frozenset(
[
# Text guided image variation with an image mask
'prompt',
'image',
'mask_image',
'height',
'width',
'guidance_scale',
'negative_prompt',
'prompt_embeds',
'negative_prompt_embeds',
]
)
A = frozenset(['prompt', 'image', 'mask_image', 'negative_prompt'])
A = frozenset(
[
# image variation with an image mask
'image',
'mask_image',
'height',
'width',
'guidance_scale',
]
)
A = frozenset(['image', 'mask_image'])
A = frozenset(
[
'example_image',
'image',
'mask_image',
'height',
'width',
'guidance_scale',
]
)
A = frozenset(['example_image', 'image', 'mask_image'])
A = frozenset(['class_labels'])
A = frozenset(['class_labels'])
A = frozenset(['batch_size'])
A = frozenset([])
A = frozenset(['batch_size'])
A = frozenset([])
A = frozenset(
[
'prompt',
'audio_length_in_s',
'guidance_scale',
'negative_prompt',
'prompt_embeds',
'negative_prompt_embeds',
'cross_attention_kwargs',
]
)
A = frozenset(['prompt', 'negative_prompt'])
A = frozenset(['input_tokens'])
A = frozenset(['input_tokens'])
| 320
|
'''simple docstring'''
from __future__ import annotations
from typing import Any
class __snake_case :
def __init__( self, A, A, A = 0 ):
"""simple docstring"""
lowerCamelCase , lowerCamelCase : str = row, column
lowerCamelCase : List[str] = [[default_value for c in range(A )] for r in range(A )]
def __str__( self ):
"""simple docstring"""
lowerCamelCase : Any = F'''Matrix consist of {self.row} rows and {self.column} columns\n'''
# Make string identifier
lowerCamelCase : List[Any] = 0
for row_vector in self.array:
for obj in row_vector:
lowerCamelCase : List[Any] = max(A, len(str(A ) ) )
lowerCamelCase : int = F'''%{max_element_length}s'''
# Make string and return
def single_line(A ) -> str:
nonlocal string_format_identifier
lowerCamelCase : List[str] = '['
line += ", ".join(string_format_identifier % (obj,) for obj in row_vector )
line += "]"
return line
s += "\n".join(single_line(A ) for row_vector in self.array )
return s
def __repr__( self ):
"""simple docstring"""
return str(self )
def UpperCAmelCase_ ( self, A ):
"""simple docstring"""
if not (isinstance(A, (list, tuple) ) and len(A ) == 2):
return False
elif not (0 <= loc[0] < self.row and 0 <= loc[1] < self.column):
return False
else:
return True
def __getitem__( self, A ):
"""simple docstring"""
assert self.validate_indicies(A )
return self.array[loc[0]][loc[1]]
def __setitem__( self, A, A ):
"""simple docstring"""
assert self.validate_indicies(A )
lowerCamelCase : List[Any] = value
def __add__( self, A ):
"""simple docstring"""
assert isinstance(A, A )
assert self.row == another.row and self.column == another.column
# Add
lowerCamelCase : Optional[int] = Matrix(self.row, self.column )
for r in range(self.row ):
for c in range(self.column ):
lowerCamelCase : Any = self[r, c] + another[r, c]
return result
def __neg__( self ):
"""simple docstring"""
lowerCamelCase : List[str] = Matrix(self.row, self.column )
for r in range(self.row ):
for c in range(self.column ):
lowerCamelCase : Union[str, Any] = -self[r, c]
return result
def __sub__( self, A ):
"""simple docstring"""
return self + (-another)
def __mul__( self, A ):
"""simple docstring"""
if isinstance(A, (int, float) ): # Scalar multiplication
lowerCamelCase : List[Any] = Matrix(self.row, self.column )
for r in range(self.row ):
for c in range(self.column ):
lowerCamelCase : Optional[Any] = self[r, c] * another
return result
elif isinstance(A, A ): # Matrix multiplication
assert self.column == another.row
lowerCamelCase : List[Any] = Matrix(self.row, another.column )
for r in range(self.row ):
for c in range(another.column ):
for i in range(self.column ):
result[r, c] += self[r, i] * another[i, c]
return result
else:
lowerCamelCase : Tuple = F'''Unsupported type given for another ({type(A )})'''
raise TypeError(A )
def UpperCAmelCase_ ( self ):
"""simple docstring"""
lowerCamelCase : Optional[int] = Matrix(self.column, self.row )
for r in range(self.row ):
for c in range(self.column ):
lowerCamelCase : str = self[r, c]
return result
def UpperCAmelCase_ ( self, A, A ):
"""simple docstring"""
assert isinstance(A, A ) and isinstance(A, A )
assert self.row == self.column == u.row == v.row # u, v should be column vector
assert u.column == v.column == 1 # u, v should be column vector
# Calculate
lowerCamelCase : Union[str, Any] = v.transpose()
lowerCamelCase : Optional[int] = (v_t * self * u)[0, 0] + 1
if numerator_factor == 0:
return None # It's not invertable
return self - ((self * u) * (v_t * self) * (1.0 / numerator_factor))
# Testing
if __name__ == "__main__":
def UpperCAmelCase ( ):
# a^(-1)
lowerCamelCase : Optional[Any] = Matrix(3 , 3 , 0)
for i in range(3):
lowerCamelCase : Union[str, Any] = 1
print(F'''a^(-1) is {ainv}''')
# u, v
lowerCamelCase : Dict = Matrix(3 , 1 , 0)
lowerCamelCase , lowerCamelCase , lowerCamelCase : Optional[int] = 1, 2, -3
lowerCamelCase : Tuple = Matrix(3 , 1 , 0)
lowerCamelCase , lowerCamelCase , lowerCamelCase : Union[str, Any] = 4, -2, 5
print(F'''u is {u}''')
print(F'''v is {v}''')
print(F'''uv^T is {u * v.transpose()}''')
# Sherman Morrison
print(F'''(a + uv^T)^(-1) is {ainv.sherman_morrison(UpperCAmelCase__ , UpperCAmelCase__)}''')
def UpperCAmelCase ( ):
import doctest
doctest.testmod()
testa()
| 320
| 1
|
'''simple docstring'''
import json
import os
import unittest
from transformers import OpenAIGPTTokenizer, OpenAIGPTTokenizerFast
from transformers.models.openai.tokenization_openai import VOCAB_FILES_NAMES
from transformers.testing_utils import require_ftfy, require_spacy, require_tokenizers
from ...test_tokenization_common import TokenizerTesterMixin
@require_tokenizers
class UpperCamelCase__ ( a , unittest.TestCase ):
'''simple docstring'''
_snake_case = OpenAIGPTTokenizer
_snake_case = OpenAIGPTTokenizerFast
_snake_case = True
_snake_case = False
def snake_case ( self ) -> Optional[Any]:
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
__lowerCAmelCase : Union[str, Any] = [
'l',
'o',
'w',
'e',
'r',
's',
't',
'i',
'd',
'n',
'w</w>',
'r</w>',
't</w>',
'lo',
'low',
'er</w>',
'low</w>',
'lowest</w>',
'newer</w>',
'wider</w>',
'<unk>',
]
__lowerCAmelCase : Tuple = dict(zip(SCREAMING_SNAKE_CASE , range(len(SCREAMING_SNAKE_CASE ) ) ) )
__lowerCAmelCase : Union[str, Any] = ['#version: 0.2', 'l o', 'lo w', 'e r</w>', '']
__lowerCAmelCase : int = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['vocab_file'] )
__lowerCAmelCase : List[str] = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['merges_file'] )
with open(self.vocab_file , 'w' ) as fp:
fp.write(json.dumps(SCREAMING_SNAKE_CASE ) )
with open(self.merges_file , 'w' ) as fp:
fp.write('\n'.join(SCREAMING_SNAKE_CASE ) )
def snake_case ( self , SCREAMING_SNAKE_CASE ) -> Any:
return "lower newer", "lower newer"
def snake_case ( self ) -> List[str]:
__lowerCAmelCase : List[str] = OpenAIGPTTokenizer(self.vocab_file , self.merges_file )
__lowerCAmelCase : List[str] = 'lower'
__lowerCAmelCase : Union[str, Any] = ['low', 'er</w>']
__lowerCAmelCase : List[Any] = tokenizer.tokenize(SCREAMING_SNAKE_CASE )
self.assertListEqual(SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE )
__lowerCAmelCase : Tuple = tokens + ['<unk>']
__lowerCAmelCase : int = [14, 15, 20]
self.assertListEqual(tokenizer.convert_tokens_to_ids(SCREAMING_SNAKE_CASE ) , SCREAMING_SNAKE_CASE )
def snake_case ( self , SCREAMING_SNAKE_CASE=15 ) -> int:
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(F"""{tokenizer.__class__.__name__} ({pretrained_name})""" ):
__lowerCAmelCase : Dict = self.rust_tokenizer_class.from_pretrained(SCREAMING_SNAKE_CASE , **SCREAMING_SNAKE_CASE )
# Simple input
__lowerCAmelCase : Optional[Any] = 'This is a simple input'
__lowerCAmelCase : Union[str, Any] = ['This is a simple input 1', 'This is a simple input 2']
__lowerCAmelCase : int = ('This is a simple input', 'This is a pair')
__lowerCAmelCase : Optional[Any] = [
('This is a simple input 1', 'This is a simple input 2'),
('This is a simple pair 1', 'This is a simple pair 2'),
]
# Simple input tests
self.assertRaises(SCREAMING_SNAKE_CASE , tokenizer_r.encode , SCREAMING_SNAKE_CASE , max_length=SCREAMING_SNAKE_CASE , padding='max_length' )
# Simple input
self.assertRaises(SCREAMING_SNAKE_CASE , tokenizer_r.encode_plus , SCREAMING_SNAKE_CASE , max_length=SCREAMING_SNAKE_CASE , padding='max_length' )
# Simple input
self.assertRaises(
SCREAMING_SNAKE_CASE , tokenizer_r.batch_encode_plus , SCREAMING_SNAKE_CASE , max_length=SCREAMING_SNAKE_CASE , padding='max_length' , )
# Pair input
self.assertRaises(SCREAMING_SNAKE_CASE , tokenizer_r.encode , SCREAMING_SNAKE_CASE , max_length=SCREAMING_SNAKE_CASE , padding='max_length' )
# Pair input
self.assertRaises(SCREAMING_SNAKE_CASE , tokenizer_r.encode_plus , SCREAMING_SNAKE_CASE , max_length=SCREAMING_SNAKE_CASE , padding='max_length' )
# Pair input
self.assertRaises(
SCREAMING_SNAKE_CASE , tokenizer_r.batch_encode_plus , SCREAMING_SNAKE_CASE , max_length=SCREAMING_SNAKE_CASE , padding='max_length' , )
def snake_case ( self ) -> int:
pass
@require_ftfy
@require_spacy
@require_tokenizers
class UpperCamelCase__ ( a ):
'''simple docstring'''
pass
| 123
|
'''simple docstring'''
from math import factorial
A_ = {str(digit): factorial(digit) for digit in range(10)}
def A ( _UpperCAmelCase : int ) -> int:
'''simple docstring'''
if not isinstance(_UpperCAmelCase ,_UpperCAmelCase ):
raise TypeError('Parameter number must be int' )
if number < 0:
raise ValueError('Parameter number must be greater than or equal to 0' )
