code
stringlengths 86
54.5k
| code_codestyle
int64 0
371
| style_context
stringlengths 87
49.2k
| style_context_codestyle
int64 0
349
| label
int64 0
1
|
|---|---|---|---|---|
"""simple docstring"""
import copy
import inspect
import unittest
import numpy as np
from huggingface_hub import hf_hub_download
from transformers import TimesformerConfig
from transformers.models.auto import get_values
from transformers.testing_utils import require_torch, require_vision, slow, torch_device
from transformers.utils import cached_property, is_torch_available, is_vision_available
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from torch import nn
from transformers import (
MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING,
TimesformerForVideoClassification,
TimesformerModel,
)
from transformers.models.timesformer.modeling_timesformer import TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST
if is_vision_available():
from transformers import VideoMAEImageProcessor
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self , _a , _a=1_3 , _a=1_0 , _a=3 , _a=2 , _a=2 , _a=True , _a=True , _a=3_2 , _a=5 , _a=4 , _a=3_7 , _a="gelu" , _a=0.1 , _a=0.1 , _a=1_0 , _a=0.02 , _a="divided_space_time" , _a=None , ) -> List[str]:
_a : Tuple = parent
_a : str = batch_size
_a : Optional[int] = image_size
_a : int = num_channels
_a : Any = patch_size
_a : List[str] = num_frames
_a : int = is_training
_a : Optional[int] = use_labels
_a : Union[str, Any] = hidden_size
_a : Union[str, Any] = num_hidden_layers
_a : List[str] = num_attention_heads
_a : Optional[int] = intermediate_size
_a : List[str] = hidden_act
_a : List[str] = hidden_dropout_prob
_a : List[Any] = attention_probs_dropout_prob
_a : Union[str, Any] = attention_type
_a : Any = initializer_range
_a : Any = scope
_a : List[Any] = num_labels
# in TimeSformer, the number of spatial tokens equals num_frames * num_patches per frame + 1 CLS token
_a : Any = (image_size // patch_size) ** 2
_a : List[str] = (num_frames) * self.num_patches_per_frame + 1
def __lowercase ( self ) -> Dict:
_a : str = floats_tensor(
[self.batch_size, self.num_frames, self.num_channels, self.image_size, self.image_size] )
_a : str = None
if self.use_labels:
_a : List[str] = ids_tensor([self.batch_size] , self.num_labels )
_a : Tuple = self.get_config()
return config, pixel_values, labels
def __lowercase ( self ) -> Optional[Any]:
_a : Any = TimesformerConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , num_frames=self.num_frames , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , initializer_range=self.initializer_range , attention_type=self.attention_type , )
_a : List[Any] = self.num_labels
return config
def __lowercase ( self , _a , _a , _a ) -> int:
_a : List[str] = TimesformerModel(config=lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
_a : List[str] = model(lowerCAmelCase_ )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def __lowercase ( self , _a , _a , _a ) -> Tuple:
_a : Union[str, Any] = TimesformerForVideoClassification(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
_a : str = model(lowerCAmelCase_ )
# verify the logits shape
_a : Tuple = torch.Size((self.batch_size, self.num_labels) )
self.parent.assertEqual(result.logits.shape , lowerCAmelCase_ )
def __lowercase ( self ) -> Optional[int]:
_a : Union[str, Any] = self.prepare_config_and_inputs()
_a , _a , _a : List[Any] = config_and_inputs
_a : Optional[int] = {'''pixel_values''': pixel_values}
return config, inputs_dict
@require_torch
class UpperCAmelCase_ ( __lowerCAmelCase , __lowerCAmelCase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : Any = (TimesformerModel, TimesformerForVideoClassification) if is_torch_available() else ()
UpperCAmelCase__ : List[str] = (
{'''feature-extraction''': TimesformerModel, '''video-classification''': TimesformerForVideoClassification}
if is_torch_available()
else {}
)
UpperCAmelCase__ : Any = False
UpperCAmelCase__ : Dict = False
UpperCAmelCase__ : List[str] = False
UpperCAmelCase__ : Tuple = False
def __lowercase ( self ) -> Optional[Any]:
_a : Any = TimesformerModelTester(self )
_a : str = ConfigTester(
self , config_class=lowerCAmelCase_ , has_text_modality=lowerCAmelCase_ , hidden_size=3_7 )
def __lowercase ( self , _a , _a , _a=False ) -> Optional[Any]:
_a : Optional[Any] = copy.deepcopy(lowerCAmelCase_ )
if return_labels:
if model_class in get_values(lowerCAmelCase_ ):
_a : Optional[Any] = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=lowerCAmelCase_ )
return inputs_dict
def __lowercase ( self ) -> Dict:
self.config_tester.run_common_tests()
@unittest.skip(reason='''TimeSformer does not use inputs_embeds''' )
def __lowercase ( self ) -> Tuple:
pass
def __lowercase ( self ) -> List[str]:
_a , _a : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
_a : Optional[int] = model_class(lowerCAmelCase_ )
self.assertIsInstance(model.get_input_embeddings() , (nn.Module) )
_a : Optional[int] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(lowerCAmelCase_ , nn.Linear ) )
def __lowercase ( self ) -> str:
_a , _a : int = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
_a : Dict = model_class(lowerCAmelCase_ )
_a : str = inspect.signature(model.forward )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
_a : Tuple = [*signature.parameters.keys()]
_a : int = ['''pixel_values''']
self.assertListEqual(arg_names[:1] , lowerCAmelCase_ )
def __lowercase ( self ) -> List[Any]:
_a : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*lowerCAmelCase_ )
def __lowercase ( self ) -> Tuple:
_a : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_video_classification(*lowerCAmelCase_ )
@slow
def __lowercase ( self ) -> Union[str, Any]:
for model_name in TIMESFORMER_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
_a : Union[str, Any] = TimesformerModel.from_pretrained(lowerCAmelCase_ )
self.assertIsNotNone(lowerCAmelCase_ )
def __lowercase ( self ) -> Optional[int]:
if not self.has_attentions:
pass
else:
_a , _a : str = self.model_tester.prepare_config_and_inputs_for_common()
_a : Any = True
for model_class in self.all_model_classes:
_a : Union[str, Any] = self.model_tester.seq_length
_a : List[Any] = self.model_tester.num_frames
_a : Dict = True
_a : List[str] = False
_a : Union[str, Any] = True
_a : Any = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
_a : List[Any] = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
_a : List[str] = outputs.attentions
self.assertEqual(len(lowerCAmelCase_ ) , self.model_tester.num_hidden_layers )
# check that output_attentions also work using config
del inputs_dict["output_attentions"]
_a : List[Any] = True
_a : Optional[int] = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
_a : Any = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
_a : List[Any] = outputs.attentions
self.assertEqual(len(lowerCAmelCase_ ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
_a : List[str] = len(lowerCAmelCase_ )
# Check attention is always last and order is fine
_a : Dict = True
_a : int = True
_a : Optional[int] = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
_a : Tuple = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
self.assertEqual(out_len + 1 , len(lowerCAmelCase_ ) )
_a : Optional[int] = outputs.attentions
self.assertEqual(len(lowerCAmelCase_ ) , self.model_tester.num_hidden_layers )
# attentions has shape (batch_size x num_frames) x num_heads x (num_patches per frame + 1) x (num_patches per frame + 1)
self.assertListEqual(
list(self_attentions[0].shape[-3:] ) , [self.model_tester.num_attention_heads, seq_len // num_frames + 1, seq_len // num_frames + 1] , )
def __lowercase ( self ) -> List[Any]:
def check_hidden_states_output(_a , _a , _a ):
_a : List[str] = model_class(lowerCAmelCase_ )
model.to(lowerCAmelCase_ )
model.eval()
with torch.no_grad():
_a : List[str] = model(**self._prepare_for_class(lowerCAmelCase_ , lowerCAmelCase_ ) )
_a : Union[str, Any] = outputs.hidden_states
_a : Union[str, Any] = self.model_tester.num_hidden_layers + 1
self.assertEqual(len(lowerCAmelCase_ ) , lowerCAmelCase_ )
_a : Optional[int] = self.model_tester.seq_length
self.assertListEqual(
list(hidden_states[0].shape[-2:] ) , [seq_length, self.model_tester.hidden_size] , )
_a , _a : Any = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
_a : Any = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
# check that output_hidden_states also work using config
del inputs_dict["output_hidden_states"]
_a : str = True
check_hidden_states_output(lowerCAmelCase_ , lowerCAmelCase_ , lowerCAmelCase_ )
def __UpperCAmelCase ( ) -> int:
"""simple docstring"""
_a : int = hf_hub_download(
repo_id='''hf-internal-testing/spaghetti-video''' ,filename='''eating_spaghetti.npy''' ,repo_type='''dataset''' )
_a : Union[str, Any] = np.load(snake_case__ )
return list(snake_case__ )
@require_torch
@require_vision
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def __lowercase ( self ) -> List[str]:
# logits were tested with a different mean and std, so we use the same here
return (
VideoMAEImageProcessor(image_mean=[0.5, 0.5, 0.5] , image_std=[0.5, 0.5, 0.5] )
if is_vision_available()
else None
)
@slow
def __lowercase ( self ) -> Union[str, Any]:
_a : int = TimesformerForVideoClassification.from_pretrained('''facebook/timesformer-base-finetuned-k400''' ).to(
lowerCAmelCase_ )
_a : Any = self.default_image_processor
_a : Any = prepare_video()
_a : Any = image_processor(video[:8] , return_tensors='''pt''' ).to(lowerCAmelCase_ )
# forward pass
with torch.no_grad():
_a : Any = model(**lowerCAmelCase_ )
# verify the logits
_a : Dict = torch.Size((1, 4_0_0) )
self.assertEqual(outputs.logits.shape , lowerCAmelCase_ )
_a : List[str] = torch.tensor([-0.3016, -0.7713, -0.4205] ).to(lowerCAmelCase_ )
self.assertTrue(torch.allclose(outputs.logits[0, :3] , lowerCAmelCase_ , atol=1e-4 ) )
| 364
|
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a__ = {
'''configuration_xmod''': [
'''XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP''',
'''XmodConfig''',
'''XmodOnnxConfig''',
],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''XMOD_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''XmodForCausalLM''',
'''XmodForMaskedLM''',
'''XmodForMultipleChoice''',
'''XmodForQuestionAnswering''',
'''XmodForSequenceClassification''',
'''XmodForTokenClassification''',
'''XmodModel''',
'''XmodPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_xmod import XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP, XmodConfig, XmodOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_xmod import (
XMOD_PRETRAINED_MODEL_ARCHIVE_LIST,
XmodForCausalLM,
XmodForMaskedLM,
XmodForMultipleChoice,
XmodForQuestionAnswering,
XmodForSequenceClassification,
XmodForTokenClassification,
XmodModel,
XmodPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 15
| 0
|
def __UpperCAmelCase ( __a : int ,__a : int ) -> Optional[Any]:
"""simple docstring"""
while second != 0:
_a : Tuple = first & second
first ^= second
_a : int = c << 1
return first
if __name__ == "__main__":
import doctest
doctest.testmod()
a__ = int(input('''Enter the first number: ''').strip())
a__ = int(input('''Enter the second number: ''').strip())
print(f'''{add(first, second) = }''')
| 365
|
import re
import tempfile
from pathlib import Path
import pytest
import yaml
from datasets.utils.readme import ReadMe
# @pytest.fixture
# def example_yaml_structure():
a__ = yaml.safe_load(
'''\
name: ""
allow_empty: false
allow_empty_text: true
subsections:
- name: "Dataset Card for X" # First-level markdown heading
allow_empty: false
allow_empty_text: true
subsections:
- name: "Table of Contents"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Dataset Description"
allow_empty: false
allow_empty_text: false
subsections:
- name: "Dataset Summary"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Supported Tasks and Leaderboards"
allow_empty: true
allow_empty_text: true
subsections: null
- name: Languages
allow_empty: false
allow_empty_text: true
subsections: null
'''
)
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
#### Extra Ignored Subsection
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Extra Ignored Subsection''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
}
],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tEmpty YAML markers are present in the README.'''
)
a__ = '''\
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tNo YAML markers are present in the README.'''
)
a__ = '''\
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tOnly the start of YAML tags present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Summary` but it is empty.\n-\tExpected some text in section `Dataset Summary` but it is empty (text in subsections are ignored).'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Card for My Dataset` but it is empty.\n-\tSection `Dataset Card for My Dataset` expected the following subsections: `Table of Contents`, `Dataset Description`. Found \'None\'.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tSection `Dataset Description` is missing subsection: `Supported Tasks and Leaderboards`.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Languages` but it is empty.'''
a__ = '''\
---
language:
- zh
- en
---
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
# Dataset Card My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has several first-level headings: `Dataset Card for My Dataset`, `Dataset Card My Dataset`. Only one heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tNo first-level heading starting with `Dataset Card for` found in README. Skipping further validation for this README.'''
a__ = ''''''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.\n-\tNo YAML markers are present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found while parsing the README at `{path}`:\n-\tMultiple sections with the same heading `Dataset Card for My Dataset` have been found. Please keep only one of these sections.'''
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : List[str] ) -> Optional[int]:
"""simple docstring"""
assert ReadMe.from_string(__a ,__a ).to_dict() == expected_dict
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[str] ,__a : Optional[Any] ) -> Union[str, Any]:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
_a : List[Any] = ReadMe.from_string(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Dict ,__a : Dict ) -> Tuple:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
ReadMe.from_string(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> Tuple:
"""simple docstring"""
ReadMe.from_string(__a ,__a ,suppress_parsing_errors=__a )
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : Any ) -> Optional[int]:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Tuple = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[Any] = ReadMe.from_readme(__a ,__a ).to_dict()
assert out["name"] == path
assert out["text"] == ""
assert out["is_empty_text"]
assert out["subsections"] == expected_dict["subsections"]
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[Any] ,__a : List[Any] ) -> int:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[int] = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
_a : Any = ReadMe.from_readme(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : str ,__a : Union[str, Any] ) -> Dict:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Optional[Any] = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : str = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
ReadMe.from_readme(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> str:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
ReadMe.from_readme(__a ,__a ,suppress_parsing_errors=__a )
| 15
| 0
|
from __future__ import annotations
from collections import Counter
from random import random
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self ) -> int:
_a : str = {}
def __lowercase ( self , _a ) -> None:
_a : str = {}
def __lowercase ( self , _a , _a , _a ) -> None:
if nodea not in self.connections:
self.add_node(UpperCamelCase__ )
if nodea not in self.connections:
self.add_node(UpperCamelCase__ )
_a : Any = probability
def __lowercase ( self ) -> list[str]:
return list(self.connections )
def __lowercase ( self , _a ) -> str:
_a : int = 0
_a : Optional[int] = random()
for dest in self.connections[node]:
current_probability += self.connections[node][dest]
if current_probability > random_value:
return dest
return ""
def __UpperCAmelCase ( __a : List[Any] ,__a : Union[str, Any] ,__a : str ) -> dict[str, int]:
"""simple docstring"""
_a : List[Any] = MarkovChainGraphUndirectedUnweighted()
for nodea, nodea, probability in transitions:
graph.add_transition_probability(SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ ,SCREAMING_SNAKE_CASE__ )
_a : List[str] = Counter(graph.get_nodes() )
_a : Any = start
for _ in range(SCREAMING_SNAKE_CASE__ ):
_a : Any = graph.transition(SCREAMING_SNAKE_CASE__ )
visited[node] += 1
return visited
if __name__ == "__main__":
import doctest
doctest.testmod()
| 366
|
from __future__ import annotations
def __UpperCAmelCase ( __a : list ) -> float:
"""simple docstring"""
if not nums:
raise ValueError('''List is empty''' )
return sum(__a ) / len(__a )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 15
| 0
|
import argparse
import json
import os
from collections import OrderedDict
import numpy as np
import tensorflow as tf
import torch
def __UpperCAmelCase ( __a : int ) -> List[str]:
"""simple docstring"""
_a : List[str] = os.path.join(args.tf_model_dir ,'''parameters.json''' )
_a : Optional[int] = json.loads(open(lowerCamelCase_ ).read() )
if not params:
raise ValueError(
F"""It seems that the json file at {parameter_file} is empty. Make sure you have a correct json file.""" )
if not args.output.endswith('''.pt''' ):
_a : List[str] = args.output + '.pt'
_a : Tuple = OrderedDict()
with tf.device('''/CPU:0''' ):
_a : Any = tf.train.load_checkpoint(args.tf_model_dir )
_a : List[str] = reader.get_variable_to_shape_map()
for key_name in shapes.keys():
_a : Union[str, Any] = reader.get_tensor(lowerCamelCase_ ).astype(np.floataa )
if key_name.endswith('''/adam_m''' ) or key_name.endswith('''/adam_v''' ):
continue
if key_name.startswith('''pasts/''' ):
if key_name.startswith('''pasts/mlp''' ):
_a : Optional[int] = int(key_name[9] )
elif key_name.startswith('''pasts/out''' ):
_a : int = 8
_a : List[Any] = 'model.sqout.%d.weight' % (player * 2) # enter to nn.Sequencial with Tanh, so 2 at a time
_a : Optional[Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix
_a : List[Any] = torch.tensor(lowerCamelCase_ )
elif key_name.startswith('''model/moe''' ):
_a : List[str] = int(key_name[9:].split('''/''' )[0] )
if key_name.endswith('''/switch_gating/kernel''' ):
_a : Union[str, Any] = 'model.blocks.%d.feed_forward.mlp.router.classifier.weight' % player
_a : Optional[int] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix
_a : Any = torch.tensor(lowerCamelCase_ )
elif key_name.endswith('''/softmlp/kernel''' ):
_a : Any = 'model.blocks.%d.feed_forward.soft_bypass_mlp.weight' % player
_a : List[Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix
_a : List[str] = torch.tensor(lowerCamelCase_ )
elif key_name.endswith('''/wo/kernel''' ) or key_name.endswith('''/wi/kernel''' ):
_a : List[Any] = key_name[-9:-7]
for i in range(16 ):
_a : Tuple = 'model.blocks.%d.feed_forward.mlp.experts.expert_%d.%s.weight' % (player, i, nlayer)
_a : List[str] = (
vnp[i].transpose([1, 0] ).copy()
) # In Mesh-Tensorflow, it is one array, so it is divided
_a : Optional[int] = torch.tensor(lowerCamelCase_ )
elif key_name.startswith('''model/mlp''' ):
_a : Optional[int] = int(key_name[9:].split('''/''' )[0] )
if key_name.endswith('''/p1/kernel''' ):
_a : Dict = 'model.blocks.%d.feed_forward.mlp.wi.weight' % player
_a : Optional[Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix
_a : Any = torch.tensor(lowerCamelCase_ )
elif key_name.endswith('''/p1/bias''' ):
_a : Optional[int] = 'model.blocks.%d.feed_forward.mlp.wi.bias' % player
_a : int = vnp.copy() # same because it is one dimensional
_a : Optional[int] = torch.tensor(lowerCamelCase_ )
elif key_name.endswith('''/p2/kernel''' ):
_a : List[str] = 'model.blocks.%d.feed_forward.mlp.wo.weight' % player
_a : Union[str, Any] = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix
_a : Union[str, Any] = torch.tensor(lowerCamelCase_ )
elif key_name.endswith('''/p2/bias''' ):
_a : Tuple = 'model.blocks.%d.feed_forward.mlp.wo.bias' % player
_a : Optional[int] = vnp.copy() # same because it is one dimensional
_a : int = torch.tensor(lowerCamelCase_ )
elif key_name.startswith('''model/ln''' ):
_a : Union[str, Any] = int(key_name[8:].split('''/''' )[0] )
if key_name.endswith('''/b''' ):
_a : List[str] = 'model.blocks.%d.feed_forward.norm.bias' % player
_a : Optional[Any] = vnp.copy() # same because it is one dimensional
_a : Dict = torch.tensor(lowerCamelCase_ )
elif key_name.endswith('''/g''' ):
_a : Tuple = 'model.blocks.%d.feed_forward.norm.weight' % player
_a : str = vnp.copy() # same because it is one dimensional
_a : str = torch.tensor(lowerCamelCase_ )
elif key_name.startswith('''model/att''' ):
_a : Dict = int(key_name[9:].split('''/''' )[0] )
if key_name.endswith('''/qkv/kernel''' ):
_a : Optional[int] = vnp.copy() # Compute same dimension as Mesh-tensorflow using einsum
_a : Union[str, Any] = state[:, 0, :, :]
_a : str = state[:, 1, :, :]
_a : Any = state[:, 2, :, :]
_a : Optional[int] = (
state_q.reshape([state_q.shape[0], state_q.shape[1] * state_q.shape[2]] )
.transpose([1, 0] )
.copy()
) # Mesh-Tensorflow is a diagonal matrix
_a : Union[str, Any] = (
state_k.reshape([state_k.shape[0], state_k.shape[1] * state_k.shape[2]] )
.transpose([1, 0] )
.copy()
) # Mesh-Tensorflow is a diagonal matrix
_a : Optional[int] = (
state_v.reshape([state_v.shape[0], state_v.shape[1] * state_v.shape[2]] )
.transpose([1, 0] )
.copy()
) # Mesh-Tensorflow is a diagonal matrix
_a : Optional[Any] = 'model.blocks.%d.self_attn.self_attn.q_proj.weight' % player
_a : Any = torch.tensor(lowerCamelCase_ )
_a : Tuple = 'model.blocks.%d.self_attn.self_attn.k_proj.weight' % player
_a : Optional[int] = torch.tensor(lowerCamelCase_ )
_a : Optional[int] = 'model.blocks.%d.self_attn.self_attn.v_proj.weight' % player
_a : Dict = torch.tensor(lowerCamelCase_ )
elif key_name.endswith('''/o/kernel''' ):
_a : Optional[Any] = 'model.blocks.%d.self_attn.self_attn.out_proj.weight' % player
_a : Tuple = (
vnp.reshape([vnp.shape[0] * vnp.shape[1], vnp.shape[2]] ).transpose([1, 0] ).copy()
) # Mesh-Tensorflow is a diagonal matrix
_a : Optional[Any] = torch.tensor(lowerCamelCase_ )
elif key_name.startswith('''model/an''' ):
_a : Tuple = int(key_name[8:].split('''/''' )[0] )
if key_name.endswith('''/b''' ):
_a : Optional[Any] = 'model.blocks.%d.self_attn.norm.bias' % player
_a : Dict = vnp.copy() # same because it is one dimensional
_a : Optional[int] = torch.tensor(lowerCamelCase_ )
elif key_name.endswith('''/g''' ):
_a : Optional[Any] = 'model.blocks.%d.self_attn.norm.weight' % player
_a : int = vnp.copy() # same because it is one dimensional
_a : List[str] = torch.tensor(lowerCamelCase_ )
elif (
key_name.startswith('''model/wte''' )
or key_name.startswith('''model/wpe''' )
or key_name.startswith('''model/ete''' )
):
_a : Optional[int] = {'wte': 'embed_tokens', 'wpe': 'position_embeddings', 'ete': 'extra_position_embeddings'}[
key_name[-3:]
]
_a : str = 'model.%s.weight' % nlayer
_a : Optional[Any] = vnp.copy() # same in embedded
_a : List[str] = torch.tensor(lowerCamelCase_ )
if key_name.startswith('''model/wte''' ):
_a : Dict = 'lm_head.weight'
_a : Optional[int] = vnp.copy() # same in embedded
_a : Tuple = torch.tensor(lowerCamelCase_ )
elif key_name.startswith('''model/wob''' ):
_a : Union[str, Any] = 'final_logits_bias'
_a : Tuple = vnp.copy() # same in embedded
_a : Union[str, Any] = state.reshape((1, -1) )
_a : List[Any] = torch.tensor(lowerCamelCase_ )
elif key_name == "model/dense/kernel":
_a : Dict = 'model.last_project.weight'
_a : str = vnp.transpose([1, 0] ).copy() # Mesh-Tensorflow is a diagonal matrix
_a : Optional[int] = torch.tensor(lowerCamelCase_ )
elif key_name == "model/dense_1/bias":
_a : Optional[Any] = 'model.last_project.bias'
_a : Optional[Any] = vnp.copy() # same because it is one dimensional
_a : Optional[int] = torch.tensor(lowerCamelCase_ )
torch.save(lowerCamelCase_ ,args.output )
if __name__ == "__main__":
a__ = argparse.ArgumentParser(
description='''model converter.''', formatter_class=argparse.ArgumentDefaultsHelpFormatter
)
parser.add_argument('''--tf_model_dir''', metavar='''PATH''', type=str, required=True, help='''import model''')
parser.add_argument('''--output''', metavar='''PATH''', type=str, required=True, help='''output model''')
a__ = parser.parse_args()
convert_tf_gptsan_to_pt(args)
| 367
|
import argparse
import os
import torch
from transformers.utils import WEIGHTS_NAME
a__ = ['''small''', '''medium''', '''large''']
a__ = '''lm_head.decoder.weight'''
a__ = '''lm_head.weight'''
def __UpperCAmelCase ( __a : str ,__a : str ) -> List[str]:
"""simple docstring"""
_a : Any = torch.load(__a )
_a : List[str] = d.pop(__a )
os.makedirs(__a ,exist_ok=__a )
torch.save(__a ,os.path.join(__a ,__a ) )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--dialogpt_path''', default='''.''', type=str)
a__ = parser.parse_args()
for MODEL in DIALOGPT_MODELS:
a__ = os.path.join(args.dialogpt_path, f'''{MODEL}_ft.pkl''')
a__ = f'''./DialoGPT-{MODEL}'''
convert_dialogpt_checkpoint(
checkpoint_path,
pytorch_dump_folder_path,
)
| 15
| 0
|
import shutil
import tempfile
import unittest
from unittest.mock import patch
from transformers import (
DefaultFlowCallback,
IntervalStrategy,
PrinterCallback,
ProgressCallback,
Trainer,
TrainerCallback,
TrainingArguments,
is_torch_available,
)
from transformers.testing_utils import require_torch
if is_torch_available():
from transformers.trainer import DEFAULT_CALLBACKS
from .test_trainer import RegressionDataset, RegressionModelConfig, RegressionPreTrainedModel
class UpperCAmelCase_ ( __a ):
"""simple docstring"""
def __init__( self ) -> Dict:
_a : List[str] = []
def __lowercase ( self , _a , _a , _a , **_a ) -> Optional[Any]:
self.events.append('''on_init_end''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> int:
self.events.append('''on_train_begin''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> Optional[Any]:
self.events.append('''on_train_end''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> Dict:
self.events.append('''on_epoch_begin''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> str:
self.events.append('''on_epoch_end''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> Optional[Any]:
self.events.append('''on_step_begin''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> List[Any]:
self.events.append('''on_step_end''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> Optional[Any]:
self.events.append('''on_evaluate''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> Dict:
self.events.append('''on_predict''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> Optional[int]:
self.events.append('''on_save''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> List[Any]:
self.events.append('''on_log''' )
def __lowercase ( self , _a , _a , _a , **_a ) -> Optional[int]:
self.events.append('''on_prediction_step''' )
@require_torch
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> str:
_a : Optional[Any] = tempfile.mkdtemp()
def __lowercase ( self ) -> List[str]:
shutil.rmtree(self.output_dir )
def __lowercase ( self , _a=0 , _a=0 , _a=6_4 , _a=6_4 , _a=None , _a=False , **_a ) -> Union[str, Any]:
_a : Optional[int] = RegressionDataset(length=UpperCamelCase__ )
_a : str = RegressionDataset(length=UpperCamelCase__ )
_a : Tuple = RegressionModelConfig(a=UpperCamelCase__ , b=UpperCamelCase__ )
_a : int = RegressionPreTrainedModel(UpperCamelCase__ )
_a : Optional[Any] = TrainingArguments(self.output_dir , disable_tqdm=UpperCamelCase__ , report_to=[] , **UpperCamelCase__ )
return Trainer(
UpperCamelCase__ , UpperCamelCase__ , train_dataset=UpperCamelCase__ , eval_dataset=UpperCamelCase__ , callbacks=UpperCamelCase__ , )
def __lowercase ( self , _a , _a ) -> Any:
self.assertEqual(len(UpperCamelCase__ ) , len(UpperCamelCase__ ) )
# Order doesn't matter
_a : Optional[Any] = sorted(UpperCamelCase__ , key=lambda _a : cb.__name__ if isinstance(UpperCamelCase__ , UpperCamelCase__ ) else cb.__class__.__name__ )
_a : int = sorted(UpperCamelCase__ , key=lambda _a : cb.__name__ if isinstance(UpperCamelCase__ , UpperCamelCase__ ) else cb.__class__.__name__ )
for cba, cba in zip(UpperCamelCase__ , UpperCamelCase__ ):
if isinstance(UpperCamelCase__ , UpperCamelCase__ ) and isinstance(UpperCamelCase__ , UpperCamelCase__ ):
self.assertEqual(UpperCamelCase__ , UpperCamelCase__ )
elif isinstance(UpperCamelCase__ , UpperCamelCase__ ) and not isinstance(UpperCamelCase__ , UpperCamelCase__ ):
self.assertEqual(UpperCamelCase__ , cba.__class__ )
elif not isinstance(UpperCamelCase__ , UpperCamelCase__ ) and isinstance(UpperCamelCase__ , UpperCamelCase__ ):
self.assertEqual(cba.__class__ , UpperCamelCase__ )
else:
self.assertEqual(UpperCamelCase__ , UpperCamelCase__ )
def __lowercase ( self , _a ) -> List[Any]:
_a : List[Any] = ['''on_init_end''', '''on_train_begin''']
_a : int = 0
_a : Optional[Any] = len(trainer.get_eval_dataloader() )
_a : str = ['''on_prediction_step'''] * len(trainer.get_eval_dataloader() ) + ['''on_log''', '''on_evaluate''']
for _ in range(trainer.state.num_train_epochs ):
expected_events.append('''on_epoch_begin''' )
for _ in range(UpperCamelCase__ ):
step += 1
expected_events += ["on_step_begin", "on_step_end"]
if step % trainer.args.logging_steps == 0:
expected_events.append('''on_log''' )
if trainer.args.evaluation_strategy == IntervalStrategy.STEPS and step % trainer.args.eval_steps == 0:
expected_events += evaluation_events.copy()
if step % trainer.args.save_steps == 0:
expected_events.append('''on_save''' )
expected_events.append('''on_epoch_end''' )
if trainer.args.evaluation_strategy == IntervalStrategy.EPOCH:
expected_events += evaluation_events.copy()
expected_events += ["on_log", "on_train_end"]
return expected_events
def __lowercase ( self ) -> List[str]:
_a : List[Any] = self.get_trainer()
_a : List[Any] = DEFAULT_CALLBACKS.copy() + [ProgressCallback]
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
# Callbacks passed at init are added to the default callbacks
_a : Tuple = self.get_trainer(callbacks=[MyTestTrainerCallback] )
expected_callbacks.append(UpperCamelCase__ )
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
# TrainingArguments.disable_tqdm controls if use ProgressCallback or PrinterCallback
_a : int = self.get_trainer(disable_tqdm=UpperCamelCase__ )
_a : Union[str, Any] = DEFAULT_CALLBACKS.copy() + [PrinterCallback]
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
def __lowercase ( self ) -> int:
_a : int = DEFAULT_CALLBACKS.copy() + [ProgressCallback]
_a : Tuple = self.get_trainer()
# We can add, pop, or remove by class name
trainer.remove_callback(UpperCamelCase__ )
expected_callbacks.remove(UpperCamelCase__ )
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
_a : str = self.get_trainer()
_a : List[Any] = trainer.pop_callback(UpperCamelCase__ )
self.assertEqual(cb.__class__ , UpperCamelCase__ )
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
trainer.add_callback(UpperCamelCase__ )
expected_callbacks.insert(0 , UpperCamelCase__ )
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
# We can also add, pop, or remove by instance
_a : int = self.get_trainer()
_a : Optional[Any] = trainer.callback_handler.callbacks[0]
trainer.remove_callback(UpperCamelCase__ )
expected_callbacks.remove(UpperCamelCase__ )
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
_a : Dict = self.get_trainer()
_a : Optional[Any] = trainer.callback_handler.callbacks[0]
_a : Any = trainer.pop_callback(UpperCamelCase__ )
self.assertEqual(UpperCamelCase__ , UpperCamelCase__ )
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
trainer.add_callback(UpperCamelCase__ )
expected_callbacks.insert(0 , UpperCamelCase__ )
self.check_callbacks_equality(trainer.callback_handler.callbacks , UpperCamelCase__ )
def __lowercase ( self ) -> Dict:
import warnings
# XXX: for now ignore scatter_gather warnings in this test since it's not relevant to what's being tested
warnings.simplefilter(action='''ignore''' , category=UpperCamelCase__ )
_a : int = self.get_trainer(callbacks=[MyTestTrainerCallback] )
trainer.train()
_a : List[Any] = trainer.callback_handler.callbacks[-2].events
self.assertEqual(UpperCamelCase__ , self.get_expected_events(UpperCamelCase__ ) )
# Independent log/save/eval
_a : str = self.get_trainer(callbacks=[MyTestTrainerCallback] , logging_steps=5 )
trainer.train()
_a : int = trainer.callback_handler.callbacks[-2].events
self.assertEqual(UpperCamelCase__ , self.get_expected_events(UpperCamelCase__ ) )
_a : List[Any] = self.get_trainer(callbacks=[MyTestTrainerCallback] , save_steps=5 )
trainer.train()
_a : int = trainer.callback_handler.callbacks[-2].events
self.assertEqual(UpperCamelCase__ , self.get_expected_events(UpperCamelCase__ ) )
_a : str = self.get_trainer(callbacks=[MyTestTrainerCallback] , eval_steps=5 , evaluation_strategy='''steps''' )
trainer.train()
_a : Tuple = trainer.callback_handler.callbacks[-2].events
self.assertEqual(UpperCamelCase__ , self.get_expected_events(UpperCamelCase__ ) )
_a : List[Any] = self.get_trainer(callbacks=[MyTestTrainerCallback] , evaluation_strategy='''epoch''' )
trainer.train()
_a : Union[str, Any] = trainer.callback_handler.callbacks[-2].events
self.assertEqual(UpperCamelCase__ , self.get_expected_events(UpperCamelCase__ ) )
# A bit of everything
_a : Dict = self.get_trainer(
callbacks=[MyTestTrainerCallback] , logging_steps=3 , save_steps=1_0 , eval_steps=5 , evaluation_strategy='''steps''' , )
trainer.train()
_a : int = trainer.callback_handler.callbacks[-2].events
self.assertEqual(UpperCamelCase__ , self.get_expected_events(UpperCamelCase__ ) )
# warning should be emitted for duplicated callbacks
with patch('''transformers.trainer_callback.logger.warning''' ) as warn_mock:
_a : str = self.get_trainer(
callbacks=[MyTestTrainerCallback, MyTestTrainerCallback] , )
assert str(UpperCamelCase__ ) in warn_mock.call_args[0][0]
| 368
|
import enum
import warnings
from .. import MODEL_FOR_CAUSAL_LM_MAPPING, TF_MODEL_FOR_CAUSAL_LM_MAPPING
from ..utils import add_end_docstrings, is_tf_available
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
class UpperCAmelCase_ ( enum.Enum ):
"""simple docstring"""
UpperCAmelCase__ : int = 0
UpperCAmelCase__ : Union[str, Any] = 1
UpperCAmelCase__ : Optional[Any] = 2
@add_end_docstrings(__lowercase )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = "\n In 1991, the remains of Russian Tsar Nicholas II and his family (except for Alexei and Maria) are discovered. The\n voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the remainder of the story. 1883 Western\n Siberia, a young Grigori Rasputin is asked by his father and a group of men to perform magic. Rasputin has a vision\n and denounces one of the men as a horse thief. Although his father initially slaps him for making such an\n accusation, Rasputin watches as the man is chased outside and beaten. Twenty years later, Rasputin sees a vision of\n the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous, with people, even a bishop,\n begging for his blessing. <eod> </s> <eos>\n "
def __init__( self , *_a , **_a ) -> List[str]:
super().__init__(*_a , **_a )
self.check_model_type(
TF_MODEL_FOR_CAUSAL_LM_MAPPING if self.framework == '''tf''' else MODEL_FOR_CAUSAL_LM_MAPPING )
if "prefix" not in self._preprocess_params:
# This is very specific. The logic is quite complex and needs to be done
# as a "default".
# It also defines both some preprocess_kwargs and generate_kwargs
# which is why we cannot put them in their respective methods.
_a : Dict = None
if self.model.config.prefix is not None:
_a : List[Any] = self.model.config.prefix
if prefix is None and self.model.__class__.__name__ in [
"XLNetLMHeadModel",
"TransfoXLLMHeadModel",
"TFXLNetLMHeadModel",
"TFTransfoXLLMHeadModel",
]:
# For XLNet and TransformerXL we add an article to the prompt to give more state to the model.
_a : Optional[Any] = self.XL_PREFIX
if prefix is not None:
# Recalculate some generate_kwargs linked to prefix.
_a , _a , _a : str = self._sanitize_parameters(prefix=_a , **self._forward_params )
_a : Optional[Any] = {**self._preprocess_params, **preprocess_params}
_a : List[Any] = {**self._forward_params, **forward_params}
def __lowercase ( self , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , **_a , ) -> Optional[int]:
_a : List[Any] = {}
if prefix is not None:
_a : Optional[Any] = prefix
if prefix:
_a : Dict = self.tokenizer(
_a , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Tuple = prefix_inputs['''input_ids'''].shape[-1]
if handle_long_generation is not None:
if handle_long_generation not in {"hole"}:
raise ValueError(
F"""{handle_long_generation} is not a valid value for `handle_long_generation` parameter expected"""
''' [None, \'hole\']''' )
_a : Dict = handle_long_generation
preprocess_params.update(_a )
_a : Tuple = generate_kwargs
_a : Any = {}
if return_full_text is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_full_text`''' )
if return_tensors is not None:
raise ValueError('''`return_full_text` is mutually exclusive with `return_tensors`''' )
_a : List[str] = ReturnType.FULL_TEXT if return_full_text else ReturnType.NEW_TEXT
if return_tensors is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_tensors`''' )
_a : Any = ReturnType.TENSORS
if return_type is not None:
_a : Any = return_type
if clean_up_tokenization_spaces is not None:
_a : List[Any] = clean_up_tokenization_spaces
if stop_sequence is not None:
_a : Tuple = self.tokenizer.encode(_a , add_special_tokens=_a )
if len(_a ) > 1:
warnings.warn(
'''Stopping on a multiple token sequence is not yet supported on transformers. The first token of'''
''' the stop sequence will be used as the stop sequence string in the interim.''' )
_a : List[Any] = stop_sequence_ids[0]
return preprocess_params, forward_params, postprocess_params
def __lowercase ( self , *_a , **_a ) -> Union[str, Any]:
# Parse arguments
if self.model.__class__.__name__ in ["TransfoXLLMHeadModel"]:
kwargs.update({'''add_space_before_punct_symbol''': True} )
return super()._parse_and_tokenize(*_a , **_a )
def __call__( self , _a , **_a ) -> List[str]:
return super().__call__(_a , **_a )
def __lowercase ( self , _a , _a="" , _a=None , **_a ) -> List[Any]:
_a : Optional[int] = self.tokenizer(
prefix + prompt_text , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Union[str, Any] = prompt_text
if handle_long_generation == "hole":
_a : List[str] = inputs['''input_ids'''].shape[-1]
if "max_new_tokens" in generate_kwargs:
_a : int = generate_kwargs['''max_new_tokens''']
else:
_a : List[Any] = generate_kwargs.get('''max_length''' , self.model.config.max_length ) - cur_len
if new_tokens < 0:
raise ValueError('''We cannot infer how many new tokens are expected''' )
if cur_len + new_tokens > self.tokenizer.model_max_length:
_a : List[str] = self.tokenizer.model_max_length - new_tokens
if keep_length <= 0:
raise ValueError(
'''We cannot use `hole` to handle this generation the number of desired tokens exceeds the'''
''' models max length''' )
_a : List[Any] = inputs['''input_ids'''][:, -keep_length:]
if "attention_mask" in inputs:
_a : List[str] = inputs['''attention_mask'''][:, -keep_length:]
return inputs
def __lowercase ( self , _a , **_a ) -> Optional[int]:
_a : Any = model_inputs['''input_ids''']
_a : Optional[Any] = model_inputs.get('''attention_mask''' , _a )
# Allow empty prompts
if input_ids.shape[1] == 0:
_a : int = None
_a : int = None
_a : List[str] = 1
else:
_a : List[Any] = input_ids.shape[0]
_a : Union[str, Any] = model_inputs.pop('''prompt_text''' )
# If there is a prefix, we may need to adjust the generation length. Do so without permanently modifying
# generate_kwargs, as some of the parameterization may come from the initialization of the pipeline.
_a : int = generate_kwargs.pop('''prefix_length''' , 0 )
if prefix_length > 0:
_a : Tuple = '''max_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].max_new_tokens is not None
)
if not has_max_new_tokens:
_a : int = generate_kwargs.get('''max_length''' ) or self.model.config.max_length
generate_kwargs["max_length"] += prefix_length
_a : Dict = '''min_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].min_new_tokens is not None
)
if not has_min_new_tokens and "min_length" in generate_kwargs:
generate_kwargs["min_length"] += prefix_length
# BS x SL
_a : Optional[Any] = self.model.generate(input_ids=_a , attention_mask=_a , **_a )
_a : int = generated_sequence.shape[0]
if self.framework == "pt":
_a : Tuple = generated_sequence.reshape(_a , out_b // in_b , *generated_sequence.shape[1:] )
elif self.framework == "tf":
_a : List[Any] = tf.reshape(_a , (in_b, out_b // in_b, *generated_sequence.shape[1:]) )
return {"generated_sequence": generated_sequence, "input_ids": input_ids, "prompt_text": prompt_text}
def __lowercase ( self , _a , _a=ReturnType.FULL_TEXT , _a=True ) -> int:
_a : Tuple = model_outputs['''generated_sequence'''][0]
_a : int = model_outputs['''input_ids''']
_a : Any = model_outputs['''prompt_text''']
_a : Any = generated_sequence.numpy().tolist()
_a : Any = []
for sequence in generated_sequence:
if return_type == ReturnType.TENSORS:
_a : Optional[int] = {'''generated_token_ids''': sequence}
elif return_type in {ReturnType.NEW_TEXT, ReturnType.FULL_TEXT}:
# Decode text
_a : str = self.tokenizer.decode(
_a , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , )
# Remove PADDING prompt of the sequence if XLNet or Transfo-XL model is used
if input_ids is None:
_a : Union[str, Any] = 0
else:
_a : str = len(
self.tokenizer.decode(
input_ids[0] , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , ) )
if return_type == ReturnType.FULL_TEXT:
_a : str = prompt_text + text[prompt_length:]
else:
_a : List[str] = text[prompt_length:]
_a : Union[str, Any] = {'''generated_text''': all_text}
records.append(_a )
return records
| 15
| 0
|
def __UpperCAmelCase ( __a : str ,__a : str ) -> List[Any]:
"""simple docstring"""
def get_matched_characters(__a : str ,__a : str ) -> str:
_a : Optional[int] = []
_a : List[str] = min(len(_stra ) ,len(_stra ) ) // 2
for i, l in enumerate(_stra ):
_a : Dict = int(max(0 ,i - limit ) )
_a : List[Any] = int(min(i + limit + 1 ,len(_stra ) ) )
if l in _stra[left:right]:
matched.append(snake_case_ )
_a : List[Any] = F"""{_stra[0:_stra.index(snake_case_ )]} {_stra[_stra.index(snake_case_ ) + 1:]}"""
return "".join(snake_case_ )
# matching characters
_a : Optional[Any] = get_matched_characters(snake_case_ ,snake_case_ )
_a : int = get_matched_characters(snake_case_ ,snake_case_ )
_a : str = len(snake_case_ )
# transposition
_a : List[str] = (
len([(ca, ca) for ca, ca in zip(snake_case_ ,snake_case_ ) if ca != ca] ) // 2
)
if not match_count:
_a : List[str] = 0.0
else:
_a : Tuple = (
1
/ 3
* (
match_count / len(snake_case_ )
+ match_count / len(snake_case_ )
+ (match_count - transpositions) / match_count
)
)
# common prefix up to 4 characters
_a : List[Any] = 0
for ca, ca in zip(stra[:4] ,stra[:4] ):
if ca == ca:
prefix_len += 1
else:
break
return jaro + 0.1 * prefix_len * (1 - jaro)
if __name__ == "__main__":
import doctest
doctest.testmod()
print(jaro_winkler('''hello''', '''world'''))
| 369
|
# 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
import os
from accelerate.test_utils import execute_subprocess_async
def __UpperCAmelCase ( __a : Dict=None ) -> str:
"""simple docstring"""
if subparsers is not None:
_a : Union[str, Any] = subparsers.add_parser('''test''' )
else:
_a : List[str] = argparse.ArgumentParser('''Accelerate test command''' )
parser.add_argument(
'''--config_file''' ,default=__a ,help=(
'''The path to use to store the config file. Will default to a file named default_config.yaml in the cache '''
'''location, which is the content of the environment `HF_HOME` suffixed with \'accelerate\', or if you don\'t have '''
'''such an environment variable, your cache directory (\'~/.cache\' or the content of `XDG_CACHE_HOME`) suffixed '''
'''with \'huggingface\'.'''
) ,)
if subparsers is not None:
parser.set_defaults(func=__a )
return parser
def __UpperCAmelCase ( __a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
_a : Dict = os.path.sep.join(__file__.split(os.path.sep )[:-2] + ['''test_utils''', '''scripts''', '''test_script.py'''] )
if args.config_file is None:
_a : List[Any] = script_name
else:
_a : Union[str, Any] = F"""--config_file={args.config_file} {script_name}"""
_a : str = ['''accelerate-launch'''] + test_args.split()
_a : str = execute_subprocess_async(__a ,env=os.environ.copy() )
if result.returncode == 0:
print('''Test is a success! You are ready for your distributed training!''' )
def __UpperCAmelCase ( ) -> List[Any]:
"""simple docstring"""
_a : Optional[int] = test_command_parser()
_a : List[Any] = parser.parse_args()
test_command(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
# flake8: noqa
# Lint as: python3
a__ = [
'''VerificationMode''',
'''Version''',
'''disable_progress_bar''',
'''enable_progress_bar''',
'''is_progress_bar_enabled''',
'''experimental''',
]
from .info_utils import VerificationMode
from .logging import disable_progress_bar, enable_progress_bar, is_progress_bar_enabled
from .version import Version
from .experimental import experimental
| 370
|
import json
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import BertTokenizerFast
from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer
from transformers.testing_utils import require_tokenizers, require_vision
from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available
if is_vision_available():
from PIL import Image
from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor
@require_tokenizers
@require_vision
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = tempfile.mkdtemp()
# fmt: off
_a : Optional[int] = ['''[UNK]''', '''[CLS]''', '''[SEP]''', '''[PAD]''', '''[MASK]''', '''want''', '''##want''', '''##ed''', '''wa''', '''un''', '''runn''', '''##ing''', ''',''', '''low''', '''lowest''']
# fmt: on
_a : Tuple = 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] ) )
_a : Any = {
'''do_resize''': True,
'''size''': {'''height''': 1_8, '''width''': 1_8},
'''do_normalize''': True,
'''image_mean''': [0.5, 0.5, 0.5],
'''image_std''': [0.5, 0.5, 0.5],
}
_a : str = os.path.join(self.tmpdirname , _a )
with open(self.image_processor_file , '''w''' , encoding='''utf-8''' ) as fp:
json.dump(_a , _a )
def __lowercase ( self , **_a ) -> Any:
return BertTokenizer.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self , **_a ) -> str:
return ViTImageProcessor.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self ) -> List[Any]:
shutil.rmtree(self.tmpdirname )
def __lowercase ( self ) -> Any:
_a : Union[str, Any] = [np.random.randint(2_5_5 , size=(3, 3_0, 4_0_0) , dtype=np.uinta )]
_a : Tuple = [Image.fromarray(np.moveaxis(_a , 0 , -1 ) ) for x in image_inputs]
return image_inputs
def __lowercase ( self ) -> str:
_a : List[str] = self.get_tokenizer()
_a : Tuple = self.get_image_processor()
_a : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
processor.save_pretrained(self.tmpdirname )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Dict:
_a : List[str] = VisionTextDualEncoderProcessor(
tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() )
processor.save_pretrained(self.tmpdirname )
_a : Any = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' )
_a : List[Any] = self.get_image_processor(do_normalize=_a , padding_value=1.0 )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(
self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=_a , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Any:
_a : Dict = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : int = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[str] = self.prepare_image_inputs()
_a : List[Any] = image_processor(_a , return_tensors='''np''' )
_a : Dict = processor(images=_a , return_tensors='''np''' )
for key in input_feat_extract.keys():
self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1e-2 )
def __lowercase ( self ) -> List[str]:
_a : Union[str, Any] = self.get_image_processor()
_a : Dict = self.get_tokenizer()
_a : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Tuple = '''lower newer'''
_a : int = processor(text=_a )
_a : str = tokenizer(_a )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def __lowercase ( self ) -> List[Any]:
_a : Any = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : Tuple = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[Any] = '''lower newer'''
_a : Union[str, Any] = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , ['''input_ids''', '''token_type_ids''', '''attention_mask''', '''pixel_values'''] )
# test if it raises when no input is passed
with self.assertRaises(_a ):
processor()
def __lowercase ( self ) -> Optional[int]:
_a : Union[str, Any] = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : Any = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Any = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
_a : int = processor.batch_decode(_a )
_a : int = tokenizer.batch_decode(_a )
self.assertListEqual(_a , _a )
def __lowercase ( self ) -> List[Any]:
_a : Tuple = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : str = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Optional[int] = '''lower newer'''
_a : Dict = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
| 15
| 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
a__ = datasets.utils.logging.get_logger(__name__)
a__ = ['''names''', '''prefix''']
a__ = ['''warn_bad_lines''', '''error_bad_lines''', '''mangle_dupe_cols''']
a__ = ['''encoding_errors''', '''on_bad_lines''']
a__ = ['''date_format''']
@dataclass
class UpperCAmelCase_ ( datasets.BuilderConfig ):
"""simple docstring"""
UpperCAmelCase__ : str = ","
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : Optional[Union[int, List[int], str]] = "infer"
UpperCAmelCase__ : Optional[List[str]] = None
UpperCAmelCase__ : Optional[List[str]] = None
UpperCAmelCase__ : Optional[Union[int, str, List[int], List[str]]] = None
UpperCAmelCase__ : Optional[Union[List[int], List[str]]] = None
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : bool = True
UpperCAmelCase__ : Optional[Literal["c", "python", "pyarrow"]] = None
UpperCAmelCase__ : Dict[Union[int, str], Callable[[Any], Any]] = None
UpperCAmelCase__ : Optional[list] = None
UpperCAmelCase__ : Optional[list] = None
UpperCAmelCase__ : bool = False
UpperCAmelCase__ : Optional[Union[int, List[int]]] = None
UpperCAmelCase__ : Optional[int] = None
UpperCAmelCase__ : Optional[Union[str, List[str]]] = None
UpperCAmelCase__ : bool = True
UpperCAmelCase__ : bool = True
UpperCAmelCase__ : bool = False
UpperCAmelCase__ : bool = True
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : str = "."
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : str = '"'
UpperCAmelCase__ : int = 0
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : bool = True
UpperCAmelCase__ : bool = True
UpperCAmelCase__ : int = 0
UpperCAmelCase__ : bool = True
UpperCAmelCase__ : bool = False
UpperCAmelCase__ : Optional[str] = None
UpperCAmelCase__ : int = 10000
UpperCAmelCase__ : Optional[datasets.Features] = None
UpperCAmelCase__ : Optional[str] = "strict"
UpperCAmelCase__ : Literal["error", "warn", "skip"] = "error"
UpperCAmelCase__ : Optional[str] = None
def __lowercase ( self ) -> Any:
if self.delimiter is not None:
_a : int = self.delimiter
if self.column_names is not None:
_a : List[str] = self.column_names
@property
def __lowercase ( self ) -> Optional[Any]:
_a : Any = {
'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() , _A ):
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 UpperCAmelCase_ ( datasets.ArrowBasedBuilder ):
"""simple docstring"""
UpperCAmelCase__ : Tuple = CsvConfig
def __lowercase ( self ) -> str:
return datasets.DatasetInfo(features=self.config.features )
def __lowercase ( self , _a ) -> str:
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}""" )
_a : Union[str, Any] = dl_manager.download_and_extract(self.config.data_files )
if isinstance(_A , (str, list, tuple) ):
_a : Tuple = data_files
if isinstance(_A , _A ):
_a : int = [files]
_a : Union[str, Any] = [dl_manager.iter_files(_A ) for file in files]
return [datasets.SplitGenerator(name=datasets.Split.TRAIN , gen_kwargs={'''files''': files} )]
_a : List[str] = []
for split_name, files in data_files.items():
if isinstance(_A , _A ):
_a : Optional[Any] = [files]
_a : int = [dl_manager.iter_files(_A ) for file in files]
splits.append(datasets.SplitGenerator(name=_A , gen_kwargs={'''files''': files} ) )
return splits
def __lowercase ( self , _a ) -> pa.Table:
if self.config.features is not None:
_a : List[str] = self.config.features.arrow_schema
if all(not require_storage_cast(_A ) for feature in self.config.features.values() ):
# cheaper cast
_a : List[Any] = pa.Table.from_arrays([pa_table[field.name] for field in schema] , schema=_A )
else:
# more expensive cast; allows str <-> int/float or str to Audio for example
_a : str = table_cast(_A , _A )
return pa_table
def __lowercase ( self , _a ) -> List[str]:
_a : List[Any] = self.config.features.arrow_schema if self.config.features else None
# dtype allows reading an int column as str
_a : Any = (
{
name: dtype.to_pandas_dtype() if not require_storage_cast(_A ) 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(_A ) ):
_a : List[str] = pd.read_csv(_A , iterator=_A , dtype=_A , **self.config.pd_read_csv_kwargs )
try:
for batch_idx, df in enumerate(_A ):
_a : str = pa.Table.from_pandas(_A )
# 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(_A )
except ValueError as e:
logger.error(F"""Failed to read file '{file}' with error {type(_A )}: {e}""" )
raise
| 371
|
import argparse
import json
import os
import fairseq
import torch
from fairseq.data import Dictionary
from transformers import (
HubertConfig,
HubertForCTC,
HubertModel,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaProcessor,
logging,
)
logging.set_verbosity_info()
a__ = logging.get_logger(__name__)
a__ = {
'''post_extract_proj''': '''feature_projection.projection''',
'''encoder.pos_conv.0''': '''encoder.pos_conv_embed.conv''',
'''self_attn.k_proj''': '''encoder.layers.*.attention.k_proj''',
'''self_attn.v_proj''': '''encoder.layers.*.attention.v_proj''',
'''self_attn.q_proj''': '''encoder.layers.*.attention.q_proj''',
'''self_attn.out_proj''': '''encoder.layers.*.attention.out_proj''',
'''self_attn_layer_norm''': '''encoder.layers.*.layer_norm''',
'''fc1''': '''encoder.layers.*.feed_forward.intermediate_dense''',
'''fc2''': '''encoder.layers.*.feed_forward.output_dense''',
'''final_layer_norm''': '''encoder.layers.*.final_layer_norm''',
'''encoder.layer_norm''': '''encoder.layer_norm''',
'''w2v_model.layer_norm''': '''feature_projection.layer_norm''',
'''w2v_encoder.proj''': '''lm_head''',
'''mask_emb''': '''masked_spec_embed''',
}
def __UpperCAmelCase ( __a : List[Any] ,__a : Optional[int] ,__a : Optional[int] ,__a : List[str] ,__a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
for attribute in key.split('''.''' ):
_a : Optional[Any] = getattr(__a ,__a )
if weight_type is not None:
_a : Dict = getattr(__a ,__a ).shape
else:
_a : Optional[int] = 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":
_a : List[Any] = value
elif weight_type == "weight_g":
_a : Any = value
elif weight_type == "weight_v":
_a : Union[str, Any] = value
elif weight_type == "bias":
_a : Optional[int] = value
else:
_a : List[Any] = value
logger.info(F"""{key + '.' + weight_type if weight_type is not None else ''} was initialized from {full_name}.""" )
def __UpperCAmelCase ( __a : Any ,__a : Union[str, Any] ,__a : Union[str, Any] ) -> int:
"""simple docstring"""
_a : Union[str, Any] = []
_a : Union[str, Any] = fairseq_model.state_dict()
_a : Union[str, Any] = hf_model.hubert.feature_extractor if is_finetuned else hf_model.feature_extractor
for name, value in fairseq_dict.items():
_a : int = False
if "conv_layers" in name:
load_conv_layer(
__a ,__a ,__a ,__a ,hf_model.config.feat_extract_norm == '''group''' ,)
_a : Optional[Any] = True
else:
for key, mapped_key in MAPPING.items():
_a : Union[str, Any] = '''hubert.''' + 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] and not is_finetuned):
_a : Any = True
if "*" in mapped_key:
_a : Optional[int] = name.split(__a )[0].split('''.''' )[-2]
_a : Any = mapped_key.replace('''*''' ,__a )
if "weight_g" in name:
_a : List[Any] = '''weight_g'''
elif "weight_v" in name:
_a : List[str] = '''weight_v'''
elif "weight" in name:
_a : Any = '''weight'''
elif "bias" in name:
_a : str = '''bias'''
else:
_a : Any = None
set_recursively(__a ,__a ,__a ,__a ,__a )
continue
if not is_used:
unused_weights.append(__a )
logger.warning(F"""Unused weights: {unused_weights}""" )
def __UpperCAmelCase ( __a : int ,__a : Optional[Any] ,__a : Dict ,__a : List[str] ,__a : Any ) -> Tuple:
"""simple docstring"""
_a : int = full_name.split('''conv_layers.''' )[-1]
_a : Any = name.split('''.''' )
_a : List[Any] = int(items[0] )
_a : Optional[int] = 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."""
)
_a : Optional[int] = 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."""
)
_a : Optional[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:
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."
)
_a : int = 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."""
)
_a : Any = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(__a )
@torch.no_grad()
def __UpperCAmelCase ( __a : Dict ,__a : List[Any] ,__a : List[str]=None ,__a : Optional[int]=None ,__a : int=True ) -> List[Any]:
"""simple docstring"""
if config_path is not None:
_a : Tuple = HubertConfig.from_pretrained(__a )
else:
_a : Any = HubertConfig()
if is_finetuned:
if dict_path:
_a : Tuple = Dictionary.load(__a )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
_a : Any = target_dict.pad_index
_a : Tuple = target_dict.bos_index
_a : Optional[int] = target_dict.eos_index
_a : Optional[Any] = len(target_dict.symbols )
_a : Tuple = os.path.join(__a ,'''vocab.json''' )
if not os.path.isdir(__a ):
logger.error('''--pytorch_dump_folder_path ({}) should be a directory'''.format(__a ) )
return
os.makedirs(__a ,exist_ok=__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ) as vocab_handle:
json.dump(target_dict.indices ,__a )
_a : Tuple = WavaVecaCTCTokenizer(
__a ,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=__a ,)
_a : Tuple = True if config.feat_extract_norm == '''layer''' else False
_a : List[Any] = WavaVecaFeatureExtractor(
feature_size=1 ,sampling_rate=16_000 ,padding_value=0 ,do_normalize=__a ,return_attention_mask=__a ,)
_a : List[Any] = WavaVecaProcessor(feature_extractor=__a ,tokenizer=__a )
processor.save_pretrained(__a )
_a : Tuple = HubertForCTC(__a )
else:
_a : Tuple = HubertModel(__a )
if is_finetuned:
_a , _a , _a : int = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
_a , _a , _a : str = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] )
_a : Any = model[0].eval()
recursively_load_weights(__a ,__a ,__a )
hf_wavavec.save_pretrained(__a )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
parser.add_argument('''--checkpoint_path''', default=None, type=str, help='''Path to fairseq checkpoint''')
parser.add_argument('''--dict_path''', default=None, type=str, help='''Path to dict of fine-tuned model''')
parser.add_argument('''--config_path''', default=None, type=str, help='''Path to hf config.json of model to convert''')
parser.add_argument(
'''--not_finetuned''', action='''store_true''', help='''Whether the model to convert is a fine-tuned model or not'''
)
a__ = parser.parse_args()
convert_hubert_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 15
| 0
|
from math import acos, sin
from typing import List, Tuple, Union
import numpy as np
import torch
from PIL import Image
from ...models import AutoencoderKL, UNetaDConditionModel
from ...schedulers import DDIMScheduler, DDPMScheduler
from ...utils import randn_tensor
from ..pipeline_utils import AudioPipelineOutput, BaseOutput, DiffusionPipeline, ImagePipelineOutput
from .mel import Mel
class UpperCAmelCase_ ( _SCREAMING_SNAKE_CASE ):
"""simple docstring"""
UpperCAmelCase__ : Dict = ["vqvae"]
def __init__( self , _a , _a , _a , _a , ) -> Union[str, Any]:
super().__init__()
self.register_modules(unet=_a , scheduler=_a , mel=_a , vqvae=_a )
def __lowercase ( self ) -> int:
return 5_0 if isinstance(self.scheduler , _a ) else 1_0_0_0
@torch.no_grad()
def __call__( self , _a = 1 , _a = None , _a = None , _a = 0 , _a = 0 , _a = None , _a = None , _a = 0 , _a = 0 , _a = None , _a = 0 , _a = None , _a = None , _a=True , ) -> Union[
Union[AudioPipelineOutput, ImagePipelineOutput],
Tuple[List[Image.Image], Tuple[int, List[np.ndarray]]],
]:
_a : List[str] = steps or self.get_default_steps()
self.scheduler.set_timesteps(_a )
_a : str = step_generator or generator
# For backwards compatibility
if type(self.unet.config.sample_size ) == int:
_a : List[Any] = (self.unet.config.sample_size, self.unet.config.sample_size)
if noise is None:
_a : List[str] = randn_tensor(
(
batch_size,
self.unet.config.in_channels,
self.unet.config.sample_size[0],
self.unet.config.sample_size[1],
) , generator=_a , device=self.device , )
_a : Optional[int] = noise
_a : int = None
if audio_file is not None or raw_audio is not None:
self.mel.load_audio(_a , _a )
_a : int = self.mel.audio_slice_to_image(_a )
_a : List[Any] = np.frombuffer(input_image.tobytes() , dtype='''uint8''' ).reshape(
(input_image.height, input_image.width) )
_a : List[str] = (input_image / 2_5_5) * 2 - 1
_a : Tuple = torch.tensor(input_image[np.newaxis, :, :] , dtype=torch.float ).to(self.device )
if self.vqvae is not None:
_a : Optional[Any] = self.vqvae.encode(torch.unsqueeze(_a , 0 ) ).latent_dist.sample(
generator=_a )[0]
_a : Tuple = self.vqvae.config.scaling_factor * input_images
if start_step > 0:
_a : str = self.scheduler.add_noise(_a , _a , self.scheduler.timesteps[start_step - 1] )
_a : Optional[Any] = (
self.unet.config.sample_size[1] * self.mel.get_sample_rate() / self.mel.x_res / self.mel.hop_length
)
_a : Optional[int] = int(mask_start_secs * pixels_per_second )
_a : Optional[Any] = int(mask_end_secs * pixels_per_second )
_a : Tuple = self.scheduler.add_noise(_a , _a , torch.tensor(self.scheduler.timesteps[start_step:] ) )
for step, t in enumerate(self.progress_bar(self.scheduler.timesteps[start_step:] ) ):
if isinstance(self.unet , _a ):
_a : List[str] = self.unet(_a , _a , _a )["""sample"""]
else:
_a : Dict = self.unet(_a , _a )["""sample"""]
if isinstance(self.scheduler , _a ):
_a : List[Any] = self.scheduler.step(
model_output=_a , timestep=_a , sample=_a , eta=_a , generator=_a , )["""prev_sample"""]
else:
_a : List[str] = self.scheduler.step(
model_output=_a , timestep=_a , sample=_a , generator=_a , )["""prev_sample"""]
if mask is not None:
if mask_start > 0:
_a : List[str] = mask[:, step, :, :mask_start]
if mask_end > 0:
_a : Optional[int] = mask[:, step, :, -mask_end:]
if self.vqvae is not None:
# 0.18215 was scaling factor used in training to ensure unit variance
_a : List[str] = 1 / self.vqvae.config.scaling_factor * images
_a : Optional[int] = self.vqvae.decode(_a )["""sample"""]
_a : Optional[int] = (images / 2 + 0.5).clamp(0 , 1 )
_a : Any = images.cpu().permute(0 , 2 , 3 , 1 ).numpy()
_a : Optional[Any] = (images * 2_5_5).round().astype('''uint8''' )
_a : Optional[Any] = list(
(Image.fromarray(_[:, :, 0] ) for _ in images)
if images.shape[3] == 1
else (Image.fromarray(_a , mode='''RGB''' ).convert('''L''' ) for _ in images) )
_a : List[str] = [self.mel.image_to_audio(_a ) for _ in images]
if not return_dict:
return images, (self.mel.get_sample_rate(), audios)
return BaseOutput(**AudioPipelineOutput(np.array(_a )[:, np.newaxis, :] ) , **ImagePipelineOutput(_a ) )
@torch.no_grad()
def __lowercase ( self , _a , _a = 5_0 ) -> np.ndarray:
assert isinstance(self.scheduler , _a )
self.scheduler.set_timesteps(_a )
_a : str = np.array(
[np.frombuffer(image.tobytes() , dtype='''uint8''' ).reshape((1, image.height, image.width) ) for image in images] )
_a : List[Any] = (sample / 2_5_5) * 2 - 1
_a : Any = torch.Tensor(_a ).to(self.device )
for t in self.progress_bar(torch.flip(self.scheduler.timesteps , (0,) ) ):
_a : Optional[Any] = t - self.scheduler.config.num_train_timesteps // self.scheduler.num_inference_steps
_a : Optional[Any] = self.scheduler.alphas_cumprod[t]
_a : Union[str, Any] = (
self.scheduler.alphas_cumprod[prev_timestep]
if prev_timestep >= 0
else self.scheduler.final_alpha_cumprod
)
_a : int = 1 - alpha_prod_t
_a : List[str] = self.unet(_a , _a )["""sample"""]
_a : List[str] = (1 - alpha_prod_t_prev) ** 0.5 * model_output
_a : Dict = (sample - pred_sample_direction) * alpha_prod_t_prev ** (-0.5)
_a : Optional[int] = sample * alpha_prod_t ** 0.5 + beta_prod_t ** 0.5 * model_output
return sample
@staticmethod
def __lowercase ( _a , _a , _a ) -> torch.Tensor:
_a : Union[str, Any] = acos(torch.dot(torch.flatten(_a ) , torch.flatten(_a ) ) / torch.norm(_a ) / torch.norm(_a ) )
return sin((1 - alpha) * theta ) * xa / sin(_a ) + sin(alpha * theta ) * xa / sin(_a )
| 350
|
import warnings
from typing import List, Optional, Union
from ...processing_utils import ProcessorMixin
from ...tokenization_utils_base import BatchEncoding, PaddingStrategy, PreTokenizedInput, TextInput, TruncationStrategy
from ...utils import TensorType
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[str] = ["image_processor", "tokenizer"]
UpperCAmelCase__ : str = "ViltImageProcessor"
UpperCAmelCase__ : Union[str, Any] = ("BertTokenizer", "BertTokenizerFast")
def __init__( self , _a=None , _a=None , **_a ) -> Any:
_a : Union[str, Any] = None
if "feature_extractor" in kwargs:
warnings.warn(
'''The `feature_extractor` argument is deprecated and will be removed in v5, use `image_processor`'''
''' instead.''' , _a , )
_a : Dict = kwargs.pop('''feature_extractor''' )
_a : Optional[int] = image_processor if image_processor is not None else feature_extractor
if image_processor is None:
raise ValueError('''You need to specify an `image_processor`.''' )
if tokenizer is None:
raise ValueError('''You need to specify a `tokenizer`.''' )
super().__init__(_a , _a )
_a : int = self.image_processor
def __call__( self , _a , _a = None , _a = True , _a = False , _a = None , _a = None , _a = 0 , _a = None , _a = None , _a = None , _a = False , _a = False , _a = False , _a = False , _a = True , _a = None , **_a , ) -> BatchEncoding:
_a : Tuple = self.tokenizer(
text=_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , stride=_a , pad_to_multiple_of=_a , return_token_type_ids=_a , return_attention_mask=_a , return_overflowing_tokens=_a , return_special_tokens_mask=_a , return_offsets_mapping=_a , return_length=_a , verbose=_a , return_tensors=_a , **_a , )
# add pixel_values + pixel_mask
_a : str = self.image_processor(_a , return_tensors=_a )
encoding.update(_a )
return encoding
def __lowercase ( self , *_a , **_a ) -> Optional[Any]:
return self.tokenizer.batch_decode(*_a , **_a )
def __lowercase ( self , *_a , **_a ) -> str:
return self.tokenizer.decode(*_a , **_a )
@property
def __lowercase ( self ) -> Optional[int]:
_a : str = self.tokenizer.model_input_names
_a : Optional[Any] = self.image_processor.model_input_names
return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names ) )
@property
def __lowercase ( self ) -> Optional[Any]:
warnings.warn(
'''`feature_extractor_class` is deprecated and will be removed in v5. Use `image_processor_class` instead.''' , _a , )
return self.image_processor_class
@property
def __lowercase ( self ) -> Any:
warnings.warn(
'''`feature_extractor` is deprecated and will be removed in v5. Use `image_processor` instead.''' , _a , )
return self.image_processor
| 15
| 0
|
import importlib
import inspect
import json
import os
import re
import shutil
import sys
from pathlib import Path
from typing import Dict, Optional, Union
from urllib import request
from huggingface_hub import HfFolder, cached_download, hf_hub_download, model_info
from packaging import version
from .. import __version__
from . import DIFFUSERS_DYNAMIC_MODULE_NAME, HF_MODULES_CACHE, logging
a__ = (
'''https://raw.githubusercontent.com/huggingface/diffusers/{revision}/examples/community/{pipeline}.py'''
)
a__ = logging.get_logger(__name__) # pylint: disable=invalid-name
def __UpperCAmelCase ( ) -> Dict:
"""simple docstring"""
_a : Dict = '''https://pypi.org/pypi/diffusers/json'''
_a : Union[str, Any] = json.loads(request.urlopen(lowerCAmelCase__ ).read() )['''releases'''].keys()
return sorted(lowerCAmelCase__ ,key=lambda __a : version.Version(lowerCAmelCase__ ) )
def __UpperCAmelCase ( ) -> str:
"""simple docstring"""
if HF_MODULES_CACHE in sys.path:
return
sys.path.append(lowerCAmelCase__ )
os.makedirs(lowerCAmelCase__ ,exist_ok=lowerCAmelCase__ )
_a : Optional[Any] = Path(lowerCAmelCase__ ) / '''__init__.py'''
if not init_path.exists():
init_path.touch()
def __UpperCAmelCase ( __a : Union[str, os.PathLike] ) -> int:
"""simple docstring"""
init_hf_modules()
_a : str = Path(lowerCAmelCase__ ) / name
# If the parent module does not exist yet, recursively create it.
if not dynamic_module_path.parent.exists():
create_dynamic_module(dynamic_module_path.parent )
os.makedirs(lowerCAmelCase__ ,exist_ok=lowerCAmelCase__ )
_a : List[Any] = dynamic_module_path / '''__init__.py'''
if not init_path.exists():
init_path.touch()
def __UpperCAmelCase ( __a : List[Any] ) -> str:
"""simple docstring"""
with open(lowerCAmelCase__ ,'''r''' ,encoding='''utf-8''' ) as f:
_a : Any = f.read()
# Imports of the form `import .xxx`
_a : Any = re.findall('''^\s*import\s+\.(\S+)\s*$''' ,lowerCAmelCase__ ,flags=re.MULTILINE )
# Imports of the form `from .xxx import yyy`
relative_imports += re.findall('''^\s*from\s+\.(\S+)\s+import''' ,lowerCAmelCase__ ,flags=re.MULTILINE )
# Unique-ify
return list(set(lowerCAmelCase__ ) )
def __UpperCAmelCase ( __a : Dict ) -> Any:
"""simple docstring"""
_a : List[str] = False
_a : str = [module_file]
_a : List[str] = []
# Let's recurse through all relative imports
while not no_change:
_a : Dict = []
for f in files_to_check:
new_imports.extend(get_relative_imports(lowerCAmelCase__ ) )
_a : str = Path(lowerCAmelCase__ ).parent
_a : Dict = [str(module_path / m ) for m in new_imports]
_a : Optional[Any] = [f for f in new_import_files if f not in all_relative_imports]
_a : Dict = [F"""{f}.py""" for f in new_import_files]
_a : List[Any] = len(lowerCAmelCase__ ) == 0
all_relative_imports.extend(lowerCAmelCase__ )
return all_relative_imports
def __UpperCAmelCase ( __a : Dict ) -> int:
"""simple docstring"""
with open(lowerCAmelCase__ ,'''r''' ,encoding='''utf-8''' ) as f:
_a : Union[str, Any] = f.read()
# Imports of the form `import xxx`
_a : Optional[Any] = re.findall('''^\s*import\s+(\S+)\s*$''' ,lowerCAmelCase__ ,flags=re.MULTILINE )
# Imports of the form `from xxx import yyy`
imports += re.findall('''^\s*from\s+(\S+)\s+import''' ,lowerCAmelCase__ ,flags=re.MULTILINE )
# Only keep the top-level module
_a : Optional[Any] = [imp.split('''.''' )[0] for imp in imports if not imp.startswith('''.''' )]
# Unique-ify and test we got them all
_a : Union[str, Any] = list(set(lowerCAmelCase__ ) )
_a : List[str] = []
for imp in imports:
try:
importlib.import_module(lowerCAmelCase__ )
except ImportError:
missing_packages.append(lowerCAmelCase__ )
if len(lowerCAmelCase__ ) > 0:
raise ImportError(
'''This modeling file requires the following packages that were not found in your environment: '''
F"""{', '.join(lowerCAmelCase__ )}. Run `pip install {' '.join(lowerCAmelCase__ )}`""" )
return get_relative_imports(lowerCAmelCase__ )
def __UpperCAmelCase ( __a : int ,__a : Dict ) -> List[Any]:
"""simple docstring"""
_a : List[str] = module_path.replace(os.path.sep ,'''.''' )
_a : Optional[int] = importlib.import_module(lowerCAmelCase__ )
if class_name is None:
return find_pipeline_class(lowerCAmelCase__ )
return getattr(lowerCAmelCase__ ,lowerCAmelCase__ )
def __UpperCAmelCase ( __a : str ) -> Union[str, Any]:
"""simple docstring"""
from ..pipelines import DiffusionPipeline
_a : Tuple = dict(inspect.getmembers(lowerCAmelCase__ ,inspect.isclass ) )
_a : Optional[Any] = None
for cls_name, cls in cls_members.items():
if (
cls_name != DiffusionPipeline.__name__
and issubclass(cls ,lowerCAmelCase__ )
and cls.__module__.split('''.''' )[0] != "diffusers"
):
if pipeline_class is not None:
raise ValueError(
F"""Multiple classes that inherit from {DiffusionPipeline.__name__} have been found:"""
F""" {pipeline_class.__name__}, and {cls_name}. Please make sure to define only one in"""
F""" {loaded_module}.""" )
_a : Any = cls
return pipeline_class
def __UpperCAmelCase ( __a : Union[str, os.PathLike] ,__a : str ,__a : Optional[Union[str, os.PathLike]] = None ,__a : bool = False ,__a : bool = False ,__a : Optional[Dict[str, str]] = None ,__a : Optional[Union[bool, str]] = None ,__a : Optional[str] = None ,__a : bool = False ,) -> Tuple:
"""simple docstring"""
_a : List[Any] = str(lowerCAmelCase__ )
_a : List[Any] = os.path.join(lowerCAmelCase__ ,lowerCAmelCase__ )
if os.path.isfile(lowerCAmelCase__ ):
_a : Dict = module_file_or_url
_a : List[Any] = '''local'''
elif pretrained_model_name_or_path.count('''/''' ) == 0:
_a : List[str] = get_diffusers_versions()
# cut ".dev0"
_a : List[Any] = '''v''' + '''.'''.join(__version__.split('''.''' )[:3] )
# retrieve github version that matches
if revision is None:
_a : str = latest_version if latest_version[1:] in available_versions else '''main'''
logger.info(F"""Defaulting to latest_version: {revision}.""" )
elif revision in available_versions:
_a : Union[str, Any] = F"""v{revision}"""
elif revision == "main":
_a : str = revision
else:
raise ValueError(
F"""`custom_revision`: {revision} does not exist. Please make sure to choose one of"""
F""" {', '.join(available_versions + ['main'] )}.""" )
# community pipeline on GitHub
_a : Tuple = COMMUNITY_PIPELINES_URL.format(revision=lowerCAmelCase__ ,pipeline=lowerCAmelCase__ )
try:
_a : Union[str, Any] = cached_download(
lowerCAmelCase__ ,cache_dir=lowerCAmelCase__ ,force_download=lowerCAmelCase__ ,proxies=lowerCAmelCase__ ,resume_download=lowerCAmelCase__ ,local_files_only=lowerCAmelCase__ ,use_auth_token=lowerCAmelCase__ ,)
_a : List[str] = '''git'''
_a : Tuple = pretrained_model_name_or_path + '''.py'''
except EnvironmentError:
logger.error(F"""Could not locate the {module_file} inside {pretrained_model_name_or_path}.""" )
raise
else:
try:
# Load from URL or cache if already cached
_a : Any = hf_hub_download(
lowerCAmelCase__ ,lowerCAmelCase__ ,cache_dir=lowerCAmelCase__ ,force_download=lowerCAmelCase__ ,proxies=lowerCAmelCase__ ,resume_download=lowerCAmelCase__ ,local_files_only=lowerCAmelCase__ ,use_auth_token=lowerCAmelCase__ ,)
_a : Any = os.path.join('''local''' ,'''--'''.join(pretrained_model_name_or_path.split('''/''' ) ) )
except EnvironmentError:
logger.error(F"""Could not locate the {module_file} inside {pretrained_model_name_or_path}.""" )
raise
# Check we have all the requirements in our environment
_a : List[Any] = check_imports(lowerCAmelCase__ )
# Now we move the module inside our cached dynamic modules.
_a : List[Any] = DIFFUSERS_DYNAMIC_MODULE_NAME + os.path.sep + submodule
create_dynamic_module(lowerCAmelCase__ )
_a : List[Any] = Path(lowerCAmelCase__ ) / full_submodule
if submodule == "local" or submodule == "git":
# We always copy local files (we could hash the file to see if there was a change, and give them the name of
# that hash, to only copy when there is a modification but it seems overkill for now).
# The only reason we do the copy is to avoid putting too many folders in sys.path.
shutil.copy(lowerCAmelCase__ ,submodule_path / module_file )
for module_needed in modules_needed:
_a : Tuple = F"""{module_needed}.py"""
shutil.copy(os.path.join(lowerCAmelCase__ ,lowerCAmelCase__ ) ,submodule_path / module_needed )
else:
# Get the commit hash
# TODO: we will get this info in the etag soon, so retrieve it from there and not here.
if isinstance(lowerCAmelCase__ ,lowerCAmelCase__ ):
_a : List[str] = use_auth_token
elif use_auth_token is True:
_a : str = HfFolder.get_token()
else:
_a : Any = None
_a : List[Any] = model_info(lowerCAmelCase__ ,revision=lowerCAmelCase__ ,token=lowerCAmelCase__ ).sha
# The module file will end up being placed in a subfolder with the git hash of the repo. This way we get the
# benefit of versioning.
_a : List[str] = submodule_path / commit_hash
_a : int = full_submodule + os.path.sep + commit_hash
create_dynamic_module(lowerCAmelCase__ )
if not (submodule_path / module_file).exists():
shutil.copy(lowerCAmelCase__ ,submodule_path / module_file )
# Make sure we also have every file with relative
for module_needed in modules_needed:
if not (submodule_path / module_needed).exists():
get_cached_module_file(
lowerCAmelCase__ ,F"""{module_needed}.py""" ,cache_dir=lowerCAmelCase__ ,force_download=lowerCAmelCase__ ,resume_download=lowerCAmelCase__ ,proxies=lowerCAmelCase__ ,use_auth_token=lowerCAmelCase__ ,revision=lowerCAmelCase__ ,local_files_only=lowerCAmelCase__ ,)
return os.path.join(lowerCAmelCase__ ,lowerCAmelCase__ )
def __UpperCAmelCase ( __a : Union[str, os.PathLike] ,__a : str ,__a : Optional[str] = None ,__a : Optional[Union[str, os.PathLike]] = None ,__a : bool = False ,__a : bool = False ,__a : Optional[Dict[str, str]] = None ,__a : Optional[Union[bool, str]] = None ,__a : Optional[str] = None ,__a : bool = False ,**__a : Optional[int] ,) -> int:
"""simple docstring"""
_a : Dict = get_cached_module_file(
lowerCAmelCase__ ,lowerCAmelCase__ ,cache_dir=lowerCAmelCase__ ,force_download=lowerCAmelCase__ ,resume_download=lowerCAmelCase__ ,proxies=lowerCAmelCase__ ,use_auth_token=lowerCAmelCase__ ,revision=lowerCAmelCase__ ,local_files_only=lowerCAmelCase__ ,)
return get_class_in_module(lowerCAmelCase__ ,final_module.replace('''.py''' ,'''''' ) )
| 351
|
from math import ceil
def __UpperCAmelCase ( __a : int = 1_001 ) -> int:
"""simple docstring"""
_a : Dict = 1
for i in range(1 ,int(ceil(n / 2.0 ) ) ):
_a : int = 2 * i + 1
_a : str = 2 * i
_a : Any = total + 4 * odd**2 - 6 * even
return total
if __name__ == "__main__":
import sys
if len(sys.argv) == 1:
print(solution())
else:
try:
a__ = int(sys.argv[1])
print(solution(n))
except ValueError:
print('''Invalid entry - please enter a number''')
| 15
| 0
|
"""simple docstring"""
import dataclasses
import json
import warnings
from dataclasses import dataclass, field
from time import time
from typing import List
from ..utils import logging
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : int=None ,__a : List[Any]=None ) -> int:
"""simple docstring"""
return field(default_factory=lambda: default ,metadata=__a )
@dataclass
class UpperCAmelCase_ :
"""simple docstring"""
UpperCAmelCase__ : List[str] = list_field(
default=[] , metadata={
"help": (
"Model checkpoints to be provided to the AutoModel classes. Leave blank to benchmark the base version"
" of all available models"
)
} , )
UpperCAmelCase__ : List[int] = list_field(
default=[8] , metadata={"help": "List of batch sizes for which memory and time performance will be evaluated"} )
UpperCAmelCase__ : List[int] = list_field(
default=[8, 32, 128, 512] , metadata={"help": "List of sequence lengths for which memory and time performance will be evaluated"} , )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={"help": "Whether to benchmark inference of model. Inference can be disabled via --no-inference."} , )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={"help": "Whether to run on available cuda devices. Cuda can be disabled via --no-cuda."} , )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={"help": "Whether to run on available tpu devices. TPU can be disabled via --no-tpu."} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Use FP16 to accelerate inference."} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Benchmark training of model"} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Verbose memory tracing"} )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={"help": "Whether to perform speed measurements. Speed measurements can be disabled via --no-speed."} , )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={
"help": "Whether to perform memory measurements. Memory measurements can be disabled via --no-memory"
} , )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Trace memory line by line"} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Save result to a CSV file"} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Save all print statements in a log file"} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Whether to print environment information"} )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={
"help": (
"Whether to use multiprocessing for memory and speed measurement. It is highly recommended to use"
" multiprocessing for accurate CPU and GPU memory measurements. This option should only be disabled"
" for debugging / testing and on TPU."
)
} , )
UpperCAmelCase__ : str = field(
default=F'''inference_time_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving time results to csv."} , )
UpperCAmelCase__ : str = field(
default=F'''inference_memory_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving memory results to csv."} , )
UpperCAmelCase__ : str = field(
default=F'''train_time_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving time results to csv for training."} , )
UpperCAmelCase__ : str = field(
default=F'''train_memory_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving memory results to csv for training."} , )
UpperCAmelCase__ : str = field(
default=F'''env_info_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving environment information."} , )
UpperCAmelCase__ : str = field(
default=F'''log_{round(time() )}.csv''' , metadata={"help": "Log filename used if print statements are saved in log."} , )
UpperCAmelCase__ : int = field(default=3 , metadata={"help": "Times an experiment will be run."} )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={
"help": (
"Instead of loading the model as defined in `config.architectures` if exists, just load the pretrain"
" model weights."