# Converts number in string to iterate on its digits and adds its factorial.
return sum(DIGIT_FACTORIAL[digit] for digit in str(_UpperCAmelCase ) )
def A ( _UpperCAmelCase : int = 6_0 ,_UpperCAmelCase : int = 1_0_0_0_0_0_0 ) -> int:
'''simple docstring'''
if not isinstance(_UpperCAmelCase ,_UpperCAmelCase ) or not isinstance(_UpperCAmelCase ,_UpperCAmelCase ):
raise TypeError('Parameters chain_length and number_limit must be int' )
if chain_length <= 0 or number_limit <= 0:
raise ValueError(
'Parameters chain_length and number_limit must be greater than 0' )
# the counter for the chains with the exact desired length
__lowerCAmelCase : Any = 0
# the cached sizes of the previous chains
__lowerCAmelCase : dict[int, int] = {}
for start_chain_element in range(1 ,_UpperCAmelCase ):
# The temporary set will contain the elements of the chain
__lowerCAmelCase : Union[str, Any] = set()
__lowerCAmelCase : Union[str, Any] = 0
# Stop computing the chain when you find a cached size, a repeating item or the
# length is greater then the desired one.
__lowerCAmelCase : List[str] = start_chain_element
while (
chain_element not in chain_sets_lengths
and chain_element not in chain_set
and chain_set_length <= chain_length
):
chain_set.add(_UpperCAmelCase )
chain_set_length += 1
__lowerCAmelCase : Optional[Any] = digit_factorial_sum(_UpperCAmelCase )
if chain_element in chain_sets_lengths:
chain_set_length += chain_sets_lengths[chain_element]
__lowerCAmelCase : Any = chain_set_length
# If chain contains the exact amount of elements increase the counter
if chain_set_length == chain_length:
chains_counter += 1
return chains_counter
if __name__ == "__main__":
import doctest
doctest.testmod()
print(F'''{solution()}''')
| 123
| 1
|
'''simple docstring'''
def a_ ( __snake_case : list ) -> list:
"""simple docstring"""
if len(__snake_case ) <= 1:
return [tuple(__snake_case )]
lowerCamelCase_ =[]
def generate(__snake_case : int , __snake_case : list ):
if k == 1:
res.append(tuple(arr[:] ) )
return
generate(k - 1 , __snake_case )
for i in range(k - 1 ):
if k % 2 == 0: # k is even
lowerCamelCase_, lowerCamelCase_ =arr[k - 1], arr[i]
else: # k is odd
lowerCamelCase_, lowerCamelCase_ =arr[k - 1], arr[0]
generate(k - 1 , __snake_case )
generate(len(__snake_case ) , __snake_case )
return res
if __name__ == "__main__":
a_ : Tuple = input("""Enter numbers separated by a comma:\n""").strip()
a_ : List[str] = [int(item) for item in user_input.split(""",""")]
print(heaps(arr))
| 676
|
'''simple docstring'''
def a_ ( __snake_case : str , __snake_case : str ) -> str:
"""simple docstring"""
lowerCamelCase_ =len(__snake_case )
lowerCamelCase_ =len(__snake_case )
lowerCamelCase_ =(
first_str_length if first_str_length > second_str_length else second_str_length
)
lowerCamelCase_ =[]
for char_count in range(__snake_case ):
if char_count < first_str_length:
output_list.append(first_str[char_count] )
if char_count < second_str_length:
output_list.append(second_str[char_count] )
return "".join(__snake_case )
if __name__ == "__main__":
print(alternative_string_arrange("""AB""", """XYZ"""), end=""" """)
| 676
| 1
|
def A ( lowercase__ : Dict ) -> int:
UpperCamelCase__ :List[Any] = [0] * len(lowercase__ )
UpperCamelCase__ :Any = []
UpperCamelCase__ :List[Any] = [1] * len(lowercase__ )
for values in graph.values():
for i in values:
indegree[i] += 1
for i in range(len(lowercase__ ) ):
if indegree[i] == 0:
queue.append(lowercase__ )
while queue:
UpperCamelCase__ :Union[str, Any] = queue.pop(0 )
for x in graph[vertex]:
indegree[x] -= 1
if long_dist[vertex] + 1 > long_dist[x]:
UpperCamelCase__ :Dict = long_dist[vertex] + 1
if indegree[x] == 0:
queue.append(lowercase__ )
print(max(lowercase__ ) )
# Adjacency list of Graph
UpperCamelCase = {0: [2, 3, 4], 1: [2, 7], 2: [5], 3: [5, 7], 4: [7], 5: [6], 6: [7], 7: []}
longest_distance(graph)
| 704
|
import numpy as np
class lowerCAmelCase_ :
"""simple docstring"""
def __init__( self :List[str] ):
UpperCamelCase__ :List[str] = (0, 0)
UpperCamelCase__ :Dict = None
UpperCamelCase__ :List[str] = 0
UpperCamelCase__ :Any = 0
UpperCamelCase__ :Optional[int] = 0
def __eq__( self :Optional[int] , lowerCamelCase__ :Dict ):
return self.position == cell.position
def __a ( self :str ):
print(self.position )
class lowerCAmelCase_ :
"""simple docstring"""
def __init__( self :Tuple , lowerCamelCase__ :List[Any]=(5, 5) ):
UpperCamelCase__ :Optional[int] = np.zeros(lowerCamelCase__ )
UpperCamelCase__ :List[Any] = world_size[0]
UpperCamelCase__ :Optional[Any] = world_size[1]
def __a ( self :Optional[int] ):
print(self.w )
def __a ( self :Dict , lowerCamelCase__ :Optional[Any] ):
UpperCamelCase__ :Tuple = [
(-1, -1),
(-1, 0),
(-1, 1),
(0, -1),
(0, 1),
(1, -1),
(1, 0),
(1, 1),
]
UpperCamelCase__ :List[str] = cell.position[0]
UpperCamelCase__ :List[Any] = cell.position[1]
UpperCamelCase__ :Tuple = []
for n in neughbour_cord:
UpperCamelCase__ :int = current_x + n[0]
UpperCamelCase__ :List[Any] = current_y + n[1]
if 0 <= x < self.world_x_limit and 0 <= y < self.world_y_limit:
UpperCamelCase__ :List[Any] = Cell()
UpperCamelCase__ :Optional[int] = (x, y)
UpperCamelCase__ :str = cell
neighbours.append(lowerCamelCase__ )
return neighbours
def A ( lowercase__ : Optional[int] , lowercase__ : str , lowercase__ : Optional[int] ) -> Optional[Any]:
UpperCamelCase__ :Tuple = []
UpperCamelCase__ :int = []
_open.append(lowercase__ )
while _open:
UpperCamelCase__ :int = np.argmin([n.f for n in _open] )
UpperCamelCase__ :str = _open[min_f]
_closed.append(_open.pop(lowercase__ ) )
if current == goal:
break
for n in world.get_neigbours(lowercase__ ):
for c in _closed:
if c == n:
continue
UpperCamelCase__ :Optional[int] = current.g + 1
UpperCamelCase__ , UpperCamelCase__ :Union[str, Any] = n.position
UpperCamelCase__ , UpperCamelCase__ :int = goal.position
UpperCamelCase__ :List[Any] = (ya - ya) ** 2 + (xa - xa) ** 2
UpperCamelCase__ :Union[str, Any] = n.h + n.g
for c in _open:
if c == n and c.f < n.f:
continue
_open.append(lowercase__ )
UpperCamelCase__ :Optional[int] = []
while current.parent is not None:
path.append(current.position )
UpperCamelCase__ :Optional[int] = current.parent
path.append(current.position )
return path[::-1]
if __name__ == "__main__":
UpperCamelCase = Gridworld()
# Start position and goal
UpperCamelCase = Cell()
UpperCamelCase = (0, 0)
UpperCamelCase = Cell()
UpperCamelCase = (4, 4)
print(f'''path from {start.position} to {goal.position}''')
UpperCamelCase = astar(world, start, goal)