)
} , )
def __lowercase ( self ) -> Tuple:
warnings.warn(
F"""The class {self.__class__} is deprecated. Hugging Face Benchmarking utils"""
''' are deprecated in general and it is advised to use external Benchmarking libraries '''
''' to benchmark Transformer models.''' , _a , )
def __lowercase ( self ) -> List[str]:
return json.dumps(dataclasses.asdict(self ) , indent=2 )
@property
def __lowercase ( self ) -> List[str]:
if len(self.models ) <= 0:
raise ValueError(
'''Please make sure you provide at least one model name / model identifier, *e.g.* `--models'''
''' bert-base-cased` or `args.models = [\'bert-base-cased\'].''' )
return self.models
@property
def __lowercase ( self ) -> Union[str, Any]:
if not self.multi_process:
return False
elif self.is_tpu:
logger.info('''Multiprocessing is currently not possible on TPU.''' )
return False
else:
return True
| 352
|
from typing import Dict, Iterable, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format, to_pil_image
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_pytesseract_available, is_vision_available, logging, requires_backends
if is_vision_available():
import PIL
# soft dependency
if is_pytesseract_available():
import pytesseract
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : str ,__a : Union[str, Any] ) -> List[str]:
"""simple docstring"""
return [
int(1_000 * (box[0] / width) ),
int(1_000 * (box[1] / height) ),
int(1_000 * (box[2] / width) ),
int(1_000 * (box[3] / height) ),
]
def __UpperCAmelCase ( __a : np.ndarray ,__a : Optional[str] ,__a : Optional[str] ) -> List[Any]:
"""simple docstring"""
_a : str = to_pil_image(__a )
_a , _a : Optional[Any] = pil_image.size
_a : Tuple = pytesseract.image_to_data(__a ,lang=__a ,output_type='''dict''' ,config=__a )
_a , _a , _a , _a , _a : List[str] = data['''text'''], data['''left'''], data['''top'''], data['''width'''], data['''height''']
# filter empty words and corresponding coordinates
_a : Dict = [idx for idx, word in enumerate(__a ) if not word.strip()]
_a : str = [word for idx, word in enumerate(__a ) if idx not in irrelevant_indices]
_a : List[str] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : str = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
# turn coordinates into (left, top, left+width, top+height) format
_a : int = []
for x, y, w, h in zip(__a ,__a ,__a ,__a ):
_a : List[str] = [x, y, x + w, y + h]
actual_boxes.append(__a )
# finally, normalize the bounding boxes
_a : Dict = []
for box in actual_boxes:
normalized_boxes.append(normalize_box(__a ,__a ,__a ) )
assert len(__a ) == len(__a ), "Not as many words as there are bounding boxes"
return words, normalized_boxes
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[int] = ["pixel_values"]
def __init__( self , _a = True , _a = None , _a = PILImageResampling.BILINEAR , _a = True , _a = 1 / 2_5_5 , _a = True , _a = None , _a = None , _a = True , _a = None , _a = "" , **_a , ) -> None:
super().__init__(**_a )
_a : List[str] = size if size is not None else {'''height''': 2_2_4, '''width''': 2_2_4}
_a : Union[str, Any] = get_size_dict(_a )
_a : int = do_resize
_a : Optional[int] = size
_a : str = resample
_a : str = do_rescale
_a : Any = rescale_value
_a : Optional[Any] = do_normalize
_a : int = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
_a : List[str] = image_std if image_std is not None else IMAGENET_STANDARD_STD
_a : List[Any] = apply_ocr
_a : Optional[int] = ocr_lang
_a : Tuple = tesseract_config
def __lowercase ( self , _a , _a , _a = PILImageResampling.BILINEAR , _a = None , **_a , ) -> np.ndarray:
_a : Any = get_size_dict(_a )
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()}""" )
_a : Optional[int] = (size['''height'''], size['''width'''])
return resize(_a , size=_a , resample=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
return rescale(_a , scale=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a , _a = None , **_a , ) -> np.ndarray:
return normalize(_a , mean=_a , std=_a , data_format=_a , **_a )
def __lowercase ( 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:
_a : Optional[int] = do_resize if do_resize is not None else self.do_resize
_a : Union[str, Any] = size if size is not None else self.size
_a : Any = get_size_dict(_a )
_a : List[str] = resample if resample is not None else self.resample
_a : int = do_rescale if do_rescale is not None else self.do_rescale
_a : Union[str, Any] = rescale_factor if rescale_factor is not None else self.rescale_factor
_a : int = do_normalize if do_normalize is not None else self.do_normalize
_a : str = image_mean if image_mean is not None else self.image_mean
_a : Tuple = image_std if image_std is not None else self.image_std
_a : Any = apply_ocr if apply_ocr is not None else self.apply_ocr
_a : int = ocr_lang if ocr_lang is not None else self.ocr_lang
_a : Optional[int] = tesseract_config if tesseract_config is not None else self.tesseract_config
_a : List[Any] = make_list_of_images(_a )
if not valid_images(_a ):
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_normalize and (image_mean is None or image_std is None):
raise ValueError('''If do_normalize is True, image_mean and image_std must be specified.''' )
# All transformations expect numpy arrays.
_a : Any = [to_numpy_array(_a ) for image in images]
# Tesseract OCR to get words + normalized bounding boxes
if apply_ocr:
requires_backends(self , '''pytesseract''' )
_a : str = []
_a : str = []
for image in images:
_a , _a : Union[str, Any] = apply_tesseract(_a , _a , _a )
words_batch.append(_a )
boxes_batch.append(_a )
if do_resize:
_a : List[str] = [self.resize(image=_a , size=_a , resample=_a ) for image in images]
if do_rescale:
_a : Optional[Any] = [self.rescale(image=_a , scale=_a ) for image in images]
if do_normalize:
_a : List[Any] = [self.normalize(image=_a , mean=_a , std=_a ) for image in images]
_a : List[str] = [to_channel_dimension_format(_a , _a ) for image in images]
_a : List[str] = BatchFeature(data={'''pixel_values''': images} , tensor_type=_a )
if apply_ocr:
_a : Optional[int] = words_batch
_a : List[Any] = boxes_batch
return data
| 15
| 0
|
"""simple docstring"""
from typing import Any, Dict, List, Union
from ..utils import add_end_docstrings, is_torch_available, is_vision_available, logging, requires_backends
from .base import PIPELINE_INIT_ARGS, ChunkPipeline
if is_vision_available():
from PIL import Image
from ..image_utils import load_image
if is_torch_available():
import torch
from transformers.modeling_outputs import BaseModelOutput
from ..models.auto.modeling_auto import MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING
a__ = logging.get_logger(__name__)
@add_end_docstrings(lowercase_ )
class UpperCAmelCase_ ( lowercase_ ):
"""simple docstring"""
def __init__( self , **_a ) -> Any:
super().__init__(**a__ )
if self.framework == "tf":
raise ValueError(F"""The {self.__class__} is only available in PyTorch.""" )
requires_backends(self , '''vision''' )
self.check_model_type(a__ )
def __call__( self , _a , _a = None , **_a , ) -> Optional[int]:
if "text_queries" in kwargs:
_a : str = kwargs.pop('''text_queries''' )
if isinstance(a__ , (str, Image.Image) ):
_a : int = {'''image''': image, '''candidate_labels''': candidate_labels}
else:
_a : int = image
_a : Optional[int] = super().__call__(a__ , **a__ )
return results
def __lowercase ( self , **_a ) -> int:
_a : Union[str, Any] = {}
if "threshold" in kwargs:
_a : Union[str, Any] = kwargs['''threshold''']
if "top_k" in kwargs:
_a : Optional[int] = kwargs['''top_k''']
return {}, {}, postprocess_params
def __lowercase ( self , _a ) -> int:
_a : int = load_image(inputs['''image'''] )
_a : Optional[int] = inputs['''candidate_labels''']
if isinstance(a__ , a__ ):
_a : List[Any] = candidate_labels.split(''',''' )
_a : Any = torch.tensor([[image.height, image.width]] , dtype=torch.intaa )
for i, candidate_label in enumerate(a__ ):
_a : str = self.tokenizer(a__ , return_tensors=self.framework )
_a : Optional[Any] = self.image_processor(a__ , return_tensors=self.framework )
yield {
"is_last": i == len(a__ ) - 1,
"target_size": target_size,
"candidate_label": candidate_label,
**text_inputs,
**image_features,
}
def __lowercase ( self , _a ) -> Union[str, Any]:
_a : Tuple = model_inputs.pop('''target_size''' )
_a : Optional[int] = model_inputs.pop('''candidate_label''' )
_a : Any = model_inputs.pop('''is_last''' )
_a : str = self.model(**a__ )
_a : Dict = {'''target_size''': target_size, '''candidate_label''': candidate_label, '''is_last''': is_last, **outputs}
return model_outputs
def __lowercase ( self , _a , _a=0.1 , _a=None ) -> Tuple:
_a : str = []
for model_output in model_outputs:
_a : Union[str, Any] = model_output['''candidate_label''']
_a : Dict = BaseModelOutput(a__ )
_a : Union[str, Any] = self.image_processor.post_process_object_detection(
outputs=a__ , threshold=a__ , target_sizes=model_output['''target_size'''] )[0]
for index in outputs["scores"].nonzero():
_a : str = outputs['''scores'''][index].item()
_a : int = self._get_bounding_box(outputs['''boxes'''][index][0] )
_a : Tuple = {'''score''': score, '''label''': label, '''box''': box}
results.append(a__ )
_a : Tuple = sorted(a__ , key=lambda _a : x["score"] , reverse=a__ )
if top_k:
_a : List[str] = results[:top_k]
return results
def __lowercase ( self , _a ) -> Dict[str, int]:
if self.framework != "pt":
raise ValueError('''The ZeroShotObjectDetectionPipeline is only available in PyTorch.''' )
_a , _a , _a , _a : Optional[Any] = box.int().tolist()
_a : Tuple = {
'''xmin''': xmin,
'''ymin''': ymin,
'''xmax''': xmax,
'''ymax''': ymax,
}
return bbox
| 353
|
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from argparse import ArgumentParser
from accelerate.commands.config import get_config_parser
from accelerate.commands.env import env_command_parser
from accelerate.commands.launch import launch_command_parser
from accelerate.commands.test import test_command_parser
from accelerate.commands.tpu import tpu_command_parser
def __UpperCAmelCase ( ) -> Optional[Any]:
"""simple docstring"""
_a : int = ArgumentParser('''Accelerate CLI tool''' ,usage='''accelerate <command> [<args>]''' ,allow_abbrev=__a )
_a : Optional[int] = parser.add_subparsers(help='''accelerate command helpers''' )
# Register commands
get_config_parser(subparsers=__a )
env_command_parser(subparsers=__a )
launch_command_parser(subparsers=__a )
tpu_command_parser(subparsers=__a )
test_command_parser(subparsers=__a )
# Let's go
_a : Dict = parser.parse_args()
if not hasattr(__a ,'''func''' ):
parser.print_help()
exit(1 )
# Run
args.func(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
"""simple docstring"""
# Lint as: python3
import os
import re
import urllib.parse
from pathlib import Path
from typing import Callable, List, Optional, Union
from zipfile import ZipFile
from ..utils.file_utils import cached_path, hf_github_url
from ..utils.logging import get_logger
from ..utils.version import Version
a__ = get_logger(__name__)
class UpperCAmelCase_ :
"""simple docstring"""
UpperCAmelCase__ : Tuple = '''dummy_data'''
UpperCAmelCase__ : List[str] = '''datasets'''
UpperCAmelCase__ : Dict = False
def __init__( self , _a , _a , _a , _a = None , _a = False , _a = True , _a = None , ) -> Tuple:
_a : Optional[int] = 0
_a : str = dataset_name
_a : Any = cache_dir
_a : int = use_local_dummy_data
_a : Optional[int] = config
# download_callbacks take a single url as input
_a : Any = download_callbacks or []
# if False, it doesn't load existing files and it returns the paths of the dummy files relative
# to the dummy_data zip file root
_a : List[Any] = load_existing_dummy_data
# TODO(PVP, QL) might need to make this more general
_a : str = str(UpperCamelCase_ )
# to be downloaded
_a : List[str] = None
_a : int = None
@property
def __lowercase ( self ) -> Optional[int]:
if self._dummy_file is None:
_a : Optional[int] = self.download_dummy_data()
return self._dummy_file
@property
def __lowercase ( self ) -> Dict:
if self.config is not None:
# structure is dummy / config_name / version_name
return os.path.join('''dummy''' , self.config.name , self.version_name )
# structure is dummy / version_name
return os.path.join('''dummy''' , self.version_name )
@property
def __lowercase ( self ) -> List[Any]:
return os.path.join(self.dummy_data_folder , '''dummy_data.zip''' )
def __lowercase ( self ) -> Tuple:
_a : Optional[Any] = (
self.local_path_to_dummy_data if self.use_local_dummy_data is True else self.github_path_to_dummy_data
)
_a : List[Any] = cached_path(
UpperCamelCase_ , cache_dir=self.cache_dir , extract_compressed_file=UpperCamelCase_ , force_extract=UpperCamelCase_ )
return os.path.join(UpperCamelCase_ , self.dummy_file_name )
@property
def __lowercase ( self ) -> str:
return os.path.join(self.datasets_scripts_dir , self.dataset_name , self.dummy_zip_file )
@property
def __lowercase ( self ) -> List[Any]:
if self._bucket_url is None:
_a : int = hf_github_url(self.dataset_name , self.dummy_zip_file.replace(os.sep , '''/''' ) )
return self._bucket_url
@property
def __lowercase ( self ) -> List[str]:
if os.path.isdir(self.dummy_file ):
return self.dummy_file
# else cut off path to file -> example `xsum`.
return "/".join(self.dummy_file.replace(os.sep , '''/''' ).split('''/''' )[:-1] )
def __lowercase ( self , _a , *_a ) -> Tuple:
if self.load_existing_dummy_data:
# dummy data is downloaded and tested
_a : List[Any] = self.dummy_file
else:
# dummy data cannot be downloaded and only the path to dummy file is returned
_a : int = self.dummy_file_name
# special case when data_url is a dict
if isinstance(UpperCamelCase_ , UpperCamelCase_ ):
return self.create_dummy_data_dict(UpperCamelCase_ , UpperCamelCase_ )
elif isinstance(UpperCamelCase_ , (list, tuple) ):
return self.create_dummy_data_list(UpperCamelCase_ , UpperCamelCase_ )
else:
return self.create_dummy_data_single(UpperCamelCase_ , UpperCamelCase_ )
def __lowercase ( self , _a , *_a ) -> List[Any]:
return self.download_and_extract(UpperCamelCase_ )
def __lowercase ( self , _a , _a ) -> List[Any]:
return self.download_and_extract(UpperCamelCase_ )
def __lowercase ( self , _a , *_a , **_a ) -> Tuple:
return path
def __lowercase ( self ) -> Tuple:
return {}
def __lowercase ( self , _a , _a ) -> List[str]:
_a : Optional[Any] = {}
for key, single_urls in data_url.items():
for download_callback in self.download_callbacks:
if isinstance(UpperCamelCase_ , UpperCamelCase_ ):
for single_url in single_urls:
download_callback(UpperCamelCase_ )
else:
_a : List[Any] = single_urls
download_callback(UpperCamelCase_ )
# we force the name of each key to be the last file / folder name of the url path
# if the url has arguments, we need to encode them with urllib.parse.quote_plus
if isinstance(UpperCamelCase_ , UpperCamelCase_ ):
_a : Tuple = [os.path.join(UpperCamelCase_ , urllib.parse.quote_plus(Path(UpperCamelCase_ ).name ) ) for x in single_urls]
else:
_a : List[str] = single_urls
_a : int = os.path.join(UpperCamelCase_ , urllib.parse.quote_plus(Path(UpperCamelCase_ ).name ) )
_a : Tuple = value
# make sure that values are unique
if all(isinstance(UpperCamelCase_ , UpperCamelCase_ ) for i in dummy_data_dict.values() ) and len(set(dummy_data_dict.values() ) ) < len(
dummy_data_dict.values() ):
# append key to value to make its name unique
_a : int = {key: value + key for key, value in dummy_data_dict.items()}
return dummy_data_dict
def __lowercase ( self , _a , _a ) -> Union[str, Any]:
_a : List[Any] = []
# trick: if there are many shards named like `data.txt-000001-of-00300`, only use the first one
_a : int = all(bool(re.findall('''[0-9]{3,}-of-[0-9]{3,}''' , UpperCamelCase_ ) ) for url in data_url )
_a : Union[str, Any] = all(
url.startswith('''https://ftp.ncbi.nlm.nih.gov/pubmed/baseline/pubmed''' ) for url in data_url )
if data_url and (is_tf_records or is_pubmed_records):
_a : Dict = [data_url[0]] * len(UpperCamelCase_ )
for single_url in data_url:
for download_callback in self.download_callbacks:
download_callback(UpperCamelCase_ )
# we force the name of each key to be the last file / folder name of the url path
# if the url has arguments, we need to encode them with urllib.parse.quote_plus
_a : List[Any] = os.path.join(UpperCamelCase_ , urllib.parse.quote_plus(single_url.split('''/''' )[-1] ) )
dummy_data_list.append(UpperCamelCase_ )
return dummy_data_list
def __lowercase ( self , _a , _a ) -> Optional[int]:
for download_callback in self.download_callbacks:
download_callback(UpperCamelCase_ )
# we force the name of each key to be the last file / folder name of the url path
# if the url has arguments, we need to encode them with urllib.parse.quote_plus
_a : Tuple = os.path.join(UpperCamelCase_ , urllib.parse.quote_plus(data_url.split('''/''' )[-1] ) )
if os.path.exists(UpperCamelCase_ ) or not self.load_existing_dummy_data:
return value
else:
# Backward compatibility, maybe deprecate at one point.
# For many datasets with single url calls to dl_manager.download_and_extract,
# the dummy_data.zip file is actually the zipped downloaded file
# while now we expected the dummy_data.zip file to be a directory containing
# the downloaded file.
return path_to_dummy_data
def __lowercase ( self ) -> Dict:
pass
def __lowercase ( self ) -> Optional[Any]:
pass
def __lowercase ( self , _a ) -> List[str]:
def _iter_archive_members(_a ):
# this preserves the order of the members inside the ZIP archive
_a : List[Any] = Path(self.dummy_file ).parent
_a : Tuple = path.relative_to(UpperCamelCase_ )
with ZipFile(self.local_path_to_dummy_data ) as zip_file:
_a : Optional[int] = zip_file.namelist()
for member in members:
if member.startswith(relative_path.as_posix() ):
yield dummy_parent_path.joinpath(UpperCamelCase_ )
_a : List[str] = Path(UpperCamelCase_ )
_a : Optional[Any] = _iter_archive_members(UpperCamelCase_ ) if self.use_local_dummy_data else path.rglob('''*''' )
for file_path in file_paths:
if file_path.is_file() and not file_path.name.startswith(('''.''', '''__''') ):
yield file_path.relative_to(UpperCamelCase_ ).as_posix(), file_path.open('''rb''' )
def __lowercase ( self , _a ) -> Optional[int]:
if not isinstance(UpperCamelCase_ , UpperCamelCase_ ):
_a : Tuple = [paths]
for path in paths:
if os.path.isfile(UpperCamelCase_ ):
if os.path.basename(UpperCamelCase_ ).startswith(('''.''', '''__''') ):
return
yield path
else:
for dirpath, dirnames, filenames in os.walk(UpperCamelCase_ ):
if os.path.basename(UpperCamelCase_ ).startswith(('''.''', '''__''') ):
continue
dirnames.sort()
for filename in sorted(UpperCamelCase_ ):
if filename.startswith(('''.''', '''__''') ):
continue
yield os.path.join(UpperCamelCase_ , UpperCamelCase_ )
| 354
|
import itertools
import os
import random
import tempfile
import unittest
import numpy as np
from transformers import TvltFeatureExtractor, is_datasets_available
from transformers.testing_utils import check_json_file_has_correct_format, require_torch, require_torchaudio
from transformers.utils.import_utils import is_torch_available
from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
if is_torch_available():
import torch
if is_datasets_available():
from datasets import load_dataset
a__ = random.Random()
def __UpperCAmelCase ( __a : Tuple ,__a : str=1.0 ,__a : Optional[int]=None ,__a : List[Any]=None ) -> Any:
"""simple docstring"""
if rng is None:
_a : Dict = global_rng
_a : Optional[Any] = []
for batch_idx in range(shape[0] ):
values.append([] )
for _ in range(shape[1] ):
values[-1].append(rng.random() * scale )
return values
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __init__( self , _a , _a=7 , _a=4_0_0 , _a=2_0_0_0 , _a=2_0_4_8 , _a=1_2_8 , _a=1 , _a=5_1_2 , _a=3_0 , _a=4_4_1_0_0 , ) -> List[Any]:
_a : Optional[Any] = parent
_a : str = batch_size
_a : List[str] = min_seq_length
_a : str = max_seq_length
_a : Dict = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
_a : List[Any] = spectrogram_length
_a : List[str] = feature_size
_a : List[Any] = num_audio_channels
_a : Tuple = hop_length
_a : Optional[int] = chunk_length
_a : int = sampling_rate
def __lowercase ( self ) -> Union[str, Any]:
return {
"spectrogram_length": self.spectrogram_length,
"feature_size": self.feature_size,
"num_audio_channels": self.num_audio_channels,
"hop_length": self.hop_length,
"chunk_length": self.chunk_length,
"sampling_rate": self.sampling_rate,
}
def __lowercase ( self , _a=False , _a=False ) -> List[Any]:
def _flatten(_a ):
return list(itertools.chain(*_a ) )
if equal_length:
_a : List[Any] = [floats_list((self.max_seq_length, self.feature_size) ) for _ in range(self.batch_size )]
else:
# make sure that inputs increase in size
_a : List[Any] = [
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 : str = [np.asarray(_a ) for x in speech_inputs]
return speech_inputs
@require_torch
@require_torchaudio
class UpperCAmelCase_ ( __lowercase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = TvltFeatureExtractor
def __lowercase ( self ) -> Dict:
_a : List[str] = TvltFeatureExtractionTester(self )
def __lowercase ( self ) -> Any:
_a : List[Any] = self.feature_extraction_class(**self.feat_extract_dict )
self.assertTrue(hasattr(_a , '''spectrogram_length''' ) )
self.assertTrue(hasattr(_a , '''feature_size''' ) )
self.assertTrue(hasattr(_a , '''num_audio_channels''' ) )
self.assertTrue(hasattr(_a , '''hop_length''' ) )
self.assertTrue(hasattr(_a , '''chunk_length''' ) )
self.assertTrue(hasattr(_a , '''sampling_rate''' ) )
def __lowercase ( self ) -> Optional[int]:
_a : Optional[Any] = self.feature_extraction_class(**self.feat_extract_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : int = feat_extract_first.save_pretrained(_a )[0]
check_json_file_has_correct_format(_a )
_a : Dict = self.feature_extraction_class.from_pretrained(_a )
_a : List[Any] = feat_extract_first.to_dict()
_a : Union[str, Any] = feat_extract_second.to_dict()
_a : Any = dict_first.pop('''mel_filters''' )
_a : int = dict_second.pop('''mel_filters''' )
self.assertTrue(np.allclose(_a , _a ) )
self.assertEqual(_a , _a )
def __lowercase ( self ) -> Optional[int]:
_a : Any = self.feature_extraction_class(**self.feat_extract_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : Optional[int] = os.path.join(_a , '''feat_extract.json''' )
feat_extract_first.to_json_file(_a )
_a : List[str] = self.feature_extraction_class.from_json_file(_a )
_a : List[Any] = feat_extract_first.to_dict()
_a : Dict = feat_extract_second.to_dict()
_a : str = dict_first.pop('''mel_filters''' )
_a : str = dict_second.pop('''mel_filters''' )
self.assertTrue(np.allclose(_a , _a ) )
self.assertEqual(_a , _a )
def __lowercase ( self ) -> Union[str, Any]:
# Initialize feature_extractor
_a : Union[str, Any] = self.feature_extraction_class(**self.feat_extract_dict )
# create three inputs of length 800, 1000, and 1200
_a : Any = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
_a : List[str] = [np.asarray(_a ) for speech_input in speech_inputs]
# Test not batched input
_a : Tuple = feature_extractor(np_speech_inputs[0] , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test batched
_a : Dict = feature_extractor(_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test audio masking
_a : Union[str, Any] = feature_extractor(
_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 , mask_audio=_a ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test 2-D numpy arrays are batched.
_a : Optional[Any] = [floats_list((1, x) )[0] for x in (8_0_0, 8_0_0, 8_0_0)]
_a : int = np.asarray(_a )
_a : Tuple = feature_extractor(_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[Any] = load_dataset('''hf-internal-testing/librispeech_asr_dummy''' , '''clean''' , split='''validation''' )
# automatic decoding with librispeech
_a : Optional[int] = ds.sort('''id''' ).select(range(_a ) )[:num_samples]['''audio''']
return [x["array"] for x in speech_samples]
def __lowercase ( self ) -> int:
_a : Union[str, Any] = self._load_datasamples(1 )
_a : int = TvltFeatureExtractor()
_a : Union[str, Any] = feature_extractor(_a , return_tensors='''pt''' ).audio_values
self.assertEquals(audio_values.shape , (1, 1, 1_9_2, 1_2_8) )
_a : Union[str, Any] = torch.tensor([[-0.3032, -0.2708], [-0.4434, -0.4007]] )
self.assertTrue(torch.allclose(audio_values[0, 0, :2, :2] , _a , atol=1e-4 ) )
| 15
| 0
|
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a__ = logging.get_logger(__name__)
a__ = {
'microsoft/markuplm-base': 'https://huggingface.co/microsoft/markuplm-base/resolve/main/config.json',
'microsoft/markuplm-large': 'https://huggingface.co/microsoft/markuplm-large/resolve/main/config.json',
}
class UpperCAmelCase_ ( lowerCamelCase__ ):
"""simple docstring"""
UpperCAmelCase__ : Optional[int] = "markuplm"
def __init__( self , _a=3_0_5_2_2 , _a=7_6_8 , _a=1_2 , _a=1_2 , _a=3_0_7_2 , _a="gelu" , _a=0.1 , _a=0.1 , _a=5_1_2 , _a=2 , _a=0.02 , _a=1e-1_2 , _a=0 , _a=0 , _a=2 , _a=2_5_6 , _a=1_0_2_4 , _a=2_1_6 , _a=1_0_0_1 , _a=3_2 , _a=5_0 , _a="absolute" , _a=True , _a=None , **_a , ) -> List[Any]:
super().__init__(
pad_token_id=__lowerCamelCase , bos_token_id=__lowerCamelCase , eos_token_id=__lowerCamelCase , **__lowerCamelCase , )
_a : int = vocab_size
_a : Optional[Any] = hidden_size
_a : Any = num_hidden_layers
_a : Optional[int] = num_attention_heads
_a : str = hidden_act
_a : Tuple = intermediate_size
_a : Any = hidden_dropout_prob
_a : Tuple = attention_probs_dropout_prob
_a : List[str] = max_position_embeddings
_a : Union[str, Any] = type_vocab_size
_a : List[str] = initializer_range
_a : str = layer_norm_eps
_a : Union[str, Any] = position_embedding_type
_a : List[Any] = use_cache
_a : Dict = classifier_dropout
# additional properties
_a : int = max_depth
_a : List[Any] = max_xpath_tag_unit_embeddings
_a : Optional[Any] = max_xpath_subs_unit_embeddings
_a : Tuple = tag_pad_id
_a : Any = subs_pad_id
_a : Dict = xpath_unit_hidden_size
| 355
|
from typing import Dict
import numpy as np
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, GenericTensor, Pipeline, PipelineException
if is_tf_available():
import tensorflow as tf
from ..tf_utils import stable_softmax
if is_torch_available():
import torch
a__ = logging.get_logger(__name__)
@add_end_docstrings(
__lowercase , r"\n top_k (`int`, defaults to 5):\n The number of predictions to return.\n targets (`str` or `List[str]`, *optional*):\n When passed, the model will limit the scores to the passed targets instead of looking up in the whole\n vocab. If the provided targets are not in the model vocab, they will be tokenized and the first resulting\n token will be used (with a warning, and that might be slower).\n\n " , )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self , _a ) -> np.ndarray:
if self.framework == "tf":
_a : List[str] = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()
elif self.framework == "pt":
_a : Tuple = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a )
else:
raise ValueError('''Unsupported framework''' )
return masked_index
def __lowercase ( self , _a ) -> np.ndarray:
_a : int = self.get_masked_index(_a )
_a : Tuple = np.prod(masked_index.shape )
if numel < 1:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , F"""No mask_token ({self.tokenizer.mask_token}) found on the input""" , )
def __lowercase ( self , _a ) -> Optional[int]:
if isinstance(_a , _a ):
for model_input in model_inputs:
self._ensure_exactly_one_mask_token(model_input['''input_ids'''][0] )
else:
for input_ids in model_inputs["input_ids"]:
self._ensure_exactly_one_mask_token(_a )
def __lowercase ( self , _a , _a=None , **_a ) -> Dict[str, GenericTensor]:
if return_tensors is None:
_a : Union[str, Any] = self.framework
_a : str = self.tokenizer(_a , return_tensors=_a )
self.ensure_exactly_one_mask_token(_a )
return model_inputs
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[str] = self.model(**_a )
_a : Any = model_inputs['''input_ids''']
return model_outputs
def __lowercase ( self , _a , _a=5 , _a=None ) -> str:
# Cap top_k if there are targets
if target_ids is not None and target_ids.shape[0] < top_k:
_a : List[Any] = target_ids.shape[0]
_a : Any = model_outputs['''input_ids'''][0]
_a : List[str] = model_outputs['''logits''']
if self.framework == "tf":
_a : Tuple = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()[:, 0]
_a : List[str] = outputs.numpy()
_a : Dict = outputs[0, masked_index, :]
_a : str = stable_softmax(_a , axis=-1 )
if target_ids is not None:
_a : Any = tf.gather_nd(tf.squeeze(_a , 0 ) , target_ids.reshape(-1 , 1 ) )
_a : Union[str, Any] = tf.expand_dims(_a , 0 )
_a : Optional[int] = tf.math.top_k(_a , k=_a )
_a , _a : Optional[Any] = topk.values.numpy(), topk.indices.numpy()
else:
_a : Optional[Any] = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a ).squeeze(-1 )
# Fill mask pipeline supports only one ${mask_token} per sample
_a : List[str] = outputs[0, masked_index, :]
_a : List[Any] = logits.softmax(dim=-1 )
if target_ids is not None:
_a : List[Any] = probs[..., target_ids]
_a , _a : Optional[Any] = probs.topk(_a )
_a : Dict = []
_a : List[Any] = values.shape[0] == 1
for i, (_values, _predictions) in enumerate(zip(values.tolist() , predictions.tolist() ) ):
_a : Optional[Any] = []
for v, p in zip(_values , _predictions ):
# Copy is important since we're going to modify this array in place
_a : Optional[int] = input_ids.numpy().copy()
if target_ids is not None:
_a : Tuple = target_ids[p].tolist()
_a : List[str] = p
# Filter padding out:
_a : List[Any] = tokens[np.where(tokens != self.tokenizer.pad_token_id )]
# Originally we skip special tokens to give readable output.
# For multi masks though, the other [MASK] would be removed otherwise
# making the output look odd, so we add them back
_a : List[str] = self.tokenizer.decode(_a , skip_special_tokens=_a )
_a : List[Any] = {'''score''': v, '''token''': p, '''token_str''': self.tokenizer.decode([p] ), '''sequence''': sequence}
row.append(_a )
result.append(_a )
if single_mask:
return result[0]
return result
def __lowercase ( self , _a , _a=None ) -> Dict:
if isinstance(_a , _a ):
_a : Tuple = [targets]
try:
_a : int = self.tokenizer.get_vocab()
except Exception:
_a : Any = {}
_a : List[Any] = []
for target in targets:
_a : List[Any] = vocab.get(_a , _a )
if id_ is None:
_a : Tuple = self.tokenizer(
_a , add_special_tokens=_a , return_attention_mask=_a , return_token_type_ids=_a , max_length=1 , truncation=_a , )['''input_ids''']
if len(_a ) == 0:
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
'''We cannot replace it with anything meaningful, ignoring it''' )
continue
_a : Tuple = input_ids[0]
# XXX: If users encounter this pass
# it becomes pretty slow, so let's make sure
# The warning enables them to fix the input to
# get faster performance.
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
F"""Replacing with `{self.tokenizer.convert_ids_to_tokens(id_ )}`.""" )
target_ids.append(id_ )
_a : List[str] = list(set(_a ) )
if len(_a ) == 0:
raise ValueError('''At least one target must be provided when passed.''' )
_a : int = np.array(_a )
return target_ids
def __lowercase ( self , _a=None , _a=None ) -> Tuple:
_a : str = {}
if targets is not None:
_a : List[Any] = self.get_target_ids(_a , _a )
_a : Optional[Any] = target_ids
if top_k is not None:
_a : Union[str, Any] = top_k
if self.tokenizer.mask_token_id is None:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , '''The tokenizer does not define a `mask_token`.''' )
return {}, {}, postprocess_params
def __call__( self , _a , *_a , **_a ) -> int:
_a : Optional[Any] = super().__call__(_a , **_a )
if isinstance(_a , _a ) and len(_a ) == 1:
return outputs[0]
return outputs
| 15
| 0
|
import gc
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import TransformeraDModel, VQDiffusionPipeline, VQDiffusionScheduler, VQModel
from diffusers.pipelines.vq_diffusion.pipeline_vq_diffusion import LearnedClassifierFreeSamplingEmbeddings
from diffusers.utils import load_numpy, slow, torch_device
from diffusers.utils.testing_utils import require_torch_gpu
a__ = False
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> List[Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
@property
def __lowercase ( self ) -> Union[str, Any]:
return 1_2
@property
def __lowercase ( self ) -> Union[str, Any]:
return 1_2
@property
def __lowercase ( self ) -> Tuple:
return 3_2
@property
def __lowercase ( self ) -> Dict:
torch.manual_seed(0 )
_a : Any = VQModel(
block_out_channels=[3_2, 6_4] , in_channels=3 , out_channels=3 , down_block_types=['''DownEncoderBlock2D''', '''DownEncoderBlock2D'''] , up_block_types=['''UpDecoderBlock2D''', '''UpDecoderBlock2D'''] , latent_channels=3 , num_vq_embeddings=self.num_embed , vq_embed_dim=3 , )
return model
@property
def __lowercase ( self ) -> Optional[int]:
_a : Dict = CLIPTokenizer.from_pretrained('''hf-internal-testing/tiny-random-clip''' )
return tokenizer
@property
def __lowercase ( self ) -> Optional[int]:
torch.manual_seed(0 )
_a : Any = CLIPTextConfig(
bos_token_id=0 , eos_token_id=2 , hidden_size=self.text_embedder_hidden_size , 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 , )
return CLIPTextModel(_A )
@property
def __lowercase ( self ) -> Dict:
torch.manual_seed(0 )
_a : Any = 1_2
_a : int = 1_2
_a : List[Any] = {
'attention_bias': True,
'cross_attention_dim': 3_2,
'attention_head_dim': height * width,
'num_attention_heads': 1,
'num_vector_embeds': self.num_embed,
'num_embeds_ada_norm': self.num_embeds_ada_norm,
'norm_num_groups': 3_2,
'sample_size': width,
'activation_fn': 'geglu-approximate',
}
_a : Union[str, Any] = TransformeraDModel(**_A )
return model
def __lowercase ( self ) -> Dict:
_a : Tuple = 'cpu'
_a : List[str] = self.dummy_vqvae
_a : str = self.dummy_text_encoder
_a : Optional[Any] = self.dummy_tokenizer
_a : Dict = self.dummy_transformer
_a : Optional[int] = VQDiffusionScheduler(self.num_embed )
_a : List[Any] = LearnedClassifierFreeSamplingEmbeddings(learnable=_A )
_a : List[Any] = VQDiffusionPipeline(
vqvae=_A , text_encoder=_A , tokenizer=_A , transformer=_A , scheduler=_A , learned_classifier_free_sampling_embeddings=_A , )
_a : List[Any] = pipe.to(_A )
pipe.set_progress_bar_config(disable=_A )
_a : Optional[Any] = 'teddy bear playing in the pool'
_a : str = torch.Generator(device=_A ).manual_seed(0 )
_a : Union[str, Any] = pipe([prompt] , generator=_A , num_inference_steps=2 , output_type='''np''' )
_a : Optional[int] = output.images
_a : int = torch.Generator(device=_A ).manual_seed(0 )
_a : Tuple = pipe(
[prompt] , generator=_A , output_type='''np''' , return_dict=_A , num_inference_steps=2 )[0]
_a : str = image[0, -3:, -3:, -1]
_a : Union[str, Any] = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 2_4, 2_4, 3)
_a : str = np.array([0.6551, 0.6168, 0.5008, 0.5676, 0.5659, 0.4295, 0.6073, 0.5599, 0.4992] )
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 __lowercase ( self ) -> Optional[int]:
_a : Optional[Any] = 'cpu'
_a : Optional[int] = self.dummy_vqvae
_a : List[str] = self.dummy_text_encoder
_a : Optional[int] = self.dummy_tokenizer
_a : Optional[Any] = self.dummy_transformer
_a : Union[str, Any] = VQDiffusionScheduler(self.num_embed )
_a : Optional[int] = LearnedClassifierFreeSamplingEmbeddings(
learnable=_A , hidden_size=self.text_embedder_hidden_size , length=tokenizer.model_max_length )
_a : Union[str, Any] = VQDiffusionPipeline(
vqvae=_A , text_encoder=_A , tokenizer=_A , transformer=_A , scheduler=_A , learned_classifier_free_sampling_embeddings=_A , )
_a : Union[str, Any] = pipe.to(_A )
pipe.set_progress_bar_config(disable=_A )
_a : Union[str, Any] = 'teddy bear playing in the pool'
_a : Optional[int] = torch.Generator(device=_A ).manual_seed(0 )
_a : Tuple = pipe([prompt] , generator=_A , num_inference_steps=2 , output_type='''np''' )
_a : Optional[Any] = output.images
_a : str = torch.Generator(device=_A ).manual_seed(0 )
_a : Dict = pipe(
[prompt] , generator=_A , output_type='''np''' , return_dict=_A , num_inference_steps=2 )[0]
_a : int = image[0, -3:, -3:, -1]
_a : Tuple = image_from_tuple[0, -3:, -3:, -1]
assert image.shape == (1, 2_4, 2_4, 3)
_a : Optional[Any] = np.array([0.6693, 0.6075, 0.4959, 0.5701, 0.5583, 0.4333, 0.6171, 0.5684, 0.4988] )
assert np.abs(image_slice.flatten() - expected_slice ).max() < 2.0
assert np.abs(image_from_tuple_slice.flatten() - expected_slice ).max() < 1e-2
@slow
@require_torch_gpu
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Union[str, Any]:
# clean up the VRAM after each test
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def __lowercase ( self ) -> Optional[int]:
_a : List[Any] = load_numpy(
'''https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main'''
'''/vq_diffusion/teddy_bear_pool_classifier_free_sampling.npy''' )
_a : Union[str, Any] = VQDiffusionPipeline.from_pretrained('''microsoft/vq-diffusion-ithq''' )
_a : Tuple = pipeline.to(_A )
pipeline.set_progress_bar_config(disable=_A )
# requires GPU generator for gumbel softmax
# don't use GPU generator in tests though
_a : Optional[int] = torch.Generator(device=_A ).manual_seed(0 )
_a : int = pipeline(
'''teddy bear playing in the pool''' , num_images_per_prompt=1 , generator=_A , output_type='''np''' , )
_a : int = output.images[0]
assert image.shape == (2_5_6, 2_5_6, 3)
assert np.abs(expected_image - image ).max() < 2.0
| 356
|
import argparse
import json
import logging
import os
import sys
from unittest.mock import patch
from transformers.testing_utils import TestCasePlus, get_gpu_count, slow
a__ = [
os.path.join(os.path.dirname(__file__), dirname)
for dirname in [
'''text-classification''',
'''language-modeling''',
'''summarization''',
'''token-classification''',
'''question-answering''',
]
]
sys.path.extend(SRC_DIRS)
if SRC_DIRS is not None:
import run_clm_flax
import run_flax_glue
import run_flax_ner
import run_mlm_flax
import run_qa
import run_summarization_flax
import run_ta_mlm_flax
logging.basicConfig(level=logging.DEBUG)
a__ = logging.getLogger()
def __UpperCAmelCase ( ) -> Optional[int]:
"""simple docstring"""
_a : Any = argparse.ArgumentParser()
parser.add_argument('''-f''' )
_a : Dict = parser.parse_args()
return args.f
def __UpperCAmelCase ( __a : Optional[int] ,__a : List[str]="eval" ) -> Any:
"""simple docstring"""
_a : Any = os.path.join(__a ,F"""{split}_results.json""" )
if os.path.exists(__a ):
with open(__a ,'''r''' ) as f:
return json.load(__a )
raise ValueError(F"""can't find {path}""" )
a__ = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self ) -> str:
_a : Any = self.get_auto_remove_tmp_dir()
_a : Optional[Any] = F"""
run_glue.py
--model_name_or_path distilbert-base-uncased
--output_dir {tmp_dir}
--train_file ./tests/fixtures/tests_samples/MRPC/train.csv
--validation_file ./tests/fixtures/tests_samples/MRPC/dev.csv
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
--learning_rate=1e-4
--eval_steps=2
--warmup_steps=2
--seed=42
--max_seq_length=128
""".split()
with patch.object(_a , '''argv''' , _a ):
run_flax_glue.main()
_a : Any = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.75 )
@slow
def __lowercase ( self ) -> Dict:
_a : Tuple = self.get_auto_remove_tmp_dir()
_a : Tuple = F"""
run_clm_flax.py
--model_name_or_path distilgpt2
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--do_train
--do_eval
--block_size 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--num_train_epochs 2
--logging_steps 2 --eval_steps 2
--output_dir {tmp_dir}
--overwrite_output_dir
""".split()
with patch.object(_a , '''argv''' , _a ):
run_clm_flax.main()
_a : List[str] = get_results(_a )
self.assertLess(result['''eval_perplexity'''] , 1_0_0 )
@slow
def __lowercase ( self ) -> Optional[int]:
_a : str = self.get_auto_remove_tmp_dir()
_a : Optional[int] = F"""
run_summarization.py
--model_name_or_path t5-small
--train_file tests/fixtures/tests_samples/xsum/sample.json
--validation_file tests/fixtures/tests_samples/xsum/sample.json
--test_file tests/fixtures/tests_samples/xsum/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--num_train_epochs=3
--warmup_steps=8
--do_train
--do_eval
--do_predict
--learning_rate=2e-4
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
--predict_with_generate
""".split()
with patch.object(_a , '''argv''' , _a ):
run_summarization_flax.main()
_a : Optional[int] = get_results(_a , split='''test''' )
self.assertGreaterEqual(result['''test_rouge1'''] , 1_0 )
self.assertGreaterEqual(result['''test_rouge2'''] , 2 )
self.assertGreaterEqual(result['''test_rougeL'''] , 7 )
self.assertGreaterEqual(result['''test_rougeLsum'''] , 7 )
@slow
def __lowercase ( self ) -> Tuple:
_a : List[str] = self.get_auto_remove_tmp_dir()
_a : List[Any] = F"""
run_mlm.py
--model_name_or_path distilroberta-base
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--output_dir {tmp_dir}
--overwrite_output_dir
--max_seq_length 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--logging_steps 2 --eval_steps 2
--do_train
--do_eval
--num_train_epochs=1
""".split()
with patch.object(_a , '''argv''' , _a ):
run_mlm_flax.main()
_a : List[Any] = get_results(_a )
self.assertLess(result['''eval_perplexity'''] , 4_2 )
@slow
def __lowercase ( self ) -> Dict:
_a : Optional[Any] = self.get_auto_remove_tmp_dir()
_a : int = F"""
run_t5_mlm_flax.py
--model_name_or_path t5-small
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--do_train
--do_eval
--max_seq_length 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--num_train_epochs 2
--logging_steps 2 --eval_steps 2
--output_dir {tmp_dir}
--overwrite_output_dir
""".split()
with patch.object(_a , '''argv''' , _a ):
run_ta_mlm_flax.main()
_a : List[Any] = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.42 )
@slow
def __lowercase ( self ) -> Optional[Any]:
# with so little data distributed training needs more epochs to get the score on par with 0/1 gpu
_a : Any = 7 if get_gpu_count() > 1 else 2
_a : List[Any] = self.get_auto_remove_tmp_dir()
_a : List[Any] = F"""
run_flax_ner.py
--model_name_or_path bert-base-uncased
--train_file tests/fixtures/tests_samples/conll/sample.json
--validation_file tests/fixtures/tests_samples/conll/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--do_train
--do_eval
--warmup_steps=2
--learning_rate=2e-4
--logging_steps 2 --eval_steps 2
--per_device_train_batch_size=2
--per_device_eval_batch_size=2
--num_train_epochs={epochs}
--seed 7
""".split()
with patch.object(_a , '''argv''' , _a ):
run_flax_ner.main()
_a : Dict = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.75 )
self.assertGreaterEqual(result['''eval_f1'''] , 0.3 )
@slow
def __lowercase ( self ) -> Any:
_a : Optional[int] = self.get_auto_remove_tmp_dir()
_a : Union[str, Any] = F"""
run_qa.py
--model_name_or_path bert-base-uncased
--version_2_with_negative
--train_file tests/fixtures/tests_samples/SQUAD/sample.json
--validation_file tests/fixtures/tests_samples/SQUAD/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--num_train_epochs=3
--warmup_steps=2
--do_train
--do_eval
--logging_steps 2 --eval_steps 2
--learning_rate=2e-4
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
""".split()
with patch.object(_a , '''argv''' , _a ):
run_qa.main()
_a : Any = get_results(_a )
self.assertGreaterEqual(result['''eval_f1'''] , 3_0 )
self.assertGreaterEqual(result['''eval_exact'''] , 3_0 )
| 15
| 0
|
import os
import tempfile
import unittest
from transformers.models.marian.convert_marian_tatoeba_to_pytorch import DEFAULT_REPO, TatoebaConverter
from transformers.testing_utils import slow
from transformers.utils import cached_property
@unittest.skipUnless(os.path.exists(lowerCamelCase__ ) , "Tatoeba directory does not exist." )
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def __lowercase ( self ) -> Dict:
_a : List[Any] = tempfile.mkdtemp()
return TatoebaConverter(save_dir=__snake_case )
@slow
def __lowercase ( self ) -> List[Any]:
self.resolver.convert_models(['''heb-eng'''] )
@slow
def __lowercase ( self ) -> List[str]:
_a : Optional[int] = self.resolver.write_model_card('''opus-mt-he-en''' , dry_run=__snake_case )
assert mmeta["long_pair"] == "heb-eng"
| 357
|
import argparse
import os
import re
import packaging.version
a__ = '''examples/'''
a__ = {
'''examples''': (re.compile(R'''^check_min_version\("[^"]+"\)\s*$''', re.MULTILINE), '''check_min_version("VERSION")\n'''),
'''init''': (re.compile(R'''^__version__\s+=\s+"([^"]+)"\s*$''', re.MULTILINE), '''__version__ = "VERSION"\n'''),
'''setup''': (re.compile(R'''^(\s*)version\s*=\s*"[^"]+",''', re.MULTILINE), R'''\1version="VERSION",'''),
'''doc''': (re.compile(R'''^(\s*)release\s*=\s*"[^"]+"$''', re.MULTILINE), '''release = "VERSION"\n'''),
}
a__ = {
'''init''': '''src/transformers/__init__.py''',
'''setup''': '''setup.py''',
}
a__ = '''README.md'''
def __UpperCAmelCase ( __a : List[str] ,__a : int ,__a : Optional[Any] ) -> int:
"""simple docstring"""
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Tuple = f.read()
_a , _a : str = REPLACE_PATTERNS[pattern]
_a : List[str] = replace.replace('''VERSION''' ,__a )
_a : List[Any] = re_pattern.sub(__a ,__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.write(__a )
def __UpperCAmelCase ( __a : Any ) -> List[Any]:
"""simple docstring"""
for folder, directories, fnames in os.walk(__a ):
# Removing some of the folders with non-actively maintained examples from the walk
if "research_projects" in directories:
directories.remove('''research_projects''' )
if "legacy" in directories:
directories.remove('''legacy''' )
for fname in fnames:
if fname.endswith('''.py''' ):
update_version_in_file(os.path.join(__a ,__a ) ,__a ,pattern='''examples''' )
def __UpperCAmelCase ( __a : List[Any] ,__a : List[str]=False ) -> int:
"""simple docstring"""
for pattern, fname in REPLACE_FILES.items():
update_version_in_file(__a ,__a ,__a )
if not patch:
update_version_in_examples(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
_a : Optional[Any] = '''🤗 Transformers currently provides the following architectures'''
_a : str = '''1. Want to contribute a new model?'''