# Just for visual reasons.
for i in s:
UpperCamelCase = 1
print(world.w)
| 383
| 0
|
"""simple docstring"""
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
lowerCamelCase__ = logging.getLogger(__name__)
lowerCamelCase__ = "pytorch_model.bin"
@dataclasses.dataclass
class __SCREAMING_SNAKE_CASE :
'''simple docstring'''
SCREAMING_SNAKE_CASE__ :str = dataclasses.field(
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models."} )
SCREAMING_SNAKE_CASE__ :Optional[str] = dataclasses.field(
default=__A , metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co."} , )
@dataclasses.dataclass
class __SCREAMING_SNAKE_CASE :
'''simple docstring'''
SCREAMING_SNAKE_CASE__ :str = dataclasses.field(metadata={"help": "A csv or a json file containing the training data."} )
SCREAMING_SNAKE_CASE__ :str = dataclasses.field(metadata={"help": "A csv or a json file containing the data to predict on."} )
SCREAMING_SNAKE_CASE__ :Optional[str] = dataclasses.field(
default=__A , metadata={"help": "A csv or a json file containing the validation data."} )
SCREAMING_SNAKE_CASE__ :Optional[str] = dataclasses.field(
default=__A , metadata={"help": "The name of the task to train on."} , )
SCREAMING_SNAKE_CASE__ :Optional[List[str]] = dataclasses.field(
default=__A , metadata={"help": "The list of labels for the task."} )
@dataclasses.dataclass
class __SCREAMING_SNAKE_CASE :
'''simple docstring'''
SCREAMING_SNAKE_CASE__ :str = dataclasses.field(
metadata={"help": "The output directory where the model predictions and checkpoints will be written."} )
SCREAMING_SNAKE_CASE__ :Optional[str] = dataclasses.field(
default="accuracy" , metadata={"help": "The evaluation metric used for the task."} )
SCREAMING_SNAKE_CASE__ :Optional[str] = dataclasses.field(
default="no" , metadata={
"help": "The evaluation strategy to adopt during training. Possible values are: [\"no\", \"step\", \"epoch]"
} , )
SCREAMING_SNAKE_CASE__ :Optional[int] = dataclasses.field(
default=10 , metadata={"help": "Number of evaluation calls with no improvement after which training will be stopped."} , )
SCREAMING_SNAKE_CASE__ :Optional[float] = dataclasses.field(
default=0.0 , metadata={
"help": "How much the specified evaluation metric must improve to satisfy early stopping conditions."
} , )
SCREAMING_SNAKE_CASE__ :Optional[bool] = dataclasses.field(
default=__A , metadata={"help": "Whether to filter the pseudo-labeled data based on the confidence score."} , )
SCREAMING_SNAKE_CASE__ :Optional[bool] = dataclasses.field(
default=__A , metadata={"help": "Whether to filter the pseudo-labeled data based on the validation performance."} , )
SCREAMING_SNAKE_CASE__ :Optional[bool] = dataclasses.field(
default=__A , metadata={"help": "Whether to fine-tune on labeled data after pseudo training."} , )
SCREAMING_SNAKE_CASE__ :Optional[float] = dataclasses.field(
default=0.0 , metadata={"help": "Confidence threshold for pseudo-labeled data filtering."} , )
SCREAMING_SNAKE_CASE__ :Optional[int] = dataclasses.field(
default=100 , metadata={"help": "Number of evaluation calls with no improvement after which training will be stopped."} , )
SCREAMING_SNAKE_CASE__ :Optional[int] = dataclasses.field(
default=__A , metadata={"help": "Random seed for initialization."} , )
def lowercase__ ( lowercase_ ,lowercase_ ,lowercase_ ,lowercase_ ,lowercase_ ,lowercase_ ) -> Optional[int]:
"""simple docstring"""
_UpperCamelCase : Union[str, Any] = datasets.concatenate_datasets([infer_input, infer_output] ,axis=1 )
if args.do_filter_by_confidence:
_UpperCamelCase : Union[str, Any] = dataset.filter(lambda lowercase_ : example["probability"] > args.confidence_threshold )
if args.do_filter_by_val_performance:
assert eval_result >= 0.0 and eval_result <= 1.0
_UpperCamelCase : Tuple = int(eval_result * len(SCREAMING_SNAKE_CASE__ ) )
print(SCREAMING_SNAKE_CASE__ )
_UpperCamelCase : List[str] = dataset.sort("probability" ,reverse=SCREAMING_SNAKE_CASE__ )
_UpperCamelCase : int = dataset.select(range(SCREAMING_SNAKE_CASE__ ) )
_UpperCamelCase : Any = dataset.remove_columns(["label", "probability"] )
_UpperCamelCase : Dict = dataset.rename_column("prediction" ,"label" )
_UpperCamelCase : Dict = dataset.map(lambda lowercase_ : {"label": idalabel[example["label"]]} )
_UpperCamelCase : Any = dataset.shuffle(seed=args.seed )
_UpperCamelCase : Dict = os.path.join(SCREAMING_SNAKE_CASE__ ,F'''train_pseudo.{args.data_file_extension}''' )
if args.data_file_extension == "csv":
dataset.to_csv(SCREAMING_SNAKE_CASE__ ,index=SCREAMING_SNAKE_CASE__ )
else:
dataset.to_json(SCREAMING_SNAKE_CASE__ )
def lowercase__ ( lowercase_ ,lowercase_ ,lowercase_ ,lowercase_ ,**lowercase_ ) -> List[str]:
"""simple docstring"""
_UpperCamelCase : str = 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()
_UpperCamelCase : Dict = STModelArguments(model_name_or_path=SCREAMING_SNAKE_CASE__ )
_UpperCamelCase : List[str] = STDataArguments(train_file=SCREAMING_SNAKE_CASE__ ,infer_file=SCREAMING_SNAKE_CASE__ )
_UpperCamelCase : List[Any] = STTrainingArguments(output_dir=SCREAMING_SNAKE_CASE__ )
_UpperCamelCase : str = argparse.Namespace()
for arg_class in (model_args, data_args, training_args):
for key, value in vars(SCREAMING_SNAKE_CASE__ ).items():
setattr(SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ )
for key, value in kwargs.items():
if hasattr(SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ):
setattr(SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ )
# Sanity checks
_UpperCamelCase : str = {}
_UpperCamelCase : Any = 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
_UpperCamelCase : Union[str, Any] = args.train_file
_UpperCamelCase : Tuple = args.infer_file
if args.evaluation_strategy != IntervalStrategy.NO.value:
assert args.eval_file is not None
_UpperCamelCase : Union[str, Any] = args.eval_file
for key in data_files:
_UpperCamelCase : Optional[int] = 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:
_UpperCamelCase : str = 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..." )
_UpperCamelCase : List[str] = F'''{args.output_dir}/self-train_iter-{{}}'''.format
_UpperCamelCase : str = data_dir_format(0 )
if accelerator.is_main_process:
if args.output_dir is not None:
os.makedirs(args.output_dir ,exist_ok=SCREAMING_SNAKE_CASE__ )
os.makedirs(SCREAMING_SNAKE_CASE__ ,exist_ok=SCREAMING_SNAKE_CASE__ )
accelerator.wait_for_everyone()
_UpperCamelCase : Optional[int] = None
_UpperCamelCase : List[Any] = None
_UpperCamelCase : Union[str, Any] = 0
_UpperCamelCase : Tuple = False
# Show the progress bar
_UpperCamelCase : Optional[Any] = 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 ) ):
_UpperCamelCase : int = data_dir_format(SCREAMING_SNAKE_CASE__ )
assert os.path.exists(SCREAMING_SNAKE_CASE__ )
# Stage 1: initial fine-tuning for iteration = 0 or pseudo-training for
# iteration > 0
_UpperCamelCase : Optional[int] = os.path.join(SCREAMING_SNAKE_CASE__ ,"stage-1" )
_UpperCamelCase : Any = {
"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(SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ):
arguments_dict.update({key: value} )
_UpperCamelCase : Dict = os.path.join(SCREAMING_SNAKE_CASE__ ,"best-checkpoint" ,SCREAMING_SNAKE_CASE__ )
if os.path.exists(SCREAMING_SNAKE_CASE__ ):
logger.info(
"Found existing model checkpoint at %s. Skipping self-training: iteration: %d, stage: 1." ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,)
else:
logger.info("***** Running self-training: iteration: %d, stage: 1 *****" ,SCREAMING_SNAKE_CASE__ )
finetune(**SCREAMING_SNAKE_CASE__ )
accelerator.wait_for_everyone()
assert os.path.exists(SCREAMING_SNAKE_CASE__ )
logger.info("Self-training job completed: iteration: %d, stage: 1." ,SCREAMING_SNAKE_CASE__ )
if iteration > 0 and args.finetune_on_labeled_data:
# Stage 2 (optional): fine-tuning on the original labeled data
_UpperCamelCase : List[Any] = os.path.join(SCREAMING_SNAKE_CASE__ ,"best-checkpoint" )
_UpperCamelCase : List[str] = os.path.join(SCREAMING_SNAKE_CASE__ ,"stage-2" )
# Update arguments_dict
_UpperCamelCase : Optional[int] = model_path
_UpperCamelCase : Optional[Any] = data_files["train"]
_UpperCamelCase : List[Any] = current_output_dir
_UpperCamelCase : List[str] = os.path.join(SCREAMING_SNAKE_CASE__ ,"best-checkpoint" ,SCREAMING_SNAKE_CASE__ )
if os.path.exists(SCREAMING_SNAKE_CASE__ ):
logger.info(
"Found existing model checkpoint at %s. Skipping self-training: iteration: %d, stage: 2." ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,)
else:
logger.info("***** Running self-training: iteration: %d, stage: 2 *****" ,SCREAMING_SNAKE_CASE__ )
finetune(**SCREAMING_SNAKE_CASE__ )
accelerator.wait_for_everyone()
assert os.path.exists(SCREAMING_SNAKE_CASE__ )
logger.info("Self-training job completed: iteration: %d, stage: 2." ,SCREAMING_SNAKE_CASE__ )
_UpperCamelCase : int = iteration
_UpperCamelCase : str = data_dir_format(iteration + 1 )
_UpperCamelCase : Optional[Any] = AutoConfig.from_pretrained(os.path.join(SCREAMING_SNAKE_CASE__ ,"best-checkpoint" ) )
_UpperCamelCase : Any = config.idalabel
_UpperCamelCase : List[Any] = os.path.join(SCREAMING_SNAKE_CASE__ ,"eval_results_best-checkpoint.json" )
_UpperCamelCase : Dict = os.path.join(SCREAMING_SNAKE_CASE__ ,"test_results_best-checkpoint.json" )
assert os.path.exists(SCREAMING_SNAKE_CASE__ )
with open(SCREAMING_SNAKE_CASE__ ,"r" ) as f:
_UpperCamelCase : str = float(json.load(SCREAMING_SNAKE_CASE__ )[args.eval_metric] )
_UpperCamelCase : Any = os.path.join(SCREAMING_SNAKE_CASE__ ,"infer_output_best-checkpoint.csv" )
assert os.path.exists(SCREAMING_SNAKE_CASE__ )