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Optional[int] = f.readlines()
# Find the start of the list.
_a : Optional[int] = 0
while not lines[start_index].startswith(_start_prompt ):
start_index += 1
start_index += 1
_a : List[Any] = start_index
# Update the lines in the model list.
while not lines[index].startswith(_end_prompt ):
if lines[index].startswith('''1.''' ):
_a : Tuple = lines[index].replace(
'''https://huggingface.co/docs/transformers/main/model_doc''' ,'''https://huggingface.co/docs/transformers/model_doc''' ,)
index += 1
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.writelines(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
with open(REPLACE_FILES['''init'''] ,'''r''' ) as f:
_a : Optional[Any] = f.read()
_a : Optional[Any] = REPLACE_PATTERNS['''init'''][0].search(__a ).groups()[0]
return packaging.version.parse(__a )
def __UpperCAmelCase ( __a : Dict=False ) -> str:
"""simple docstring"""
_a : Optional[Any] = get_version()
if patch and default_version.is_devrelease:
raise ValueError('''Can\'t create a patch version from the dev branch, checkout a released version!''' )
if default_version.is_devrelease:
_a : List[Any] = default_version.base_version
elif patch:
_a : str = F"""{default_version.major}.{default_version.minor}.{default_version.micro + 1}"""
else:
_a : List[str] = F"""{default_version.major}.{default_version.minor + 1}.0"""
# Now let's ask nicely if that's the right one.
_a : Dict = input(F"""Which version are you releasing? [{default_version}]""" )
if len(__a ) == 0:
_a : int = default_version
print(F"""Updating version to {version}.""" )
global_version_update(__a ,patch=__a )
if not patch:
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
def __UpperCAmelCase ( ) -> Tuple:
"""simple docstring"""
_a : str = get_version()
_a : int = F"""{current_version.major}.{current_version.minor + 1}.0.dev0"""
_a : List[Any] = current_version.base_version
# Check with the user we got that right.
_a : Union[str, Any] = input(F"""Which version are we developing now? [{dev_version}]""" )
if len(__a ) == 0:
_a : List[str] = dev_version
print(F"""Updating version to {version}.""" )
global_version_update(__a )
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--post_release''', action='''store_true''', help='''Whether this is pre or post release.''')
parser.add_argument('''--patch''', action='''store_true''', help='''Whether or not this is a patch release.''')
a__ = parser.parse_args()
if not args.post_release:
pre_release_work(patch=args.patch)
elif args.patch:
print('''Nothing to do after a patch :-)''')
else:
post_release_work()
| 15
| 0
|
from datetime import datetime
import matplotlib.pyplot as plt
import torch
def __UpperCAmelCase ( __a : Optional[Any] ) -> List[str]:
"""simple docstring"""
for param in module.parameters():
_a : List[str] = False
def __UpperCAmelCase ( ) -> Dict:
"""simple docstring"""
_a : Union[str, Any] = '''cuda''' if torch.cuda.is_available() else '''cpu'''
if torch.backends.mps.is_available() and torch.backends.mps.is_built():
_a : Optional[int] = '''mps'''
if device == "mps":
print(
'''WARNING: MPS currently doesn\'t seem to work, and messes up backpropagation without any visible torch'''
''' errors. I recommend using CUDA on a colab notebook or CPU instead if you\'re facing inexplicable issues'''
''' with generations.''' )
return device
def __UpperCAmelCase ( __a : List[Any] ) -> int:
"""simple docstring"""
_a : Tuple = plt.imshow(lowerCAmelCase__ )
fig.axes.get_xaxis().set_visible(lowerCAmelCase__ )
fig.axes.get_yaxis().set_visible(lowerCAmelCase__ )
plt.show()
def __UpperCAmelCase ( ) -> Optional[int]:
"""simple docstring"""
_a : Optional[Any] = datetime.now()
_a : int = current_time.strftime('''%H:%M:%S''' )
return timestamp
| 358
|
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
if n == 1 or not isinstance(__a ,__a ):
return 0
elif n == 2:
return 1
else:
_a : Any = [0, 1]
for i in range(2 ,n + 1 ):
sequence.append(sequence[i - 1] + sequence[i - 2] )
return sequence[n]
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
_a : Any = 0
_a : Dict = 2
while digits < n:
index += 1
_a : Dict = len(str(fibonacci(__a ) ) )
return index
def __UpperCAmelCase ( __a : int = 1_000 ) -> int:
"""simple docstring"""
return fibonacci_digits_index(__a )
if __name__ == "__main__":
print(solution(int(str(input()).strip())))
| 15
| 0
|
from statistics import mean
import numpy as np
def __UpperCAmelCase ( __a : Optional[int] ,__a : Optional[Any] ,__a : List[Any] ,__a : int ) -> list:
"""simple docstring"""
_a : Any = 0
# Number of processes finished
_a : Tuple = 0
# Displays the finished process.
# If it is 0, the performance is completed if it is 1, before the performance.
_a : Optional[int] = [0] * no_of_process
# List to include calculation results
_a : Optional[Any] = [0] * no_of_process
# Sort by arrival time.
_a : Optional[int] = [burst_time[i] for i in np.argsort(__a )]
_a : Any = [process_name[i] for i in np.argsort(__a )]
arrival_time.sort()
while no_of_process > finished_process_count:
_a : str = 0
while finished_process[i] == 1:
i += 1
if current_time < arrival_time[i]:
_a : str = arrival_time[i]
_a : Optional[Any] = 0
# Index showing the location of the process being performed
_a : Optional[Any] = 0
# Saves the current response ratio.
_a : Dict = 0
for i in range(0 ,__a ):
if finished_process[i] == 0 and arrival_time[i] <= current_time:
_a : int = (burst_time[i] + (current_time - arrival_time[i])) / burst_time[
i
]
if response_ratio < temp:
_a : str = temp
_a : int = i
# Calculate the turn around time
_a : Union[str, Any] = current_time + burst_time[loc] - arrival_time[loc]
current_time += burst_time[loc]
# Indicates that the process has been performed.
_a : Dict = 1
# Increase finished_process_count by 1
finished_process_count += 1
return turn_around_time
def __UpperCAmelCase ( __a : Tuple ,__a : Optional[Any] ,__a : Union[str, Any] ,__a : List[str] ) -> list:
"""simple docstring"""
_a : Union[str, Any] = [0] * no_of_process
for i in range(0 ,__a ):
_a : Optional[Any] = turn_around_time[i] - burst_time[i]
return waiting_time
if __name__ == "__main__":
a__ = 5
a__ = ['''A''', '''B''', '''C''', '''D''', '''E''']
a__ = [1, 2, 3, 4, 5]
a__ = [1, 2, 3, 4, 5]
a__ = calculate_turn_around_time(
process_name, arrival_time, burst_time, no_of_process
)
a__ = calculate_waiting_time(
process_name, turn_around_time, burst_time, no_of_process
)
print('''Process name \tArrival time \tBurst time \tTurn around time \tWaiting time''')
for i in range(0, no_of_process):
print(
f'''{process_name[i]}\t\t{arrival_time[i]}\t\t{burst_time[i]}\t\t'''
f'''{turn_around_time[i]}\t\t\t{waiting_time[i]}'''
)
print(f'''average waiting time : {mean(waiting_time):.5f}''')
print(f'''average turn around time : {mean(turn_around_time):.5f}''')
| 359
|
from sklearn.metrics import fa_score, matthews_corrcoef
import datasets
from .record_evaluation import evaluate as evaluate_record
a__ = '''\
@article{wang2019superglue,
title={SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems},
author={Wang, Alex and Pruksachatkun, Yada and Nangia, Nikita and Singh, Amanpreet and Michael, Julian and Hill, Felix and Levy, Omer and Bowman, Samuel R},
journal={arXiv preprint arXiv:1905.00537},
year={2019}
}
'''
a__ = '''\
SuperGLUE (https://super.gluebenchmark.com/) is a new benchmark styled after
GLUE with a new set of more difficult language understanding tasks, improved
resources, and a new public leaderboard.
'''
a__ = '''
Compute SuperGLUE evaluation metric associated to each SuperGLUE dataset.
Args:
predictions: list of predictions to score. Depending on the SuperGlUE subset:
- for \'record\': list of question-answer dictionaries with the following keys:
- \'idx\': index of the question as specified by the dataset
- \'prediction_text\': the predicted answer text
- for \'multirc\': list of question-answer dictionaries with the following keys:
- \'idx\': index of the question-answer pair as specified by the dataset
- \'prediction\': the predicted answer label
- otherwise: list of predicted labels
references: list of reference labels. Depending on the SuperGLUE subset:
- for \'record\': list of question-answers dictionaries with the following keys:
- \'idx\': index of the question as specified by the dataset
- \'answers\': list of possible answers
- otherwise: list of reference labels
Returns: depending on the SuperGLUE subset:
- for \'record\':
- \'exact_match\': Exact match between answer and gold answer
- \'f1\': F1 score
- for \'multirc\':
- \'exact_match\': Exact match between answer and gold answer
- \'f1_m\': Per-question macro-F1 score
- \'f1_a\': Average F1 score over all answers
- for \'axb\':
\'matthews_correlation\': Matthew Correlation
- for \'cb\':
- \'accuracy\': Accuracy
- \'f1\': F1 score
- for all others:
- \'accuracy\': Accuracy
Examples:
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'copa\') # any of ["copa", "rte", "wic", "wsc", "wsc.fixed", "boolq", "axg"]
>>> predictions = [0, 1]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'accuracy\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'cb\')
>>> predictions = [0, 1]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'accuracy\': 1.0, \'f1\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'record\')
>>> predictions = [{\'idx\': {\'passage\': 0, \'query\': 0}, \'prediction_text\': \'answer\'}]
>>> references = [{\'idx\': {\'passage\': 0, \'query\': 0}, \'answers\': [\'answer\', \'another_answer\']}]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 1.0, \'f1\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'multirc\')
>>> predictions = [{\'idx\': {\'answer\': 0, \'paragraph\': 0, \'question\': 0}, \'prediction\': 0}, {\'idx\': {\'answer\': 1, \'paragraph\': 2, \'question\': 3}, \'prediction\': 1}]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 1.0, \'f1_m\': 1.0, \'f1_a\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'axb\')
>>> references = [0, 1]
>>> predictions = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'matthews_correlation\': 1.0}
'''
def __UpperCAmelCase ( __a : int ,__a : List[str] ) -> Optional[Any]:
"""simple docstring"""
return float((preds == labels).mean() )
def __UpperCAmelCase ( __a : List[Any] ,__a : Union[str, Any] ,__a : List[str]="binary" ) -> Optional[int]:
"""simple docstring"""
_a : List[str] = simple_accuracy(__a ,__a )
_a : Any = float(fa_score(y_true=__a ,y_pred=__a ,average=__a ) )
return {
"accuracy": acc,
"f1": fa,
}
def __UpperCAmelCase ( __a : Optional[Any] ,__a : str ) -> List[Any]:
"""simple docstring"""
_a : Union[str, Any] = {}
for id_pred, label in zip(__a ,__a ):
_a : Optional[int] = F"""{id_pred['idx']['paragraph']}-{id_pred['idx']['question']}"""
_a : Optional[Any] = id_pred['''prediction''']
if question_id in question_map:
question_map[question_id].append((pred, label) )
else:
_a : str = [(pred, label)]
_a , _a : Any = [], []
for question, preds_labels in question_map.items():
_a , _a : Any = zip(*__a )
_a : List[Any] = fa_score(y_true=__a ,y_pred=__a ,average='''macro''' )
fas.append(__a )
_a : List[str] = int(sum(pred == label for pred, label in preds_labels ) == len(__a ) )
ems.append(__a )
_a : List[str] = float(sum(__a ) / len(__a ) )
_a : str = sum(__a ) / len(__a )
_a : Optional[int] = float(fa_score(y_true=__a ,y_pred=[id_pred['''prediction'''] for id_pred in ids_preds] ) )
return {"exact_match": em, "f1_m": fa_m, "f1_a": fa_a}
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> List[Any]:
if self.config_name not in [
"boolq",
"cb",
"copa",
"multirc",
"record",
"rte",
"wic",
"wsc",
"wsc.fixed",
"axb",
"axg",
]:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["boolq", "cb", "copa", "multirc", "record", "rte", "wic", "wsc", "wsc.fixed", "axb", "axg",]''' )
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(self._get_feature_types() ) , codebase_urls=[] , reference_urls=[] , format='''numpy''' if not self.config_name == '''record''' and not self.config_name == '''multirc''' else None , )
def __lowercase ( self ) -> Any:
if self.config_name == "record":
return {
"predictions": {
"idx": {
"passage": datasets.Value('''int64''' ),
"query": datasets.Value('''int64''' ),
},
"prediction_text": datasets.Value('''string''' ),
},
"references": {
"idx": {
"passage": datasets.Value('''int64''' ),
"query": datasets.Value('''int64''' ),
},
"answers": datasets.Sequence(datasets.Value('''string''' ) ),
},
}
elif self.config_name == "multirc":
return {
"predictions": {
"idx": {
"answer": datasets.Value('''int64''' ),
"paragraph": datasets.Value('''int64''' ),
"question": datasets.Value('''int64''' ),
},
"prediction": datasets.Value('''int64''' ),
},
"references": datasets.Value('''int64''' ),
}
else:
return {
"predictions": datasets.Value('''int64''' ),
"references": datasets.Value('''int64''' ),
}
def __lowercase ( self , _a , _a ) -> Optional[Any]:
if self.config_name == "axb":
return {"matthews_correlation": matthews_corrcoef(_a , _a )}
elif self.config_name == "cb":
return acc_and_fa(_a , _a , fa_avg='''macro''' )
elif self.config_name == "record":
_a : Any = [
{
'''qas''': [
{'''id''': ref['''idx''']['''query'''], '''answers''': [{'''text''': ans} for ans in ref['''answers''']]}
for ref in references
]
}
]
_a : Any = {pred['''idx''']['''query''']: pred['''prediction_text'''] for pred in predictions}
return evaluate_record(_a , _a )[0]
elif self.config_name == "multirc":
return evaluate_multirc(_a , _a )
elif self.config_name in ["copa", "rte", "wic", "wsc", "wsc.fixed", "boolq", "axg"]:
return {"accuracy": simple_accuracy(_a , _a )}
else:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["boolq", "cb", "copa", "multirc", "record", "rte", "wic", "wsc", "wsc.fixed", "axb", "axg",]''' )
| 15
| 0
|
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_tokenizers_available, is_torch_available
a__ = {
"configuration_bloom": ["BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP", "BloomConfig", "BloomOnnxConfig"],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = ["BloomTokenizerFast"]
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
"BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST",
"BloomForCausalLM",
"BloomModel",
"BloomPreTrainedModel",
"BloomForSequenceClassification",
"BloomForTokenClassification",
"BloomForQuestionAnswering",
]
if TYPE_CHECKING:
from .configuration_bloom import BLOOM_PRETRAINED_CONFIG_ARCHIVE_MAP, BloomConfig, BloomOnnxConfig
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_bloom_fast import BloomTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_bloom import (
BLOOM_PRETRAINED_MODEL_ARCHIVE_LIST,
BloomForCausalLM,
BloomForQuestionAnswering,
BloomForSequenceClassification,
BloomForTokenClassification,
BloomModel,
BloomPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 360
|
import numpy as np
def __UpperCAmelCase ( __a : np.ndarray ,__a : np.ndarray ,__a : float = 1E-12 ,__a : int = 100 ,) -> tuple[float, np.ndarray]:
"""simple docstring"""
assert np.shape(__a )[0] == np.shape(__a )[1]
# Ensure proper dimensionality.
assert np.shape(__a )[0] == np.shape(__a )[0]
# Ensure inputs are either both complex or both real
assert np.iscomplexobj(__a ) == np.iscomplexobj(__a )
_a : List[str] = np.iscomplexobj(__a )
if is_complex:
# Ensure complex input_matrix is Hermitian
assert np.array_equal(__a ,input_matrix.conj().T )
# Set convergence to False. Will define convergence when we exceed max_iterations
# or when we have small changes from one iteration to next.
_a : List[str] = False
_a : List[str] = 0
_a : Tuple = 0
_a : str = 1E12
while not convergence:
# Multiple matrix by the vector.
_a : str = np.dot(__a ,__a )
# Normalize the resulting output vector.
_a : List[Any] = w / np.linalg.norm(__a )
# Find rayleigh quotient
# (faster than usual b/c we know vector is normalized already)
_a : Dict = vector.conj().T if is_complex else vector.T
_a : Tuple = np.dot(__a ,np.dot(__a ,__a ) )
# Check convergence.
_a : List[str] = np.abs(lambda_ - lambda_previous ) / lambda_
iterations += 1
if error <= error_tol or iterations >= max_iterations:
_a : Dict = True
_a : str = lambda_
if is_complex:
_a : Tuple = np.real(lambda_ )
return lambda_, vector
def __UpperCAmelCase ( ) -> None:
"""simple docstring"""
_a : List[str] = np.array([[41, 4, 20], [4, 26, 30], [20, 30, 50]] )
_a : int = np.array([41, 4, 20] )
_a : Optional[Any] = real_input_matrix.astype(np.complexaaa )
_a : int = np.triu(1j * complex_input_matrix ,1 )
complex_input_matrix += imag_matrix
complex_input_matrix += -1 * imag_matrix.T
_a : Union[str, Any] = np.array([41, 4, 20] ).astype(np.complexaaa )
for problem_type in ["real", "complex"]:
if problem_type == "real":
_a : Optional[int] = real_input_matrix
_a : Union[str, Any] = real_vector
elif problem_type == "complex":
_a : str = complex_input_matrix
_a : str = complex_vector
# Our implementation.
_a , _a : Optional[Any] = power_iteration(__a ,__a )
# Numpy implementation.
# Get eigenvalues and eigenvectors using built-in numpy
# eigh (eigh used for symmetric or hermetian matrices).
_a , _a : List[str] = np.linalg.eigh(__a )
# Last eigenvalue is the maximum one.
_a : Tuple = eigen_values[-1]
# Last column in this matrix is eigenvector corresponding to largest eigenvalue.
_a : List[Any] = eigen_vectors[:, -1]
# Check our implementation and numpy gives close answers.
assert np.abs(eigen_value - eigen_value_max ) <= 1E-6
# Take absolute values element wise of each eigenvector.
# as they are only unique to a minus sign.
assert np.linalg.norm(np.abs(__a ) - np.abs(__a ) ) <= 1E-6
if __name__ == "__main__":
import doctest
doctest.testmod()
test_power_iteration()
| 15
| 0
|
import pickle
import unittest
import torch
from accelerate import Accelerator
from accelerate.state import AcceleratorState
from accelerate.test_utils import require_cpu
@require_cpu
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> int:
_a : int = torch.nn.Linear(1_0 , 1_0 )
_a : Any = torch.optim.SGD(model.parameters() , 0.1 )
_a : Tuple = Accelerator()
_a : List[Any] = accelerator.prepare(__a )
try:
pickle.loads(pickle.dumps(__a ) )
except Exception as e:
self.fail(F"""Accelerated optimizer pickling failed with {e}""" )
AcceleratorState._reset_state()
| 361
|
import itertools
from dataclasses import dataclass
from typing import Optional
import pandas as pd
import pyarrow as pa
import datasets
from datasets.table import table_cast
@dataclass
class UpperCAmelCase_ ( datasets.BuilderConfig ):
"""simple docstring"""
UpperCAmelCase__ : Optional[datasets.Features] = None
class UpperCAmelCase_ ( datasets.ArrowBasedBuilder ):
"""simple docstring"""
UpperCAmelCase__ : Any = PandasConfig
def __lowercase ( self ) -> Any:
return datasets.DatasetInfo(features=self.config.features )
def __lowercase ( self , _a ) -> List[Any]:
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}""" )
_a : str = dl_manager.download_and_extract(self.config.data_files )
if isinstance(_a , (str, list, tuple) ):
_a : Dict = data_files
if isinstance(_a , _a ):
_a : Dict = [files]
# Use `dl_manager.iter_files` to skip hidden files in an extracted archive
_a : int = [dl_manager.iter_files(_a ) for file in files]
return [datasets.SplitGenerator(name=datasets.Split.TRAIN , gen_kwargs={'''files''': files} )]
_a : Optional[Any] = []
for split_name, files in data_files.items():
if isinstance(_a , _a ):
_a : List[str] = [files]
# Use `dl_manager.iter_files` to skip hidden files in an extracted archive
_a : Any = [dl_manager.iter_files(_a ) for file in files]
splits.append(datasets.SplitGenerator(name=_a , gen_kwargs={'''files''': files} ) )
return splits
def __lowercase ( self , _a ) -> pa.Table:
if self.config.features is not None:
# more expensive cast to support nested features with keys in a different order
# allows str <-> int/float or str to Audio for example
_a : Optional[Any] = table_cast(_a , self.config.features.arrow_schema )
return pa_table
def __lowercase ( self , _a ) -> List[str]:
for i, file in enumerate(itertools.chain.from_iterable(_a ) ):
with open(_a , '''rb''' ) as f:
_a : str = pa.Table.from_pandas(pd.read_pickle(_a ) )
yield i, self._cast_table(_a )
| 15
| 0
|
from typing import List, Optional, Union
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a__ = logging.get_logger(__name__)
a__ = {
'''huggingface/time-series-transformer-tourism-monthly''': (
'''https://huggingface.co/huggingface/time-series-transformer-tourism-monthly/resolve/main/config.json'''
),
# See all TimeSeriesTransformer models at https://huggingface.co/models?filter=time_series_transformer
}
class UpperCAmelCase_ ( _snake_case ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = "time_series_transformer"
UpperCAmelCase__ : str = {
"hidden_size": "d_model",
"num_attention_heads": "encoder_attention_heads",
"num_hidden_layers": "encoder_layers",
}
def __init__( self , _a = None , _a = None , _a = "student_t" , _a = "nll" , _a = 1 , _a = [1, 2, 3, 4, 5, 6, 7] , _a = "mean" , _a = 0 , _a = 0 , _a = 0 , _a = 0 , _a = None , _a = None , _a = 3_2 , _a = 3_2 , _a = 2 , _a = 2 , _a = 2 , _a = 2 , _a = True , _a = "gelu" , _a = 6_4 , _a = 0.1 , _a = 0.1 , _a = 0.1 , _a = 0.1 , _a = 0.1 , _a = 1_0_0 , _a = 0.02 , _a=True , **_a , ) -> Optional[int]:
_a : Optional[int] = prediction_length
_a : Tuple = context_length or prediction_length
_a : Union[str, Any] = distribution_output
_a : Any = loss
_a : Optional[Any] = input_size
_a : Optional[int] = num_time_features
_a : Tuple = lags_sequence
_a : Any = scaling
_a : int = num_dynamic_real_features
_a : Dict = num_static_real_features
_a : Union[str, Any] = num_static_categorical_features
if cardinality and num_static_categorical_features > 0:
if len(UpperCamelCase__ ) != num_static_categorical_features:
raise ValueError(
'''The cardinality should be a list of the same length as `num_static_categorical_features`''' )
_a : int = cardinality
else:
_a : Tuple = [0]
if embedding_dimension and num_static_categorical_features > 0:
if len(UpperCamelCase__ ) != num_static_categorical_features:
raise ValueError(
'''The embedding dimension should be a list of the same length as `num_static_categorical_features`''' )
_a : Tuple = embedding_dimension
else:
_a : Dict = [min(5_0 , (cat + 1) // 2 ) for cat in self.cardinality]
_a : str = num_parallel_samples
# Transformer architecture configuration
_a : Tuple = input_size * len(UpperCamelCase__ ) + self._number_of_features
_a : Union[str, Any] = d_model
_a : str = encoder_attention_heads
_a : str = decoder_attention_heads
_a : Dict = encoder_ffn_dim
_a : List[str] = decoder_ffn_dim
_a : List[Any] = encoder_layers
_a : Union[str, Any] = decoder_layers
_a : Optional[int] = dropout
_a : int = attention_dropout
_a : Any = activation_dropout
_a : Optional[Any] = encoder_layerdrop
_a : Union[str, Any] = decoder_layerdrop
_a : Tuple = activation_function
_a : Union[str, Any] = init_std
_a : Union[str, Any] = use_cache
super().__init__(is_encoder_decoder=UpperCamelCase__ , **UpperCamelCase__ )
@property
def __lowercase ( self ) -> int:
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
)
| 362
|
def __UpperCAmelCase ( __a : int ,__a : int ,__a : int ) -> int:
"""simple docstring"""
if exponent == 1:
return base
if exponent % 2 == 0:
_a : List[Any] = _modexpt(__a ,exponent // 2 ,__a ) % modulo_value
return (x * x) % modulo_value
else:
return (base * _modexpt(__a ,exponent - 1 ,__a )) % modulo_value
def __UpperCAmelCase ( __a : int = 1_777 ,__a : int = 1_855 ,__a : int = 8 ) -> int:
"""simple docstring"""
_a : List[Any] = base
for _ in range(1 ,__a ):
_a : Any = _modexpt(__a ,__a ,10**digits )
return result
if __name__ == "__main__":
print(f'''{solution() = }''')
| 15
| 0
|
import inspect
import os
import unittest
import torch
import accelerate
from accelerate import debug_launcher
from accelerate.test_utils import (
execute_subprocess_async,
require_cpu,
require_huggingface_suite,
require_multi_gpu,
require_single_gpu,
)
from accelerate.utils import patch_environment
@require_huggingface_suite
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Dict:
_a : Union[str, Any] = inspect.getfile(accelerate.test_utils )
_a : List[Any] = os.path.sep.join(
mod_file.split(os.path.sep )[:-1] + ['''scripts''', '''external_deps''', '''test_metrics.py'''] )
from accelerate.test_utils.scripts.external_deps import test_metrics # noqa: F401
_a : List[Any] = test_metrics
@require_cpu
def __lowercase ( self ) -> Optional[Any]:
debug_launcher(self.test_metrics.main , num_processes=1 )
@require_cpu
def __lowercase ( self ) -> Optional[Any]:
debug_launcher(self.test_metrics.main )
@require_single_gpu
def __lowercase ( self ) -> List[Any]:
self.test_metrics.main()
@require_multi_gpu
def __lowercase ( self ) -> Dict:
print(F"""Found {torch.cuda.device_count()} devices.""" )
_a : int = ['''torchrun''', F"""--nproc_per_node={torch.cuda.device_count()}""", self.test_file_path]
with patch_environment(omp_num_threads=1 ):
execute_subprocess_async(_A , env=os.environ.copy() )
| 363
|
import numpy as np
import torch
from torch.nn import CrossEntropyLoss
from transformers import AutoModelForCausalLM, AutoTokenizer
import datasets
from datasets import logging
a__ = '''\
'''
a__ = '''
Perplexity (PPL) is one of the most common metrics for evaluating language models.
It is defined as the exponentiated average negative log-likelihood of a sequence.
For more information, see https://huggingface.co/docs/transformers/perplexity
'''
a__ = '''
Args:
model_id (str): model used for calculating Perplexity
NOTE: Perplexity can only be calculated for causal language models.
This includes models such as gpt2, causal variations of bert,
causal versions of t5, and more (the full list can be found
in the AutoModelForCausalLM documentation here:
https://huggingface.co/docs/transformers/master/en/model_doc/auto#transformers.AutoModelForCausalLM )
input_texts (list of str): input text, each separate text snippet
is one list entry.
batch_size (int): the batch size to run texts through the model. Defaults to 16.
add_start_token (bool): whether to add the start token to the texts,
so the perplexity can include the probability of the first word. Defaults to True.
device (str): device to run on, defaults to \'cuda\' when available
Returns:
perplexity: dictionary containing the perplexity scores for the texts
in the input list, as well as the mean perplexity. If one of the input texts is
longer than the max input length of the model, then it is truncated to the
max length for the perplexity computation.
Examples:
Example 1:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = ["lorem ipsum", "Happy Birthday!", "Bienvenue"]
>>> results = perplexity.compute(model_id=\'gpt2\',
... add_start_token=False,
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
78.22
>>> print(round(results["perplexities"][0], 2))
11.11
Example 2:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = datasets.load_dataset("wikitext",
... "wikitext-2-raw-v1",
... split="test")["text"][:50] # doctest:+ELLIPSIS
[...]
>>> input_texts = [s for s in input_texts if s!=\'\']
>>> results = perplexity.compute(model_id=\'gpt2\',
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
60.35
>>> print(round(results["perplexities"][0], 2))
81.12
'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> Any:
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(
{
'''input_texts''': datasets.Value('''string''' ),
} ) , reference_urls=['''https://huggingface.co/docs/transformers/perplexity'''] , )
def __lowercase ( self , _a , _a , _a = 1_6 , _a = True , _a=None ) -> List[Any]:
if device is not None:
assert device in ["gpu", "cpu", "cuda"], "device should be either gpu or cpu."
if device == "gpu":
_a : List[str] = '''cuda'''
else:
_a : Optional[Any] = '''cuda''' if torch.cuda.is_available() else '''cpu'''
_a : Dict = AutoModelForCausalLM.from_pretrained(_a )
_a : List[Any] = model.to(_a )
_a : List[str] = AutoTokenizer.from_pretrained(_a )
# if batch_size > 1 (which generally leads to padding being required), and
# if there is not an already assigned pad_token, assign an existing
# special token to also be the padding token
if tokenizer.pad_token is None and batch_size > 1:
_a : str = list(tokenizer.special_tokens_map_extended.values() )
# check that the model already has at least one special token defined
assert (
len(_a ) > 0
), "If batch_size > 1, model must have at least one special token to use for padding. Please use a different model or set batch_size=1."
# assign one of the special tokens to also be the pad token
tokenizer.add_special_tokens({'''pad_token''': existing_special_tokens[0]} )
if add_start_token:
# leave room for <BOS> token to be added:
assert (
tokenizer.bos_token is not None
), "Input model must already have a BOS token if using add_start_token=True. Please use a different model, or set add_start_token=False"
_a : List[Any] = model.config.max_length - 1
else:
_a : List[str] = model.config.max_length
_a : Union[str, Any] = tokenizer(
_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , return_tensors='''pt''' , return_attention_mask=_a , ).to(_a )
_a : List[Any] = encodings['''input_ids''']
_a : int = encodings['''attention_mask''']
# check that each input is long enough:
if add_start_token:
assert torch.all(torch.ge(attn_masks.sum(1 ) , 1 ) ), "Each input text must be at least one token long."
else:
assert torch.all(
torch.ge(attn_masks.sum(1 ) , 2 ) ), "When add_start_token=False, each input text must be at least two tokens long. Run with add_start_token=True if inputting strings of only one token, and remove all empty input strings."
_a : Optional[int] = []
_a : Dict = CrossEntropyLoss(reduction='''none''' )
for start_index in logging.tqdm(range(0 , len(_a ) , _a ) ):
_a : Dict = min(start_index + batch_size , len(_a ) )
_a : Union[str, Any] = encoded_texts[start_index:end_index]
_a : int = attn_masks[start_index:end_index]
if add_start_token:
_a : Dict = torch.tensor([[tokenizer.bos_token_id]] * encoded_batch.size(dim=0 ) ).to(_a )
_a : List[str] = torch.cat([bos_tokens_tensor, encoded_batch] , dim=1 )
_a : Dict = torch.cat(
[torch.ones(bos_tokens_tensor.size() , dtype=torch.intaa ).to(_a ), attn_mask] , dim=1 )
_a : Dict = encoded_batch
with torch.no_grad():
_a : Any = model(_a , attention_mask=_a ).logits
_a : List[str] = out_logits[..., :-1, :].contiguous()
_a : Union[str, Any] = labels[..., 1:].contiguous()
_a : Optional[int] = attn_mask[..., 1:].contiguous()
_a : Union[str, Any] = torch.expa(
(loss_fct(shift_logits.transpose(1 , 2 ) , _a ) * shift_attention_mask_batch).sum(1 )
/ shift_attention_mask_batch.sum(1 ) )
ppls += perplexity_batch.tolist()
return {"perplexities": ppls, "mean_perplexity": np.mean(_a )}
| 15
| 0
|
"""simple docstring"""
from __future__ import annotations
def __UpperCAmelCase ( __a : list[int] ) -> list[int]:
"""simple docstring"""
if len(snake_case__ ) == 0:
return array
_a : int = min(snake_case__ ), max(snake_case__ )
# Compute the variables
_a : Dict = _max - _min + 1
_a : Union[str, Any] = [0] * holes_range, [0] * holes_range
# Make the sorting.
for i in array:
_a : Union[str, Any] = i - _min
_a : Optional[int] = i
holes_repeat[index] += 1
# Makes the array back by replacing the numbers.
_a : Tuple = 0
for i in range(snake_case__ ):
while holes_repeat[i] > 0:
_a : int = holes[i]
index += 1
holes_repeat[i] -= 1
# Returns the sorted array.
return array
if __name__ == "__main__":
import doctest
doctest.testmod()
a__ = input('''Enter numbers separated by comma:\n''')
a__ = [int(x) for x in user_input.split(''',''')]
print(pigeon_sort(unsorted))
| 364
|
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a__ = {
'''configuration_xmod''': [
'''XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP''',
'''XmodConfig''',
'''XmodOnnxConfig''',
],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''XMOD_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''XmodForCausalLM''',
'''XmodForMaskedLM''',
'''XmodForMultipleChoice''',
'''XmodForQuestionAnswering''',
'''XmodForSequenceClassification''',
'''XmodForTokenClassification''',
'''XmodModel''',
'''XmodPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_xmod import XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP, XmodConfig, XmodOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_xmod import (
XMOD_PRETRAINED_MODEL_ARCHIVE_LIST,
XmodForCausalLM,
XmodForMaskedLM,
XmodForMultipleChoice,
XmodForQuestionAnswering,
XmodForSequenceClassification,
XmodForTokenClassification,
XmodModel,
XmodPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 15
| 0
|
from __future__ import annotations
import copy
import inspect
import json
import math
import os
import tempfile
import unittest
from importlib import import_module
import numpy as np
from transformers import ViTMAEConfig
from transformers.file_utils import cached_property, is_tf_available, is_vision_available
from transformers.testing_utils import require_tf, require_vision, slow
from ...test_configuration_common import ConfigTester
from ...test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor
from ...test_pipeline_mixin import PipelineTesterMixin
if is_tf_available():
import tensorflow as tf
from transformers import TFViTMAEForPreTraining, TFViTMAEModel
if is_vision_available():
from PIL import Image
from transformers import ViTImageProcessor
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self , _a , _a=1_3 , _a=3_0 , _a=2 , _a=3 , _a=True , _a=True , _a=3_2 , _a=2 , _a=4 , _a=3_7 , _a="gelu" , _a=0.1 , _a=0.1 , _a=1_0 , _a=0.02 , _a=3 , _a=0.6 , _a=None , ) -> int:
_a : Union[str, Any] = parent
_a : Optional[Any] = batch_size
_a : List[Any] = image_size
_a : int = patch_size
_a : Optional[int] = num_channels
_a : Optional[Any] = is_training
_a : Union[str, Any] = use_labels
_a : Any = hidden_size
_a : Tuple = num_hidden_layers
_a : int = num_attention_heads
_a : Optional[Any] = intermediate_size
_a : Dict = hidden_act
_a : int = hidden_dropout_prob
_a : List[str] = attention_probs_dropout_prob
_a : Tuple = type_sequence_label_size
_a : Optional[int] = initializer_range
_a : List[str] = mask_ratio
_a : Tuple = scope
# in ViTMAE, the expected sequence length = (num_patches + 1) * (1 - config.mask_ratio), rounded above
# (we add 1 for the [CLS] token)
_a : str = (image_size // patch_size) ** 2
_a : Optional[Any] = int(math.ceil((1 - mask_ratio) * (num_patches + 1) ) )
def __lowercase ( self ) -> Optional[Any]:
_a : Tuple = floats_tensor([self.batch_size, self.num_channels, self.image_size, self.image_size] )
_a : Union[str, Any] = None
if self.use_labels:
_a : Dict = ids_tensor([self.batch_size] , self.type_sequence_label_size )
_a : Optional[Any] = self.get_config()
return config, pixel_values, labels
def __lowercase ( self ) -> Any:
return ViTMAEConfig(
image_size=self.image_size , patch_size=self.patch_size , num_channels=self.num_channels , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , decoder_hidden_size=self.hidden_size , decoder_num_hidden_layers=self.num_hidden_layers , decoder_num_attention_heads=self.num_attention_heads , decoder_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=_a , initializer_range=self.initializer_range , mask_ratio=self.mask_ratio , )
def __lowercase ( self , _a , _a , _a ) -> Union[str, Any]:
_a : Tuple = TFViTMAEModel(config=_a )
_a : List[str] = model(_a , training=_a )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def __lowercase ( self , _a , _a , _a ) -> str:
_a : Optional[Any] = TFViTMAEForPreTraining(_a )
_a : str = model(_a , training=_a )
# expected sequence length = num_patches
_a : str = (self.image_size // self.patch_size) ** 2
_a : int = self.patch_size**2 * self.num_channels
self.parent.assertEqual(result.logits.shape , (self.batch_size, num_patches, expected_num_channels) )
# test greyscale images
_a : str = 1
_a : str = TFViTMAEForPreTraining(_a )
_a : Optional[Any] = floats_tensor([self.batch_size, 1, self.image_size, self.image_size] )
_a : int = model(_a , training=_a )
_a : int = self.patch_size**2
self.parent.assertEqual(result.logits.shape , (self.batch_size, num_patches, expected_num_channels) )
def __lowercase ( self ) -> List[Any]:
_a : str = self.prepare_config_and_inputs()
((_a) , (_a) , (_a)) : Optional[int] = config_and_inputs
_a : Dict = {'''pixel_values''': pixel_values}
return config, inputs_dict
@require_tf
class UpperCAmelCase_ ( _SCREAMING_SNAKE_CASE , _SCREAMING_SNAKE_CASE , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[int] = (TFViTMAEModel, TFViTMAEForPreTraining) if is_tf_available() else ()
UpperCAmelCase__ : List[Any] = {"feature-extraction": TFViTMAEModel} if is_tf_available() else {}
UpperCAmelCase__ : Union[str, Any] = False
UpperCAmelCase__ : int = False
UpperCAmelCase__ : Optional[Any] = False
UpperCAmelCase__ : Any = False
def __lowercase ( self ) -> Union[str, Any]:
_a : Dict = TFViTMAEModelTester(self )
_a : List[Any] = ConfigTester(self , config_class=_a , has_text_modality=_a , hidden_size=3_7 )
def __lowercase ( self ) -> Tuple:
self.config_tester.run_common_tests()
@unittest.skip(reason='''ViTMAE does not use inputs_embeds''' )
def __lowercase ( self ) -> str:
pass
def __lowercase ( self ) -> List[str]:
_a , _a : str = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
_a : Tuple = model_class(_a )
self.assertIsInstance(model.get_input_embeddings() , (tf.keras.layers.Layer) )
_a : Optional[int] = model.get_output_embeddings()
self.assertTrue(x is None or isinstance(_a , tf.keras.layers.Layer ) )
def __lowercase ( self ) -> List[Any]:
_a , _a : Dict = self.model_tester.prepare_config_and_inputs_for_common()
for model_class in self.all_model_classes:
_a : Dict = model_class(_a )
_a : str = inspect.signature(model.call )
# signature.parameters is an OrderedDict => so arg_names order is deterministic
_a : Any = [*signature.parameters.keys()]
_a : Dict = ['''pixel_values''']
self.assertListEqual(arg_names[:1] , _a )
def __lowercase ( self ) -> List[str]:
_a : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*_a )
def __lowercase ( self ) -> Any:
_a : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_pretraining(*_a )
def __lowercase ( self ) -> List[Any]:
# make the mask reproducible
np.random.seed(2 )
_a , _a : str = self.model_tester.prepare_config_and_inputs_for_common()
_a : Any = int((config.image_size // config.patch_size) ** 2 )
_a : Tuple = np.random.uniform(size=(self.model_tester.batch_size, num_patches) )
for model_class in self.all_model_classes:
_a : Optional[Any] = model_class(_a )
_a : int = self._prepare_for_class(_a , _a )
_a : List[str] = model(_a , noise=_a )
_a : Dict = copy.deepcopy(self._prepare_for_class(_a , _a ) )
_a : str = model(**_a , noise=_a )
_a : List[str] = outputs_dict[0].numpy()
_a : Union[str, Any] = outputs_keywords[0].numpy()
self.assertLess(np.sum(np.abs(output_dict - output_keywords ) ) , 1e-6 )
def __lowercase ( self ) -> Dict:
# make the mask reproducible
np.random.seed(2 )
_a , _a : str = self.model_tester.prepare_config_and_inputs_for_common()
_a : Optional[Any] = int((config.image_size // config.patch_size) ** 2 )
_a : Optional[Any] = np.random.uniform(size=(self.model_tester.batch_size, num_patches) )
def prepare_numpy_arrays(_a ):
_a : Optional[int] = {}
for k, v in inputs_dict.items():
if tf.is_tensor(_a ):
_a : Optional[Any] = v.numpy()
else:
_a : Dict = np.array(_a )
return inputs_np_dict
for model_class in self.all_model_classes:
_a : List[str] = model_class(_a )
_a : List[str] = self._prepare_for_class(_a , _a )
_a : Tuple = prepare_numpy_arrays(_a )
_a : Tuple = model(_a , noise=_a )
_a : int = model(**_a , noise=_a )
self.assert_outputs_same(_a , _a )
def __lowercase ( self , _a , _a , _a ) -> int:
# make masks reproducible
np.random.seed(2 )
_a : Union[str, Any] = int((tf_model.config.image_size // tf_model.config.patch_size) ** 2 )
_a : List[str] = np.random.uniform(size=(self.model_tester.batch_size, num_patches) )
_a : Union[str, Any] = tf.constant(_a )
# Add `noise` argument.
# PT inputs will be prepared in `super().check_pt_tf_models()` with this added `noise` argument
_a : Union[str, Any] = tf_noise
super().check_pt_tf_models(_a , _a , _a )
def __lowercase ( self ) -> Tuple:
# make mask reproducible
np.random.seed(2 )
_a , _a : Tuple = self.model_tester.prepare_config_and_inputs_for_common()
_a : Any = {
module_member
for model_class in self.all_model_classes
for module in (import_module(model_class.__module__ ),)
for module_member_name in dir(_a )
if module_member_name.endswith('''MainLayer''' )
# This condition is required, since `modeling_tf_clip.py` has 3 classes whose names end with `MainLayer`.
and module_member_name[: -len('''MainLayer''' )] == model_class.__name__[: -len('''Model''' )]
for module_member in (getattr(_a , _a ),)
if isinstance(_a , _a )
and tf.keras.layers.Layer in module_member.__bases__
and getattr(_a , '''_keras_serializable''' , _a )
}
_a : Tuple = int((config.image_size // config.patch_size) ** 2 )
_a : str = np.random.uniform(size=(self.model_tester.batch_size, num_patches) )
_a : Any = tf.convert_to_tensor(_a )
inputs_dict.update({'''noise''': noise} )
for main_layer_class in tf_main_layer_classes:
_a : Optional[Any] = main_layer_class(_a )
_a : Union[str, Any] = {
name: tf.keras.Input(tensor.shape[1:] , dtype=tensor.dtype ) for name, tensor in inputs_dict.items()
}
_a : Tuple = tf.keras.Model(_a , outputs=main_layer(_a ) )
_a : List[str] = model(_a )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : Union[str, Any] = os.path.join(_a , '''keras_model.h5''' )
model.save(_a )
_a : Tuple = tf.keras.models.load_model(
_a , custom_objects={main_layer_class.__name__: main_layer_class} )
assert isinstance(_a , tf.keras.Model )
_a : Any = model(_a )
self.assert_outputs_same(_a , _a )
@slow
def __lowercase ( self ) -> Optional[int]:
# make mask reproducible
np.random.seed(2 )
_a , _a : Optional[int] = self.model_tester.prepare_config_and_inputs_for_common()
_a : Optional[Any] = int((config.image_size // config.patch_size) ** 2 )
_a : str = np.random.uniform(size=(self.model_tester.batch_size, num_patches) )
for model_class in self.all_model_classes:
_a : str = model_class(_a )
_a : List[str] = self._prepare_for_class(_a , _a )
_a : List[str] = model(_a , noise=_a )
if model_class.__name__ == "TFViTMAEModel":
_a : Any = outputs.last_hidden_state.numpy()
_a : int = 0
else:
_a : Dict = outputs.logits.numpy()
_a : Tuple = 0
with tempfile.TemporaryDirectory() as tmpdirname:
model.save_pretrained(_a , saved_model=_a )
_a : List[Any] = model_class.from_pretrained(_a )
_a : List[str] = model(_a , noise=_a )
if model_class.__name__ == "TFViTMAEModel":
_a : Optional[int] = after_outputs['''last_hidden_state'''].numpy()
_a : List[Any] = 0
else:
_a : List[str] = after_outputs['''logits'''].numpy()
_a : Dict = 0
_a : Any = np.amax(np.abs(out_a - out_a ) )
self.assertLessEqual(_a , 1e-5 )
def __lowercase ( self ) -> Optional[Any]:
# make mask reproducible
np.random.seed(2 )
_a , _a : List[str] = self.model_tester.prepare_config_and_inputs_for_common()
_a : Tuple = int((config.image_size // config.patch_size) ** 2 )
_a : Optional[int] = np.random.uniform(size=(self.model_tester.batch_size, num_patches) )
for model_class in self.all_model_classes:
_a : List[str] = model_class(_a )
_a : str = self._prepare_for_class(_a , _a )
_a : Optional[Any] = model(_a , noise=_a )
_a : int = model.get_config()
# make sure that returned config is jsonifiable, which is required by keras
json.dumps(_a )
_a : Any = model_class.from_config(model.get_config() )
# make sure it also accepts a normal config
_a : Union[str, Any] = model_class.from_config(model.config )
_a : List[Any] = new_model(_a ) # Build model
new_model.set_weights(model.get_weights() )
_a : Tuple = new_model(_a , noise=_a )
self.assert_outputs_same(_a , _a )
@unittest.skip(
reason='''ViTMAE returns a random mask + ids_restore in each forward pass. See test_save_load\n to get deterministic results.''' )
def __lowercase ( self ) -> str:
pass
@unittest.skip(reason='''ViTMAE returns a random mask + ids_restore in each forward pass. See test_save_load''' )
def __lowercase ( self ) -> int:
pass
@slow
def __lowercase ( self ) -> List[Any]:
_a : Union[str, Any] = TFViTMAEModel.from_pretrained('''google/vit-base-patch16-224''' )
self.assertIsNotNone(_a )
def __UpperCAmelCase ( ) -> Optional[int]:
"""simple docstring"""
_a : Optional[Any] = Image.open('''./tests/fixtures/tests_samples/COCO/000000039769.png''' )
return image
@require_tf
@require_vision
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
@cached_property
def __lowercase ( self ) -> int:
return ViTImageProcessor.from_pretrained('''facebook/vit-mae-base''' ) if is_vision_available() else None
@slow
def __lowercase ( self ) -> Union[str, Any]:
# make random mask reproducible across the PT and TF model
np.random.seed(2 )
_a : Optional[Any] = TFViTMAEForPreTraining.from_pretrained('''facebook/vit-mae-base''' )
_a : Dict = self.default_image_processor
_a : int = prepare_img()
_a : Dict = image_processor(images=_a , return_tensors='''tf''' )
# prepare a noise vector that will be also used for testing the TF model
# (this way we can ensure that the PT and TF models operate on the same inputs)
_a : List[Any] = ViTMAEConfig()
_a : Optional[Any] = int((vit_mae_config.image_size // vit_mae_config.patch_size) ** 2 )
_a : List[str] = np.random.uniform(size=(1, num_patches) )
# forward pass
_a : Dict = model(**_a , noise=_a )
# verify the logits
_a : Optional[Any] = tf.convert_to_tensor([1, 1_9_6, 7_6_8] )
self.assertEqual(outputs.logits.shape , _a )
_a : int = tf.convert_to_tensor(
[[-0.0548, -1.7023, -0.9325], [0.3721, -0.5670, -0.2233], [0.8235, -1.3878, -0.3524]] )
tf.debugging.assert_near(outputs.logits[0, :3, :3] , _a , atol=1e-4 )
| 365
|
import re
import tempfile
from pathlib import Path
import pytest
import yaml
from datasets.utils.readme import ReadMe
# @pytest.fixture
# def example_yaml_structure():
a__ = yaml.safe_load(
'''\
name: ""
allow_empty: false
allow_empty_text: true
subsections:
- name: "Dataset Card for X" # First-level markdown heading
allow_empty: false
allow_empty_text: true
subsections:
- name: "Table of Contents"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Dataset Description"
allow_empty: false
allow_empty_text: false
subsections:
- name: "Dataset Summary"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Supported Tasks and Leaderboards"
allow_empty: true
allow_empty_text: true
subsections: null
- name: Languages
allow_empty: false
allow_empty_text: true
subsections: null
'''
)
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
#### Extra Ignored Subsection
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Extra Ignored Subsection''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
}
],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tEmpty YAML markers are present in the README.'''