# Loading the dataset from local csv or json files.
_UpperCamelCase : Optional[int] = load_dataset(args.data_file_extension ,data_files={"data": data_files["infer"]} )["data"]
_UpperCamelCase : Any = load_dataset("csv" ,data_files={"data": infer_output_file} )["data"]
if accelerator.is_main_process:
os.makedirs(SCREAMING_SNAKE_CASE__ ,exist_ok=SCREAMING_SNAKE_CASE__ )
shutil.copy(SCREAMING_SNAKE_CASE__ ,os.path.join(SCREAMING_SNAKE_CASE__ ,F'''eval_results_iter-{iteration}.json''' ) )
if os.path.exists(SCREAMING_SNAKE_CASE__ ):
shutil.copy(SCREAMING_SNAKE_CASE__ ,os.path.join(SCREAMING_SNAKE_CASE__ ,F'''test_results_iter-{iteration}.json''' ) )
create_pseudo_labeled_data(SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ )
accelerator.wait_for_everyone()
_UpperCamelCase : Union[str, Any] = os.path.join(SCREAMING_SNAKE_CASE__ ,F'''train_pseudo.{args.data_file_extension}''' )
if args.evaluation_strategy != IntervalStrategy.NO.value:
_UpperCamelCase : str = eval_result
if best_iteration is None:
_UpperCamelCase : Optional[Any] = new_iteration
_UpperCamelCase : Optional[Any] = new_eval_result
else:
if new_eval_result - best_eval_result > args.early_stopping_threshold:
_UpperCamelCase : Union[str, Any] = new_iteration
_UpperCamelCase : Any = new_eval_result
_UpperCamelCase : Any = 0
else:
if new_eval_result == best_eval_result:
_UpperCamelCase : Optional[int] = new_iteration
_UpperCamelCase : Optional[int] = new_eval_result
early_stopping_patience_counter += 1
if early_stopping_patience_counter >= args.early_stopping_patience:
_UpperCamelCase : int = 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" ,SCREAMING_SNAKE_CASE__ )
logger.info("Best evaluation result: %s = %f" ,args.eval_metric ,SCREAMING_SNAKE_CASE__ )
accelerator.wait_for_everyone()
if accelerator.is_main_process:
shutil.copy(
os.path.join(SCREAMING_SNAKE_CASE__ ,F'''eval_results_iter-{iteration}.json''' ) ,os.path.join(SCREAMING_SNAKE_CASE__ ,"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 ,SCREAMING_SNAKE_CASE__ )
accelerator.wait_for_everyone()
if accelerator.is_main_process:
shutil.copy(
os.path.join(SCREAMING_SNAKE_CASE__ ,F'''eval_results_iter-{args.max_selftrain_iterations - 1}.json''' ) ,os.path.join(SCREAMING_SNAKE_CASE__ ,"eval_results_best-iteration.json" ) ,)
| 624
|
from __future__ import annotations
def __SCREAMING_SNAKE_CASE (SCREAMING_SNAKE_CASE__ ):
if not nums:
raise ValueError('''List is empty''' )
return sum(SCREAMING_SNAKE_CASE__ ) / len(SCREAMING_SNAKE_CASE__ )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 39
| 0
|
'''simple docstring'''
import argparse
import torch
from torch import nn
from transformers import MBartConfig, MBartForConditionalGeneration
def SCREAMING_SNAKE_CASE( UpperCamelCase ) -> Any:
UpperCAmelCase_ : List[str] = [
'encoder.version',
'decoder.version',
'model.encoder.version',
'model.decoder.version',
'_float_tensor',
'decoder.output_projection.weight',
]
for k in ignore_keys:
state_dict.pop(UpperCamelCase ,UpperCamelCase )
def SCREAMING_SNAKE_CASE( UpperCamelCase ) -> Tuple:
UpperCAmelCase_ , UpperCAmelCase_ : int = emb.weight.shape
UpperCAmelCase_ : List[str] = nn.Linear(UpperCamelCase ,UpperCamelCase ,bias=UpperCamelCase )
UpperCAmelCase_ : Optional[Any] = emb.weight.data
return lin_layer
def SCREAMING_SNAKE_CASE( UpperCamelCase ,UpperCamelCase="facebook/mbart-large-en-ro" ,UpperCamelCase=False ,UpperCamelCase=False ) -> List[str]:
UpperCAmelCase_ : Tuple = torch.load(UpperCamelCase ,map_location='cpu' )['model']
remove_ignore_keys_(UpperCamelCase )
UpperCAmelCase_ : List[Any] = state_dict['encoder.embed_tokens.weight'].shape[0]
UpperCAmelCase_ : str = MBartConfig.from_pretrained(UpperCamelCase ,vocab_size=UpperCamelCase )
if mbart_aa and finetuned:
UpperCAmelCase_ : str = 'relu'
UpperCAmelCase_ : List[str] = state_dict['decoder.embed_tokens.weight']
UpperCAmelCase_ : Dict = MBartForConditionalGeneration(UpperCamelCase )
model.model.load_state_dict(UpperCamelCase )
if finetuned:
UpperCAmelCase_ : Any = make_linear_from_emb(model.model.shared )
return model
if __name__ == "__main__":
lowerCAmelCase__ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"fairseq_path", type=str, help="bart.large, bart.large.cnn or a path to a model.pt on local filesystem."
)
parser.add_argument("pytorch_dump_folder_path", default=None, type=str, help="Path to the output PyTorch model.")
parser.add_argument(
"--hf_config",
default="facebook/mbart-large-cc25",
type=str,
help="Which huggingface architecture to use: mbart-large",
)
parser.add_argument("--mbart_50", action="store_true", help="whether the model is mMART-50 checkpoint")
parser.add_argument("--finetuned", action="store_true", help="whether the model is a fine-tuned checkpoint")
lowerCAmelCase__ = parser.parse_args()
lowerCAmelCase__ = convert_fairseq_mbart_checkpoint_from_disk(
args.fairseq_path, hf_config_path=args.hf_config, finetuned=args.finetuned, mbart_aa=args.mbart_aa
)
model.save_pretrained(args.pytorch_dump_folder_path)
| 471
|
'''simple docstring'''
from collections import OrderedDict
from typing import Mapping
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...utils import logging
lowerCAmelCase__ = logging.get_logger(__name__)
lowerCAmelCase__ = {
"facebook/xlm-roberta-xl": "https://huggingface.co/facebook/xlm-roberta-xl/resolve/main/config.json",
"facebook/xlm-roberta-xxl": "https://huggingface.co/facebook/xlm-roberta-xxl/resolve/main/config.json",
# See all XLM-RoBERTa-XL models at https://huggingface.co/models?filter=xlm-roberta-xl
}
class lowercase ( a_ ):
_lowerCamelCase : Any= "xlm-roberta-xl"
def __init__( self , _snake_case=25_0880 , _snake_case=2560 , _snake_case=36 , _snake_case=32 , _snake_case=1_0240 , _snake_case="gelu" , _snake_case=0.1 , _snake_case=0.1 , _snake_case=514 , _snake_case=1 , _snake_case=0.02 , _snake_case=1e-05 , _snake_case=1 , _snake_case=0 , _snake_case=2 , _snake_case="absolute" , _snake_case=True , _snake_case=None , **_snake_case , ) -> Optional[Any]:
super().__init__(pad_token_id=_snake_case , bos_token_id=_snake_case , eos_token_id=_snake_case , **_snake_case)
UpperCAmelCase_ : List[str] = vocab_size
UpperCAmelCase_ : int = hidden_size
UpperCAmelCase_ : List[Any] = num_hidden_layers
UpperCAmelCase_ : Optional[int] = num_attention_heads
UpperCAmelCase_ : Union[str, Any] = hidden_act
UpperCAmelCase_ : int = intermediate_size
UpperCAmelCase_ : Tuple = hidden_dropout_prob
UpperCAmelCase_ : List[str] = attention_probs_dropout_prob
UpperCAmelCase_ : str = max_position_embeddings
UpperCAmelCase_ : Union[str, Any] = type_vocab_size
UpperCAmelCase_ : str = initializer_range
UpperCAmelCase_ : List[str] = layer_norm_eps
UpperCAmelCase_ : Dict = position_embedding_type
UpperCAmelCase_ : int = use_cache
UpperCAmelCase_ : List[Any] = classifier_dropout
class lowercase ( a_ ):
@property
def _snake_case ( self) -> Mapping[str, Mapping[int, str]]:
if self.task == "multiple-choice":
UpperCAmelCase_ : Any = {0: 'batch', 1: 'choice', 2: 'sequence'}
else:
UpperCAmelCase_ : Optional[int] = {0: 'batch', 1: 'sequence'}
return OrderedDict(
[
('input_ids', dynamic_axis),
('attention_mask', dynamic_axis),
])
| 471
| 1
|
"""simple docstring"""
from __future__ import annotations
from math import pi, sqrt
def _snake_case ( __snake_case : float , __snake_case : float ):
"""simple docstring"""
if inductance <= 0:
raise ValueError("""Inductance cannot be 0 or negative""" )
elif capacitance <= 0:
raise ValueError("""Capacitance cannot be 0 or negative""" )
else:
return (
"Resonant frequency",
float(1 / (2 * pi * (sqrt(inductance * capacitance ))) ),
)
if __name__ == "__main__":
import doctest
doctest.testmod()
| 88
|
import json
import os
import shutil
import tempfile