)
a__ = '''\
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tNo YAML markers are present in the README.'''
)
a__ = '''\
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tOnly the start of YAML tags present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Summary` but it is empty.\n-\tExpected some text in section `Dataset Summary` but it is empty (text in subsections are ignored).'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Card for My Dataset` but it is empty.\n-\tSection `Dataset Card for My Dataset` expected the following subsections: `Table of Contents`, `Dataset Description`. Found \'None\'.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tSection `Dataset Description` is missing subsection: `Supported Tasks and Leaderboards`.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Languages` but it is empty.'''
a__ = '''\
---
language:
- zh
- en
---
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
# Dataset Card My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has several first-level headings: `Dataset Card for My Dataset`, `Dataset Card My Dataset`. Only one heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tNo first-level heading starting with `Dataset Card for` found in README. Skipping further validation for this README.'''
a__ = ''''''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.\n-\tNo YAML markers are present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found while parsing the README at `{path}`:\n-\tMultiple sections with the same heading `Dataset Card for My Dataset` have been found. Please keep only one of these sections.'''
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : List[str] ) -> Optional[int]:
"""simple docstring"""
assert ReadMe.from_string(__a ,__a ).to_dict() == expected_dict
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[str] ,__a : Optional[Any] ) -> Union[str, Any]:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
_a : List[Any] = ReadMe.from_string(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Dict ,__a : Dict ) -> Tuple:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
ReadMe.from_string(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> Tuple:
"""simple docstring"""
ReadMe.from_string(__a ,__a ,suppress_parsing_errors=__a )
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : Any ) -> Optional[int]:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Tuple = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[Any] = ReadMe.from_readme(__a ,__a ).to_dict()
assert out["name"] == path
assert out["text"] == ""
assert out["is_empty_text"]
assert out["subsections"] == expected_dict["subsections"]
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[Any] ,__a : List[Any] ) -> int:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[int] = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
_a : Any = ReadMe.from_readme(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : str ,__a : Union[str, Any] ) -> Dict:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Optional[Any] = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : str = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
ReadMe.from_readme(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> str:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
ReadMe.from_readme(__a ,__a ,suppress_parsing_errors=__a )
| 15
| 0
|
import unittest
from transformers import BertGenerationTokenizer
from transformers.testing_utils import get_tests_dir, require_sentencepiece, require_torch, slow
from transformers.utils import cached_property
from ...test_tokenization_common import TokenizerTesterMixin
a__ = '''▁'''
a__ = get_tests_dir('''fixtures/test_sentencepiece.model''')
@require_sentencepiece
class UpperCAmelCase_ ( UpperCAmelCase_ , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : int = BertGenerationTokenizer
UpperCAmelCase__ : Dict = False
UpperCAmelCase__ : str = True
def __lowercase ( self ) -> List[str]:
super().setUp()
_a : Optional[Any] = BertGenerationTokenizer(__lowercase , keep_accents=__lowercase )
tokenizer.save_pretrained(self.tmpdirname )
def __lowercase ( self ) -> int:
_a : Dict = '''<s>'''
_a : Dict = 1
self.assertEqual(self.get_tokenizer()._convert_token_to_id(__lowercase ) , __lowercase )
self.assertEqual(self.get_tokenizer()._convert_id_to_token(__lowercase ) , __lowercase )
def __lowercase ( self ) -> Optional[int]:
_a : Optional[Any] = list(self.get_tokenizer().get_vocab().keys() )
self.assertEqual(vocab_keys[0] , '''<unk>''' )
self.assertEqual(vocab_keys[1] , '''<s>''' )
self.assertEqual(vocab_keys[-1] , '''<pad>''' )
self.assertEqual(len(__lowercase ) , 1_0_0_2 )
def __lowercase ( self ) -> Dict:
self.assertEqual(self.get_tokenizer().vocab_size , 1_0_0_0 )
def __lowercase ( self ) -> Optional[Any]:
_a : str = BertGenerationTokenizer(__lowercase , keep_accents=__lowercase )
_a : Optional[Any] = tokenizer.tokenize('''This is a test''' )
self.assertListEqual(__lowercase , ['''▁This''', '''▁is''', '''▁a''', '''▁t''', '''est'''] )
self.assertListEqual(
tokenizer.convert_tokens_to_ids(__lowercase ) , [2_8_5, 4_6, 1_0, 1_7_0, 3_8_2] , )
_a : List[Any] = tokenizer.tokenize('''I was born in 92000, and this is falsé.''' )
self.assertListEqual(
__lowercase , [
SPIECE_UNDERLINE + '''I''',
SPIECE_UNDERLINE + '''was''',
SPIECE_UNDERLINE + '''b''',
'''or''',
'''n''',
SPIECE_UNDERLINE + '''in''',
SPIECE_UNDERLINE + '''''',
'''9''',
'''2''',
'''0''',
'''0''',
'''0''',
''',''',
SPIECE_UNDERLINE + '''and''',
SPIECE_UNDERLINE + '''this''',
SPIECE_UNDERLINE + '''is''',
SPIECE_UNDERLINE + '''f''',
'''al''',
'''s''',
'''é''',
'''.''',
] , )
_a : Any = tokenizer.convert_tokens_to_ids(__lowercase )
self.assertListEqual(
__lowercase , [8, 2_1, 8_4, 5_5, 2_4, 1_9, 7, 0, 6_0_2, 3_4_7, 3_4_7, 3_4_7, 3, 1_2, 6_6, 4_6, 7_2, 8_0, 6, 0, 4] , )
_a : Tuple = tokenizer.convert_ids_to_tokens(__lowercase )
self.assertListEqual(
__lowercase , [
SPIECE_UNDERLINE + '''I''',
SPIECE_UNDERLINE + '''was''',
SPIECE_UNDERLINE + '''b''',
'''or''',
'''n''',
SPIECE_UNDERLINE + '''in''',
SPIECE_UNDERLINE + '''''',
'''<unk>''',
'''2''',
'''0''',
'''0''',
'''0''',
''',''',
SPIECE_UNDERLINE + '''and''',
SPIECE_UNDERLINE + '''this''',
SPIECE_UNDERLINE + '''is''',
SPIECE_UNDERLINE + '''f''',
'''al''',
'''s''',
'''<unk>''',
'''.''',
] , )
@cached_property
def __lowercase ( self ) -> int:
return BertGenerationTokenizer.from_pretrained('''google/bert_for_seq_generation_L-24_bbc_encoder''' )
@slow
def __lowercase ( self ) -> List[str]:
_a : Optional[Any] = '''Hello World!'''
_a : Optional[int] = [1_8_5_3_6, 2_2_6_0, 1_0_1]
self.assertListEqual(__lowercase , self.big_tokenizer.encode(__lowercase ) )
@slow
def __lowercase ( self ) -> List[str]:
_a : Optional[Any] = (
'''This is a very long text with a lot of weird characters, such as: . , ~ ? ( ) " [ ] ! : - . Also we will'''
''' add words that should not exsist and be tokenized to <unk>, such as saoneuhaoesuth'''
)
_a : Union[str, Any] = [
8_7_1,
4_1_9,
3_5_8,
9_4_6,
9_9_1,
2_5_2_1,
4_5_2,
3_5_8,
1_3_5_7,
3_8_7,
7_7_5_1,
3_5_3_6,
1_1_2,
9_8_5,
4_5_6,
1_2_6,
8_6_5,
9_3_8,
5_4_0_0,
5_7_3_4,
4_5_8,
1_3_6_8,
4_6_7,
7_8_6,
2_4_6_2,
5_2_4_6,
1_1_5_9,
6_3_3,
8_6_5,
4_5_1_9,
4_5_7,
5_8_2,
8_5_2,
2_5_5_7,
4_2_7,
9_1_6,
5_0_8,
4_0_5,
3_4_3_2_4,
4_9_7,
3_9_1,
4_0_8,
1_1_3_4_2,
1_2_4_4,
3_8_5,
1_0_0,
9_3_8,
9_8_5,
4_5_6,
5_7_4,
3_6_2,
1_2_5_9_7,
3_2_0_0,
3_1_2_9,
1_1_7_2,
]
self.assertListEqual(__lowercase , self.big_tokenizer.encode(__lowercase ) )
@require_torch
@slow
def __lowercase ( self ) -> Optional[int]:
import torch
from transformers import BertGenerationConfig, BertGenerationEncoder
# Build sequence
_a : Optional[Any] = list(self.big_tokenizer.get_vocab().keys() )[:1_0]
_a : Optional[int] = ''' '''.join(__lowercase )
_a : Optional[int] = self.big_tokenizer.encode_plus(__lowercase , return_tensors='''pt''' , return_token_type_ids=__lowercase )
_a : List[str] = self.big_tokenizer.batch_encode_plus(
[sequence + ''' ''' + sequence] , return_tensors='''pt''' , return_token_type_ids=__lowercase )
_a : List[Any] = BertGenerationConfig()
_a : Optional[Any] = BertGenerationEncoder(__lowercase )
assert model.get_input_embeddings().weight.shape[0] >= self.big_tokenizer.vocab_size
with torch.no_grad():
model(**__lowercase )
model(**__lowercase )
@slow
def __lowercase ( self ) -> Dict:
# fmt: off
_a : List[str] = {'''input_ids''': [[3_9_2_8_6, 4_5_8, 3_6_3_3_5, 2_0_0_1, 4_5_6, 1_3_0_7_3, 1_3_2_6_6, 4_5_5, 1_1_3, 7_7_4_6, 1_7_4_1, 1_1_1_5_7, 3_9_1, 1_3_0_7_3, 1_3_2_6_6, 4_5_5, 1_1_3, 3_9_6_7, 3_5_4_1_2, 1_1_3, 4_9_3_6, 1_0_9, 3_8_7_0, 2_3_7_7, 1_1_3, 3_0_0_8_4, 4_5_7_2_0, 4_5_8, 1_3_4, 1_7_4_9_6, 1_1_2, 5_0_3, 1_1_6_7_2, 1_1_3, 1_1_8, 1_1_2, 5_6_6_5, 1_3_3_4_7, 3_8_6_8_7, 1_1_2, 1_4_9_6, 3_1_3_8_9, 1_1_2, 3_2_6_8, 4_7_2_6_4, 1_3_4, 9_6_2, 1_1_2, 1_6_3_7_7, 8_0_3_5, 2_3_1_3_0, 4_3_0, 1_2_1_6_9, 1_5_5_1_8, 2_8_5_9_2, 4_5_8, 1_4_6, 4_1_6_9_7, 1_0_9, 3_9_1, 1_2_1_6_9, 1_5_5_1_8, 1_6_6_8_9, 4_5_8, 1_4_6, 4_1_3_5_8, 1_0_9, 4_5_2, 7_2_6, 4_0_3_4, 1_1_1, 7_6_3, 3_5_4_1_2, 5_0_8_2, 3_8_8, 1_9_0_3, 1_1_1, 9_0_5_1, 3_9_1, 2_8_7_0, 4_8_9_1_8, 1_9_0_0, 1_1_2_3, 5_5_0, 9_9_8, 1_1_2, 9_5_8_6, 1_5_9_8_5, 4_5_5, 3_9_1, 4_1_0, 2_2_9_5_5, 3_7_6_3_6, 1_1_4], [4_4_8, 1_7_4_9_6, 4_1_9, 3_6_6_3, 3_8_5, 7_6_3, 1_1_3, 2_7_5_3_3, 2_8_7_0, 3_2_8_3, 1_3_0_4_3, 1_6_3_9, 2_4_7_1_3, 5_2_3, 6_5_6, 2_4_0_1_3, 1_8_5_5_0, 2_5_2_1, 5_1_7, 2_7_0_1_4, 2_1_2_4_4, 4_2_0, 1_2_1_2, 1_4_6_5, 3_9_1, 9_2_7, 4_8_3_3, 3_8_8, 5_7_8, 1_1_7_8_6, 1_1_4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [4_8_4, 2_1_6_9, 7_6_8_7, 2_1_9_3_2, 1_8_1_4_6, 7_2_6, 3_6_3, 1_7_0_3_2, 3_3_9_1, 1_1_4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]], '''attention_mask''': [[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]]} # noqa: E501
# fmt: on
self.tokenizer_integration_test_util(
expected_encoding=__lowercase , model_name='''google/bert_for_seq_generation_L-24_bbc_encoder''' , revision='''c817d1fd1be2ffa69431227a1fe320544943d4db''' , )
| 366
|
from __future__ import annotations
def __UpperCAmelCase ( __a : list ) -> float:
"""simple docstring"""
if not nums:
raise ValueError('''List is empty''' )
return sum(__a ) / len(__a )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 15
| 0
|
from __future__ import annotations
a__ = 1.6021E-19 # units = C
def __UpperCAmelCase ( __a : float ,__a : float ,__a : float ,) -> tuple[str, float]:
"""simple docstring"""
if (conductivity, electron_conc, mobility).count(0 ) != 1:
raise ValueError('''You cannot supply more or less than 2 values''' )
elif conductivity < 0:
raise ValueError('''Conductivity cannot be negative''' )
elif electron_conc < 0:
raise ValueError('''Electron concentration cannot be negative''' )
elif mobility < 0:
raise ValueError('''mobility cannot be negative''' )
elif conductivity == 0:
return (
"conductivity",
mobility * electron_conc * ELECTRON_CHARGE,
)
elif electron_conc == 0:
return (
"electron_conc",
conductivity / (mobility * ELECTRON_CHARGE),
)
else:
return (
"mobility",
conductivity / (electron_conc * ELECTRON_CHARGE),
)
if __name__ == "__main__":
import doctest
doctest.testmod()
| 367
|
import argparse
import os
import torch
from transformers.utils import WEIGHTS_NAME
a__ = ['''small''', '''medium''', '''large''']
a__ = '''lm_head.decoder.weight'''
a__ = '''lm_head.weight'''
def __UpperCAmelCase ( __a : str ,__a : str ) -> List[str]:
"""simple docstring"""
_a : Any = torch.load(__a )
_a : List[str] = d.pop(__a )
os.makedirs(__a ,exist_ok=__a )
torch.save(__a ,os.path.join(__a ,__a ) )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--dialogpt_path''', default='''.''', type=str)
a__ = parser.parse_args()
for MODEL in DIALOGPT_MODELS:
a__ = os.path.join(args.dialogpt_path, f'''{MODEL}_ft.pkl''')
a__ = f'''./DialoGPT-{MODEL}'''
convert_dialogpt_checkpoint(
checkpoint_path,
pytorch_dump_folder_path,
)
| 15
| 0
|
import warnings
from ...utils import logging
from .image_processing_donut import DonutImageProcessor
a__ = logging.get_logger(__name__)
class UpperCAmelCase_ ( _UpperCamelCase ):
"""simple docstring"""
def __init__( self , *_a , **_a ) -> Any:
warnings.warn(
'''The class DonutFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please'''
''' use DonutImageProcessor instead.''' , _UpperCAmelCase , )
super().__init__(*_UpperCAmelCase , **_UpperCAmelCase )
| 368
|
import enum
import warnings
from .. import MODEL_FOR_CAUSAL_LM_MAPPING, TF_MODEL_FOR_CAUSAL_LM_MAPPING
from ..utils import add_end_docstrings, is_tf_available
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
class UpperCAmelCase_ ( enum.Enum ):
"""simple docstring"""
UpperCAmelCase__ : int = 0
UpperCAmelCase__ : Union[str, Any] = 1
UpperCAmelCase__ : Optional[Any] = 2
@add_end_docstrings(__lowercase )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = "\n In 1991, the remains of Russian Tsar Nicholas II and his family (except for Alexei and Maria) are discovered. The\n voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the remainder of the story. 1883 Western\n Siberia, a young Grigori Rasputin is asked by his father and a group of men to perform magic. Rasputin has a vision\n and denounces one of the men as a horse thief. Although his father initially slaps him for making such an\n accusation, Rasputin watches as the man is chased outside and beaten. Twenty years later, Rasputin sees a vision of\n the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous, with people, even a bishop,\n begging for his blessing. <eod> </s> <eos>\n "
def __init__( self , *_a , **_a ) -> List[str]:
super().__init__(*_a , **_a )
self.check_model_type(
TF_MODEL_FOR_CAUSAL_LM_MAPPING if self.framework == '''tf''' else MODEL_FOR_CAUSAL_LM_MAPPING )
if "prefix" not in self._preprocess_params:
# This is very specific. The logic is quite complex and needs to be done
# as a "default".
# It also defines both some preprocess_kwargs and generate_kwargs
# which is why we cannot put them in their respective methods.
_a : Dict = None
if self.model.config.prefix is not None:
_a : List[Any] = self.model.config.prefix
if prefix is None and self.model.__class__.__name__ in [
"XLNetLMHeadModel",
"TransfoXLLMHeadModel",
"TFXLNetLMHeadModel",
"TFTransfoXLLMHeadModel",
]:
# For XLNet and TransformerXL we add an article to the prompt to give more state to the model.
_a : Optional[Any] = self.XL_PREFIX
if prefix is not None:
# Recalculate some generate_kwargs linked to prefix.
_a , _a , _a : str = self._sanitize_parameters(prefix=_a , **self._forward_params )
_a : Optional[Any] = {**self._preprocess_params, **preprocess_params}
_a : List[Any] = {**self._forward_params, **forward_params}
def __lowercase ( self , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , **_a , ) -> Optional[int]:
_a : List[Any] = {}
if prefix is not None:
_a : Optional[Any] = prefix
if prefix:
_a : Dict = self.tokenizer(
_a , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Tuple = prefix_inputs['''input_ids'''].shape[-1]
if handle_long_generation is not None:
if handle_long_generation not in {"hole"}:
raise ValueError(
F"""{handle_long_generation} is not a valid value for `handle_long_generation` parameter expected"""
''' [None, \'hole\']''' )
_a : Dict = handle_long_generation
preprocess_params.update(_a )
_a : Tuple = generate_kwargs
_a : Any = {}
if return_full_text is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_full_text`''' )
if return_tensors is not None:
raise ValueError('''`return_full_text` is mutually exclusive with `return_tensors`''' )
_a : List[str] = ReturnType.FULL_TEXT if return_full_text else ReturnType.NEW_TEXT
if return_tensors is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_tensors`''' )
_a : Any = ReturnType.TENSORS
if return_type is not None:
_a : Any = return_type
if clean_up_tokenization_spaces is not None:
_a : List[Any] = clean_up_tokenization_spaces
if stop_sequence is not None:
_a : Tuple = self.tokenizer.encode(_a , add_special_tokens=_a )
if len(_a ) > 1:
warnings.warn(
'''Stopping on a multiple token sequence is not yet supported on transformers. The first token of'''
''' the stop sequence will be used as the stop sequence string in the interim.''' )
_a : List[Any] = stop_sequence_ids[0]
return preprocess_params, forward_params, postprocess_params
def __lowercase ( self , *_a , **_a ) -> Union[str, Any]:
# Parse arguments
if self.model.__class__.__name__ in ["TransfoXLLMHeadModel"]:
kwargs.update({'''add_space_before_punct_symbol''': True} )
return super()._parse_and_tokenize(*_a , **_a )
def __call__( self , _a , **_a ) -> List[str]:
return super().__call__(_a , **_a )
def __lowercase ( self , _a , _a="" , _a=None , **_a ) -> List[Any]:
_a : Optional[int] = self.tokenizer(
prefix + prompt_text , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Union[str, Any] = prompt_text
if handle_long_generation == "hole":
_a : List[str] = inputs['''input_ids'''].shape[-1]
if "max_new_tokens" in generate_kwargs:
_a : int = generate_kwargs['''max_new_tokens''']
else:
_a : List[Any] = generate_kwargs.get('''max_length''' , self.model.config.max_length ) - cur_len
if new_tokens < 0:
raise ValueError('''We cannot infer how many new tokens are expected''' )
if cur_len + new_tokens > self.tokenizer.model_max_length:
_a : List[str] = self.tokenizer.model_max_length - new_tokens
if keep_length <= 0:
raise ValueError(
'''We cannot use `hole` to handle this generation the number of desired tokens exceeds the'''
''' models max length''' )
_a : List[Any] = inputs['''input_ids'''][:, -keep_length:]
if "attention_mask" in inputs:
_a : List[str] = inputs['''attention_mask'''][:, -keep_length:]
return inputs
def __lowercase ( self , _a , **_a ) -> Optional[int]:
_a : Any = model_inputs['''input_ids''']
_a : Optional[Any] = model_inputs.get('''attention_mask''' , _a )
# Allow empty prompts
if input_ids.shape[1] == 0:
_a : int = None
_a : int = None
_a : List[str] = 1
else:
_a : List[Any] = input_ids.shape[0]
_a : Union[str, Any] = model_inputs.pop('''prompt_text''' )
# If there is a prefix, we may need to adjust the generation length. Do so without permanently modifying
# generate_kwargs, as some of the parameterization may come from the initialization of the pipeline.
_a : int = generate_kwargs.pop('''prefix_length''' , 0 )
if prefix_length > 0:
_a : Tuple = '''max_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].max_new_tokens is not None
)
if not has_max_new_tokens:
_a : int = generate_kwargs.get('''max_length''' ) or self.model.config.max_length
generate_kwargs["max_length"] += prefix_length
_a : Dict = '''min_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].min_new_tokens is not None
)
if not has_min_new_tokens and "min_length" in generate_kwargs:
generate_kwargs["min_length"] += prefix_length
# BS x SL
_a : Optional[Any] = self.model.generate(input_ids=_a , attention_mask=_a , **_a )
_a : int = generated_sequence.shape[0]
if self.framework == "pt":
_a : Tuple = generated_sequence.reshape(_a , out_b // in_b , *generated_sequence.shape[1:] )
elif self.framework == "tf":
_a : List[Any] = tf.reshape(_a , (in_b, out_b // in_b, *generated_sequence.shape[1:]) )
return {"generated_sequence": generated_sequence, "input_ids": input_ids, "prompt_text": prompt_text}
def __lowercase ( self , _a , _a=ReturnType.FULL_TEXT , _a=True ) -> int:
_a : Tuple = model_outputs['''generated_sequence'''][0]
_a : int = model_outputs['''input_ids''']
_a : Any = model_outputs['''prompt_text''']
_a : Any = generated_sequence.numpy().tolist()
_a : Any = []
for sequence in generated_sequence:
if return_type == ReturnType.TENSORS:
_a : Optional[int] = {'''generated_token_ids''': sequence}
elif return_type in {ReturnType.NEW_TEXT, ReturnType.FULL_TEXT}:
# Decode text
_a : str = self.tokenizer.decode(
_a , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , )
# Remove PADDING prompt of the sequence if XLNet or Transfo-XL model is used
if input_ids is None:
_a : Union[str, Any] = 0
else:
_a : str = len(
self.tokenizer.decode(
input_ids[0] , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , ) )
if return_type == ReturnType.FULL_TEXT:
_a : str = prompt_text + text[prompt_length:]
else:
_a : List[str] = text[prompt_length:]
_a : Union[str, Any] = {'''generated_text''': all_text}
records.append(_a )
return records
| 15
| 0
|
import math
import time
from typing import Dict, List, Optional
from torch.utils.data import Dataset
from transformers import SeqaSeqTrainer, is_torch_tpu_available
from transformers.trainer_utils import PredictionOutput, speed_metrics
if is_torch_tpu_available(check_device=False):
import torch_xla.core.xla_model as xm
import torch_xla.debug.metrics as met
class UpperCAmelCase_ ( __snake_case ):
"""simple docstring"""
def __init__( self , *_a , _a=None , _a=None , **_a ) -> Optional[int]:
super().__init__(*UpperCamelCase__ , **UpperCamelCase__ )
_a : Optional[int] = eval_examples
_a : Optional[Any] = post_process_function
def __lowercase ( self , _a = None , _a=None , _a = None , _a = "eval" , **_a , ) -> Dict[str, float]:
_a : Any = gen_kwargs.copy()
_a : str = (
gen_kwargs["max_length"] if gen_kwargs.get('''max_length''' ) is not None else self.args.generation_max_length
)
_a : Any = (
gen_kwargs["num_beams"] if gen_kwargs.get('''num_beams''' ) is not None else self.args.generation_num_beams
)
_a : Union[str, Any] = gen_kwargs
_a : int = self.eval_dataset if eval_dataset is None else eval_dataset
_a : List[str] = self.get_eval_dataloader(UpperCamelCase__ )
_a : int = self.eval_examples if eval_examples is None else eval_examples
# Temporarily disable metric computation, we will do it in the loop here.
_a : Any = self.compute_metrics
_a : Any = None
_a : Optional[int] = time.time()
_a : Any = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop
try:
_a : List[str] = eval_loop(
UpperCamelCase__ , description='''Evaluation''' , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCamelCase__ , metric_key_prefix=UpperCamelCase__ , )
finally:
_a : List[Any] = compute_metrics
_a : List[Any] = self.args.eval_batch_size * self.args.world_size
if F"""{metric_key_prefix}_jit_compilation_time""" in output.metrics:
start_time += output.metrics[F"""{metric_key_prefix}_jit_compilation_time"""]
output.metrics.update(
speed_metrics(
UpperCamelCase__ , UpperCamelCase__ , num_samples=output.num_samples , num_steps=math.ceil(output.num_samples / total_batch_size ) , ) )
if self.post_process_function is not None and self.compute_metrics is not None and self.args.should_save:
# Only the main node write the results by default
_a : int = self.post_process_function(UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ )
_a : Tuple = self.compute_metrics(UpperCamelCase__ )
# Prefix all keys with metric_key_prefix + '_'
for key in list(metrics.keys() ):
if not key.startswith(F"""{metric_key_prefix}_""" ):
_a : Any = metrics.pop(UpperCamelCase__ )
metrics.update(output.metrics )
else:
_a : int = output.metrics
if self.args.should_log:
# Only the main node log the results by default
self.log(UpperCamelCase__ )
if self.args.tpu_metrics_debug or self.args.debug:
# tpu-comment: Logging debug metrics for PyTorch/XLA (compile, execute times, ops, etc.)
xm.master_print(met.metrics_report() )
_a : str = self.callback_handler.on_evaluate(self.args , self.state , self.control , UpperCamelCase__ )
return metrics
def __lowercase ( self , _a , _a , _a=None , _a = "test" , **_a ) -> Union[str, Any]:
_a : Dict = gen_kwargs.copy()
_a : List[str] = self.get_test_dataloader(UpperCamelCase__ )
# Temporarily disable metric computation, we will do it in the loop here.
_a : Optional[int] = self.compute_metrics
_a : Optional[int] = None
_a : Any = time.time()
_a : Dict = self.prediction_loop if self.args.use_legacy_prediction_loop else self.evaluation_loop
try:
_a : int = eval_loop(
UpperCamelCase__ , description='''Prediction''' , prediction_loss_only=True if compute_metrics is None else None , ignore_keys=UpperCamelCase__ , metric_key_prefix=UpperCamelCase__ , )
finally:
_a : Tuple = compute_metrics
_a : Optional[int] = self.args.eval_batch_size * self.args.world_size
if F"""{metric_key_prefix}_jit_compilation_time""" in output.metrics:
start_time += output.metrics[F"""{metric_key_prefix}_jit_compilation_time"""]
output.metrics.update(
speed_metrics(
UpperCamelCase__ , UpperCamelCase__ , num_samples=output.num_samples , num_steps=math.ceil(output.num_samples / total_batch_size ) , ) )
if self.post_process_function is None or self.compute_metrics is None:
return output
_a : List[str] = self.post_process_function(UpperCamelCase__ , UpperCamelCase__ , UpperCamelCase__ , '''predict''' )
_a : List[Any] = self.compute_metrics(UpperCamelCase__ )
# Prefix all keys with metric_key_prefix + '_'
for key in list(metrics.keys() ):
if not key.startswith(F"""{metric_key_prefix}_""" ):
_a : int = metrics.pop(UpperCamelCase__ )
metrics.update(output.metrics )
return PredictionOutput(predictions=predictions.predictions , label_ids=predictions.label_ids , metrics=UpperCamelCase__ )
| 369
|
# 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
import os
from accelerate.test_utils import execute_subprocess_async
def __UpperCAmelCase ( __a : Dict=None ) -> str:
"""simple docstring"""
if subparsers is not None:
_a : Union[str, Any] = subparsers.add_parser('''test''' )
else:
_a : List[str] = argparse.ArgumentParser('''Accelerate test command''' )
parser.add_argument(
'''--config_file''' ,default=__a ,help=(
'''The path to use to store the config file. Will default to a file named default_config.yaml in the cache '''
'''location, which is the content of the environment `HF_HOME` suffixed with \'accelerate\', or if you don\'t have '''
'''such an environment variable, your cache directory (\'~/.cache\' or the content of `XDG_CACHE_HOME`) suffixed '''
'''with \'huggingface\'.'''
) ,)
if subparsers is not None:
parser.set_defaults(func=__a )
return parser
def __UpperCAmelCase ( __a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
_a : Dict = os.path.sep.join(__file__.split(os.path.sep )[:-2] + ['''test_utils''', '''scripts''', '''test_script.py'''] )
if args.config_file is None:
_a : List[Any] = script_name
else:
_a : Union[str, Any] = F"""--config_file={args.config_file} {script_name}"""
_a : str = ['''accelerate-launch'''] + test_args.split()
_a : str = execute_subprocess_async(__a ,env=os.environ.copy() )
if result.returncode == 0:
print('''Test is a success! You are ready for your distributed training!''' )
def __UpperCAmelCase ( ) -> List[Any]:
"""simple docstring"""
_a : Optional[int] = test_command_parser()
_a : List[Any] = parser.parse_args()
test_command(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_sentencepiece_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
a__ = {'''configuration_mbart''': ['''MBART_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''MBartConfig''', '''MBartOnnxConfig''']}
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = ['''MBartTokenizer''']
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = ['''MBartTokenizerFast''']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''MBART_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''MBartForCausalLM''',
'''MBartForConditionalGeneration''',
'''MBartForQuestionAnswering''',
'''MBartForSequenceClassification''',
'''MBartModel''',
'''MBartPreTrainedModel''',
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''TFMBartForConditionalGeneration''',
'''TFMBartModel''',
'''TFMBartPreTrainedModel''',
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''FlaxMBartForConditionalGeneration''',
'''FlaxMBartForQuestionAnswering''',
'''FlaxMBartForSequenceClassification''',
'''FlaxMBartModel''',
'''FlaxMBartPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_mbart import MBART_PRETRAINED_CONFIG_ARCHIVE_MAP, MBartConfig, MBartOnnxConfig
try:
if not is_sentencepiece_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_mbart import MBartTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_mbart_fast import MBartTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_mbart import (
MBART_PRETRAINED_MODEL_ARCHIVE_LIST,
MBartForCausalLM,
MBartForConditionalGeneration,
MBartForQuestionAnswering,
MBartForSequenceClassification,
MBartModel,
MBartPreTrainedModel,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_mbart import TFMBartForConditionalGeneration, TFMBartModel, TFMBartPreTrainedModel
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_mbart import (
FlaxMBartForConditionalGeneration,
FlaxMBartForQuestionAnswering,
FlaxMBartForSequenceClassification,
FlaxMBartModel,
FlaxMBartPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 370
|
import json
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import BertTokenizerFast
from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer
from transformers.testing_utils import require_tokenizers, require_vision
from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available
if is_vision_available():
from PIL import Image
from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor
@require_tokenizers
@require_vision
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = tempfile.mkdtemp()
# fmt: off
_a : Optional[int] = ['''[UNK]''', '''[CLS]''', '''[SEP]''', '''[PAD]''', '''[MASK]''', '''want''', '''##want''', '''##ed''', '''wa''', '''un''', '''runn''', '''##ing''', ''',''', '''low''', '''lowest''']
# fmt: on
_a : Tuple = 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] ) )
_a : Any = {
'''do_resize''': True,
'''size''': {'''height''': 1_8, '''width''': 1_8},
'''do_normalize''': True,
'''image_mean''': [0.5, 0.5, 0.5],
'''image_std''': [0.5, 0.5, 0.5],
}
_a : str = os.path.join(self.tmpdirname , _a )
with open(self.image_processor_file , '''w''' , encoding='''utf-8''' ) as fp:
json.dump(_a , _a )
def __lowercase ( self , **_a ) -> Any:
return BertTokenizer.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self , **_a ) -> str:
return ViTImageProcessor.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self ) -> List[Any]:
shutil.rmtree(self.tmpdirname )
def __lowercase ( self ) -> Any:
_a : Union[str, Any] = [np.random.randint(2_5_5 , size=(3, 3_0, 4_0_0) , dtype=np.uinta )]
_a : Tuple = [Image.fromarray(np.moveaxis(_a , 0 , -1 ) ) for x in image_inputs]
return image_inputs
def __lowercase ( self ) -> str:
_a : List[str] = self.get_tokenizer()
_a : Tuple = self.get_image_processor()
_a : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
processor.save_pretrained(self.tmpdirname )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Dict:
_a : List[str] = VisionTextDualEncoderProcessor(
tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() )
processor.save_pretrained(self.tmpdirname )
_a : Any = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' )
_a : List[Any] = self.get_image_processor(do_normalize=_a , padding_value=1.0 )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(
self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=_a , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Any:
_a : Dict = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : int = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[str] = self.prepare_image_inputs()
_a : List[Any] = image_processor(_a , return_tensors='''np''' )
_a : Dict = processor(images=_a , return_tensors='''np''' )
for key in input_feat_extract.keys():
self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1e-2 )
def __lowercase ( self ) -> List[str]:
_a : Union[str, Any] = self.get_image_processor()
_a : Dict = self.get_tokenizer()
_a : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Tuple = '''lower newer'''
_a : int = processor(text=_a )
_a : str = tokenizer(_a )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def __lowercase ( self ) -> List[Any]:
_a : Any = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : Tuple = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[Any] = '''lower newer'''
_a : Union[str, Any] = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , ['''input_ids''', '''token_type_ids''', '''attention_mask''', '''pixel_values'''] )
# test if it raises when no input is passed
with self.assertRaises(_a ):
processor()
def __lowercase ( self ) -> Optional[int]:
_a : Union[str, Any] = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : Any = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Any = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
_a : int = processor.batch_decode(_a )
_a : int = tokenizer.batch_decode(_a )
self.assertListEqual(_a , _a )
def __lowercase ( self ) -> List[Any]:
_a : Tuple = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : str = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Optional[int] = '''lower newer'''
_a : Dict = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
| 15
| 0
|
from ..utils import DummyObject, requires_backends
class UpperCAmelCase_ ( metaclass=_SCREAMING_SNAKE_CASE ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = ["torch", "scipy"]
def __init__( self , *_a , **_a ) -> Any:
requires_backends(self , ['''torch''', '''scipy'''] )
@classmethod
def __lowercase ( cls , *_a , **_a ) -> Tuple:
requires_backends(cls , ['''torch''', '''scipy'''] )
@classmethod
def __lowercase ( cls , *_a , **_a ) -> Tuple:
requires_backends(cls , ['''torch''', '''scipy'''] )
| 371
|
import argparse
import json
import os
import fairseq
import torch
from fairseq.data import Dictionary
from transformers import (
HubertConfig,
HubertForCTC,
HubertModel,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaProcessor,
logging,
)
logging.set_verbosity_info()
a__ = logging.get_logger(__name__)
a__ = {
'''post_extract_proj''': '''feature_projection.projection''',
'''encoder.pos_conv.0''': '''encoder.pos_conv_embed.conv''',
'''self_attn.k_proj''': '''encoder.layers.*.attention.k_proj''',
'''self_attn.v_proj''': '''encoder.layers.*.attention.v_proj''',
'''self_attn.q_proj''': '''encoder.layers.*.attention.q_proj''',
'''self_attn.out_proj''': '''encoder.layers.*.attention.out_proj''',
'''self_attn_layer_norm''': '''encoder.layers.*.layer_norm''',
'''fc1''': '''encoder.layers.*.feed_forward.intermediate_dense''',
'''fc2''': '''encoder.layers.*.feed_forward.output_dense''',
'''final_layer_norm''': '''encoder.layers.*.final_layer_norm''',
'''encoder.layer_norm''': '''encoder.layer_norm''',
'''w2v_model.layer_norm''': '''feature_projection.layer_norm''',
'''w2v_encoder.proj''': '''lm_head''',
'''mask_emb''': '''masked_spec_embed''',
}
def __UpperCAmelCase ( __a : List[Any] ,__a : Optional[int] ,__a : Optional[int] ,__a : List[str] ,__a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
for attribute in key.split('''.''' ):
_a : Optional[Any] = getattr(__a ,__a )
if weight_type is not None:
_a : Dict = getattr(__a ,__a ).shape
else:
_a : Optional[int] = 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":
_a : List[Any] = value
elif weight_type == "weight_g":
_a : Any = value
elif weight_type == "weight_v":
_a : Union[str, Any] = value
elif weight_type == "bias":
_a : Optional[int] = value
else:
_a : List[Any] = value
logger.info(F"""{key + '.' + weight_type if weight_type is not None else ''} was initialized from {full_name}.""" )
def __UpperCAmelCase ( __a : Any ,__a : Union[str, Any] ,__a : Union[str, Any] ) -> int:
"""simple docstring"""
_a : Union[str, Any] = []
_a : Union[str, Any] = fairseq_model.state_dict()
_a : Union[str, Any] = hf_model.hubert.feature_extractor if is_finetuned else hf_model.feature_extractor
for name, value in fairseq_dict.items():
_a : int = False
if "conv_layers" in name:
load_conv_layer(
__a ,__a ,__a ,__a ,hf_model.config.feat_extract_norm == '''group''' ,)
_a : Optional[Any] = True
else:
for key, mapped_key in MAPPING.items():
_a : Union[str, Any] = '''hubert.''' + 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] and not is_finetuned):
_a : Any = True
if "*" in mapped_key:
_a : Optional[int] = name.split(__a )[0].split('''.''' )[-2]
_a : Any = mapped_key.replace('''*''' ,__a )
if "weight_g" in name:
_a : List[Any] = '''weight_g'''
elif "weight_v" in name:
_a : List[str] = '''weight_v'''
elif "weight" in name:
_a : Any = '''weight'''
elif "bias" in name:
_a : str = '''bias'''
else:
_a : Any = None
set_recursively(__a ,__a ,__a ,__a ,__a )
continue
if not is_used:
unused_weights.append(__a )
logger.warning(F"""Unused weights: {unused_weights}""" )
def __UpperCAmelCase ( __a : int ,__a : Optional[Any] ,__a : Dict ,__a : List[str] ,__a : Any ) -> Tuple:
"""simple docstring"""
_a : int = full_name.split('''conv_layers.''' )[-1]
_a : Any = name.split('''.''' )
_a : List[Any] = int(items[0] )
_a : Optional[int] = 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."""
)
_a : Optional[int] = 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."""
)
_a : Optional[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:
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."
)
_a : int = 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."""
)
_a : Any = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(__a )
@torch.no_grad()
def __UpperCAmelCase ( __a : Dict ,__a : List[Any] ,__a : List[str]=None ,__a : Optional[int]=None ,__a : int=True ) -> List[Any]:
"""simple docstring"""
if config_path is not None:
_a : Tuple = HubertConfig.from_pretrained(__a )
else:
_a : Any = HubertConfig()
if is_finetuned:
if dict_path:
_a : Tuple = Dictionary.load(__a )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
_a : Any = target_dict.pad_index
_a : Tuple = target_dict.bos_index
_a : Optional[int] = target_dict.eos_index
_a : Optional[Any] = len(target_dict.symbols )
_a : Tuple = os.path.join(__a ,'''vocab.json''' )
if not os.path.isdir(__a ):
logger.error('''--pytorch_dump_folder_path ({}) should be a directory'''.format(__a ) )
return
os.makedirs(__a ,exist_ok=__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ) as vocab_handle:
json.dump(target_dict.indices ,__a )
_a : Tuple = WavaVecaCTCTokenizer(
__a ,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=__a ,)
_a : Tuple = True if config.feat_extract_norm == '''layer''' else False
_a : List[Any] = WavaVecaFeatureExtractor(
feature_size=1 ,sampling_rate=16_000 ,padding_value=0 ,do_normalize=__a ,return_attention_mask=__a ,)
_a : List[Any] = WavaVecaProcessor(feature_extractor=__a ,tokenizer=__a )
processor.save_pretrained(__a )
_a : Tuple = HubertForCTC(__a )
else:
_a : Tuple = HubertModel(__a )
if is_finetuned:
_a , _a , _a : int = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
_a , _a , _a : str = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] )
_a : Any = model[0].eval()
recursively_load_weights(__a ,__a ,__a )
hf_wavavec.save_pretrained(__a )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
parser.add_argument('''--checkpoint_path''', default=None, type=str, help='''Path to fairseq checkpoint''')
parser.add_argument('''--dict_path''', default=None, type=str, help='''Path to dict of fine-tuned model''')
parser.add_argument('''--config_path''', default=None, type=str, help='''Path to hf config.json of model to convert''')
parser.add_argument(
'''--not_finetuned''', action='''store_true''', help='''Whether the model to convert is a fine-tuned model or not'''
)
a__ = parser.parse_args()
convert_hubert_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 15
| 0
|
import unittest
from transformers import SqueezeBertConfig, is_torch_available
from transformers.testing_utils import require_sentencepiece, require_tokenizers, 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 (
SQUEEZEBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
SqueezeBertForMaskedLM,
SqueezeBertForMultipleChoice,
SqueezeBertForQuestionAnswering,
SqueezeBertForSequenceClassification,
SqueezeBertForTokenClassification,
SqueezeBertModel,
)
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __init__( self , _a , _a=1_3 , _a=7 , _a=True , _a=True , _a=False , _a=True , _a=9_9 , _a=3_2 , _a=5 , _a=4 , _a=6_4 , _a="gelu" , _a=0.1 , _a=0.1 , _a=5_1_2 , _a=1_6 , _a=2 , _a=0.02 , _a=3 , _a=4 , _a=None , _a=2 , _a=2 , _a=2 , _a=2 , _a=4 , _a=1 , ) -> str:
_a : Optional[Any] = parent
_a : Tuple = batch_size
_a : Any = seq_length
_a : Optional[Any] = is_training
_a : List[Any] = use_input_mask
_a : str = use_token_type_ids
_a : Dict = use_labels
_a : Dict = vocab_size
_a : Any = hidden_size
_a : List[Any] = num_hidden_layers
_a : Any = num_attention_heads
_a : Tuple = intermediate_size
_a : Dict = hidden_act
_a : str = hidden_dropout_prob
_a : str = attention_probs_dropout_prob
_a : Optional[int] = max_position_embeddings
_a : Any = type_vocab_size
_a : Union[str, Any] = type_sequence_label_size
_a : str = initializer_range
_a : Dict = num_labels
_a : Optional[Any] = num_choices
_a : Optional[int] = scope
_a : int = q_groups
_a : List[str] = k_groups
_a : str = v_groups
_a : str = post_attention_groups
_a : Dict = intermediate_groups
_a : Tuple = output_groups
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[int] = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
_a : Optional[int] = None
if self.use_input_mask:
_a : List[str] = random_attention_mask([self.batch_size, self.seq_length] )
_a : Any = None
_a : Optional[int] = None
_a : Any = None
if self.use_labels:
_a : str = ids_tensor([self.batch_size] , self.type_sequence_label_size )
_a : str = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
_a : str = ids_tensor([self.batch_size] , self.num_choices )
_a : Optional[int] = self.get_config()
return config, input_ids, input_mask, sequence_labels, token_labels, choice_labels
def __lowercase ( self ) -> str:
return SqueezeBertConfig(
embedding_size=self.hidden_size , 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 , attention_probs_dropout_prob=self.hidden_dropout_prob , attention_dropout=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , initializer_range=self.initializer_range , q_groups=self.q_groups , k_groups=self.k_groups , v_groups=self.v_groups , post_attention_groups=self.post_attention_groups , intermediate_groups=self.intermediate_groups , output_groups=self.output_groups , )
def __lowercase ( self , _a , _a , _a , _a , _a , _a ) -> str:
_a : Tuple = SqueezeBertModel(config=_a )
model.to(_a )
model.eval()
_a : Union[str, Any] = model(_a , _a )
_a : Union[str, Any] = model(_a )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a ) -> Any:
_a : int = SqueezeBertForMaskedLM(config=_a )
model.to(_a )
model.eval()
_a : List[Any] = model(_a , attention_mask=_a , labels=_a )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a ) -> Tuple:
_a : List[str] = SqueezeBertForQuestionAnswering(config=_a )
model.to(_a )
model.eval()
_a : Union[str, Any] = model(
_a , attention_mask=_a , start_positions=_a , end_positions=_a )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a ) -> Optional[int]:
_a : Dict = self.num_labels
_a : str = SqueezeBertForSequenceClassification(_a )
model.to(_a )
model.eval()
_a : Union[str, Any] = model(_a , attention_mask=_a , labels=_a )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a ) -> Optional[Any]:
_a : int = self.num_labels
_a : Any = SqueezeBertForTokenClassification(config=_a )
model.to(_a )
model.eval()
_a : Optional[int] = model(_a , attention_mask=_a , labels=_a )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a ) -> Optional[int]:
_a : Dict = self.num_choices
_a : List[Any] = SqueezeBertForMultipleChoice(config=_a )
model.to(_a )
model.eval()
_a : str = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
_a : Dict = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
_a : Optional[Any] = model(
_a , attention_mask=_a , labels=_a , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def __lowercase ( self ) -> str:
_a : Optional[Any] = self.prepare_config_and_inputs()
(_a) : Tuple = config_and_inputs
_a : int = {'''input_ids''': input_ids, '''attention_mask''': input_mask}
return config, inputs_dict
@require_torch
class UpperCAmelCase_ ( __lowercase , __lowercase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : Tuple = (
(
SqueezeBertModel,
SqueezeBertForMaskedLM,
SqueezeBertForMultipleChoice,
SqueezeBertForQuestionAnswering,
SqueezeBertForSequenceClassification,
SqueezeBertForTokenClassification,
)
if is_torch_available()
else None
)
UpperCAmelCase__ : Union[str, Any] = (
{
"feature-extraction": SqueezeBertModel,
"fill-mask": SqueezeBertForMaskedLM,
"question-answering": SqueezeBertForQuestionAnswering,
"text-classification": SqueezeBertForSequenceClassification,
"token-classification": SqueezeBertForTokenClassification,
"zero-shot": SqueezeBertForSequenceClassification,
}
if is_torch_available()
else {}
)
UpperCAmelCase__ : List[str] = False
UpperCAmelCase__ : Any = True
UpperCAmelCase__ : Optional[Any] = False
def __lowercase ( self ) -> Optional[Any]:
_a : Union[str, Any] = SqueezeBertModelTester(self )
_a : List[Any] = ConfigTester(self , config_class=_a , dim=3_7 )
def __lowercase ( self ) -> Union[str, Any]:
self.config_tester.run_common_tests()
def __lowercase ( self ) -> Tuple:
_a : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_squeezebert_model(*_a )
def __lowercase ( self ) -> Optional[Any]:
_a : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_squeezebert_for_masked_lm(*_a )
def __lowercase ( self ) -> Union[str, Any]:
_a : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_squeezebert_for_question_answering(*_a )
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_squeezebert_for_sequence_classification(*_a )
def __lowercase ( self ) -> str:
_a : Any = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_squeezebert_for_token_classification(*_a )
def __lowercase ( self ) -> Optional[Any]:
_a : str = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_squeezebert_for_multiple_choice(*_a )
@slow
def __lowercase ( self ) -> List[str]:
for model_name in SQUEEZEBERT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
_a : Optional[int] = SqueezeBertModel.from_pretrained(_a )
self.assertIsNotNone(_a )
@require_sentencepiece
@require_tokenizers
@require_torch
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
@slow
def __lowercase ( self ) -> List[str]:
_a : int = SqueezeBertForSequenceClassification.from_pretrained('''squeezebert/squeezebert-mnli''' )
_a : Optional[int] = torch.tensor([[1, 2_9_4_1_4, 2_3_2, 3_2_8, 7_4_0, 1_1_4_0, 1_2_6_9_5, 6_9, 1_3, 1_5_8_8, 2]] )
_a : List[str] = model(_a )[0]
_a : Any = torch.Size((1, 3) )
self.assertEqual(output.shape , _a )
_a : Optional[int] = torch.tensor([[0.6401, -0.0349, -0.6041]] )
self.assertTrue(torch.allclose(_a , _a , atol=1e-4 ) )
| 350
|
import warnings
from typing import List, Optional, Union
from ...processing_utils import ProcessorMixin
from ...tokenization_utils_base import BatchEncoding, PaddingStrategy, PreTokenizedInput, TextInput, TruncationStrategy
from ...utils import TensorType
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[str] = ["image_processor", "tokenizer"]
UpperCAmelCase__ : str = "ViltImageProcessor"
UpperCAmelCase__ : Union[str, Any] = ("BertTokenizer", "BertTokenizerFast")
def __init__( self , _a=None , _a=None , **_a ) -> Any:
_a : Union[str, Any] = None
if "feature_extractor" in kwargs:
warnings.warn(
'''The `feature_extractor` argument is deprecated and will be removed in v5, use `image_processor`'''
''' instead.''' , _a , )
_a : Dict = kwargs.pop('''feature_extractor''' )
_a : Optional[int] = image_processor if image_processor is not None else feature_extractor
if image_processor is None:
raise ValueError('''You need to specify an `image_processor`.''' )
if tokenizer is None:
raise ValueError('''You need to specify a `tokenizer`.''' )
super().__init__(_a , _a )
_a : int = self.image_processor
def __call__( self , _a , _a = None , _a = True , _a = False , _a = None , _a = None , _a = 0 , _a = None , _a = None , _a = None , _a = False , _a = False , _a = False , _a = False , _a = True , _a = None , **_a , ) -> BatchEncoding:
_a : Tuple = self.tokenizer(
text=_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , stride=_a , pad_to_multiple_of=_a , return_token_type_ids=_a , return_attention_mask=_a , return_overflowing_tokens=_a , return_special_tokens_mask=_a , return_offsets_mapping=_a , return_length=_a , verbose=_a , return_tensors=_a , **_a , )
# add pixel_values + pixel_mask
_a : str = self.image_processor(_a , return_tensors=_a )
encoding.update(_a )
return encoding
def __lowercase ( self , *_a , **_a ) -> Optional[Any]:
return self.tokenizer.batch_decode(*_a , **_a )
def __lowercase ( self , *_a , **_a ) -> str:
return self.tokenizer.decode(*_a , **_a )
@property
def __lowercase ( self ) -> Optional[int]:
_a : str = self.tokenizer.model_input_names
_a : Optional[Any] = self.image_processor.model_input_names
return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names ) )
@property
def __lowercase ( self ) -> Optional[Any]:
warnings.warn(
'''`feature_extractor_class` is deprecated and will be removed in v5. Use `image_processor_class` instead.''' , _a , )
return self.image_processor_class
@property
def __lowercase ( self ) -> Any:
warnings.warn(
'''`feature_extractor` is deprecated and will be removed in v5. Use `image_processor` instead.''' , _a , )
return self.image_processor
| 15
| 0
|
from __future__ import annotations
from math import pi
def __UpperCAmelCase ( __a : float ,__a : float ,__a : float ) -> dict[str, float]:
"""simple docstring"""
if (inductance, frequency, reactance).count(0 ) != 1:
raise ValueError('''One and only one argument must be 0''' )
if inductance < 0:
raise ValueError('''Inductance cannot be negative''' )
if frequency < 0:
raise ValueError('''Frequency cannot be negative''' )
if reactance < 0:
raise ValueError('''Inductive reactance cannot be negative''' )
if inductance == 0:
return {"inductance": reactance / (2 * pi * frequency)}
elif frequency == 0:
return {"frequency": reactance / (2 * pi * inductance)}
elif reactance == 0:
return {"reactance": 2 * pi * frequency * inductance}
else:
raise ValueError('''Exactly one argument must be 0''' )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 351
|
from math import ceil
def __UpperCAmelCase ( __a : int = 1_001 ) -> int:
"""simple docstring"""
_a : Dict = 1
for i in range(1 ,int(ceil(n / 2.0 ) ) ):
_a : int = 2 * i + 1
_a : str = 2 * i
_a : Any = total + 4 * odd**2 - 6 * even
return total
if __name__ == "__main__":
import sys
if len(sys.argv) == 1:
print(solution())
else:
try:
a__ = int(sys.argv[1])
print(solution(n))
except ValueError:
print('''Invalid entry - please enter a number''')
| 15
| 0
|
"""simple docstring"""
from operator import delitem, getitem, setitem
import pytest
from data_structures.hashing.hash_map import HashMap
def __UpperCAmelCase ( __a : List[str] ) -> str:
"""simple docstring"""
return getitem, k
def __UpperCAmelCase ( __a : int ,__a : List[Any] ) -> Optional[int]:
"""simple docstring"""
return setitem, k, v
def __UpperCAmelCase ( __a : Any ) -> Optional[Any]:
"""simple docstring"""
return delitem, k
def __UpperCAmelCase ( __a : Any ,__a : List[str] ,*__a : Any ) -> Dict:
"""simple docstring"""
try:
return fun(__a ,*__a ), 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 __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
_a : str = HashMap(initial_block_size=4 )
_a : Union[str, Any] = {}
for _, (fun, *args) in enumerate(__a ):
_a : List[Any] = _run_operation(__a ,__a ,*__a )
_a : List[Any] = _run_operation(__a ,__a ,*__a )
assert my_res == py_res
assert str(__a ) == str(__a )
assert set(__a ) == set(__a )
assert len(__a ) == len(__a )
assert set(my.items() ) == set(py.items() )
def __UpperCAmelCase ( ) -> Optional[Any]:
"""simple docstring"""
def is_public(__a : str ) -> bool:
return not name.startswith('''_''' )
_a : List[Any] = {name for name in dir({} ) if is_public(__a )}
_a : Tuple = {name for name in dir(HashMap() ) if is_public(__a )}
assert dict_public_names > hash_public_names
| 352
|
from typing import Dict, Iterable, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format, to_pil_image
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_pytesseract_available, is_vision_available, logging, requires_backends
if is_vision_available():
import PIL
# soft dependency
if is_pytesseract_available():
import pytesseract
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : str ,__a : Union[str, Any] ) -> List[str]:
"""simple docstring"""
return [
int(1_000 * (box[0] / width) ),
int(1_000 * (box[1] / height) ),
int(1_000 * (box[2] / width) ),
int(1_000 * (box[3] / height) ),
]
def __UpperCAmelCase ( __a : np.ndarray ,__a : Optional[str] ,__a : Optional[str] ) -> List[Any]:
"""simple docstring"""
_a : str = to_pil_image(__a )
_a , _a : Optional[Any] = pil_image.size
_a : Tuple = pytesseract.image_to_data(__a ,lang=__a ,output_type='''dict''' ,config=__a )
_a , _a , _a , _a , _a : List[str] = data['''text'''], data['''left'''], data['''top'''], data['''width'''], data['''height''']
# filter empty words and corresponding coordinates
_a : Dict = [idx for idx, word in enumerate(__a ) if not word.strip()]
_a : str = [word for idx, word in enumerate(__a ) if idx not in irrelevant_indices]
_a : List[str] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : str = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
# turn coordinates into (left, top, left+width, top+height) format
_a : int = []
for x, y, w, h in zip(__a ,__a ,__a ,__a ):
_a : List[str] = [x, y, x + w, y + h]
actual_boxes.append(__a )
# finally, normalize the bounding boxes
_a : Dict = []
for box in actual_boxes:
normalized_boxes.append(normalize_box(__a ,__a ,__a ) )
assert len(__a ) == len(__a ), "Not as many words as there are bounding boxes"
return words, normalized_boxes
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[int] = ["pixel_values"]
def __init__( self , _a = True , _a = None , _a = PILImageResampling.BILINEAR , _a = True , _a = 1 / 2_5_5 , _a = True , _a = None , _a = None , _a = True , _a = None , _a = "" , **_a , ) -> None:
super().__init__(**_a )
_a : List[str] = size if size is not None else {'''height''': 2_2_4, '''width''': 2_2_4}
_a : Union[str, Any] = get_size_dict(_a )
_a : int = do_resize
_a : Optional[int] = size
_a : str = resample
_a : str = do_rescale
_a : Any = rescale_value
_a : Optional[Any] = do_normalize
_a : int = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
_a : List[str] = image_std if image_std is not None else IMAGENET_STANDARD_STD
_a : List[Any] = apply_ocr
_a : Optional[int] = ocr_lang
_a : Tuple = tesseract_config
def __lowercase ( self , _a , _a , _a = PILImageResampling.BILINEAR , _a = None , **_a , ) -> np.ndarray:
_a : Any = get_size_dict(_a )
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()}""" )
_a : Optional[int] = (size['''height'''], size['''width'''])
return resize(_a , size=_a , resample=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
return rescale(_a , scale=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a , _a = None , **_a , ) -> np.ndarray:
return normalize(_a , mean=_a , std=_a , data_format=_a , **_a )
def __lowercase ( 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:
_a : Optional[int] = do_resize if do_resize is not None else self.do_resize
_a : Union[str, Any] = size if size is not None else self.size
_a : Any = get_size_dict(_a )
_a : List[str] = resample if resample is not None else self.resample
_a : int = do_rescale if do_rescale is not None else self.do_rescale
_a : Union[str, Any] = rescale_factor if rescale_factor is not None else self.rescale_factor
_a : int = do_normalize if do_normalize is not None else self.do_normalize
_a : str = image_mean if image_mean is not None else self.image_mean
_a : Tuple = image_std if image_std is not None else self.image_std
_a : Any = apply_ocr if apply_ocr is not None else self.apply_ocr
_a : int = ocr_lang if ocr_lang is not None else self.ocr_lang
_a : Optional[int] = tesseract_config if tesseract_config is not None else self.tesseract_config
_a : List[Any] = make_list_of_images(_a )
if not valid_images(_a ):
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_normalize and (image_mean is None or image_std is None):
raise ValueError('''If do_normalize is True, image_mean and image_std must be specified.''' )
# All transformations expect numpy arrays.
_a : Any = [to_numpy_array(_a ) for image in images]
# Tesseract OCR to get words + normalized bounding boxes
if apply_ocr:
requires_backends(self , '''pytesseract''' )
_a : str = []
_a : str = []
for image in images:
_a , _a : Union[str, Any] = apply_tesseract(_a , _a , _a )
words_batch.append(_a )
boxes_batch.append(_a )
if do_resize:
_a : List[str] = [self.resize(image=_a , size=_a , resample=_a ) for image in images]
if do_rescale:
_a : Optional[Any] = [self.rescale(image=_a , scale=_a ) for image in images]
if do_normalize:
_a : List[Any] = [self.normalize(image=_a , mean=_a , std=_a ) for image in images]
_a : List[str] = [to_channel_dimension_format(_a , _a ) for image in images]
_a : List[str] = BatchFeature(data={'''pixel_values''': images} , tensor_type=_a )
if apply_ocr:
_a : Optional[int] = words_batch
_a : List[Any] = boxes_batch
return data
| 15
| 0
|
"""simple docstring"""
import functools
import operator
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a__ = logging.get_logger(__name__)
a__ = {
'''asapp/sew-d-tiny-100k''': '''https://huggingface.co/asapp/sew-d-tiny-100k/resolve/main/config.json''',
# See all SEW-D models at https://huggingface.co/models?filter=sew-d
}
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = "sew-d"
def __init__( self , _a=3_2 , _a=7_6_8 , _a=1_2 , _a=1_2 , _a=3_0_7_2 , _a=2 , _a=5_1_2 , _a=2_5_6 , _a=True , _a=True , _a=("p2c", "c2p") , _a="layer_norm" , _a="gelu_python" , _a=0.1 , _a=0.1 , _a=0.1 , _a=0.0 , _a=0.1 , _a=0.02 , _a=1e-7 , _a=1e-5 , _a="group" , _a="gelu" , _a=(6_4, 1_2_8, 1_2_8, 1_2_8, 1_2_8, 2_5_6, 2_5_6, 2_5_6, 2_5_6, 5_1_2, 5_1_2, 5_1_2, 5_1_2) , _a=(5, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1) , _a=(1_0, 3, 1, 3, 1, 3, 1, 3, 1, 2, 1, 2, 1) , _a=False , _a=1_2_8 , _a=1_6 , _a=True , _a=0.05 , _a=1_0 , _a=2 , _a=0.0 , _a=1_0 , _a=0 , _a="mean" , _a=False , _a=False , _a=2_5_6 , _a=0 , _a=1 , _a=2 , **_a , ) -> int:
super().__init__(**_a , pad_token_id=_a , bos_token_id=_a , eos_token_id=_a )
_a : Optional[Any] = hidden_size
_a : List[str] = feat_extract_norm
_a : List[str] = feat_extract_activation
_a : Tuple = list(_a )
_a : List[Any] = list(_a )
_a : List[str] = list(_a )
_a : Tuple = conv_bias
_a : str = num_conv_pos_embeddings
_a : int = num_conv_pos_embedding_groups
_a : List[Any] = len(self.conv_dim )
_a : Optional[int] = num_hidden_layers
_a : Optional[int] = intermediate_size
_a : int = squeeze_factor
_a : Optional[Any] = max_position_embeddings
_a : Any = position_buckets
_a : List[str] = share_att_key
_a : Optional[int] = relative_attention
_a : str = norm_rel_ebd
_a : List[Any] = list(_a )
_a : Optional[int] = hidden_act
_a : int = num_attention_heads
_a : str = hidden_dropout
_a : List[Any] = attention_dropout
_a : int = activation_dropout
_a : int = feat_proj_dropout
_a : Any = final_dropout
_a : Any = layer_norm_eps
_a : int = feature_layer_norm_eps
_a : Optional[int] = initializer_range
_a : Tuple = vocab_size
if (
(len(self.conv_stride ) != self.num_feat_extract_layers)
or (len(self.conv_kernel ) != self.num_feat_extract_layers)
or (len(self.conv_dim ) != self.num_feat_extract_layers)
):
raise ValueError(
'''Configuration for convolutional layers is incorrect.'''
'''It is required that `len(config.conv_dim)` == `len(config.conv_stride)` == `len(config.conv_kernel)`,'''
F"""but is `len(config.conv_dim) = {len(self.conv_dim )}`, `len(config.conv_stride)"""
F"""= {len(self.conv_stride )}`, `len(config.conv_kernel) = {len(self.conv_kernel )}`.""" )
# fine-tuning config parameters for SpecAugment: https://arxiv.org/abs/1904.08779
_a : Optional[Any] = apply_spec_augment
_a : Optional[int] = mask_time_prob
_a : Tuple = mask_time_length
_a : str = mask_time_min_masks
_a : Union[str, Any] = mask_feature_prob
_a : Any = mask_feature_length
_a : Optional[int] = mask_feature_min_masks
# ctc loss
_a : Tuple = ctc_loss_reduction
_a : Optional[int] = ctc_zero_infinity
# sequence classification
_a : Optional[Any] = use_weighted_layer_sum
_a : Any = classifier_proj_size
@property
def __lowercase ( self ) -> List[str]:
return functools.reduce(operator.mul , self.conv_stride , 1 )
| 353
|
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from argparse import ArgumentParser
from accelerate.commands.config import get_config_parser
from accelerate.commands.env import env_command_parser
from accelerate.commands.launch import launch_command_parser
from accelerate.commands.test import test_command_parser
from accelerate.commands.tpu import tpu_command_parser
def __UpperCAmelCase ( ) -> Optional[Any]:
"""simple docstring"""
_a : int = ArgumentParser('''Accelerate CLI tool''' ,usage='''accelerate <command> [<args>]''' ,allow_abbrev=__a )
_a : Optional[int] = parser.add_subparsers(help='''accelerate command helpers''' )
# Register commands
get_config_parser(subparsers=__a )
env_command_parser(subparsers=__a )
launch_command_parser(subparsers=__a )
tpu_command_parser(subparsers=__a )
test_command_parser(subparsers=__a )
# Let's go
_a : Dict = parser.parse_args()
if not hasattr(__a ,'''func''' ):
parser.print_help()
exit(1 )
# Run
args.func(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
"""simple docstring"""
import argparse
import os
import re
import packaging.version
a__ = '''examples/'''
a__ = {
'''examples''': (re.compile(R'''^check_min_version\("[^"]+"\)\s*$''', re.MULTILINE), '''check_min_version("VERSION")\n'''),
'''init''': (re.compile(R'''^__version__\s+=\s+"([^"]+)"\s*$''', re.MULTILINE), '''__version__ = "VERSION"\n'''),
'''setup''': (re.compile(R'''^(\s*)version\s*=\s*"[^"]+",''', re.MULTILINE), R'''\1version="VERSION",'''),
'''doc''': (re.compile(R'''^(\s*)release\s*=\s*"[^"]+"$''', re.MULTILINE), '''release = "VERSION"\n'''),
}
a__ = {
'''init''': '''src/transformers/__init__.py''',
'''setup''': '''setup.py''',
}
a__ = '''README.md'''
def __UpperCAmelCase ( __a : List[str] ,__a : int ,__a : Optional[Any] ) -> int:
"""simple docstring"""
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Tuple = f.read()
_a : str = REPLACE_PATTERNS[pattern]
_a : List[str] = replace.replace('''VERSION''' ,__a )
_a : List[Any] = re_pattern.sub(__a ,__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.write(__a )
def __UpperCAmelCase ( __a : Any ) -> List[Any]:
"""simple docstring"""
for folder, directories, fnames in os.walk(__a ):
# Removing some of the folders with non-actively maintained examples from the walk
if "research_projects" in directories:
directories.remove('''research_projects''' )
if "legacy" in directories:
directories.remove('''legacy''' )
for fname in fnames:
if fname.endswith('''.py''' ):
update_version_in_file(os.path.join(__a ,__a ) ,__a ,pattern='''examples''' )
def __UpperCAmelCase ( __a : List[Any] ,__a : List[str]=False ) -> int:
"""simple docstring"""
for pattern, fname in REPLACE_FILES.items():
update_version_in_file(__a ,__a ,__a )
if not patch:
update_version_in_examples(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
_a : Optional[Any] = '''🤗 Transformers currently provides the following architectures'''
_a : str = '''1. Want to contribute a new model?'''
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Optional[int] = f.readlines()
# Find the start of the list.
_a : Optional[int] = 0
while not lines[start_index].startswith(_start_prompt ):
start_index += 1
start_index += 1
_a : List[Any] = start_index
# Update the lines in the model list.
while not lines[index].startswith(_end_prompt ):
if lines[index].startswith('''1.''' ):
_a : Tuple = lines[index].replace(
'''https://huggingface.co/docs/transformers/main/model_doc''' ,'''https://huggingface.co/docs/transformers/model_doc''' ,)
index += 1
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.writelines(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
with open(REPLACE_FILES['''init'''] ,'''r''' ) as f:
_a : Optional[Any] = f.read()
_a : Optional[Any] = REPLACE_PATTERNS['''init'''][0].search(__a ).groups()[0]
return packaging.version.parse(__a )
def __UpperCAmelCase ( __a : Dict=False ) -> str:
"""simple docstring"""
_a : Optional[Any] = get_version()
if patch and default_version.is_devrelease:
raise ValueError('''Can\'t create a patch version from the dev branch, checkout a released version!''' )
if default_version.is_devrelease:
_a : List[Any] = default_version.base_version
elif patch:
_a : str = F"""{default_version.major}.{default_version.minor}.{default_version.micro + 1}"""
else:
_a : List[str] = F"""{default_version.major}.{default_version.minor + 1}.0"""
# Now let's ask nicely if that's the right one.
_a : Dict = input(F"""Which version are you releasing? [{default_version}]""" )
if len(__a ) == 0:
_a : int = default_version
print(F"""Updating version to {version}.""" )
global_version_update(__a ,patch=__a )
if not patch:
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
def __UpperCAmelCase ( ) -> Tuple:
"""simple docstring"""
_a : str = get_version()
_a : int = F"""{current_version.major}.{current_version.minor + 1}.0.dev0"""
_a : List[Any] = current_version.base_version
# Check with the user we got that right.
_a : Union[str, Any] = input(F"""Which version are we developing now? [{dev_version}]""" )
if len(__a ) == 0:
_a : List[str] = dev_version
print(F"""Updating version to {version}.""" )
global_version_update(__a )
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--post_release''', action='''store_true''', help='''Whether this is pre or post release.''')
parser.add_argument('''--patch''', action='''store_true''', help='''Whether or not this is a patch release.''')
a__ = parser.parse_args()
if not args.post_release:
pre_release_work(patch=args.patch)
elif args.patch:
print('''Nothing to do after a patch :-)''')
else:
post_release_work()
| 354
|
import itertools
import os
import random
import tempfile
import unittest
import numpy as np
from transformers import TvltFeatureExtractor, is_datasets_available
from transformers.testing_utils import check_json_file_has_correct_format, require_torch, require_torchaudio
from transformers.utils.import_utils import is_torch_available
from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
if is_torch_available():
import torch
if is_datasets_available():
from datasets import load_dataset
a__ = random.Random()
def __UpperCAmelCase ( __a : Tuple ,__a : str=1.0 ,__a : Optional[int]=None ,__a : List[Any]=None ) -> Any:
"""simple docstring"""
if rng is None:
_a : Dict = global_rng
_a : Optional[Any] = []
for batch_idx in range(shape[0] ):
values.append([] )
for _ in range(shape[1] ):
values[-1].append(rng.random() * scale )
return values
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __init__( self , _a , _a=7 , _a=4_0_0 , _a=2_0_0_0 , _a=2_0_4_8 , _a=1_2_8 , _a=1 , _a=5_1_2 , _a=3_0 , _a=4_4_1_0_0 , ) -> List[Any]:
_a : Optional[Any] = parent
_a : str = batch_size
_a : List[str] = min_seq_length
_a : str = max_seq_length
_a : Dict = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
_a : List[Any] = spectrogram_length
_a : List[str] = feature_size
_a : List[Any] = num_audio_channels
_a : Tuple = hop_length
_a : Optional[int] = chunk_length
_a : int = sampling_rate
def __lowercase ( self ) -> Union[str, Any]:
return {
"spectrogram_length": self.spectrogram_length,
"feature_size": self.feature_size,
"num_audio_channels": self.num_audio_channels,
"hop_length": self.hop_length,
"chunk_length": self.chunk_length,
"sampling_rate": self.sampling_rate,
}
def __lowercase ( self , _a=False , _a=False ) -> List[Any]:
def _flatten(_a ):
return list(itertools.chain(*_a ) )
if equal_length:
_a : List[Any] = [floats_list((self.max_seq_length, self.feature_size) ) for _ in range(self.batch_size )]
else:
# make sure that inputs increase in size
_a : List[Any] = [
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 : str = [np.asarray(_a ) for x in speech_inputs]
return speech_inputs
@require_torch
@require_torchaudio
class UpperCAmelCase_ ( __lowercase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = TvltFeatureExtractor
def __lowercase ( self ) -> Dict:
_a : List[str] = TvltFeatureExtractionTester(self )
def __lowercase ( self ) -> Any:
_a : List[Any] = self.feature_extraction_class(**self.feat_extract_dict )
self.assertTrue(hasattr(_a , '''spectrogram_length''' ) )
self.assertTrue(hasattr(_a , '''feature_size''' ) )
self.assertTrue(hasattr(_a , '''num_audio_channels''' ) )
self.assertTrue(hasattr(_a , '''hop_length''' ) )
self.assertTrue(hasattr(_a , '''chunk_length''' ) )
self.assertTrue(hasattr(_a , '''sampling_rate''' ) )
def __lowercase ( self ) -> Optional[int]:
_a : Optional[Any] = self.feature_extraction_class(**self.feat_extract_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : int = feat_extract_first.save_pretrained(_a )[0]
check_json_file_has_correct_format(_a )
_a : Dict = self.feature_extraction_class.from_pretrained(_a )
_a : List[Any] = feat_extract_first.to_dict()
_a : Union[str, Any] = feat_extract_second.to_dict()
_a : Any = dict_first.pop('''mel_filters''' )
_a : int = dict_second.pop('''mel_filters''' )
self.assertTrue(np.allclose(_a , _a ) )
self.assertEqual(_a , _a )
def __lowercase ( self ) -> Optional[int]:
_a : Any = self.feature_extraction_class(**self.feat_extract_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : Optional[int] = os.path.join(_a , '''feat_extract.json''' )
feat_extract_first.to_json_file(_a )
_a : List[str] = self.feature_extraction_class.from_json_file(_a )
_a : List[Any] = feat_extract_first.to_dict()
_a : Dict = feat_extract_second.to_dict()
_a : str = dict_first.pop('''mel_filters''' )
_a : str = dict_second.pop('''mel_filters''' )
self.assertTrue(np.allclose(_a , _a ) )
self.assertEqual(_a , _a )
def __lowercase ( self ) -> Union[str, Any]:
# Initialize feature_extractor
_a : Union[str, Any] = self.feature_extraction_class(**self.feat_extract_dict )
# create three inputs of length 800, 1000, and 1200
_a : Any = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
_a : List[str] = [np.asarray(_a ) for speech_input in speech_inputs]
# Test not batched input
_a : Tuple = feature_extractor(np_speech_inputs[0] , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test batched
_a : Dict = feature_extractor(_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test audio masking
_a : Union[str, Any] = feature_extractor(
_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 , mask_audio=_a ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test 2-D numpy arrays are batched.
_a : Optional[Any] = [floats_list((1, x) )[0] for x in (8_0_0, 8_0_0, 8_0_0)]
_a : int = np.asarray(_a )
_a : Tuple = feature_extractor(_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[Any] = load_dataset('''hf-internal-testing/librispeech_asr_dummy''' , '''clean''' , split='''validation''' )
# automatic decoding with librispeech
_a : Optional[int] = ds.sort('''id''' ).select(range(_a ) )[:num_samples]['''audio''']
return [x["array"] for x in speech_samples]
def __lowercase ( self ) -> int:
_a : Union[str, Any] = self._load_datasamples(1 )
_a : int = TvltFeatureExtractor()
_a : Union[str, Any] = feature_extractor(_a , return_tensors='''pt''' ).audio_values
self.assertEquals(audio_values.shape , (1, 1, 1_9_2, 1_2_8) )
_a : Union[str, Any] = torch.tensor([[-0.3032, -0.2708], [-0.4434, -0.4007]] )
self.assertTrue(torch.allclose(audio_values[0, 0, :2, :2] , _a , atol=1e-4 ) )
| 15
| 0
|
import PIL.Image
import PIL.ImageOps
from packaging import version
from PIL import Image
if version.parse(version.parse(PIL.__version__).base_version) >= version.parse('''9.1.0'''):
a__ = {
'''linear''': PIL.Image.Resampling.BILINEAR,
'''bilinear''': PIL.Image.Resampling.BILINEAR,
'''bicubic''': PIL.Image.Resampling.BICUBIC,
'''lanczos''': PIL.Image.Resampling.LANCZOS,
'''nearest''': PIL.Image.Resampling.NEAREST,
}
else:
a__ = {
'''linear''': PIL.Image.LINEAR,
'''bilinear''': PIL.Image.BILINEAR,
'''bicubic''': PIL.Image.BICUBIC,
'''lanczos''': PIL.Image.LANCZOS,
'''nearest''': PIL.Image.NEAREST,
}
def __UpperCAmelCase ( __a : Union[str, Any] ) -> Optional[Any]:
"""simple docstring"""
_a : List[Any] = (images / 2 + 0.5).clamp(0 ,1 )
_a : List[Any] = images.cpu().permute(0 ,2 ,3 ,1 ).float().numpy()
_a : Any = numpy_to_pil(__a )
return images
def __UpperCAmelCase ( __a : List[str] ) -> Optional[Any]:
"""simple docstring"""
if images.ndim == 3:
_a : Optional[int] = images[None, ...]
_a : Tuple = (images * 255).round().astype('''uint8''' )
if images.shape[-1] == 1:
# special case for grayscale (single channel) images
_a : Optional[int] = [Image.fromarray(image.squeeze() ,mode='''L''' ) for image in images]
else:
_a : Any = [Image.fromarray(__a ) for image in images]
return pil_images
| 355
|
from typing import Dict
import numpy as np
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, GenericTensor, Pipeline, PipelineException
if is_tf_available():
import tensorflow as tf
from ..tf_utils import stable_softmax
if is_torch_available():
import torch
a__ = logging.get_logger(__name__)
@add_end_docstrings(
__lowercase , r"\n top_k (`int`, defaults to 5):\n The number of predictions to return.\n targets (`str` or `List[str]`, *optional*):\n When passed, the model will limit the scores to the passed targets instead of looking up in the whole\n vocab. If the provided targets are not in the model vocab, they will be tokenized and the first resulting\n token will be used (with a warning, and that might be slower).\n\n " , )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self , _a ) -> np.ndarray:
if self.framework == "tf":
_a : List[str] = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()
elif self.framework == "pt":
_a : Tuple = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a )
else:
raise ValueError('''Unsupported framework''' )
return masked_index
def __lowercase ( self , _a ) -> np.ndarray:
_a : int = self.get_masked_index(_a )
_a : Tuple = np.prod(masked_index.shape )
if numel < 1:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , F"""No mask_token ({self.tokenizer.mask_token}) found on the input""" , )
def __lowercase ( self , _a ) -> Optional[int]:
if isinstance(_a , _a ):
for model_input in model_inputs:
self._ensure_exactly_one_mask_token(model_input['''input_ids'''][0] )
else:
for input_ids in model_inputs["input_ids"]:
self._ensure_exactly_one_mask_token(_a )
def __lowercase ( self , _a , _a=None , **_a ) -> Dict[str, GenericTensor]:
if return_tensors is None:
_a : Union[str, Any] = self.framework
_a : str = self.tokenizer(_a , return_tensors=_a )
self.ensure_exactly_one_mask_token(_a )
return model_inputs
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[str] = self.model(**_a )
_a : Any = model_inputs['''input_ids''']
return model_outputs
def __lowercase ( self , _a , _a=5 , _a=None ) -> str:
# Cap top_k if there are targets
if target_ids is not None and target_ids.shape[0] < top_k:
_a : List[Any] = target_ids.shape[0]
_a : Any = model_outputs['''input_ids'''][0]
_a : List[str] = model_outputs['''logits''']
if self.framework == "tf":
_a : Tuple = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()[:, 0]
_a : List[str] = outputs.numpy()
_a : Dict = outputs[0, masked_index, :]
_a : str = stable_softmax(_a , axis=-1 )
if target_ids is not None:
_a : Any = tf.gather_nd(tf.squeeze(_a , 0 ) , target_ids.reshape(-1 , 1 ) )
_a : Union[str, Any] = tf.expand_dims(_a , 0 )
_a : Optional[int] = tf.math.top_k(_a , k=_a )
_a , _a : Optional[Any] = topk.values.numpy(), topk.indices.numpy()
else:
_a : Optional[Any] = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a ).squeeze(-1 )
# Fill mask pipeline supports only one ${mask_token} per sample
_a : List[str] = outputs[0, masked_index, :]
_a : List[Any] = logits.softmax(dim=-1 )
if target_ids is not None:
_a : List[Any] = probs[..., target_ids]
_a , _a : Optional[Any] = probs.topk(_a )
_a : Dict = []
_a : List[Any] = values.shape[0] == 1
for i, (_values, _predictions) in enumerate(zip(values.tolist() , predictions.tolist() ) ):
_a : Optional[Any] = []
for v, p in zip(_values , _predictions ):
# Copy is important since we're going to modify this array in place
_a : Optional[int] = input_ids.numpy().copy()
if target_ids is not None:
_a : Tuple = target_ids[p].tolist()
_a : List[str] = p
# Filter padding out:
_a : List[Any] = tokens[np.where(tokens != self.tokenizer.pad_token_id )]
# Originally we skip special tokens to give readable output.
# For multi masks though, the other [MASK] would be removed otherwise
# making the output look odd, so we add them back
_a : List[str] = self.tokenizer.decode(_a , skip_special_tokens=_a )
_a : List[Any] = {'''score''': v, '''token''': p, '''token_str''': self.tokenizer.decode([p] ), '''sequence''': sequence}
row.append(_a )
result.append(_a )
if single_mask:
return result[0]
return result
def __lowercase ( self , _a , _a=None ) -> Dict:
if isinstance(_a , _a ):
_a : Tuple = [targets]
try:
_a : int = self.tokenizer.get_vocab()
except Exception:
_a : Any = {}
_a : List[Any] = []
for target in targets:
_a : List[Any] = vocab.get(_a , _a )
if id_ is None:
_a : Tuple = self.tokenizer(
_a , add_special_tokens=_a , return_attention_mask=_a , return_token_type_ids=_a , max_length=1 , truncation=_a , )['''input_ids''']
if len(_a ) == 0:
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
'''We cannot replace it with anything meaningful, ignoring it''' )
continue
_a : Tuple = input_ids[0]
# XXX: If users encounter this pass
# it becomes pretty slow, so let's make sure
# The warning enables them to fix the input to
# get faster performance.
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
F"""Replacing with `{self.tokenizer.convert_ids_to_tokens(id_ )}`.""" )
target_ids.append(id_ )
_a : List[str] = list(set(_a ) )
if len(_a ) == 0:
raise ValueError('''At least one target must be provided when passed.''' )
_a : int = np.array(_a )
return target_ids
def __lowercase ( self , _a=None , _a=None ) -> Tuple:
_a : str = {}
if targets is not None:
_a : List[Any] = self.get_target_ids(_a , _a )
_a : Optional[Any] = target_ids
if top_k is not None:
_a : Union[str, Any] = top_k
if self.tokenizer.mask_token_id is None:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , '''The tokenizer does not define a `mask_token`.''' )
return {}, {}, postprocess_params
def __call__( self , _a , *_a , **_a ) -> int:
_a : Optional[Any] = super().__call__(_a , **_a )
if isinstance(_a , _a ) and len(_a ) == 1:
return outputs[0]
return outputs
| 15
| 0
|
import re
from filelock import FileLock
try:
import nltk
a__ = True
except (ImportError, ModuleNotFoundError):
a__ = False
if NLTK_AVAILABLE:
with FileLock('''.lock''') as lock:
nltk.download('''punkt''', quiet=True)
def __UpperCAmelCase ( __a : str ) -> str:
"""simple docstring"""
re.sub('''<n>''' ,'''''' ,__a ) # remove pegasus newline char
assert NLTK_AVAILABLE, "nltk must be installed to separate newlines between sentences. (pip install nltk)"
return "\n".join(nltk.sent_tokenize(__a ) )
| 356
|
import argparse
import json
import logging
import os
import sys
from unittest.mock import patch
from transformers.testing_utils import TestCasePlus, get_gpu_count, slow
a__ = [
os.path.join(os.path.dirname(__file__), dirname)
for dirname in [
'''text-classification''',
'''language-modeling''',
'''summarization''',
'''token-classification''',
'''question-answering''',
]
]
sys.path.extend(SRC_DIRS)
if SRC_DIRS is not None:
import run_clm_flax
import run_flax_glue
import run_flax_ner
import run_mlm_flax
import run_qa
import run_summarization_flax
import run_ta_mlm_flax
logging.basicConfig(level=logging.DEBUG)
a__ = logging.getLogger()
def __UpperCAmelCase ( ) -> Optional[int]:
"""simple docstring"""
_a : Any = argparse.ArgumentParser()
parser.add_argument('''-f''' )
_a : Dict = parser.parse_args()
return args.f
def __UpperCAmelCase ( __a : Optional[int] ,__a : List[str]="eval" ) -> Any:
"""simple docstring"""
_a : Any = os.path.join(__a ,F"""{split}_results.json""" )
if os.path.exists(__a ):
with open(__a ,'''r''' ) as f:
return json.load(__a )
raise ValueError(F"""can't find {path}""" )
a__ = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self ) -> str:
_a : Any = self.get_auto_remove_tmp_dir()
_a : Optional[Any] = F"""
run_glue.py
--model_name_or_path distilbert-base-uncased
--output_dir {tmp_dir}
--train_file ./tests/fixtures/tests_samples/MRPC/train.csv
--validation_file ./tests/fixtures/tests_samples/MRPC/dev.csv
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
--learning_rate=1e-4
--eval_steps=2
--warmup_steps=2
--seed=42
--max_seq_length=128
""".split()
with patch.object(_a , '''argv''' , _a ):
run_flax_glue.main()
_a : Any = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.75 )
@slow
def __lowercase ( self ) -> Dict:
_a : Tuple = self.get_auto_remove_tmp_dir()
_a : Tuple = F"""
run_clm_flax.py
--model_name_or_path distilgpt2
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--do_train
--do_eval
--block_size 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--num_train_epochs 2
--logging_steps 2 --eval_steps 2
--output_dir {tmp_dir}
--overwrite_output_dir
""".split()
with patch.object(_a , '''argv''' , _a ):
run_clm_flax.main()
_a : List[str] = get_results(_a )
self.assertLess(result['''eval_perplexity'''] , 1_0_0 )
@slow
def __lowercase ( self ) -> Optional[int]:
_a : str = self.get_auto_remove_tmp_dir()
_a : Optional[int] = F"""
run_summarization.py
--model_name_or_path t5-small
--train_file tests/fixtures/tests_samples/xsum/sample.json
--validation_file tests/fixtures/tests_samples/xsum/sample.json
--test_file tests/fixtures/tests_samples/xsum/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--num_train_epochs=3
--warmup_steps=8
--do_train
--do_eval
--do_predict
--learning_rate=2e-4
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
--predict_with_generate
""".split()
with patch.object(_a , '''argv''' , _a ):
run_summarization_flax.main()
_a : Optional[int] = get_results(_a , split='''test''' )
self.assertGreaterEqual(result['''test_rouge1'''] , 1_0 )
self.assertGreaterEqual(result['''test_rouge2'''] , 2 )
self.assertGreaterEqual(result['''test_rougeL'''] , 7 )
self.assertGreaterEqual(result['''test_rougeLsum'''] , 7 )
@slow
def __lowercase ( self ) -> Tuple:
_a : List[str] = self.get_auto_remove_tmp_dir()
_a : List[Any] = F"""
run_mlm.py
--model_name_or_path distilroberta-base
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--output_dir {tmp_dir}
--overwrite_output_dir
--max_seq_length 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--logging_steps 2 --eval_steps 2
--do_train
--do_eval
--num_train_epochs=1
""".split()
with patch.object(_a , '''argv''' , _a ):
run_mlm_flax.main()
_a : List[Any] = get_results(_a )
self.assertLess(result['''eval_perplexity'''] , 4_2 )
@slow
def __lowercase ( self ) -> Dict:
_a : Optional[Any] = self.get_auto_remove_tmp_dir()
_a : int = F"""
run_t5_mlm_flax.py
--model_name_or_path t5-small
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--do_train
--do_eval
--max_seq_length 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--num_train_epochs 2
--logging_steps 2 --eval_steps 2
--output_dir {tmp_dir}
--overwrite_output_dir
""".split()
with patch.object(_a , '''argv''' , _a ):
run_ta_mlm_flax.main()
_a : List[Any] = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.42 )
@slow
def __lowercase ( self ) -> Optional[Any]:
# with so little data distributed training needs more epochs to get the score on par with 0/1 gpu
_a : Any = 7 if get_gpu_count() > 1 else 2
_a : List[Any] = self.get_auto_remove_tmp_dir()
_a : List[Any] = F"""
run_flax_ner.py
--model_name_or_path bert-base-uncased
--train_file tests/fixtures/tests_samples/conll/sample.json
--validation_file tests/fixtures/tests_samples/conll/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--do_train
--do_eval
--warmup_steps=2
--learning_rate=2e-4
--logging_steps 2 --eval_steps 2
--per_device_train_batch_size=2
--per_device_eval_batch_size=2
--num_train_epochs={epochs}
--seed 7
""".split()
with patch.object(_a , '''argv''' , _a ):
run_flax_ner.main()
_a : Dict = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.75 )
self.assertGreaterEqual(result['''eval_f1'''] , 0.3 )
@slow
def __lowercase ( self ) -> Any:
_a : Optional[int] = self.get_auto_remove_tmp_dir()
_a : Union[str, Any] = F"""
run_qa.py
--model_name_or_path bert-base-uncased
--version_2_with_negative
--train_file tests/fixtures/tests_samples/SQUAD/sample.json
--validation_file tests/fixtures/tests_samples/SQUAD/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--num_train_epochs=3
--warmup_steps=2
--do_train
--do_eval
--logging_steps 2 --eval_steps 2
--learning_rate=2e-4
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
""".split()
with patch.object(_a , '''argv''' , _a ):
run_qa.main()
_a : Any = get_results(_a )
self.assertGreaterEqual(result['''eval_f1'''] , 3_0 )
self.assertGreaterEqual(result['''eval_exact'''] , 3_0 )
| 15
| 0
|
# Copyright 2021 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 packaging import version
from .. import __version__
from .constants import IMAGENET_DEFAULT_MEAN, IMAGENET_DEFAULT_STD, IMAGENET_STANDARD_MEAN, IMAGENET_STANDARD_STD
from .doc import (
add_code_sample_docstrings,
add_end_docstrings,
add_start_docstrings,
add_start_docstrings_to_model_forward,
copy_func,
replace_return_docstrings,
)
from .generic import (
ContextManagers,
ExplicitEnum,
ModelOutput,
PaddingStrategy,
TensorType,
add_model_info_to_auto_map,
cached_property,
can_return_loss,
expand_dims,
find_labels,
flatten_dict,
infer_framework,
is_jax_tensor,
is_numpy_array,
is_tensor,
is_tf_symbolic_tensor,
is_tf_tensor,
is_torch_device,
is_torch_dtype,
is_torch_tensor,
reshape,
squeeze,
strtobool,
tensor_size,
to_numpy,
to_py_obj,
transpose,
working_or_temp_dir,
)
from .hub import (
CLOUDFRONT_DISTRIB_PREFIX,
DISABLE_TELEMETRY,
HF_MODULES_CACHE,
HUGGINGFACE_CO_PREFIX,
HUGGINGFACE_CO_RESOLVE_ENDPOINT,
PYTORCH_PRETRAINED_BERT_CACHE,
PYTORCH_TRANSFORMERS_CACHE,
S3_BUCKET_PREFIX,
TRANSFORMERS_CACHE,
TRANSFORMERS_DYNAMIC_MODULE_NAME,
EntryNotFoundError,
PushToHubMixin,
RepositoryNotFoundError,
RevisionNotFoundError,
cached_file,
default_cache_path,
define_sagemaker_information,
download_url,
extract_commit_hash,
get_cached_models,
get_file_from_repo,
get_full_repo_name,
has_file,
http_user_agent,
is_offline_mode,
is_remote_url,
move_cache,
send_example_telemetry,
try_to_load_from_cache,
)
from .import_utils import (
ENV_VARS_TRUE_AND_AUTO_VALUES,
ENV_VARS_TRUE_VALUES,
TORCH_FX_REQUIRED_VERSION,
USE_JAX,
USE_TF,
USE_TORCH,
DummyObject,
OptionalDependencyNotAvailable,
_LazyModule,
ccl_version,
direct_transformers_import,
get_torch_version,
is_accelerate_available,
is_apex_available,
is_bitsandbytes_available,
is_bsa_available,
is_coloredlogs_available,
is_cython_available,
is_datasets_available,
is_decord_available,
is_detectrona_available,
is_faiss_available,
is_flax_available,
is_ftfy_available,
is_in_notebook,
is_ipex_available,
is_jieba_available,
is_jumanpp_available,
is_kenlm_available,
is_keras_nlp_available,
is_librosa_available,
is_natten_available,
is_ninja_available,
is_onnx_available,
is_openai_available,
is_optimum_available,
is_pandas_available,
is_peft_available,
is_phonemizer_available,
is_protobuf_available,
is_psutil_available,
is_pyanvml_available,
is_pyctcdecode_available,
is_pytesseract_available,
is_pytest_available,
is_pytorch_quantization_available,
is_rjieba_available,
is_sacremoses_available,
is_safetensors_available,
is_sagemaker_dp_enabled,
is_sagemaker_mp_enabled,
is_scipy_available,
is_sentencepiece_available,
is_seqio_available,
is_sklearn_available,
is_soundfile_availble,
is_spacy_available,
is_speech_available,
is_sudachi_available,
is_tensorflow_probability_available,
is_tensorflow_text_available,
is_tfaonnx_available,
is_tf_available,
is_timm_available,
is_tokenizers_available,
is_torch_available,
is_torch_bfaa_available,
is_torch_bfaa_cpu_available,
is_torch_bfaa_gpu_available,
is_torch_compile_available,
is_torch_cuda_available,
is_torch_fx_available,
is_torch_fx_proxy,
is_torch_mps_available,
is_torch_neuroncore_available,
is_torch_tensorrt_fx_available,
is_torch_tfaa_available,
is_torch_tpu_available,
is_torchaudio_available,
is_torchdistx_available,
is_torchdynamo_available,
is_torchvision_available,
is_training_run_on_sagemaker,
is_vision_available,
requires_backends,
torch_only_method,
)
a__ = '''pytorch_model.bin'''
a__ = '''pytorch_model.bin.index.json'''
a__ = '''adapter_config.json'''
a__ = '''adapter_model.bin'''
a__ = '''adapter_model.safetensors'''
a__ = '''tf_model.h5'''
a__ = '''tf_model.h5.index.json'''
a__ = '''model.ckpt'''
a__ = '''flax_model.msgpack'''
a__ = '''flax_model.msgpack.index.json'''
a__ = '''model.safetensors'''
a__ = '''model.safetensors.index.json'''
a__ = '''config.json'''
a__ = '''preprocessor_config.json'''
a__ = FEATURE_EXTRACTOR_NAME
a__ = '''generation_config.json'''
a__ = '''modelcard.json'''
a__ = '''▁'''
a__ = SENTENCEPIECE_UNDERLINE # Kept for backward compatibility
a__ = [
[[0, 1, 0, 1], [1, 0, 0, 1]]
] * 2 # Needs to have 0s and 1s only since XLM uses it for langs too.
a__ = [[7, 6, 0, 0, 1], [1, 2, 3, 0, 0], [0, 0, 0, 4, 5]]
a__ = [[1, 1, 1, 1, 1], [1, 1, 1, 0, 0], [0, 0, 0, 1, 1]]
def __UpperCAmelCase ( __a : int ) -> Tuple:
"""simple docstring"""
if version.parse(__a ) < version.parse(__a ):
if "dev" in min_version:
_a : Tuple = (
'''This example requires a source install from HuggingFace Transformers (see '''
'''`https://huggingface.co/docs/transformers/installation#install-from-source`),'''
)
else:
_a : List[str] = F"""This example requires a minimum version of {min_version},"""
error_message += F""" but the version found is {__version__}.\n"""
raise ImportError(
error_message
+ '''Check out https://github.com/huggingface/transformers/tree/main/examples#important-note for the examples corresponding to other '''
'''versions of HuggingFace Transformers.''' )
| 357
|
import argparse
import os
import re
import packaging.version
a__ = '''examples/'''
a__ = {
'''examples''': (re.compile(R'''^check_min_version\("[^"]+"\)\s*$''', re.MULTILINE), '''check_min_version("VERSION")\n'''),
'''init''': (re.compile(R'''^__version__\s+=\s+"([^"]+)"\s*$''', re.MULTILINE), '''__version__ = "VERSION"\n'''),
'''setup''': (re.compile(R'''^(\s*)version\s*=\s*"[^"]+",''', re.MULTILINE), R'''\1version="VERSION",'''),
'''doc''': (re.compile(R'''^(\s*)release\s*=\s*"[^"]+"$''', re.MULTILINE), '''release = "VERSION"\n'''),
}
a__ = {
'''init''': '''src/transformers/__init__.py''',
'''setup''': '''setup.py''',
}
a__ = '''README.md'''
def __UpperCAmelCase ( __a : List[str] ,__a : int ,__a : Optional[Any] ) -> int:
"""simple docstring"""
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Tuple = f.read()
_a , _a : str = REPLACE_PATTERNS[pattern]
_a : List[str] = replace.replace('''VERSION''' ,__a )
_a : List[Any] = re_pattern.sub(__a ,__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.write(__a )
def __UpperCAmelCase ( __a : Any ) -> List[Any]:
"""simple docstring"""
for folder, directories, fnames in os.walk(__a ):
# Removing some of the folders with non-actively maintained examples from the walk
if "research_projects" in directories:
directories.remove('''research_projects''' )
if "legacy" in directories:
directories.remove('''legacy''' )
for fname in fnames:
if fname.endswith('''.py''' ):
update_version_in_file(os.path.join(__a ,__a ) ,__a ,pattern='''examples''' )
def __UpperCAmelCase ( __a : List[Any] ,__a : List[str]=False ) -> int:
"""simple docstring"""
for pattern, fname in REPLACE_FILES.items():
update_version_in_file(__a ,__a ,__a )
if not patch:
update_version_in_examples(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
_a : Optional[Any] = '''🤗 Transformers currently provides the following architectures'''
_a : str = '''1. Want to contribute a new model?'''