import unittest
import numpy as np
import pytest
from transformers import BertTokenizer, BertTokenizerFast
from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES
from transformers.testing_utils import require_vision
from transformers.utils import FEATURE_EXTRACTOR_NAME, is_vision_available
if is_vision_available():
from PIL import Image
from transformers import ChineseCLIPImageProcessor, ChineseCLIPProcessor
@require_vision
class __lowerCAmelCase ( unittest.TestCase ):
def A__ ( self ) -> Tuple:
'''simple docstring'''
_lowercase =tempfile.mkdtemp()
_lowercase =[
'[UNK]',
'[CLS]',
'[SEP]',
'[PAD]',
'[MASK]',
'的',
'价',
'格',
'是',
'15',
'便',
'alex',
'##andra',
'，',
'。',
'-',
't',
'shirt',
]
_lowercase =os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['vocab_file'] )
with open(self.vocab_file , 'w' , encoding='utf-8' ) as vocab_writer:
vocab_writer.write(''.join([x + '\n' for x in vocab_tokens] ) )
_lowercase ={
'do_resize': True,
'size': {'height': 224, 'width': 224},
'do_center_crop': True,
'crop_size': {'height': 18, 'width': 18},
'do_normalize': True,
'image_mean': [0.48145466, 0.4578275, 0.40821073],
'image_std': [0.26862954, 0.26130258, 0.27577711],
'do_convert_rgb': True,
}
_lowercase =os.path.join(self.tmpdirname , lowerCAmelCase )
with open(self.image_processor_file , 'w' , encoding='utf-8' ) as fp:
json.dump(lowerCAmelCase , lowerCAmelCase )
def A__ ( self , **lowerCAmelCase ) -> Optional[int]:
'''simple docstring'''
return BertTokenizer.from_pretrained(self.tmpdirname , **lowerCAmelCase )
def A__ ( self , **lowerCAmelCase ) -> Optional[Any]:
'''simple docstring'''
return BertTokenizerFast.from_pretrained(self.tmpdirname , **lowerCAmelCase )
def A__ ( self , **lowerCAmelCase ) -> Any:
'''simple docstring'''
return ChineseCLIPImageProcessor.from_pretrained(self.tmpdirname , **lowerCAmelCase )
def A__ ( self ) -> List[str]:
'''simple docstring'''
shutil.rmtree(self.tmpdirname )
def A__ ( self ) -> Tuple:
'''simple docstring'''
_lowercase =[np.random.randint(255 , size=(3, 30, 400) , dtype=np.uinta )]
_lowercase =[Image.fromarray(np.moveaxis(lowerCAmelCase , 0 , -1 ) ) for x in image_inputs]
return image_inputs
def A__ ( self ) -> str:
'''simple docstring'''
_lowercase =self.get_tokenizer()
_lowercase =self.get_rust_tokenizer()
_lowercase =self.get_image_processor()
_lowercase =ChineseCLIPProcessor(tokenizer=lowerCAmelCase , image_processor=lowerCAmelCase )
processor_slow.save_pretrained(self.tmpdirname )
_lowercase =ChineseCLIPProcessor.from_pretrained(self.tmpdirname , use_fast=lowerCAmelCase )
_lowercase =ChineseCLIPProcessor(tokenizer=lowerCAmelCase , image_processor=lowerCAmelCase )
processor_fast.save_pretrained(self.tmpdirname )
_lowercase =ChineseCLIPProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor_slow.tokenizer.get_vocab() , tokenizer_slow.get_vocab() )
self.assertEqual(processor_fast.tokenizer.get_vocab() , tokenizer_fast.get_vocab() )
self.assertEqual(tokenizer_slow.get_vocab() , tokenizer_fast.get_vocab() )
self.assertIsInstance(processor_slow.tokenizer , lowerCAmelCase )
self.assertIsInstance(processor_fast.tokenizer , lowerCAmelCase )
self.assertEqual(processor_slow.image_processor.to_json_string() , image_processor.to_json_string() )
self.assertEqual(processor_fast.image_processor.to_json_string() , image_processor.to_json_string() )
self.assertIsInstance(processor_slow.image_processor , lowerCAmelCase )
self.assertIsInstance(processor_fast.image_processor , lowerCAmelCase )
def A__ ( self ) -> str:
'''simple docstring'''
_lowercase =ChineseCLIPProcessor(tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() )
processor.save_pretrained(self.tmpdirname )
_lowercase =self.get_tokenizer(cls_token='(CLS)' , sep_token='(SEP)' )
_lowercase =self.get_image_processor(do_normalize=lowerCAmelCase )
_lowercase =ChineseCLIPProcessor.from_pretrained(
self.tmpdirname , cls_token='(CLS)' , sep_token='(SEP)' , do_normalize=lowerCAmelCase )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , lowerCAmelCase )
self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.image_processor , lowerCAmelCase )
def A__ ( self ) -> List[str]:
'''simple docstring'''
_lowercase =self.get_image_processor()
_lowercase =self.get_tokenizer()
_lowercase =ChineseCLIPProcessor(tokenizer=lowerCAmelCase , image_processor=lowerCAmelCase )
_lowercase =self.prepare_image_inputs()
_lowercase =image_processor(lowerCAmelCase , return_tensors='np' )
_lowercase =processor(images=lowerCAmelCase , return_tensors='np' )
for key in input_feat_extract.keys():
self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1e-2 )
def A__ ( self ) -> List[str]:
'''simple docstring'''
_lowercase =self.get_image_processor()
_lowercase =self.get_tokenizer()
_lowercase =ChineseCLIPProcessor(tokenizer=lowerCAmelCase , image_processor=lowerCAmelCase )
_lowercase ='Alexandra，T-shirt的价格是15便士。'
_lowercase =processor(text=lowerCAmelCase )
_lowercase =tokenizer(lowerCAmelCase )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def A__ ( self ) -> Dict:
'''simple docstring'''
_lowercase =self.get_image_processor()
_lowercase =self.get_tokenizer()
_lowercase =ChineseCLIPProcessor(tokenizer=lowerCAmelCase , image_processor=lowerCAmelCase )
_lowercase ='Alexandra，T-shirt的价格是15便士。'
_lowercase =self.prepare_image_inputs()
_lowercase =processor(text=lowerCAmelCase , images=lowerCAmelCase )
self.assertListEqual(list(inputs.keys() ) , ['input_ids', 'token_type_ids', 'attention_mask', 'pixel_values'] )
# test if it raises when no input is passed
with pytest.raises(lowerCAmelCase ):
processor()
def A__ ( self ) -> Optional[Any]:
'''simple docstring'''
_lowercase =self.get_image_processor()
_lowercase =self.get_tokenizer()
_lowercase =ChineseCLIPProcessor(tokenizer=lowerCAmelCase , image_processor=lowerCAmelCase )
_lowercase =[[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
_lowercase =processor.batch_decode(lowerCAmelCase )
_lowercase =tokenizer.batch_decode(lowerCAmelCase )
self.assertListEqual(lowerCAmelCase , lowerCAmelCase )
def A__ ( self ) -> str:
'''simple docstring'''
_lowercase =self.get_image_processor()
_lowercase =self.get_tokenizer()
_lowercase =ChineseCLIPProcessor(tokenizer=lowerCAmelCase , image_processor=lowerCAmelCase )
_lowercase ='Alexandra，T-shirt的价格是15便士。'
_lowercase =self.prepare_image_inputs()
_lowercase =processor(text=lowerCAmelCase , images=lowerCAmelCase )
self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
| 291
| 0
|
import warnings
from ...utils import logging
from .image_processing_beit import BeitImageProcessor
A__ = logging.get_logger(__name__)
class _lowerCAmelCase ( __SCREAMING_SNAKE_CASE ):
def __init__( self : List[Any] , *__snake_case : Optional[int] , **__snake_case : Optional[Any] ):
warnings.warn(
'''The class BeitFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please'''
''' use BeitImageProcessor instead.''' , __snake_case , )
super().__init__(*__snake_case , **__snake_case )
| 706
|
import unittest
from queue import Empty
from threading import Thread
from transformers import AutoTokenizer, TextIteratorStreamer, TextStreamer, is_torch_available
from transformers.testing_utils import CaptureStdout, require_torch, torch_device
from ..test_modeling_common import ids_tensor
if is_torch_available():
import torch
from transformers import AutoModelForCausalLM
@require_torch
class _lowerCAmelCase ( unittest.TestCase ):
def snake_case ( self : str ):
lowerCamelCase :int = AutoTokenizer.from_pretrained('''hf-internal-testing/tiny-random-gpt2''' )
lowerCamelCase :List[Any] = AutoModelForCausalLM.from_pretrained('''hf-internal-testing/tiny-random-gpt2''' ).to(__snake_case )
lowerCamelCase :Optional[Any] = -1
lowerCamelCase :List[str] = ids_tensor((1, 5) , vocab_size=model.config.vocab_size ).to(__snake_case )
lowerCamelCase :Tuple = model.generate(__snake_case , max_new_tokens=10 , do_sample=__snake_case )
lowerCamelCase :str = tokenizer.decode(greedy_ids[0] )
with CaptureStdout() as cs:
lowerCamelCase :str = TextStreamer(__snake_case )
model.generate(__snake_case , max_new_tokens=10 , do_sample=__snake_case , streamer=__snake_case )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
lowerCamelCase :Optional[int] = cs.out[:-1]
self.assertEqual(__snake_case , __snake_case )
def snake_case ( self : Dict ):
lowerCamelCase :Tuple = AutoTokenizer.from_pretrained('''hf-internal-testing/tiny-random-gpt2''' )
lowerCamelCase :int = AutoModelForCausalLM.from_pretrained('''hf-internal-testing/tiny-random-gpt2''' ).to(__snake_case )
lowerCamelCase :List[Any] = -1
lowerCamelCase :Dict = ids_tensor((1, 5) , vocab_size=model.config.vocab_size ).to(__snake_case )
lowerCamelCase :Tuple = model.generate(__snake_case , max_new_tokens=10 , do_sample=__snake_case )
lowerCamelCase :List[Any] = tokenizer.decode(greedy_ids[0] )
lowerCamelCase :List[str] = TextIteratorStreamer(__snake_case )
lowerCamelCase :List[str] = {'''input_ids''': input_ids, '''max_new_tokens''': 10, '''do_sample''': False, '''streamer''': streamer}
lowerCamelCase :Tuple = Thread(target=model.generate , kwargs=__snake_case )
thread.start()
lowerCamelCase :Any = ''''''
for new_text in streamer:
streamer_text += new_text
self.assertEqual(__snake_case , __snake_case )
def snake_case ( self : str ):
lowerCamelCase :int = AutoTokenizer.from_pretrained('''hf-internal-testing/tiny-random-gpt2''' )
lowerCamelCase :Dict = AutoModelForCausalLM.from_pretrained('''hf-internal-testing/tiny-random-gpt2''' ).to(__snake_case )
lowerCamelCase :List[str] = -1
lowerCamelCase :Optional[Any] = ids_tensor((1, 5) , vocab_size=model.config.vocab_size ).to(__snake_case )
lowerCamelCase :Optional[Any] = model.generate(__snake_case , max_new_tokens=10 , do_sample=__snake_case )
lowerCamelCase :List[str] = greedy_ids[:, input_ids.shape[1] :]
lowerCamelCase :Union[str, Any] = tokenizer.decode(new_greedy_ids[0] )
with CaptureStdout() as cs:
lowerCamelCase :List[str] = TextStreamer(__snake_case , skip_prompt=__snake_case )
model.generate(__snake_case , max_new_tokens=10 , do_sample=__snake_case , streamer=__snake_case )
# The greedy text should be printed to stdout, except for the final "\n" in the streamer
lowerCamelCase :int = cs.out[:-1]
self.assertEqual(__snake_case , __snake_case )
def snake_case ( self : Optional[int] ):
# Tests that we can pass `decode_kwargs` to the streamer to control how the tokens are decoded. Must be tested
# with actual models -- the dummy models' tokenizers are not aligned with their models, and
# `skip_special_tokens=True` has no effect on them
lowerCamelCase :List[Any] = AutoTokenizer.from_pretrained('''distilgpt2''' )
lowerCamelCase :Union[str, Any] = AutoModelForCausalLM.from_pretrained('''distilgpt2''' ).to(__snake_case )
lowerCamelCase :Optional[int] = -1
lowerCamelCase :Union[str, Any] = torch.ones((1, 5) , device=__snake_case ).long() * model.config.bos_token_id
with CaptureStdout() as cs:
lowerCamelCase :Dict = TextStreamer(__snake_case , skip_special_tokens=__snake_case )
model.generate(__snake_case , max_new_tokens=1 , do_sample=__snake_case , streamer=__snake_case )
# The prompt contains a special token, so the streamer should not print it. As such, the output text, when
# re-tokenized, must only contain one token
lowerCamelCase :Tuple = cs.out[:-1] # Remove the final "\n"
lowerCamelCase :int = tokenizer(__snake_case , return_tensors='''pt''' )
self.assertEqual(streamer_text_tokenized.input_ids.shape , (1, 1) )
def snake_case ( self : List[Any] ):
lowerCamelCase :List[str] = AutoTokenizer.from_pretrained('''hf-internal-testing/tiny-random-gpt2''' )
lowerCamelCase :Optional[int] = AutoModelForCausalLM.from_pretrained('''hf-internal-testing/tiny-random-gpt2''' ).to(__snake_case )
lowerCamelCase :Optional[int] = -1
lowerCamelCase :Tuple = ids_tensor((1, 5) , vocab_size=model.config.vocab_size ).to(__snake_case )
lowerCamelCase :List[Any] = TextIteratorStreamer(__snake_case , timeout=0.0_0_1 )
lowerCamelCase :Dict = {'''input_ids''': input_ids, '''max_new_tokens''': 10, '''do_sample''': False, '''streamer''': streamer}
lowerCamelCase :Tuple = Thread(target=model.generate , kwargs=__snake_case )
thread.start()
# The streamer will timeout after 0.001 seconds, so an exception will be raised
with self.assertRaises(__snake_case ):
lowerCamelCase :Dict = ''''''
for new_text in streamer:
streamer_text += new_text
| 49
| 0
|
import inspect
from typing import Optional, Union
import numpy as np
import PIL
import torch
from torch.nn import functional as F
from torchvision import transforms
from transformers import CLIPFeatureExtractor, CLIPModel, CLIPTextModel, CLIPTokenizer
from diffusers import (
AutoencoderKL,
DDIMScheduler,
DiffusionPipeline,
DPMSolverMultistepScheduler,
LMSDiscreteScheduler,
PNDMScheduler,
UNetaDConditionModel,
)
from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipelineOutput
from diffusers.utils import (
PIL_INTERPOLATION,
randn_tensor,
)
def __magic_name__ ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ) -> str:
if isinstance(SCREAMING_SNAKE_CASE , torch.Tensor ):
return image
elif isinstance(SCREAMING_SNAKE_CASE , PIL.Image.Image ):
_lowercase : List[Any] = [image]
if isinstance(image[0] , PIL.Image.Image ):
_lowercase : Union[str, Any] = [np.array(i.resize((w, h) , resample=PIL_INTERPOLATION['lanczos'] ) )[None, :] for i in image]
_lowercase : Dict = np.concatenate(SCREAMING_SNAKE_CASE , axis=0 )
_lowercase : Dict = np.array(SCREAMING_SNAKE_CASE ).astype(np.floataa ) / 255.0
_lowercase : Any = image.transpose(0 , 3 , 1 , 2 )
_lowercase : Optional[int] = 2.0 * image - 1.0
_lowercase : Union[str, Any] = torch.from_numpy(SCREAMING_SNAKE_CASE )
elif isinstance(image[0] , torch.Tensor ):
_lowercase : Optional[Any] = torch.cat(SCREAMING_SNAKE_CASE , dim=0 )
return image
def __magic_name__ ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE=0.9995 ) -> str:
if not isinstance(SCREAMING_SNAKE_CASE , np.ndarray ):
_lowercase : Optional[Any] = True
_lowercase : Any = va.device
_lowercase : str = va.cpu().numpy()
_lowercase : str = va.cpu().numpy()
_lowercase : Tuple = np.sum(va * va / (np.linalg.norm(SCREAMING_SNAKE_CASE ) * np.linalg.norm(SCREAMING_SNAKE_CASE )) )
if np.abs(SCREAMING_SNAKE_CASE ) > DOT_THRESHOLD:
_lowercase : Dict = (1 - t) * va + t * va
else:
_lowercase : Any = np.arccos(SCREAMING_SNAKE_CASE )
_lowercase : Dict = np.sin(SCREAMING_SNAKE_CASE )
_lowercase : Any = theta_a * t
_lowercase : Optional[Any] = np.sin(SCREAMING_SNAKE_CASE )
_lowercase : Union[str, Any] = np.sin(theta_a - theta_t ) / sin_theta_a
_lowercase : List[str] = sin_theta_t / sin_theta_a
_lowercase : Optional[Any] = sa * va + sa * va
if inputs_are_torch:
_lowercase : Optional[Any] = torch.from_numpy(SCREAMING_SNAKE_CASE ).to(SCREAMING_SNAKE_CASE )
return va
def __magic_name__ ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ) -> Optional[Any]:
_lowercase : Any = F.normalize(SCREAMING_SNAKE_CASE , dim=-1 )
_lowercase : Union[str, Any] = F.normalize(SCREAMING_SNAKE_CASE , dim=-1 )
return (x - y).norm(dim=-1 ).div(2 ).arcsin().pow(2 ).mul(2 )
def __magic_name__ ( SCREAMING_SNAKE_CASE , SCREAMING_SNAKE_CASE ) -> str:
for param in model.parameters():
_lowercase : List[str] = value
class lowerCAmelCase_ ( __snake_case ):
def __init__( self , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase=None , _lowerCAmelCase=None , _lowerCAmelCase=None , ):
super().__init__()
self.register_modules(