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Optional[int] = f.readlines()
# Find the start of the list.
_a : Optional[int] = 0
while not lines[start_index].startswith(_start_prompt ):
start_index += 1
start_index += 1
_a : List[Any] = start_index
# Update the lines in the model list.
while not lines[index].startswith(_end_prompt ):
if lines[index].startswith('''1.''' ):
_a : Tuple = lines[index].replace(
'''https://huggingface.co/docs/transformers/main/model_doc''' ,'''https://huggingface.co/docs/transformers/model_doc''' ,)
index += 1
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.writelines(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
with open(REPLACE_FILES['''init'''] ,'''r''' ) as f:
_a : Optional[Any] = f.read()
_a : Optional[Any] = REPLACE_PATTERNS['''init'''][0].search(__a ).groups()[0]
return packaging.version.parse(__a )
def __UpperCAmelCase ( __a : Dict=False ) -> str:
"""simple docstring"""
_a : Optional[Any] = get_version()
if patch and default_version.is_devrelease:
raise ValueError('''Can\'t create a patch version from the dev branch, checkout a released version!''' )
if default_version.is_devrelease:
_a : List[Any] = default_version.base_version
elif patch:
_a : str = F"""{default_version.major}.{default_version.minor}.{default_version.micro + 1}"""
else:
_a : List[str] = F"""{default_version.major}.{default_version.minor + 1}.0"""
# Now let's ask nicely if that's the right one.
_a : Dict = input(F"""Which version are you releasing? [{default_version}]""" )
if len(__a ) == 0:
_a : int = default_version
print(F"""Updating version to {version}.""" )
global_version_update(__a ,patch=__a )
if not patch:
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
def __UpperCAmelCase ( ) -> Tuple:
"""simple docstring"""
_a : str = get_version()
_a : int = F"""{current_version.major}.{current_version.minor + 1}.0.dev0"""
_a : List[Any] = current_version.base_version
# Check with the user we got that right.
_a : Union[str, Any] = input(F"""Which version are we developing now? [{dev_version}]""" )
if len(__a ) == 0:
_a : List[str] = dev_version
print(F"""Updating version to {version}.""" )
global_version_update(__a )
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--post_release''', action='''store_true''', help='''Whether this is pre or post release.''')
parser.add_argument('''--patch''', action='''store_true''', help='''Whether or not this is a patch release.''')
a__ = parser.parse_args()
if not args.post_release:
pre_release_work(patch=args.patch)
elif args.patch:
print('''Nothing to do after a patch :-)''')
else:
post_release_work()
| 15
| 0
|
import argparse
import collections
import os
import re
import tempfile
import pandas as pd
from datasets import Dataset
from huggingface_hub import hf_hub_download, upload_folder
from transformers.utils import direct_transformers_import
# All paths are set with the intent you should run this script from the root of the repo with the command
# python utils/update_metadata.py
a__ = '''src/transformers'''
# This is to make sure the transformers module imported is the one in the repo.
a__ = direct_transformers_import(TRANSFORMERS_PATH)
# Regexes that match TF/Flax/PT model names.
a__ = re.compile(R'''TF(.*)(?:Model|Encoder|Decoder|ForConditionalGeneration)''')
a__ = re.compile(R'''Flax(.*)(?:Model|Encoder|Decoder|ForConditionalGeneration)''')
# Will match any TF or Flax model too so need to be in an else branch afterthe two previous regexes.
a__ = re.compile(R'''(.*)(?:Model|Encoder|Decoder|ForConditionalGeneration)''')
# Fill this with tuples (pipeline_tag, model_mapping, auto_model)
a__ = [
('''pretraining''', '''MODEL_FOR_PRETRAINING_MAPPING_NAMES''', '''AutoModelForPreTraining'''),
('''feature-extraction''', '''MODEL_MAPPING_NAMES''', '''AutoModel'''),
('''audio-classification''', '''MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING_NAMES''', '''AutoModelForAudioClassification'''),
('''text-generation''', '''MODEL_FOR_CAUSAL_LM_MAPPING_NAMES''', '''AutoModelForCausalLM'''),
('''automatic-speech-recognition''', '''MODEL_FOR_CTC_MAPPING_NAMES''', '''AutoModelForCTC'''),
('''image-classification''', '''MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMES''', '''AutoModelForImageClassification'''),
('''image-segmentation''', '''MODEL_FOR_IMAGE_SEGMENTATION_MAPPING_NAMES''', '''AutoModelForImageSegmentation'''),
('''fill-mask''', '''MODEL_FOR_MASKED_LM_MAPPING_NAMES''', '''AutoModelForMaskedLM'''),
('''object-detection''', '''MODEL_FOR_OBJECT_DETECTION_MAPPING_NAMES''', '''AutoModelForObjectDetection'''),
(
'''zero-shot-object-detection''',
'''MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING_NAMES''',
'''AutoModelForZeroShotObjectDetection''',
),
('''question-answering''', '''MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMES''', '''AutoModelForQuestionAnswering'''),
('''text2text-generation''', '''MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES''', '''AutoModelForSeq2SeqLM'''),
('''text-classification''', '''MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMES''', '''AutoModelForSequenceClassification'''),
('''automatic-speech-recognition''', '''MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMES''', '''AutoModelForSpeechSeq2Seq'''),
(
'''table-question-answering''',
'''MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMES''',
'''AutoModelForTableQuestionAnswering''',
),
('''token-classification''', '''MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMES''', '''AutoModelForTokenClassification'''),
('''multiple-choice''', '''MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMES''', '''AutoModelForMultipleChoice'''),
(
'''next-sentence-prediction''',
'''MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMES''',
'''AutoModelForNextSentencePrediction''',
),
(
'''audio-frame-classification''',
'''MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING_NAMES''',
'''AutoModelForAudioFrameClassification''',
),
('''audio-xvector''', '''MODEL_FOR_AUDIO_XVECTOR_MAPPING_NAMES''', '''AutoModelForAudioXVector'''),
(
'''document-question-answering''',
'''MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMES''',
'''AutoModelForDocumentQuestionAnswering''',
),
(
'''visual-question-answering''',
'''MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING_NAMES''',
'''AutoModelForVisualQuestionAnswering''',
),
('''image-to-text''', '''MODEL_FOR_FOR_VISION_2_SEQ_MAPPING_NAMES''', '''AutoModelForVision2Seq'''),
(
'''zero-shot-image-classification''',
'''MODEL_FOR_ZERO_SHOT_IMAGE_CLASSIFICATION_MAPPING_NAMES''',
'''AutoModelForZeroShotImageClassification''',
),
('''depth-estimation''', '''MODEL_FOR_DEPTH_ESTIMATION_MAPPING_NAMES''', '''AutoModelForDepthEstimation'''),
('''video-classification''', '''MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING_NAMES''', '''AutoModelForVideoClassification'''),
('''mask-generation''', '''MODEL_FOR_MASK_GENERATION_MAPPING_NAMES''', '''AutoModelForMaskGeneration'''),
]
def __UpperCAmelCase ( __a : Optional[Any] ) -> Dict:
"""simple docstring"""
_a : List[Any] = re.finditer('''.+?(?:(?<=[a-z])(?=[A-Z])|(?<=[A-Z])(?=[A-Z][a-z])|$)''' ,__a )
return [m.group(0 ) for m in matches]
def __UpperCAmelCase ( ) -> Dict:
"""simple docstring"""
_a : Union[str, Any] = transformers_module.models.auto.configuration_auto.CONFIG_MAPPING_NAMES
_a : Any = {
config.replace('''Config''' ,'''''' ): model_type for model_type, config in config_maping_names.items()
}
# Dictionaries flagging if each model prefix has a backend in PT/TF/Flax.
_a : List[Any] = collections.defaultdict(__a )
_a : int = collections.defaultdict(__a )
_a : Union[str, Any] = collections.defaultdict(__a )
# Let's lookup through all transformers object (once) and find if models are supported by a given backend.
for attr_name in dir(__a ):
_a : Union[str, Any] = None
if _re_tf_models.match(__a ) is not None:
_a : Any = tf_models
_a : str = _re_tf_models.match(__a ).groups()[0]
elif _re_flax_models.match(__a ) is not None:
_a : List[str] = flax_models
_a : str = _re_flax_models.match(__a ).groups()[0]
elif _re_pt_models.match(__a ) is not None:
_a : List[str] = pt_models
_a : Dict = _re_pt_models.match(__a ).groups()[0]
if lookup_dict is not None:
while len(__a ) > 0:
if attr_name in model_prefix_to_model_type:
_a : Union[str, Any] = True
break
# Try again after removing the last word in the name
_a : List[str] = ''''''.join(camel_case_split(__a )[:-1] )
_a : int = set(list(pt_models.keys() ) + list(tf_models.keys() ) + list(flax_models.keys() ) )
_a : Optional[int] = list(__a )
all_models.sort()
_a : Any = {'''model_type''': all_models}
_a : Optional[Any] = [pt_models[t] for t in all_models]
_a : List[str] = [tf_models[t] for t in all_models]
_a : Optional[Any] = [flax_models[t] for t in all_models]
# Now let's use the auto-mapping names to make sure
_a : Dict = {}
for t in all_models:
if t in transformers_module.models.auto.processing_auto.PROCESSOR_MAPPING_NAMES:
_a : List[str] = '''AutoProcessor'''
elif t in transformers_module.models.auto.tokenization_auto.TOKENIZER_MAPPING_NAMES:
_a : Optional[Any] = '''AutoTokenizer'''
elif t in transformers_module.models.auto.feature_extraction_auto.FEATURE_EXTRACTOR_MAPPING_NAMES:
_a : List[str] = '''AutoFeatureExtractor'''
else:
# Default to AutoTokenizer if a model has nothing, for backward compatibility.
_a : str = '''AutoTokenizer'''
_a : Union[str, Any] = [processors[t] for t in all_models]
return pd.DataFrame(__a )
def __UpperCAmelCase ( __a : str ) -> str:
"""simple docstring"""
_a : Any = [
transformers_module.models.auto.modeling_auto,
transformers_module.models.auto.modeling_tf_auto,
transformers_module.models.auto.modeling_flax_auto,
]
for pipeline_tag, model_mapping, auto_class in PIPELINE_TAGS_AND_AUTO_MODELS:
_a : List[Any] = [model_mapping, F"""TF_{model_mapping}""", F"""FLAX_{model_mapping}"""]
_a : int = [auto_class, F"""TF_{auto_class}""", F"""Flax_{auto_class}"""]
# Loop through all three frameworks
for module, cls, mapping in zip(__a ,__a ,__a ):
# The type of pipeline may not exist in this framework
if not hasattr(__a ,__a ):
continue
# First extract all model_names
_a : Union[str, Any] = []
for name in getattr(__a ,__a ).values():
if isinstance(__a ,__a ):
model_names.append(__a )
else:
model_names.extend(list(__a ) )
# Add pipeline tag and auto model class for those models
table.update({model_name: (pipeline_tag, cls) for model_name in model_names} )
return table
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : List[str] ) -> Optional[Any]:
"""simple docstring"""
_a : int = get_frameworks_table()
_a : Optional[int] = Dataset.from_pandas(__a )
_a : str = hf_hub_download(
'''huggingface/transformers-metadata''' ,'''pipeline_tags.json''' ,repo_type='''dataset''' ,token=__a )
_a : Tuple = Dataset.from_json(__a )
_a : Optional[int] = {
tags_dataset[i]['''model_class''']: (tags_dataset[i]['''pipeline_tag'''], tags_dataset[i]['''auto_class'''])
for i in range(len(__a ) )
}
_a : Union[str, Any] = update_pipeline_and_auto_class_table(__a )
# Sort the model classes to avoid some nondeterministic updates to create false update commits.
_a : Dict = sorted(table.keys() )
_a : str = pd.DataFrame(
{
'''model_class''': model_classes,
'''pipeline_tag''': [table[m][0] for m in model_classes],
'''auto_class''': [table[m][1] for m in model_classes],
} )
_a : Any = Dataset.from_pandas(__a )
with tempfile.TemporaryDirectory() as tmp_dir:
frameworks_dataset.to_json(os.path.join(__a ,'''frameworks.json''' ) )
tags_dataset.to_json(os.path.join(__a ,'''pipeline_tags.json''' ) )
if commit_sha is not None:
_a : Dict = (
F"""Update with commit {commit_sha}\n\nSee: """
F"""https://github.com/huggingface/transformers/commit/{commit_sha}"""
)
else:
_a : int = '''Update'''
upload_folder(
repo_id='''huggingface/transformers-metadata''' ,folder_path=__a ,repo_type='''dataset''' ,token=__a ,commit_message=__a ,)
def __UpperCAmelCase ( ) -> int:
"""simple docstring"""
_a : Any = {tag: cls for tag, _, cls in PIPELINE_TAGS_AND_AUTO_MODELS}
_a : Union[str, Any] = transformers_module.pipelines.SUPPORTED_TASKS
_a : Optional[Any] = []
for key in pipeline_tasks:
if key not in in_table:
_a : Optional[Any] = pipeline_tasks[key]['''pt''']
if isinstance(__a ,(list, tuple) ):
_a : Union[str, Any] = model[0]
_a : Any = model.__name__
if model not in in_table.values():
missing.append(__a )
if len(__a ) > 0:
_a : Optional[int] = ''', '''.join(__a )
raise ValueError(
'''The following pipeline tags are not present in the `PIPELINE_TAGS_AND_AUTO_MODELS` constant inside '''
F"""`utils/update_metadata.py`: {msg}. Please add them!""" )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--token''', type=str, help='''The token to use to push to the transformers-metadata dataset.''')
parser.add_argument('''--commit_sha''', type=str, help='''The sha of the commit going with this update.''')
parser.add_argument('''--check-only''', action='''store_true''', help='''Activate to just check all pipelines are present.''')
a__ = parser.parse_args()
if args.check_only:
check_pipeline_tags()
else:
update_metadata(args.token, args.commit_sha)
| 358
|
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
if n == 1 or not isinstance(__a ,__a ):
return 0
elif n == 2:
return 1
else:
_a : Any = [0, 1]
for i in range(2 ,n + 1 ):
sequence.append(sequence[i - 1] + sequence[i - 2] )
return sequence[n]
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
_a : Any = 0
_a : Dict = 2
while digits < n:
index += 1
_a : Dict = len(str(fibonacci(__a ) ) )
return index
def __UpperCAmelCase ( __a : int = 1_000 ) -> int:
"""simple docstring"""
return fibonacci_digits_index(__a )
if __name__ == "__main__":
print(solution(int(str(input()).strip())))
| 15
| 0
|
import importlib.util
import os
import platform
from argparse import ArgumentParser
import huggingface_hub
from .. import __version__ as version
from ..utils import (
is_accelerate_available,
is_flax_available,
is_safetensors_available,
is_tf_available,
is_torch_available,
)
from . import BaseTransformersCLICommand
def __UpperCAmelCase ( __a : Tuple ) -> str:
"""simple docstring"""
return EnvironmentCommand()
def __UpperCAmelCase ( __a : Union[str, Any] ) -> str:
"""simple docstring"""
return EnvironmentCommand(args.accelerate_config_file )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
@staticmethod
def __lowercase ( _a ) -> int:
_a : str = parser.add_parser('''env''' )
download_parser.set_defaults(func=_a )
download_parser.add_argument(
'''--accelerate-config_file''' , default=_a , help='''The accelerate config file to use for the default values in the launching script.''' , )
download_parser.set_defaults(func=_a )
def __init__( self , _a , *_a ) -> None:
_a : Union[str, Any] = accelerate_config_file
def __lowercase ( self ) -> Optional[int]:
_a : Tuple = '''not installed'''
if is_safetensors_available():
import safetensors
_a : Dict = safetensors.__version__
elif importlib.util.find_spec('''safetensors''' ) is not None:
import safetensors
_a : str = F"""{safetensors.__version__} but is ignored because of PyTorch version too old."""
_a : Optional[Any] = '''not installed'''
_a : List[Any] = '''not found'''
if is_accelerate_available():
import accelerate
from accelerate.commands.config import default_config_file, load_config_from_file
_a : Optional[int] = accelerate.__version__
# Get the default from the config file.
if self._accelerate_config_file is not None or os.path.isfile(_a ):
_a : Optional[int] = load_config_from_file(self._accelerate_config_file ).to_dict()
_a : Optional[int] = (
'''\n'''.join([F"""\t- {prop}: {val}""" for prop, val in accelerate_config.items()] )
if isinstance(_a , _a )
else F"""\t{accelerate_config}"""
)
_a : Tuple = '''not installed'''
_a : Tuple = '''NA'''
if is_torch_available():
import torch
_a : str = torch.__version__
_a : int = torch.cuda.is_available()
_a : List[Any] = '''not installed'''
_a : Optional[Any] = '''NA'''
if is_tf_available():
import tensorflow as tf
_a : int = tf.__version__
try:
# deprecated in v2.1
_a : Dict = tf.test.is_gpu_available()
except AttributeError:
# returns list of devices, convert to bool
_a : List[str] = bool(tf.config.list_physical_devices('''GPU''' ) )
_a : Dict = '''not installed'''
_a : Optional[Any] = '''not installed'''
_a : Optional[Any] = '''not installed'''
_a : List[Any] = '''NA'''
if is_flax_available():
import flax
import jax
import jaxlib
_a : Optional[Any] = flax.__version__
_a : Union[str, Any] = jax.__version__
_a : Union[str, Any] = jaxlib.__version__
_a : Tuple = jax.lib.xla_bridge.get_backend().platform
_a : Tuple = {
'''`transformers` version''': version,
'''Platform''': platform.platform(),
'''Python version''': platform.python_version(),
'''Huggingface_hub version''': huggingface_hub.__version__,
'''Safetensors version''': F"""{safetensors_version}""",
'''Accelerate version''': F"""{accelerate_version}""",
'''Accelerate config''': F"""{accelerate_config_str}""",
'''PyTorch version (GPU?)''': F"""{pt_version} ({pt_cuda_available})""",
'''Tensorflow version (GPU?)''': F"""{tf_version} ({tf_cuda_available})""",
'''Flax version (CPU?/GPU?/TPU?)''': F"""{flax_version} ({jax_backend})""",
'''Jax version''': F"""{jax_version}""",
'''JaxLib version''': F"""{jaxlib_version}""",
'''Using GPU in script?''': '''<fill in>''',
'''Using distributed or parallel set-up in script?''': '''<fill in>''',
}
print('''\nCopy-and-paste the text below in your GitHub issue and FILL OUT the two last points.\n''' )
print(self.format_dict(_a ) )
return info
@staticmethod
def __lowercase ( _a ) -> Optional[int]:
return "\n".join([F"""- {prop}: {val}""" for prop, val in d.items()] ) + "\n"
| 359
|
from sklearn.metrics import fa_score, matthews_corrcoef
import datasets
from .record_evaluation import evaluate as evaluate_record
a__ = '''\
@article{wang2019superglue,
title={SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems},
author={Wang, Alex and Pruksachatkun, Yada and Nangia, Nikita and Singh, Amanpreet and Michael, Julian and Hill, Felix and Levy, Omer and Bowman, Samuel R},
journal={arXiv preprint arXiv:1905.00537},
year={2019}
}
'''
a__ = '''\
SuperGLUE (https://super.gluebenchmark.com/) is a new benchmark styled after
GLUE with a new set of more difficult language understanding tasks, improved
resources, and a new public leaderboard.
'''
a__ = '''
Compute SuperGLUE evaluation metric associated to each SuperGLUE dataset.
Args:
predictions: list of predictions to score. Depending on the SuperGlUE subset:
- for \'record\': list of question-answer dictionaries with the following keys:
- \'idx\': index of the question as specified by the dataset
- \'prediction_text\': the predicted answer text
- for \'multirc\': list of question-answer dictionaries with the following keys:
- \'idx\': index of the question-answer pair as specified by the dataset
- \'prediction\': the predicted answer label
- otherwise: list of predicted labels
references: list of reference labels. Depending on the SuperGLUE subset:
- for \'record\': list of question-answers dictionaries with the following keys:
- \'idx\': index of the question as specified by the dataset
- \'answers\': list of possible answers
- otherwise: list of reference labels
Returns: depending on the SuperGLUE subset:
- for \'record\':
- \'exact_match\': Exact match between answer and gold answer
- \'f1\': F1 score
- for \'multirc\':
- \'exact_match\': Exact match between answer and gold answer
- \'f1_m\': Per-question macro-F1 score
- \'f1_a\': Average F1 score over all answers
- for \'axb\':
\'matthews_correlation\': Matthew Correlation
- for \'cb\':
- \'accuracy\': Accuracy
- \'f1\': F1 score
- for all others:
- \'accuracy\': Accuracy
Examples:
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'copa\') # any of ["copa", "rte", "wic", "wsc", "wsc.fixed", "boolq", "axg"]
>>> predictions = [0, 1]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'accuracy\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'cb\')
>>> predictions = [0, 1]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'accuracy\': 1.0, \'f1\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'record\')
>>> predictions = [{\'idx\': {\'passage\': 0, \'query\': 0}, \'prediction_text\': \'answer\'}]
>>> references = [{\'idx\': {\'passage\': 0, \'query\': 0}, \'answers\': [\'answer\', \'another_answer\']}]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 1.0, \'f1\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'multirc\')
>>> predictions = [{\'idx\': {\'answer\': 0, \'paragraph\': 0, \'question\': 0}, \'prediction\': 0}, {\'idx\': {\'answer\': 1, \'paragraph\': 2, \'question\': 3}, \'prediction\': 1}]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 1.0, \'f1_m\': 1.0, \'f1_a\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'axb\')
>>> references = [0, 1]
>>> predictions = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'matthews_correlation\': 1.0}
'''
def __UpperCAmelCase ( __a : int ,__a : List[str] ) -> Optional[Any]:
"""simple docstring"""
return float((preds == labels).mean() )
def __UpperCAmelCase ( __a : List[Any] ,__a : Union[str, Any] ,__a : List[str]="binary" ) -> Optional[int]:
"""simple docstring"""
_a : List[str] = simple_accuracy(__a ,__a )
_a : Any = float(fa_score(y_true=__a ,y_pred=__a ,average=__a ) )
return {
"accuracy": acc,
"f1": fa,
}
def __UpperCAmelCase ( __a : Optional[Any] ,__a : str ) -> List[Any]:
"""simple docstring"""
_a : Union[str, Any] = {}
for id_pred, label in zip(__a ,__a ):
_a : Optional[int] = F"""{id_pred['idx']['paragraph']}-{id_pred['idx']['question']}"""
_a : Optional[Any] = id_pred['''prediction''']
if question_id in question_map:
question_map[question_id].append((pred, label) )
else:
_a : str = [(pred, label)]
_a , _a : Any = [], []
for question, preds_labels in question_map.items():
_a , _a : Any = zip(*__a )
_a : List[Any] = fa_score(y_true=__a ,y_pred=__a ,average='''macro''' )
fas.append(__a )
_a : List[str] = int(sum(pred == label for pred, label in preds_labels ) == len(__a ) )
ems.append(__a )
_a : List[str] = float(sum(__a ) / len(__a ) )
_a : str = sum(__a ) / len(__a )
_a : Optional[int] = float(fa_score(y_true=__a ,y_pred=[id_pred['''prediction'''] for id_pred in ids_preds] ) )
return {"exact_match": em, "f1_m": fa_m, "f1_a": fa_a}
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> List[Any]:
if self.config_name not in [
"boolq",
"cb",
"copa",
"multirc",
"record",
"rte",
"wic",
"wsc",
"wsc.fixed",
"axb",
"axg",
]:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["boolq", "cb", "copa", "multirc", "record", "rte", "wic", "wsc", "wsc.fixed", "axb", "axg",]''' )
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(self._get_feature_types() ) , codebase_urls=[] , reference_urls=[] , format='''numpy''' if not self.config_name == '''record''' and not self.config_name == '''multirc''' else None , )
def __lowercase ( self ) -> Any:
if self.config_name == "record":
return {
"predictions": {
"idx": {
"passage": datasets.Value('''int64''' ),
"query": datasets.Value('''int64''' ),
},
"prediction_text": datasets.Value('''string''' ),
},
"references": {
"idx": {
"passage": datasets.Value('''int64''' ),
"query": datasets.Value('''int64''' ),
},
"answers": datasets.Sequence(datasets.Value('''string''' ) ),
},
}
elif self.config_name == "multirc":
return {
"predictions": {
"idx": {
"answer": datasets.Value('''int64''' ),
"paragraph": datasets.Value('''int64''' ),
"question": datasets.Value('''int64''' ),
},
"prediction": datasets.Value('''int64''' ),
},
"references": datasets.Value('''int64''' ),
}
else:
return {
"predictions": datasets.Value('''int64''' ),
"references": datasets.Value('''int64''' ),
}
def __lowercase ( self , _a , _a ) -> Optional[Any]:
if self.config_name == "axb":
return {"matthews_correlation": matthews_corrcoef(_a , _a )}
elif self.config_name == "cb":
return acc_and_fa(_a , _a , fa_avg='''macro''' )
elif self.config_name == "record":
_a : Any = [
{
'''qas''': [
{'''id''': ref['''idx''']['''query'''], '''answers''': [{'''text''': ans} for ans in ref['''answers''']]}
for ref in references
]
}
]
_a : Any = {pred['''idx''']['''query''']: pred['''prediction_text'''] for pred in predictions}
return evaluate_record(_a , _a )[0]
elif self.config_name == "multirc":
return evaluate_multirc(_a , _a )
elif self.config_name in ["copa", "rte", "wic", "wsc", "wsc.fixed", "boolq", "axg"]:
return {"accuracy": simple_accuracy(_a , _a )}
else:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["boolq", "cb", "copa", "multirc", "record", "rte", "wic", "wsc", "wsc.fixed", "axb", "axg",]''' )
| 15
| 0
|
import argparse
import requests
import torch
from PIL import Image
from transformers import SwinConfig, SwinForMaskedImageModeling, ViTImageProcessor
def __UpperCAmelCase ( __a : int ) -> Union[str, Any]:
"""simple docstring"""
_a : Any = SwinConfig(image_size=192 )
if "base" in model_name:
_a : Optional[int] = 6
_a : int = 128
_a : str = (2, 2, 18, 2)
_a : Tuple = (4, 8, 16, 32)
elif "large" in model_name:
_a : Union[str, Any] = 12
_a : List[Any] = 192
_a : str = (2, 2, 18, 2)
_a : Optional[Any] = (6, 12, 24, 48)
else:
raise ValueError('''Model not supported, only supports base and large variants''' )
_a : List[str] = window_size
_a : str = embed_dim
_a : Union[str, Any] = depths
_a : Optional[Any] = num_heads
return config
def __UpperCAmelCase ( __a : Tuple ) -> Dict:
"""simple docstring"""
if "encoder.mask_token" in name:
_a : int = name.replace('''encoder.mask_token''' ,'''embeddings.mask_token''' )
if "encoder.patch_embed.proj" in name:
_a : List[str] = name.replace('''encoder.patch_embed.proj''' ,'''embeddings.patch_embeddings.projection''' )
if "encoder.patch_embed.norm" in name:
_a : Tuple = name.replace('''encoder.patch_embed.norm''' ,'''embeddings.norm''' )
if "attn.proj" in name:
_a : List[str] = name.replace('''attn.proj''' ,'''attention.output.dense''' )
if "attn" in name:
_a : int = name.replace('''attn''' ,'''attention.self''' )
if "norm1" in name:
_a : List[str] = name.replace('''norm1''' ,'''layernorm_before''' )
if "norm2" in name:
_a : int = name.replace('''norm2''' ,'''layernorm_after''' )
if "mlp.fc1" in name:
_a : Optional[Any] = name.replace('''mlp.fc1''' ,'''intermediate.dense''' )
if "mlp.fc2" in name:
_a : int = name.replace('''mlp.fc2''' ,'''output.dense''' )
if name == "encoder.norm.weight":
_a : Dict = '''layernorm.weight'''
if name == "encoder.norm.bias":
_a : Optional[Any] = '''layernorm.bias'''
if "decoder" in name:
pass
else:
_a : List[Any] = '''swin.''' + name
return name
def __UpperCAmelCase ( __a : Dict ,__a : Dict ) -> List[str]:
"""simple docstring"""
for key in orig_state_dict.copy().keys():
_a : Tuple = orig_state_dict.pop(__a )
if "attn_mask" in key:
pass
elif "qkv" in key:
_a : List[str] = key.split('''.''' )
_a : Dict = int(key_split[2] )
_a : Any = int(key_split[4] )
_a : Tuple = model.swin.encoder.layers[layer_num].blocks[block_num].attention.self.all_head_size
if "weight" in key:
_a : str = val[:dim, :]
_a : List[Any] = val[
dim : dim * 2, :
]
_a : Tuple = val[-dim:, :]
else:
_a : Optional[Any] = val[
:dim
]
_a : List[Any] = val[
dim : dim * 2
]
_a : str = val[
-dim:
]
else:
_a : Dict = val
return orig_state_dict
def __UpperCAmelCase ( __a : Any ,__a : Optional[Any] ,__a : Union[str, Any] ,__a : Any ) -> List[Any]:
"""simple docstring"""
_a : Any = torch.load(__a ,map_location='''cpu''' )['''model''']
_a : Optional[Any] = get_swin_config(__a )
_a : int = SwinForMaskedImageModeling(__a )
model.eval()
_a : List[str] = convert_state_dict(__a ,__a )
model.load_state_dict(__a )
_a : Tuple = '''http://images.cocodataset.org/val2017/000000039769.jpg'''
_a : Tuple = ViTImageProcessor(size={'''height''': 192, '''width''': 192} )
_a : Any = Image.open(requests.get(__a ,stream=__a ).raw )
_a : str = image_processor(images=__a ,return_tensors='''pt''' )
with torch.no_grad():
_a : List[Any] = model(**__a ).logits
print(outputs.keys() )
print('''Looks ok!''' )
if pytorch_dump_folder_path is not None:
print(F"""Saving model {model_name} to {pytorch_dump_folder_path}""" )
model.save_pretrained(__a )
print(F"""Saving image processor to {pytorch_dump_folder_path}""" )
image_processor.save_pretrained(__a )
if push_to_hub:
print(F"""Pushing model and image processor for {model_name} to hub""" )
model.push_to_hub(F"""microsoft/{model_name}""" )
image_processor.push_to_hub(F"""microsoft/{model_name}""" )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--model_name''',
default='''swin-base-simmim-window6-192''',
type=str,
choices=['''swin-base-simmim-window6-192''', '''swin-large-simmim-window12-192'''],
help='''Name of the Swin SimMIM model you\'d like to convert.''',
)
parser.add_argument(
'''--checkpoint_path''',
default='''/Users/nielsrogge/Documents/SwinSimMIM/simmim_pretrain__swin_base__img192_window6__100ep.pth''',
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 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.'''
)
a__ = parser.parse_args()
convert_swin_checkpoint(args.model_name, args.checkpoint_path, args.pytorch_dump_folder_path, args.push_to_hub)
| 360
|
import numpy as np
def __UpperCAmelCase ( __a : np.ndarray ,__a : np.ndarray ,__a : float = 1E-12 ,__a : int = 100 ,) -> tuple[float, np.ndarray]:
"""simple docstring"""
assert np.shape(__a )[0] == np.shape(__a )[1]
# Ensure proper dimensionality.
assert np.shape(__a )[0] == np.shape(__a )[0]
# Ensure inputs are either both complex or both real
assert np.iscomplexobj(__a ) == np.iscomplexobj(__a )
_a : List[str] = np.iscomplexobj(__a )
if is_complex:
# Ensure complex input_matrix is Hermitian
assert np.array_equal(__a ,input_matrix.conj().T )
# Set convergence to False. Will define convergence when we exceed max_iterations
# or when we have small changes from one iteration to next.
_a : List[str] = False
_a : List[str] = 0
_a : Tuple = 0
_a : str = 1E12
while not convergence:
# Multiple matrix by the vector.
_a : str = np.dot(__a ,__a )
# Normalize the resulting output vector.
_a : List[Any] = w / np.linalg.norm(__a )
# Find rayleigh quotient
# (faster than usual b/c we know vector is normalized already)
_a : Dict = vector.conj().T if is_complex else vector.T
_a : Tuple = np.dot(__a ,np.dot(__a ,__a ) )
# Check convergence.
_a : List[str] = np.abs(lambda_ - lambda_previous ) / lambda_
iterations += 1
if error <= error_tol or iterations >= max_iterations:
_a : Dict = True
_a : str = lambda_
if is_complex:
_a : Tuple = np.real(lambda_ )
return lambda_, vector
def __UpperCAmelCase ( ) -> None:
"""simple docstring"""
_a : List[str] = np.array([[41, 4, 20], [4, 26, 30], [20, 30, 50]] )
_a : int = np.array([41, 4, 20] )
_a : Optional[Any] = real_input_matrix.astype(np.complexaaa )
_a : int = np.triu(1j * complex_input_matrix ,1 )
complex_input_matrix += imag_matrix
complex_input_matrix += -1 * imag_matrix.T
_a : Union[str, Any] = np.array([41, 4, 20] ).astype(np.complexaaa )
for problem_type in ["real", "complex"]:
if problem_type == "real":
_a : Optional[int] = real_input_matrix
_a : Union[str, Any] = real_vector
elif problem_type == "complex":
_a : str = complex_input_matrix
_a : str = complex_vector
# Our implementation.
_a , _a : Optional[Any] = power_iteration(__a ,__a )
# Numpy implementation.
# Get eigenvalues and eigenvectors using built-in numpy
# eigh (eigh used for symmetric or hermetian matrices).
_a , _a : List[str] = np.linalg.eigh(__a )
# Last eigenvalue is the maximum one.
_a : Tuple = eigen_values[-1]
# Last column in this matrix is eigenvector corresponding to largest eigenvalue.
_a : List[Any] = eigen_vectors[:, -1]
# Check our implementation and numpy gives close answers.
assert np.abs(eigen_value - eigen_value_max ) <= 1E-6
# Take absolute values element wise of each eigenvector.
# as they are only unique to a minus sign.
assert np.linalg.norm(np.abs(__a ) - np.abs(__a ) ) <= 1E-6
if __name__ == "__main__":
import doctest
doctest.testmod()
test_power_iteration()
| 15
| 0
|
def __UpperCAmelCase ( __a : list ) -> list:
"""simple docstring"""
if len(__a ) <= 1:
return [tuple(__a )]
_a : Tuple = []
def generate(__a : int ,__a : list ):
if k == 1:
res.append(tuple(arr[:] ) )
return
generate(k - 1 ,__a )
for i in range(k - 1 ):
if k % 2 == 0: # k is even
_a : Dict = arr[k - 1], arr[i]
else: # k is odd
_a : int = arr[k - 1], arr[0]
generate(k - 1 ,__a )
generate(len(__a ) ,__a )
return res
if __name__ == "__main__":
a__ = input('''Enter numbers separated by a comma:\n''').strip()
a__ = [int(item) for item in user_input.split(''',''')]
print(heaps(arr))
| 361
|
import itertools
from dataclasses import dataclass
from typing import Optional
import pandas as pd
import pyarrow as pa
import datasets
from datasets.table import table_cast
@dataclass
class UpperCAmelCase_ ( datasets.BuilderConfig ):
"""simple docstring"""
UpperCAmelCase__ : Optional[datasets.Features] = None
class UpperCAmelCase_ ( datasets.ArrowBasedBuilder ):
"""simple docstring"""
UpperCAmelCase__ : Any = PandasConfig
def __lowercase ( self ) -> Any:
return datasets.DatasetInfo(features=self.config.features )
def __lowercase ( self , _a ) -> List[Any]:
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}""" )
_a : str = dl_manager.download_and_extract(self.config.data_files )
if isinstance(_a , (str, list, tuple) ):
_a : Dict = data_files
if isinstance(_a , _a ):
_a : Dict = [files]
# Use `dl_manager.iter_files` to skip hidden files in an extracted archive
_a : int = [dl_manager.iter_files(_a ) for file in files]
return [datasets.SplitGenerator(name=datasets.Split.TRAIN , gen_kwargs={'''files''': files} )]
_a : Optional[Any] = []
for split_name, files in data_files.items():
if isinstance(_a , _a ):
_a : List[str] = [files]
# Use `dl_manager.iter_files` to skip hidden files in an extracted archive
_a : Any = [dl_manager.iter_files(_a ) for file in files]
splits.append(datasets.SplitGenerator(name=_a , gen_kwargs={'''files''': files} ) )
return splits
def __lowercase ( self , _a ) -> pa.Table:
if self.config.features is not None:
# more expensive cast to support nested features with keys in a different order
# allows str <-> int/float or str to Audio for example
_a : Optional[Any] = table_cast(_a , self.config.features.arrow_schema )
return pa_table
def __lowercase ( self , _a ) -> List[str]:
for i, file in enumerate(itertools.chain.from_iterable(_a ) ):
with open(_a , '''rb''' ) as f:
_a : str = pa.Table.from_pandas(pd.read_pickle(_a ) )
yield i, self._cast_table(_a )
| 15
| 0
|
def __UpperCAmelCase ( __a : int ) -> bool:
"""simple docstring"""
if not isinstance(__a ,__a ):
_a : Any = F"""Input value of [number={number}] must be an integer"""
raise TypeError(__a )
if number < 0:
return False
_a : Tuple = number * number
while number > 0:
if number % 10 != number_square % 10:
return False
number //= 10
number_square //= 10
return True
if __name__ == "__main__":
import doctest
doctest.testmod()
| 362
|
def __UpperCAmelCase ( __a : int ,__a : int ,__a : int ) -> int:
"""simple docstring"""
if exponent == 1:
return base
if exponent % 2 == 0:
_a : List[Any] = _modexpt(__a ,exponent // 2 ,__a ) % modulo_value
return (x * x) % modulo_value
else:
return (base * _modexpt(__a ,exponent - 1 ,__a )) % modulo_value
def __UpperCAmelCase ( __a : int = 1_777 ,__a : int = 1_855 ,__a : int = 8 ) -> int:
"""simple docstring"""
_a : List[Any] = base
for _ in range(1 ,__a ):
_a : Any = _modexpt(__a ,__a ,10**digits )
return result
if __name__ == "__main__":
print(f'''{solution() = }''')
| 15
| 0
|
import warnings
from pathlib import Path
from typing import List, Tuple, Union
import fire
from torch import nn
from transformers import AutoModelForSeqaSeqLM, AutoTokenizer, PreTrainedModel
from transformers.utils import logging
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : nn.ModuleList ,__a : nn.ModuleList ,__a : List[int] ) -> None:
"""simple docstring"""
_a : Optional[Any] = nn.ModuleList([src_layers[i] for i in layers_to_copy] )
assert len(__a ) == len(__a ), F"""{len(__a )} != {len(__a )}"""
dest_layers.load_state_dict(layers_to_copy.state_dict() )
a__ = {
# maps num layers in teacher -> num_layers in student -> which teacher layers to copy.
# 12: bart, 16: pegasus, 6: marian/Helsinki-NLP
12: {
1: [0], # This says that if the teacher has 12 layers and the student has 1, copy layer 0 of the teacher
2: [0, 6],
3: [0, 6, 11],
4: [0, 4, 8, 11],
6: [0, 2, 4, 7, 9, 11],
9: [0, 1, 2, 4, 5, 7, 9, 10, 11],
12: list(range(12)),
},
16: { # maps num layers in student -> which teacher layers to copy
1: [0],
2: [0, 15],
3: [0, 8, 15],
4: [0, 5, 10, 15],
6: [0, 3, 6, 9, 12, 15],
8: [0, 2, 4, 6, 8, 10, 12, 15],
9: [0, 1, 3, 5, 7, 9, 11, 13, 15],
12: [0, 1, 2, 3, 4, 5, 6, 7, 9, 11, 13, 15],
16: list(range(16)),
},
6: {1: [0], 2: [0, 5], 3: [0, 2, 5], 4: [0, 1, 3, 5], 6: list(range(6))},
}
a__ = {
# maps num layers in student -> which teacher layers to copy.
6: {1: [5], 2: [3, 5], 3: [1, 4, 5], 4: [1, 2, 4, 5]},
12: {1: [11], 2: [5, 11], 3: [3, 7, 11], 6: [1, 3, 5, 8, 10, 11]},
16: {1: [15], 4: [4, 9, 12, 15], 8: [1, 3, 5, 7, 9, 11, 13, 15]},
}
def __UpperCAmelCase ( __a : str ,__a : Union[str, Any] ) -> int:
"""simple docstring"""
try:
_a : int = LAYERS_TO_COPY[n_teacher][n_student]
return val
except KeyError:
if n_student != n_teacher:
warnings.warn(
F"""no hardcoded layers to copy for teacher {n_teacher} -> student {n_student}, defaulting to first"""
F""" {n_student}""" )
return list(range(__a ) )
def __UpperCAmelCase ( __a : Dict ,__a : List[str] ) -> List[int]:
"""simple docstring"""
if n_student > n_teacher:
raise ValueError(F"""Cannot perform intermediate supervision for student {n_student} > teacher {n_teacher}""" )
elif n_teacher == n_student:
return list(range(__a ) )
elif n_student == 1:
return [n_teacher - 1]
else:
return LAYERS_TO_SUPERVISE[n_teacher][n_student]
def __UpperCAmelCase ( __a : Union[str, PreTrainedModel] ,__a : Union[str, Path] = "student" ,__a : Union[int, None] = None ,__a : Union[int, None] = None ,__a : Any=False ,__a : Any=None ,__a : int=None ,**__a : Optional[int] ,) -> Tuple[PreTrainedModel, List[int], List[int]]:
"""simple docstring"""
_a : Dict = '''encoder_layers and decoder_layers cannot be both None-- you would just have an identical teacher.'''