vae=_lowerCAmelCase , text_encoder=_lowerCAmelCase , clip_model=_lowerCAmelCase , tokenizer=_lowerCAmelCase , unet=_lowerCAmelCase , scheduler=_lowerCAmelCase , feature_extractor=_lowerCAmelCase , coca_model=_lowerCAmelCase , coca_tokenizer=_lowerCAmelCase , coca_transform=_lowerCAmelCase , )
_lowercase : str = (
feature_extractor.size
if isinstance(feature_extractor.size , _lowerCAmelCase )
else feature_extractor.size['shortest_edge']
)
_lowercase : List[str] = transforms.Normalize(mean=feature_extractor.image_mean , std=feature_extractor.image_std )
set_requires_grad(self.text_encoder , _lowerCAmelCase )
set_requires_grad(self.clip_model , _lowerCAmelCase )
def __a ( self , _lowerCAmelCase = "auto" ):
if slice_size == "auto":
# half the attention head size is usually a good trade-off between
# speed and memory
_lowercase : int = self.unet.config.attention_head_dim // 2
self.unet.set_attention_slice(_lowerCAmelCase )
def __a ( self ):
self.enable_attention_slicing(_lowerCAmelCase )
def __a ( self ):
set_requires_grad(self.vae , _lowerCAmelCase )
def __a ( self ):
set_requires_grad(self.vae , _lowerCAmelCase )
def __a ( self ):
set_requires_grad(self.unet , _lowerCAmelCase )
def __a ( self ):
set_requires_grad(self.unet , _lowerCAmelCase )
def __a ( self , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase ):
# get the original timestep using init_timestep
_lowercase : Union[str, Any] = min(int(num_inference_steps * strength ) , _lowerCAmelCase )
_lowercase : List[Any] = max(num_inference_steps - init_timestep , 0 )
_lowercase : Optional[Any] = self.scheduler.timesteps[t_start:]
return timesteps, num_inference_steps - t_start
def __a ( self , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase=None ):
if not isinstance(_lowerCAmelCase , torch.Tensor ):
raise ValueError(F"""`image` has to be of type `torch.Tensor` but is {type(_lowerCAmelCase )}""" )
_lowercase : Tuple = image.to(device=_lowerCAmelCase , dtype=_lowerCAmelCase )
if isinstance(_lowerCAmelCase , _lowerCAmelCase ):
_lowercase : int = [
self.vae.encode(image[i : i + 1] ).latent_dist.sample(generator[i] ) for i in range(_lowerCAmelCase )
]
_lowercase : int = torch.cat(_lowerCAmelCase , dim=0 )
else:
_lowercase : List[str] = self.vae.encode(_lowerCAmelCase ).latent_dist.sample(_lowerCAmelCase )
# Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor
_lowercase : Optional[int] = 0.1_82_15 * init_latents
_lowercase : int = init_latents.repeat_interleave(_lowerCAmelCase , dim=0 )
_lowercase : Union[str, Any] = randn_tensor(init_latents.shape , generator=_lowerCAmelCase , device=_lowerCAmelCase , dtype=_lowerCAmelCase )
# get latents
_lowercase : Dict = self.scheduler.add_noise(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase )
_lowercase : List[Any] = init_latents
return latents
def __a ( self , _lowerCAmelCase ):
_lowercase : Dict = self.coca_transform(_lowerCAmelCase ).unsqueeze(0 )
with torch.no_grad(), torch.cuda.amp.autocast():
_lowercase : List[str] = self.coca_model.generate(transformed_image.to(device=self.device , dtype=self.coca_model.dtype ) )
_lowercase : List[Any] = self.coca_tokenizer.decode(generated[0].cpu().numpy() )
return generated.split('<end_of_text>' )[0].replace('<start_of_text>' , '' ).rstrip(' .,' )
def __a ( self , _lowerCAmelCase , _lowerCAmelCase ):
_lowercase : Any = self.feature_extractor.preprocess(_lowerCAmelCase )
_lowercase : Any = torch.from_numpy(clip_image_input['pixel_values'][0] ).unsqueeze(0 ).to(self.device ).half()
_lowercase : int = self.clip_model.get_image_features(_lowerCAmelCase )
_lowercase : Any = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_lowerCAmelCase )
_lowercase : List[Any] = image_embeddings_clip.repeat_interleave(_lowerCAmelCase , dim=0 )
return image_embeddings_clip
@torch.enable_grad()
def __a ( self , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , ):
_lowercase : Optional[Any] = latents.detach().requires_grad_()
_lowercase : Any = self.scheduler.scale_model_input(_lowerCAmelCase , _lowerCAmelCase )
# predict the noise residual
_lowercase : Union[str, Any] = self.unet(_lowerCAmelCase , _lowerCAmelCase , encoder_hidden_states=_lowerCAmelCase ).sample
if isinstance(self.scheduler , (PNDMScheduler, DDIMScheduler, DPMSolverMultistepScheduler) ):
_lowercase : List[str] = self.scheduler.alphas_cumprod[timestep]
_lowercase : Any = 1 - alpha_prod_t
# compute predicted original sample from predicted noise also called
# "predicted x_0" of formula (12) from https://arxiv.org/pdf/2010.02502.pdf
_lowercase : Dict = (latents - beta_prod_t ** 0.5 * noise_pred) / alpha_prod_t ** 0.5
_lowercase : Any = torch.sqrt(_lowerCAmelCase )
_lowercase : Union[str, Any] = pred_original_sample * (fac) + latents * (1 - fac)
elif isinstance(self.scheduler , _lowerCAmelCase ):
_lowercase : Optional[Any] = self.scheduler.sigmas[index]
_lowercase : Optional[int] = latents - sigma * noise_pred
else:
raise ValueError(F"""scheduler type {type(self.scheduler )} not supported""" )
# Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor
_lowercase : Tuple = 1 / 0.1_82_15 * sample
_lowercase : Union[str, Any] = self.vae.decode(_lowerCAmelCase ).sample
_lowercase : Any = (image / 2 + 0.5).clamp(0 , 1 )
_lowercase : str = transforms.Resize(self.feature_extractor_size )(_lowerCAmelCase )
_lowercase : Union[str, Any] = self.normalize(_lowerCAmelCase ).to(latents.dtype )
_lowercase : Optional[Any] = self.clip_model.get_image_features(_lowerCAmelCase )
_lowercase : Optional[Any] = image_embeddings_clip / image_embeddings_clip.norm(p=2 , dim=-1 , keepdim=_lowerCAmelCase )
_lowercase : Optional[Any] = spherical_dist_loss(_lowerCAmelCase , _lowerCAmelCase ).mean() * clip_guidance_scale
_lowercase : Tuple = -torch.autograd.grad(_lowerCAmelCase , _lowerCAmelCase )[0]
if isinstance(self.scheduler , _lowerCAmelCase ):
_lowercase : List[Any] = latents.detach() + grads * (sigma**2)
_lowercase : Union[str, Any] = noise_pred_original
else:
_lowercase : int = noise_pred_original - torch.sqrt(_lowerCAmelCase ) * grads
return noise_pred, latents
@torch.no_grad()
def __call__( self , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase = None , _lowerCAmelCase = None , _lowerCAmelCase = 5_1_2 , _lowerCAmelCase = 5_1_2 , _lowerCAmelCase = 0.6 , _lowerCAmelCase = 5_0 , _lowerCAmelCase = 7.5 , _lowerCAmelCase = 1 , _lowerCAmelCase = 0.0 , _lowerCAmelCase = 1_0_0 , _lowerCAmelCase = None , _lowerCAmelCase = "pil" , _lowerCAmelCase = True , _lowerCAmelCase = 0.8 , _lowerCAmelCase = 0.1 , _lowerCAmelCase = 0.1 , ):
if isinstance(_lowerCAmelCase , _lowerCAmelCase ) and len(_lowerCAmelCase ) != batch_size:
raise ValueError(F"""You have passed {batch_size} batch_size, but only {len(_lowerCAmelCase )} generators.""" )
if height % 8 != 0 or width % 8 != 0:
raise ValueError(F"""`height` and `width` have to be divisible by 8 but are {height} and {width}.""" )
if isinstance(_lowerCAmelCase , torch.Generator ) and batch_size > 1:
_lowercase : List[Any] = [generator] + [None] * (batch_size - 1)
_lowercase : List[Any] = [
('model', self.coca_model is None),
('tokenizer', self.coca_tokenizer is None),
('transform', self.coca_transform is None),
]
_lowercase : Dict = [x[0] for x in coca_is_none if x[1]]
_lowercase : Any = ', '.join(_lowerCAmelCase )
# generate prompts with coca model if prompt is None
if content_prompt is None:
if len(_lowerCAmelCase ):
raise ValueError(
F"""Content prompt is None and CoCa [{coca_is_none_str}] is None."""
F"""Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.""" )
_lowercase : Tuple = self.get_image_description(_lowerCAmelCase )
if style_prompt is None:
if len(_lowerCAmelCase ):
raise ValueError(
F"""Style prompt is None and CoCa [{coca_is_none_str}] is None."""
F""" Set prompt or pass Coca [{coca_is_none_str}] to DiffusionPipeline.""" )
_lowercase : List[str] = self.get_image_description(_lowerCAmelCase )
# get prompt text embeddings for content and style
_lowercase : int = self.tokenizer(
_lowerCAmelCase , padding='max_length' , max_length=self.tokenizer.model_max_length , truncation=_lowerCAmelCase , return_tensors='pt' , )
_lowercase : List[str] = self.text_encoder(content_text_input.input_ids.to(self.device ) )[0]
_lowercase : Tuple = self.tokenizer(
_lowerCAmelCase , padding='max_length' , max_length=self.tokenizer.model_max_length , truncation=_lowerCAmelCase , return_tensors='pt' , )
_lowercase : Tuple = self.text_encoder(style_text_input.input_ids.to(self.device ) )[0]
_lowercase : int = slerp(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase )
# duplicate text embeddings for each generation per prompt
_lowercase : Optional[int] = text_embeddings.repeat_interleave(_lowerCAmelCase , dim=0 )
# set timesteps
_lowercase : Tuple = 'offset' in set(inspect.signature(self.scheduler.set_timesteps ).parameters.keys() )
_lowercase : Optional[Any] = {}
if accepts_offset:
_lowercase : int = 1
self.scheduler.set_timesteps(_lowerCAmelCase , **_lowerCAmelCase )
# Some schedulers like PNDM have timesteps as arrays
# It's more optimized to move all timesteps to correct device beforehand
self.scheduler.timesteps.to(self.device )
_lowercase , _lowercase : List[str] = self.get_timesteps(_lowerCAmelCase , _lowerCAmelCase , self.device )
_lowercase : str = timesteps[:1].repeat(_lowerCAmelCase )
# Preprocess image
_lowercase : List[Any] = preprocess(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase )
_lowercase : Tuple = self.prepare_latents(
_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , text_embeddings.dtype , self.device , _lowerCAmelCase )
_lowercase : List[str] = preprocess(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase )
_lowercase : List[Any] = self.prepare_latents(
_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , text_embeddings.dtype , self.device , _lowerCAmelCase )
_lowercase : int = slerp(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase )
if clip_guidance_scale > 0:
_lowercase : int = self.get_clip_image_embeddings(_lowerCAmelCase , _lowerCAmelCase )
_lowercase : str = self.get_clip_image_embeddings(_lowerCAmelCase , _lowerCAmelCase )
_lowercase : Tuple = slerp(
_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase )
# here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
# of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
# corresponds to doing no classifier free guidance.
_lowercase : int = guidance_scale > 1.0
# get unconditional embeddings for classifier free guidance
if do_classifier_free_guidance:
_lowercase : Optional[int] = content_text_input.input_ids.shape[-1]
_lowercase : Optional[int] = self.tokenizer([''] , padding='max_length' , max_length=_lowerCAmelCase , return_tensors='pt' )
_lowercase : Union[str, Any] = self.text_encoder(uncond_input.input_ids.to(self.device ) )[0]
# duplicate unconditional embeddings for each generation per prompt
_lowercase : Tuple = uncond_embeddings.repeat_interleave(_lowerCAmelCase , dim=0 )
# For classifier free guidance, we need to do two forward passes.
# Here we concatenate the unconditional and text embeddings into a single batch
# to avoid doing two forward passes
_lowercase : Tuple = torch.cat([uncond_embeddings, text_embeddings] )
# get the initial random noise unless the user supplied it
# Unlike in other pipelines, latents need to be generated in the target device
# for 1-to-1 results reproducibility with the CompVis implementation.
# However this currently doesn't work in `mps`.
_lowercase : Union[str, Any] = (batch_size, self.unet.config.in_channels, height // 8, width // 8)
_lowercase : Optional[Any] = text_embeddings.dtype
if latents is None:
if self.device.type == "mps":
# randn does not work reproducibly on mps
_lowercase : Optional[int] = torch.randn(_lowerCAmelCase , generator=_lowerCAmelCase , device='cpu' , dtype=_lowerCAmelCase ).to(
self.device )
else:
_lowercase : Tuple = torch.randn(_lowerCAmelCase , generator=_lowerCAmelCase , device=self.device , dtype=_lowerCAmelCase )
else:
if latents.shape != latents_shape:
raise ValueError(F"""Unexpected latents shape, got {latents.shape}, expected {latents_shape}""" )
_lowercase : Any = latents.to(self.device )
# scale the initial noise by the standard deviation required by the scheduler
_lowercase : Union[str, Any] = latents * self.scheduler.init_noise_sigma
# prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
# eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
# eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
# and should be between [0, 1]
_lowercase : Optional[int] = 'eta' in set(inspect.signature(self.scheduler.step ).parameters.keys() )
_lowercase : str = {}
if accepts_eta:
_lowercase : Union[str, Any] = eta
# check if the scheduler accepts generator
_lowercase : Tuple = 'generator' in set(inspect.signature(self.scheduler.step ).parameters.keys() )
if accepts_generator:
_lowercase : int = generator
with self.progress_bar(total=_lowerCAmelCase ):
for i, t in enumerate(_lowerCAmelCase ):
# expand the latents if we are doing classifier free guidance
_lowercase : Optional[Any] = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents
_lowercase : Tuple = self.scheduler.scale_model_input(_lowerCAmelCase , _lowerCAmelCase )
# predict the noise residual
_lowercase : Optional[Any] = self.unet(_lowerCAmelCase , _lowerCAmelCase , encoder_hidden_states=_lowerCAmelCase ).sample
# perform classifier free guidance
if do_classifier_free_guidance:
_lowercase , _lowercase : Any = noise_pred.chunk(2 )
_lowercase : int = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
# perform clip guidance
if clip_guidance_scale > 0:
_lowercase : Optional[Any] = (
text_embeddings.chunk(2 )[1] if do_classifier_free_guidance else text_embeddings
)
_lowercase , _lowercase : Tuple = self.cond_fn(
_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , )
# compute the previous noisy sample x_t -> x_t-1
_lowercase : Optional[Any] = self.scheduler.step(_lowerCAmelCase , _lowerCAmelCase , _lowerCAmelCase , **_lowerCAmelCase ).prev_sample
# Hardcode 0.18215 because stable-diffusion-2-base has not self.vae.config.scaling_factor
_lowercase : Optional[Any] = 1 / 0.1_82_15 * latents
_lowercase : List[Any] = self.vae.decode(_lowerCAmelCase ).sample
_lowercase : Union[str, Any] = (image / 2 + 0.5).clamp(0 , 1 )
_lowercase : List[str] = image.cpu().permute(0 , 2 , 3 , 1 ).numpy()
if output_type == "pil":
_lowercase : Tuple = self.numpy_to_pil(_lowerCAmelCase )
if not return_dict:
return (image, None)
return StableDiffusionPipelineOutput(images=_lowerCAmelCase , nsfw_content_detected=_lowerCAmelCase )
| 66
|
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available, is_vision_available
UpperCamelCase = {
"configuration_poolformer": [
"POOLFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP",
"PoolFormerConfig",
"PoolFormerOnnxConfig",
]
}
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase = ["PoolFormerFeatureExtractor"]
UpperCamelCase = ["PoolFormerImageProcessor"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
UpperCamelCase = [
"POOLFORMER_PRETRAINED_MODEL_ARCHIVE_LIST",
"PoolFormerForImageClassification",
"PoolFormerModel",
"PoolFormerPreTrainedModel",
]
if TYPE_CHECKING:
from .configuration_poolformer import (
POOLFORMER_PRETRAINED_CONFIG_ARCHIVE_MAP,
PoolFormerConfig,
PoolFormerOnnxConfig,
)
try:
if not is_vision_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .feature_extraction_poolformer import PoolFormerFeatureExtractor
from .image_processing_poolformer import PoolFormerImageProcessor
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_poolformer import (
POOLFORMER_PRETRAINED_MODEL_ARCHIVE_LIST,
PoolFormerForImageClassification,
PoolFormerModel,
PoolFormerPreTrainedModel,
)
else:
import sys
UpperCamelCase = _LazyModule(__name__, globals()["__file__"], _import_structure)
| 66
| 1
|
"""simple docstring"""
import sys
import turtle
def lowercase_ ( _lowerCamelCase: tuple[float, float] , _lowerCamelCase: tuple[float, float] ) -> Optional[int]:
'''simple docstring'''
return (pa[0] + pa[0]) / 2, (pa[1] + pa[1]) / 2
def lowercase_ ( _lowerCamelCase: tuple[float, float] , _lowerCamelCase: tuple[float, float] , _lowerCamelCase: tuple[float, float] , _lowerCamelCase: int , ) -> int:
'''simple docstring'''
my_pen.up()
my_pen.goto(vertexa[0] , vertexa[1] )
my_pen.down()
my_pen.goto(vertexa[0] , vertexa[1] )
my_pen.goto(vertexa[0] , vertexa[1] )
my_pen.goto(vertexa[0] , vertexa[1] )
if depth == 0:
return
triangle(snake_case__ , get_mid(snake_case__ , snake_case__ ) , get_mid(snake_case__ , snake_case__ ) , depth - 1 )
triangle(snake_case__ , get_mid(snake_case__ , snake_case__ ) , get_mid(snake_case__ , snake_case__ ) , depth - 1 )
triangle(snake_case__ , get_mid(snake_case__ , snake_case__ ) , get_mid(snake_case__ , snake_case__ ) , depth - 1 )
if __name__ == "__main__":
if len(sys.argv) != 2:
raise ValueError(
'''Correct format for using this script: '''
'''python fractals.py <int:depth_for_fractal>'''
)
__A = turtle.Turtle()
my_pen.ht()
my_pen.speed(5)
my_pen.pencolor('''red''')
__A = [(-175, -125), (0, 175), (175, -125)] # vertices of triangle
triangle(vertices[0], vertices[1], vertices[2], int(sys.argv[1]))
| 706
|
"""simple docstring"""
def lowercase_ ( _lowerCamelCase: int = 600851475143 ) -> int:
'''simple docstring'''
try:
__lowerCamelCase : Optional[Any] = int(_lowerCamelCase )
except (TypeError, ValueError):
raise TypeError("Parameter n must be int or castable to int." )
if n <= 0:
raise ValueError("Parameter n must be greater than or equal to one." )
__lowerCamelCase : Union[str, Any] = 2
__lowerCamelCase : int = 0
if n == 2:
return 2
while n > 2:
while n % i != 0:
i += 1
__lowerCamelCase : Dict = i
while n % i == 0:
__lowerCamelCase : Union[str, Any] = n // i
i += 1
return int(_lowerCamelCase )
if __name__ == "__main__":
print(F"""{solution() = }""")
| 366
| 0
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.