assert (e is not None) or (d is not None), _msg
if isinstance(__a ,__a ):
AutoTokenizer.from_pretrained(__a ).save_pretrained(__a ) # purely for convenience
_a : List[str] = AutoModelForSeqaSeqLM.from_pretrained(__a ).eval()
else:
assert isinstance(__a ,__a ), F"""teacher must be a model or string got type {type(__a )}"""
_a : Dict = teacher.config.to_diff_dict()
try:
_a : List[Any] = teacher.config.encoder_layers, teacher.config.decoder_layers
if e is None:
_a : Union[str, Any] = teacher_e
if d is None:
_a : List[Any] = teacher_d
init_kwargs.update({'''encoder_layers''': e, '''decoder_layers''': d} )
except AttributeError: # T5
if hasattr(teacher.config ,'''num_encoder_layers''' ):
_a : List[str] = teacher.config.num_encoder_layers, teacher.config.num_decoder_layers
else:
_a : Optional[Any] = teacher.config.num_layers, teacher.config.num_decoder_layers
if e is None:
_a : Optional[int] = teacher_e
if d is None:
_a : Optional[Any] = teacher_d
if hasattr(teacher.config ,'''num_encoder_layers''' ):
init_kwargs.update({'''num_encoder_layers''': e, '''num_decoder_layers''': d} )
else:
init_kwargs.update({'''num_layers''': e, '''num_decoder_layers''': d} )
# Kwargs to instantiate student: teacher kwargs with updated layer numbers + **extra_config_kwargs
init_kwargs.update(__a )
# Copy weights
_a : List[str] = teacher.config_class(**__a )
_a : Dict = AutoModelForSeqaSeqLM.from_config(__a )
# Start by copying the full teacher state dict this will copy the first N teacher layers to the student.
_a : str = student.load_state_dict(teacher.state_dict() ,strict=__a )
assert info.missing_keys == [], info.missing_keys # every student key should have a teacher keys.
if copy_first_teacher_layers: # Our copying is done. We just log and save
_a : List[str] = list(range(__a ) ), list(range(__a ) )
logger.info(
F"""Copied encoder layers {e_layers_to_copy} and decoder layers {d_layers_to_copy}. Saving them to"""
F""" {save_path}""" )
student.save_pretrained(__a )
return student, e_layers_to_copy, d_layers_to_copy
# Decide which layers of the teacher to copy. Not exactly alternating -- we try to keep first and last layer.
if e_layers_to_copy is None:
_a : List[int] = pick_layers_to_copy(__a ,__a )
if d_layers_to_copy is None:
_a : List[int] = pick_layers_to_copy(__a ,__a )
try:
if hasattr(
__a ,'''prophetnet''' ): # For ProphetNet, student.model.encoder.layers is called student.prophetnet.encoder.layers
copy_layers(teacher.prophetnet.encoder.layers ,student.prophetnet.encoder.layers ,__a )
copy_layers(teacher.prophetnet.decoder.layers ,student.prophetnet.decoder.layers ,__a )
else:
copy_layers(teacher.model.encoder.layers ,student.model.encoder.layers ,__a )
copy_layers(teacher.model.decoder.layers ,student.model.decoder.layers ,__a )
except AttributeError: # For t5, student.model.encoder.layers is called student.encoder.block
copy_layers(teacher.encoder.block ,student.encoder.block ,__a )
copy_layers(teacher.decoder.block ,student.decoder.block ,__a )
logger.info(
F"""Copied encoder layers {e_layers_to_copy} and decoder layers {d_layers_to_copy}. Saving them to {save_path}""" )
_a : List[Any] = {
'''teacher_type''': teacher.config.model_type,
'''copied_encoder_layers''': e_layers_to_copy,
'''copied_decoder_layers''': d_layers_to_copy,
}
student.save_pretrained(__a )
# Save information about copying for easier reproducibility
return student, e_layers_to_copy, d_layers_to_copy
if __name__ == "__main__":
fire.Fire(create_student_by_copying_alternating_layers)
| 363
|
import numpy as np
import torch
from torch.nn import CrossEntropyLoss
from transformers import AutoModelForCausalLM, AutoTokenizer
import datasets
from datasets import logging
a__ = '''\
'''
a__ = '''
Perplexity (PPL) is one of the most common metrics for evaluating language models.
It is defined as the exponentiated average negative log-likelihood of a sequence.
For more information, see https://huggingface.co/docs/transformers/perplexity
'''
a__ = '''
Args:
model_id (str): model used for calculating Perplexity
NOTE: Perplexity can only be calculated for causal language models.
This includes models such as gpt2, causal variations of bert,
causal versions of t5, and more (the full list can be found
in the AutoModelForCausalLM documentation here:
https://huggingface.co/docs/transformers/master/en/model_doc/auto#transformers.AutoModelForCausalLM )
input_texts (list of str): input text, each separate text snippet
is one list entry.
batch_size (int): the batch size to run texts through the model. Defaults to 16.
add_start_token (bool): whether to add the start token to the texts,
so the perplexity can include the probability of the first word. Defaults to True.
device (str): device to run on, defaults to \'cuda\' when available
Returns:
perplexity: dictionary containing the perplexity scores for the texts
in the input list, as well as the mean perplexity. If one of the input texts is
longer than the max input length of the model, then it is truncated to the
max length for the perplexity computation.
Examples:
Example 1:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = ["lorem ipsum", "Happy Birthday!", "Bienvenue"]
>>> results = perplexity.compute(model_id=\'gpt2\',
... add_start_token=False,
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
78.22
>>> print(round(results["perplexities"][0], 2))
11.11
Example 2:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = datasets.load_dataset("wikitext",
... "wikitext-2-raw-v1",
... split="test")["text"][:50] # doctest:+ELLIPSIS
[...]
>>> input_texts = [s for s in input_texts if s!=\'\']
>>> results = perplexity.compute(model_id=\'gpt2\',
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
60.35
>>> print(round(results["perplexities"][0], 2))
81.12
'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> Any:
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(
{
'''input_texts''': datasets.Value('''string''' ),
} ) , reference_urls=['''https://huggingface.co/docs/transformers/perplexity'''] , )
def __lowercase ( self , _a , _a , _a = 1_6 , _a = True , _a=None ) -> List[Any]:
if device is not None:
assert device in ["gpu", "cpu", "cuda"], "device should be either gpu or cpu."
if device == "gpu":
_a : List[str] = '''cuda'''
else:
_a : Optional[Any] = '''cuda''' if torch.cuda.is_available() else '''cpu'''
_a : Dict = AutoModelForCausalLM.from_pretrained(_a )
_a : List[Any] = model.to(_a )
_a : List[str] = AutoTokenizer.from_pretrained(_a )
# if batch_size > 1 (which generally leads to padding being required), and
# if there is not an already assigned pad_token, assign an existing
# special token to also be the padding token
if tokenizer.pad_token is None and batch_size > 1:
_a : str = list(tokenizer.special_tokens_map_extended.values() )
# check that the model already has at least one special token defined
assert (
len(_a ) > 0
), "If batch_size > 1, model must have at least one special token to use for padding. Please use a different model or set batch_size=1."
# assign one of the special tokens to also be the pad token
tokenizer.add_special_tokens({'''pad_token''': existing_special_tokens[0]} )
if add_start_token:
# leave room for <BOS> token to be added:
assert (
tokenizer.bos_token is not None
), "Input model must already have a BOS token if using add_start_token=True. Please use a different model, or set add_start_token=False"
_a : List[Any] = model.config.max_length - 1
else:
_a : List[str] = model.config.max_length
_a : Union[str, Any] = tokenizer(
_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , return_tensors='''pt''' , return_attention_mask=_a , ).to(_a )
_a : List[Any] = encodings['''input_ids''']
_a : int = encodings['''attention_mask''']
# check that each input is long enough:
if add_start_token:
assert torch.all(torch.ge(attn_masks.sum(1 ) , 1 ) ), "Each input text must be at least one token long."
else:
assert torch.all(
torch.ge(attn_masks.sum(1 ) , 2 ) ), "When add_start_token=False, each input text must be at least two tokens long. Run with add_start_token=True if inputting strings of only one token, and remove all empty input strings."
_a : Optional[int] = []
_a : Dict = CrossEntropyLoss(reduction='''none''' )
for start_index in logging.tqdm(range(0 , len(_a ) , _a ) ):
_a : Dict = min(start_index + batch_size , len(_a ) )
_a : Union[str, Any] = encoded_texts[start_index:end_index]
_a : int = attn_masks[start_index:end_index]
if add_start_token:
_a : Dict = torch.tensor([[tokenizer.bos_token_id]] * encoded_batch.size(dim=0 ) ).to(_a )
_a : List[str] = torch.cat([bos_tokens_tensor, encoded_batch] , dim=1 )
_a : Dict = torch.cat(
[torch.ones(bos_tokens_tensor.size() , dtype=torch.intaa ).to(_a ), attn_mask] , dim=1 )
_a : Dict = encoded_batch
with torch.no_grad():
_a : Any = model(_a , attention_mask=_a ).logits
_a : List[str] = out_logits[..., :-1, :].contiguous()
_a : Union[str, Any] = labels[..., 1:].contiguous()
_a : Optional[int] = attn_mask[..., 1:].contiguous()
_a : Union[str, Any] = torch.expa(
(loss_fct(shift_logits.transpose(1 , 2 ) , _a ) * shift_attention_mask_batch).sum(1 )
/ shift_attention_mask_batch.sum(1 ) )
ppls += perplexity_batch.tolist()
return {"perplexities": ppls, "mean_perplexity": np.mean(_a )}
| 15
| 0
|
"""simple docstring"""
from typing import Dict
import numpy as np
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, GenericTensor, Pipeline, PipelineException
if is_tf_available():
import tensorflow as tf
from ..tf_utils import stable_softmax
if is_torch_available():
import torch
a__ = logging.get_logger(__name__)
@add_end_docstrings(
__lowercase , r"\n top_k (`int`, defaults to 5):\n The number of predictions to return.\n targets (`str` or `List[str]`, *optional*):\n When passed, the model will limit the scores to the passed targets instead of looking up in the whole\n vocab. If the provided targets are not in the model vocab, they will be tokenized and the first resulting\n token will be used (with a warning, and that might be slower).\n\n " , )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self , _a ) -> np.ndarray:
if self.framework == "tf":
_a : List[str] = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()
elif self.framework == "pt":
_a : Tuple = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a )
else:
raise ValueError('''Unsupported framework''' )
return masked_index
def __lowercase ( self , _a ) -> np.ndarray:
_a : int = self.get_masked_index(_a )
_a : Tuple = np.prod(masked_index.shape )
if numel < 1:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , F"""No mask_token ({self.tokenizer.mask_token}) found on the input""" , )
def __lowercase ( self , _a ) -> Optional[int]:
if isinstance(_a , _a ):
for model_input in model_inputs:
self._ensure_exactly_one_mask_token(model_input['''input_ids'''][0] )
else:
for input_ids in model_inputs["input_ids"]:
self._ensure_exactly_one_mask_token(_a )
def __lowercase ( self , _a , _a=None , **_a ) -> Dict[str, GenericTensor]:
if return_tensors is None:
_a : Union[str, Any] = self.framework
_a : str = self.tokenizer(_a , return_tensors=_a )
self.ensure_exactly_one_mask_token(_a )
return model_inputs
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[str] = self.model(**_a )
_a : Any = model_inputs['''input_ids''']
return model_outputs
def __lowercase ( self , _a , _a=5 , _a=None ) -> str:
# Cap top_k if there are targets
if target_ids is not None and target_ids.shape[0] < top_k:
_a : List[Any] = target_ids.shape[0]
_a : Any = model_outputs['''input_ids'''][0]
_a : List[str] = model_outputs['''logits''']
if self.framework == "tf":
_a : Tuple = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()[:, 0]
_a : List[str] = outputs.numpy()
_a : Dict = outputs[0, masked_index, :]
_a : str = stable_softmax(_a , axis=-1 )
if target_ids is not None:
_a : Any = tf.gather_nd(tf.squeeze(_a , 0 ) , target_ids.reshape(-1 , 1 ) )
_a : Union[str, Any] = tf.expand_dims(_a , 0 )
_a : Optional[int] = tf.math.top_k(_a , k=_a )
_a : Optional[Any] = topk.values.numpy(), topk.indices.numpy()
else:
_a : Optional[Any] = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a ).squeeze(-1 )
# Fill mask pipeline supports only one ${mask_token} per sample
_a : List[str] = outputs[0, masked_index, :]
_a : List[Any] = logits.softmax(dim=-1 )
if target_ids is not None:
_a : List[Any] = probs[..., target_ids]
_a : Optional[Any] = probs.topk(_a )
_a : Dict = []
_a : List[Any] = values.shape[0] == 1
for i, (_values, _predictions) in enumerate(zip(values.tolist() , predictions.tolist() ) ):
_a : Optional[Any] = []
for v, p in zip(_values , _predictions ):
# Copy is important since we're going to modify this array in place
_a : Optional[int] = input_ids.numpy().copy()
if target_ids is not None:
_a : Tuple = target_ids[p].tolist()
_a : List[str] = p
# Filter padding out:
_a : List[Any] = tokens[np.where(tokens != self.tokenizer.pad_token_id )]
# Originally we skip special tokens to give readable output.
# For multi masks though, the other [MASK] would be removed otherwise
# making the output look odd, so we add them back
_a : List[str] = self.tokenizer.decode(_a , skip_special_tokens=_a )
_a : List[Any] = {'''score''': v, '''token''': p, '''token_str''': self.tokenizer.decode([p] ), '''sequence''': sequence}
row.append(_a )
result.append(_a )
if single_mask:
return result[0]
return result
def __lowercase ( self , _a , _a=None ) -> Dict:
if isinstance(_a , _a ):
_a : Tuple = [targets]
try:
_a : int = self.tokenizer.get_vocab()
except Exception:
_a : Any = {}
_a : List[Any] = []
for target in targets:
_a : List[Any] = vocab.get(_a , _a )
if id_ is None:
_a : Tuple = self.tokenizer(
_a , add_special_tokens=_a , return_attention_mask=_a , return_token_type_ids=_a , max_length=1 , truncation=_a , )['''input_ids''']
if len(_a ) == 0:
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
'''We cannot replace it with anything meaningful, ignoring it''' )
continue
_a : Tuple = input_ids[0]
# XXX: If users encounter this pass
# it becomes pretty slow, so let's make sure
# The warning enables them to fix the input to
# get faster performance.
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
F"""Replacing with `{self.tokenizer.convert_ids_to_tokens(id_ )}`.""" )
target_ids.append(id_ )
_a : List[str] = list(set(_a ) )
if len(_a ) == 0:
raise ValueError('''At least one target must be provided when passed.''' )
_a : int = np.array(_a )
return target_ids
def __lowercase ( self , _a=None , _a=None ) -> Tuple:
_a : str = {}
if targets is not None:
_a : List[Any] = self.get_target_ids(_a , _a )
_a : Optional[Any] = target_ids
if top_k is not None:
_a : Union[str, Any] = top_k
if self.tokenizer.mask_token_id is None:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , '''The tokenizer does not define a `mask_token`.''' )
return {}, {}, postprocess_params
def __call__( self , _a , *_a , **_a ) -> int:
_a : Optional[Any] = super().__call__(_a , **_a )
if isinstance(_a , _a ) and len(_a ) == 1:
return outputs[0]
return outputs
| 364
|
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a__ = {
'''configuration_xmod''': [
'''XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP''',
'''XmodConfig''',
'''XmodOnnxConfig''',
],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''XMOD_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''XmodForCausalLM''',
'''XmodForMaskedLM''',
'''XmodForMultipleChoice''',
'''XmodForQuestionAnswering''',
'''XmodForSequenceClassification''',
'''XmodForTokenClassification''',
'''XmodModel''',
'''XmodPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_xmod import XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP, XmodConfig, XmodOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_xmod import (
XMOD_PRETRAINED_MODEL_ARCHIVE_LIST,
XmodForCausalLM,
XmodForMaskedLM,
XmodForMultipleChoice,
XmodForQuestionAnswering,
XmodForSequenceClassification,
XmodForTokenClassification,
XmodModel,
XmodPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 15
| 0
|
a__ = {0: [2, 3], 1: [0], 2: [1], 3: [4], 4: []}
a__ = {0: [1, 2, 3], 1: [2], 2: [0], 3: [4], 4: [5], 5: [3]}
def __UpperCAmelCase ( __a : dict[int, list[int]] ,__a : int ,__a : list[bool] ) -> list[int]:
"""simple docstring"""
_a : Tuple = True
_a : List[Any] = []
for neighbour in graph[vert]:
if not visited[neighbour]:
order += topology_sort(__a ,__a ,__a )
order.append(__a )
return order
def __UpperCAmelCase ( __a : dict[int, list[int]] ,__a : int ,__a : list[bool] ) -> list[int]:
"""simple docstring"""
_a : Tuple = True
_a : List[str] = [vert]
for neighbour in reversed_graph[vert]:
if not visited[neighbour]:
component += find_components(__a ,__a ,__a )
return component
def __UpperCAmelCase ( __a : dict[int, list[int]] ) -> list[list[int]]:
"""simple docstring"""
_a : Tuple = len(__a ) * [False]
_a : dict[int, list[int]] = {vert: [] for vert in range(len(__a ) )}
for vert, neighbours in graph.items():
for neighbour in neighbours:
reversed_graph[neighbour].append(__a )
_a : Dict = []
for i, was_visited in enumerate(__a ):
if not was_visited:
order += topology_sort(__a ,__a ,__a )
_a : Optional[Any] = []
_a : Tuple = len(__a ) * [False]
for i in range(len(__a ) ):
_a : Dict = order[len(__a ) - i - 1]
if not visited[vert]:
_a : int = find_components(__a ,__a ,__a )
components_list.append(__a )
return components_list
| 365
|
import re
import tempfile
from pathlib import Path
import pytest
import yaml
from datasets.utils.readme import ReadMe
# @pytest.fixture
# def example_yaml_structure():
a__ = yaml.safe_load(
'''\
name: ""
allow_empty: false
allow_empty_text: true
subsections:
- name: "Dataset Card for X" # First-level markdown heading
allow_empty: false
allow_empty_text: true
subsections:
- name: "Table of Contents"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Dataset Description"
allow_empty: false
allow_empty_text: false
subsections:
- name: "Dataset Summary"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Supported Tasks and Leaderboards"
allow_empty: true
allow_empty_text: true
subsections: null
- name: Languages
allow_empty: false
allow_empty_text: true
subsections: null
'''
)
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
#### Extra Ignored Subsection
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Extra Ignored Subsection''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
}
],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tEmpty YAML markers are present in the README.'''
)
a__ = '''\
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tNo YAML markers are present in the README.'''
)
a__ = '''\
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tOnly the start of YAML tags present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Summary` but it is empty.\n-\tExpected some text in section `Dataset Summary` but it is empty (text in subsections are ignored).'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Card for My Dataset` but it is empty.\n-\tSection `Dataset Card for My Dataset` expected the following subsections: `Table of Contents`, `Dataset Description`. Found \'None\'.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tSection `Dataset Description` is missing subsection: `Supported Tasks and Leaderboards`.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Languages` but it is empty.'''
a__ = '''\
---
language:
- zh
- en
---
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
# Dataset Card My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has several first-level headings: `Dataset Card for My Dataset`, `Dataset Card My Dataset`. Only one heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tNo first-level heading starting with `Dataset Card for` found in README. Skipping further validation for this README.'''
a__ = ''''''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.\n-\tNo YAML markers are present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found while parsing the README at `{path}`:\n-\tMultiple sections with the same heading `Dataset Card for My Dataset` have been found. Please keep only one of these sections.'''
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : List[str] ) -> Optional[int]:
"""simple docstring"""
assert ReadMe.from_string(__a ,__a ).to_dict() == expected_dict
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[str] ,__a : Optional[Any] ) -> Union[str, Any]:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
_a : List[Any] = ReadMe.from_string(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Dict ,__a : Dict ) -> Tuple:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
ReadMe.from_string(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> Tuple:
"""simple docstring"""
ReadMe.from_string(__a ,__a ,suppress_parsing_errors=__a )
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : Any ) -> Optional[int]:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Tuple = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[Any] = ReadMe.from_readme(__a ,__a ).to_dict()
assert out["name"] == path
assert out["text"] == ""
assert out["is_empty_text"]
assert out["subsections"] == expected_dict["subsections"]
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[Any] ,__a : List[Any] ) -> int:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[int] = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
_a : Any = ReadMe.from_readme(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : str ,__a : Union[str, Any] ) -> Dict:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Optional[Any] = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : str = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
ReadMe.from_readme(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> str:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
ReadMe.from_readme(__a ,__a ,suppress_parsing_errors=__a )
| 15
| 0
|
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
if not isinstance(__a ,__a ):
_a : str = F"""Input value of [number={number}] must be an integer"""
raise TypeError(__a )
if number < 1:
_a : Optional[Any] = F"""Input value of [number={number}] must be > 0"""
raise ValueError(__a )
_a : Optional[int] = 1
for i in range(1 ,__a ):
current_number *= 4 * i - 2
current_number //= i + 1
return current_number
if __name__ == "__main__":
import doctest
doctest.testmod()
| 366
|
from __future__ import annotations
def __UpperCAmelCase ( __a : list ) -> float:
"""simple docstring"""
if not nums:
raise ValueError('''List is empty''' )
return sum(__a ) / len(__a )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 15
| 0
|
import unittest
import torch
from diffusers import DDIMScheduler, DDPMScheduler, UNetaDModel
from diffusers.training_utils import set_seed
from diffusers.utils.testing_utils import slow
a__ = False
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self , _a=3_2 ) -> str:
set_seed(0 )
_a : str = UNetaDModel(sample_size=_a , in_channels=3 , out_channels=3 )
_a : Tuple = torch.optim.SGD(model.parameters() , lr=0.0001 )
return model, optimizer
@slow
def __lowercase ( self ) -> Optional[int]:
_a : List[Any] = '''cpu''' # ensure full determinism without setting the CUBLAS_WORKSPACE_CONFIG env variable
_a : Any = DDPMScheduler(
num_train_timesteps=1_0_0_0 , beta_start=0.0001 , beta_end=0.02 , beta_schedule='''linear''' , clip_sample=_a , )
_a : Tuple = DDIMScheduler(
num_train_timesteps=1_0_0_0 , beta_start=0.0001 , beta_end=0.02 , beta_schedule='''linear''' , clip_sample=_a , )
assert ddpm_scheduler.config.num_train_timesteps == ddim_scheduler.config.num_train_timesteps
# shared batches for DDPM and DDIM
set_seed(0 )
_a : Dict = [torch.randn((4, 3, 3_2, 3_2) ).clip(-1 , 1 ).to(_a ) for _ in range(4 )]
_a : List[Any] = [torch.randn((4, 3, 3_2, 3_2) ).to(_a ) for _ in range(4 )]
_a : Tuple = [torch.randint(0 , 1_0_0_0 , (4,) ).long().to(_a ) for _ in range(4 )]
# train with a DDPM scheduler
_a : List[str] = self.get_model_optimizer(resolution=3_2 )
model.train().to(_a )
for i in range(4 ):
optimizer.zero_grad()
_a : Optional[int] = ddpm_scheduler.add_noise(clean_images[i] , noise[i] , timesteps[i] )
_a : Dict = model(_a , timesteps[i] ).sample
_a : Tuple = torch.nn.functional.mse_loss(_a , noise[i] )
loss.backward()
optimizer.step()
del model, optimizer
# recreate the model and optimizer, and retry with DDIM
_a : List[Any] = self.get_model_optimizer(resolution=3_2 )
model.train().to(_a )
for i in range(4 ):
optimizer.zero_grad()
_a : Dict = ddim_scheduler.add_noise(clean_images[i] , noise[i] , timesteps[i] )
_a : Tuple = model(_a , timesteps[i] ).sample
_a : List[str] = torch.nn.functional.mse_loss(_a , noise[i] )
loss.backward()
optimizer.step()
del model, optimizer
self.assertTrue(torch.allclose(_a , _a , atol=1e-5 ) )
self.assertTrue(torch.allclose(_a , _a , atol=1e-5 ) )
| 367
|
import argparse
import os
import torch
from transformers.utils import WEIGHTS_NAME
a__ = ['''small''', '''medium''', '''large''']
a__ = '''lm_head.decoder.weight'''
a__ = '''lm_head.weight'''
def __UpperCAmelCase ( __a : str ,__a : str ) -> List[str]:
"""simple docstring"""
_a : Any = torch.load(__a )
_a : List[str] = d.pop(__a )
os.makedirs(__a ,exist_ok=__a )
torch.save(__a ,os.path.join(__a ,__a ) )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--dialogpt_path''', default='''.''', type=str)
a__ = parser.parse_args()
for MODEL in DIALOGPT_MODELS:
a__ = os.path.join(args.dialogpt_path, f'''{MODEL}_ft.pkl''')
a__ = f'''./DialoGPT-{MODEL}'''
convert_dialogpt_checkpoint(
checkpoint_path,
pytorch_dump_folder_path,
)
| 15
| 0
|
import warnings
from transformers import AutoTokenizer
from transformers.utils import is_torch_available
from transformers.utils.generic import ExplicitEnum
from ...processing_utils import ProcessorMixin
if is_torch_available():
import torch
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : int = "char"
UpperCAmelCase__ : Optional[int] = "bpe"
UpperCAmelCase__ : List[str] = "wp"
a__ = (DecodeType.CHARACTER, DecodeType.BPE, DecodeType.WORDPIECE)
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = ["image_processor", "char_tokenizer"]
UpperCAmelCase__ : List[Any] = "ViTImageProcessor"
UpperCAmelCase__ : List[Any] = "MgpstrTokenizer"
def __init__( self , _a=None , _a=None , **_a ) -> Any:
_a : int = None
if "feature_extractor" in kwargs:
warnings.warn(
'''The `feature_extractor` argument is deprecated and will be removed in v5, use `image_processor`'''
''' instead.''' , _a , )
_a : List[Any] = kwargs.pop('''feature_extractor''' )
_a : Optional[int] = image_processor if image_processor is not None else feature_extractor
if image_processor is None:
raise ValueError('''You need to specify an `image_processor`.''' )
if tokenizer is None:
raise ValueError('''You need to specify a `tokenizer`.''' )
_a : List[str] = tokenizer
_a : str = AutoTokenizer.from_pretrained('''gpt2''' )
_a : Dict = AutoTokenizer.from_pretrained('''bert-base-uncased''' )
super().__init__(_a , _a )
def __call__( self , _a=None , _a=None , _a=None , **_a ) -> Any:
if images is None and text is None:
raise ValueError('''You need to specify either an `images` or `text` input to process.''' )
if images is not None:
_a : str = self.image_processor(_a , return_tensors=_a , **_a )
if text is not None:
_a : Union[str, Any] = self.char_tokenizer(_a , return_tensors=_a , **_a )
if text is None:
return inputs
elif images is None:
return encodings
else:
_a : Optional[int] = encodings['''input_ids''']
return inputs
def __lowercase ( self , _a ) -> List[str]:
_a : str = sequences
_a : Optional[int] = char_preds.size(0 )
_a : List[Any] = self._decode_helper(_a , '''char''' )
_a : Optional[Any] = self._decode_helper(_a , '''bpe''' )
_a : int = self._decode_helper(_a , '''wp''' )
_a : Tuple = []
_a : Union[str, Any] = []
for i in range(_a ):
_a : Dict = [char_scores[i], bpe_scores[i], wp_scores[i]]
_a : List[Any] = [char_strs[i], bpe_strs[i], wp_strs[i]]
_a : Optional[int] = scores.index(max(_a ) )
final_strs.append(strs[max_score_index] )
final_scores.append(scores[max_score_index] )
_a : Optional[int] = {}
_a : int = final_strs
_a : List[str] = final_scores
_a : Union[str, Any] = char_strs
_a : Optional[int] = bpe_strs
_a : List[str] = wp_strs
return out
def __lowercase ( self , _a , _a ) -> Tuple:
if format == DecodeType.CHARACTER:
_a : Tuple = self.char_decode
_a : List[Any] = 1
_a : str = '''[s]'''
elif format == DecodeType.BPE:
_a : Tuple = self.bpe_decode
_a : Union[str, Any] = 2
_a : List[str] = '''#'''
elif format == DecodeType.WORDPIECE:
_a : Optional[Any] = self.wp_decode
_a : Optional[int] = 1_0_2
_a : Optional[Any] = '''[SEP]'''
else:
raise ValueError(F"""Format {format} is not supported.""" )
_a : str = [], []
_a : List[Any] = pred_logits.size(0 )
_a : Dict = pred_logits.size(1 )
_a : List[str] = pred_logits.topk(1 , dim=-1 , largest=_a , sorted=_a )
_a : Optional[Any] = preds_index.view(-1 , _a )[:, 1:]
_a : Tuple = decoder(_a )
_a : Tuple = torch.nn.functional.softmax(_a , dim=2 ).max(dim=2 )
_a : Any = preds_max_prob[:, 1:]
for index in range(_a ):
_a : str = preds_str[index].find(_a )
_a : List[str] = preds_str[index][:pred_eos]
_a : str = preds_index[index].cpu().tolist()
_a : Tuple = pred_index.index(_a ) if eos_token in pred_index else -1
_a : Optional[Any] = preds_max_prob[index][: pred_eos_index + 1]
_a : Any = pred_max_prob.cumprod(dim=0 )[-1] if pred_max_prob.nelement() != 0 else 0.0
dec_strs.append(_a )
conf_scores.append(_a )
return dec_strs, conf_scores
def __lowercase ( self , _a ) -> Union[str, Any]:
_a : str = [seq.replace(''' ''' , '''''' ) for seq in self.char_tokenizer.batch_decode(_a )]
return decode_strs
def __lowercase ( self , _a ) -> Any:
return self.bpe_tokenizer.batch_decode(_a )
def __lowercase ( self , _a ) -> Union[str, Any]:
_a : Tuple = [seq.replace(''' ''' , '''''' ) for seq in self.wp_tokenizer.batch_decode(_a )]
return decode_strs
| 368
|
import enum
import warnings
from .. import MODEL_FOR_CAUSAL_LM_MAPPING, TF_MODEL_FOR_CAUSAL_LM_MAPPING
from ..utils import add_end_docstrings, is_tf_available
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
class UpperCAmelCase_ ( enum.Enum ):
"""simple docstring"""
UpperCAmelCase__ : int = 0
UpperCAmelCase__ : Union[str, Any] = 1
UpperCAmelCase__ : Optional[Any] = 2
@add_end_docstrings(__lowercase )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = "\n In 1991, the remains of Russian Tsar Nicholas II and his family (except for Alexei and Maria) are discovered. The\n voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the remainder of the story. 1883 Western\n Siberia, a young Grigori Rasputin is asked by his father and a group of men to perform magic. Rasputin has a vision\n and denounces one of the men as a horse thief. Although his father initially slaps him for making such an\n accusation, Rasputin watches as the man is chased outside and beaten. Twenty years later, Rasputin sees a vision of\n the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous, with people, even a bishop,\n begging for his blessing. <eod> </s> <eos>\n "
def __init__( self , *_a , **_a ) -> List[str]:
super().__init__(*_a , **_a )
self.check_model_type(
TF_MODEL_FOR_CAUSAL_LM_MAPPING if self.framework == '''tf''' else MODEL_FOR_CAUSAL_LM_MAPPING )
if "prefix" not in self._preprocess_params:
# This is very specific. The logic is quite complex and needs to be done
# as a "default".
# It also defines both some preprocess_kwargs and generate_kwargs
# which is why we cannot put them in their respective methods.
_a : Dict = None
if self.model.config.prefix is not None:
_a : List[Any] = self.model.config.prefix
if prefix is None and self.model.__class__.__name__ in [
"XLNetLMHeadModel",
"TransfoXLLMHeadModel",
"TFXLNetLMHeadModel",
"TFTransfoXLLMHeadModel",
]:
# For XLNet and TransformerXL we add an article to the prompt to give more state to the model.
_a : Optional[Any] = self.XL_PREFIX
if prefix is not None:
# Recalculate some generate_kwargs linked to prefix.
_a , _a , _a : str = self._sanitize_parameters(prefix=_a , **self._forward_params )
_a : Optional[Any] = {**self._preprocess_params, **preprocess_params}
_a : List[Any] = {**self._forward_params, **forward_params}
def __lowercase ( self , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , **_a , ) -> Optional[int]:
_a : List[Any] = {}
if prefix is not None:
_a : Optional[Any] = prefix
if prefix:
_a : Dict = self.tokenizer(
_a , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Tuple = prefix_inputs['''input_ids'''].shape[-1]
if handle_long_generation is not None:
if handle_long_generation not in {"hole"}:
raise ValueError(
F"""{handle_long_generation} is not a valid value for `handle_long_generation` parameter expected"""
''' [None, \'hole\']''' )
_a : Dict = handle_long_generation
preprocess_params.update(_a )
_a : Tuple = generate_kwargs
_a : Any = {}
if return_full_text is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_full_text`''' )
if return_tensors is not None:
raise ValueError('''`return_full_text` is mutually exclusive with `return_tensors`''' )
_a : List[str] = ReturnType.FULL_TEXT if return_full_text else ReturnType.NEW_TEXT
if return_tensors is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_tensors`''' )
_a : Any = ReturnType.TENSORS
if return_type is not None:
_a : Any = return_type
if clean_up_tokenization_spaces is not None:
_a : List[Any] = clean_up_tokenization_spaces
if stop_sequence is not None:
_a : Tuple = self.tokenizer.encode(_a , add_special_tokens=_a )
if len(_a ) > 1:
warnings.warn(
'''Stopping on a multiple token sequence is not yet supported on transformers. The first token of'''
''' the stop sequence will be used as the stop sequence string in the interim.''' )
_a : List[Any] = stop_sequence_ids[0]
return preprocess_params, forward_params, postprocess_params
def __lowercase ( self , *_a , **_a ) -> Union[str, Any]:
# Parse arguments
if self.model.__class__.__name__ in ["TransfoXLLMHeadModel"]:
kwargs.update({'''add_space_before_punct_symbol''': True} )
return super()._parse_and_tokenize(*_a , **_a )
def __call__( self , _a , **_a ) -> List[str]:
return super().__call__(_a , **_a )
def __lowercase ( self , _a , _a="" , _a=None , **_a ) -> List[Any]:
_a : Optional[int] = self.tokenizer(
prefix + prompt_text , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Union[str, Any] = prompt_text
if handle_long_generation == "hole":
_a : List[str] = inputs['''input_ids'''].shape[-1]
if "max_new_tokens" in generate_kwargs:
_a : int = generate_kwargs['''max_new_tokens''']
else:
_a : List[Any] = generate_kwargs.get('''max_length''' , self.model.config.max_length ) - cur_len
if new_tokens < 0:
raise ValueError('''We cannot infer how many new tokens are expected''' )
if cur_len + new_tokens > self.tokenizer.model_max_length:
_a : List[str] = self.tokenizer.model_max_length - new_tokens
if keep_length <= 0:
raise ValueError(
'''We cannot use `hole` to handle this generation the number of desired tokens exceeds the'''
''' models max length''' )
_a : List[Any] = inputs['''input_ids'''][:, -keep_length:]
if "attention_mask" in inputs:
_a : List[str] = inputs['''attention_mask'''][:, -keep_length:]
return inputs
def __lowercase ( self , _a , **_a ) -> Optional[int]:
_a : Any = model_inputs['''input_ids''']
_a : Optional[Any] = model_inputs.get('''attention_mask''' , _a )
# Allow empty prompts
if input_ids.shape[1] == 0:
_a : int = None
_a : int = None
_a : List[str] = 1
else:
_a : List[Any] = input_ids.shape[0]
_a : Union[str, Any] = model_inputs.pop('''prompt_text''' )
# If there is a prefix, we may need to adjust the generation length. Do so without permanently modifying
# generate_kwargs, as some of the parameterization may come from the initialization of the pipeline.
_a : int = generate_kwargs.pop('''prefix_length''' , 0 )
if prefix_length > 0:
_a : Tuple = '''max_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].max_new_tokens is not None
)
if not has_max_new_tokens:
_a : int = generate_kwargs.get('''max_length''' ) or self.model.config.max_length
generate_kwargs["max_length"] += prefix_length
_a : Dict = '''min_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].min_new_tokens is not None
)
if not has_min_new_tokens and "min_length" in generate_kwargs:
generate_kwargs["min_length"] += prefix_length
# BS x SL
_a : Optional[Any] = self.model.generate(input_ids=_a , attention_mask=_a , **_a )
_a : int = generated_sequence.shape[0]
if self.framework == "pt":
_a : Tuple = generated_sequence.reshape(_a , out_b // in_b , *generated_sequence.shape[1:] )
elif self.framework == "tf":
_a : List[Any] = tf.reshape(_a , (in_b, out_b // in_b, *generated_sequence.shape[1:]) )
return {"generated_sequence": generated_sequence, "input_ids": input_ids, "prompt_text": prompt_text}
def __lowercase ( self , _a , _a=ReturnType.FULL_TEXT , _a=True ) -> int:
_a : Tuple = model_outputs['''generated_sequence'''][0]
_a : int = model_outputs['''input_ids''']
_a : Any = model_outputs['''prompt_text''']
_a : Any = generated_sequence.numpy().tolist()
_a : Any = []
for sequence in generated_sequence:
if return_type == ReturnType.TENSORS:
_a : Optional[int] = {'''generated_token_ids''': sequence}
elif return_type in {ReturnType.NEW_TEXT, ReturnType.FULL_TEXT}:
# Decode text
_a : str = self.tokenizer.decode(
_a , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , )
# Remove PADDING prompt of the sequence if XLNet or Transfo-XL model is used
if input_ids is None:
_a : Union[str, Any] = 0
else:
_a : str = len(
self.tokenizer.decode(
input_ids[0] , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , ) )
if return_type == ReturnType.FULL_TEXT:
_a : str = prompt_text + text[prompt_length:]
else:
_a : List[str] = text[prompt_length:]
_a : Union[str, Any] = {'''generated_text''': all_text}
records.append(_a )
return records
| 15
| 0
|
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
a__ = logging.get_logger(__name__)
a__ = '''▁'''
a__ = {'''vocab_file''': '''sentencepiece.bpe.model'''}
a__ = {
'''vocab_file''': {
'''facebook/mbart-large-50-one-to-many-mmt''': (
'''https://huggingface.co/facebook/mbart-large-50-one-to-many-mmt/resolve/main/sentencepiece.bpe.model'''
),
}
}
a__ = {
'''facebook/mbart-large-50-one-to-many-mmt''': 1024,
}
# fmt: off
a__ = ['''ar_AR''', '''cs_CZ''', '''de_DE''', '''en_XX''', '''es_XX''', '''et_EE''', '''fi_FI''', '''fr_XX''', '''gu_IN''', '''hi_IN''', '''it_IT''', '''ja_XX''', '''kk_KZ''', '''ko_KR''', '''lt_LT''', '''lv_LV''', '''my_MM''', '''ne_NP''', '''nl_XX''', '''ro_RO''', '''ru_RU''', '''si_LK''', '''tr_TR''', '''vi_VN''', '''zh_CN''', '''af_ZA''', '''az_AZ''', '''bn_IN''', '''fa_IR''', '''he_IL''', '''hr_HR''', '''id_ID''', '''ka_GE''', '''km_KH''', '''mk_MK''', '''ml_IN''', '''mn_MN''', '''mr_IN''', '''pl_PL''', '''ps_AF''', '''pt_XX''', '''sv_SE''', '''sw_KE''', '''ta_IN''', '''te_IN''', '''th_TH''', '''tl_XX''', '''uk_UA''', '''ur_PK''', '''xh_ZA''', '''gl_ES''', '''sl_SI''']
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Union[str, Any] = VOCAB_FILES_NAMES
UpperCAmelCase__ : Union[str, Any] = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
UpperCAmelCase__ : Tuple = PRETRAINED_VOCAB_FILES_MAP
UpperCAmelCase__ : Any = ["input_ids", "attention_mask"]
UpperCAmelCase__ : List[int] = []
UpperCAmelCase__ : List[int] = []
def __init__( self , _a , _a=None , _a=None , _a="</s>" , _a="</s>" , _a="<s>" , _a="<unk>" , _a="<pad>" , _a="<mask>" , _a = None , **_a , ) -> None:
# Mask token behave like a normal word, i.e. include the space before it
_a : int = AddedToken(_a , lstrip=_a , rstrip=_a ) if isinstance(_a , _a ) else mask_token
_a : int = {} if sp_model_kwargs is None else sp_model_kwargs
_a : List[Any] = kwargs.get('''additional_special_tokens''' , [] )
kwargs["additional_special_tokens"] += [
code for code in FAIRSEQ_LANGUAGE_CODES if code not in kwargs["additional_special_tokens"]
]
super().__init__(
src_lang=_a , tgt_lang=_a , eos_token=_a , unk_token=_a , sep_token=_a , cls_token=_a , pad_token=_a , mask_token=_a , sp_model_kwargs=self.sp_model_kwargs , **_a , )
_a : Tuple = spm.SentencePieceProcessor(**self.sp_model_kwargs )
self.sp_model.Load(str(_a ) )
_a : int = 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>' | ',' | '.' | '▁' | 's' | '▁de' | '-'
# spm | '<unk>' | '<s>' | '</s>' | ',' | '.' | '▁' | 's' | '▁de' | '-' | '▁a'
# Mimic fairseq token-to-id alignment for the first 4 token
_a : Optional[Any] = {'''<s>''': 0, '''<pad>''': 1, '''</s>''': 2, '''<unk>''': 3}
# The first "real" token "," has position 4 in the original fairseq vocab and position 3 in the spm vocab
_a : Any = 1
_a : int = len(self.sp_model )
_a : Optional[Any] = {
code: self.sp_model_size + i + self.fairseq_offset for i, code in enumerate(_a )
}
_a : str = {v: k for k, v in self.lang_code_to_id.items()}
_a : str = len(self.sp_model ) + len(self.lang_code_to_id ) + self.fairseq_offset
self.fairseq_tokens_to_ids.update(self.lang_code_to_id )
_a : Optional[int] = {v: k for k, v in self.fairseq_tokens_to_ids.items()}
_a : Dict = src_lang if src_lang is not None else '''en_XX'''
_a : Optional[Any] = self.lang_code_to_id[self._src_lang]
_a : List[Any] = tgt_lang
self.set_src_lang_special_tokens(self._src_lang )
@property
def __lowercase ( self ) -> int:
return len(self.sp_model ) + len(self.lang_code_to_id ) + self.fairseq_offset + 1 # Plus 1 for the mask token
@property
def __lowercase ( self ) -> str:
return self._src_lang
@src_lang.setter
def __lowercase ( self , _a ) -> None:
_a : int = new_src_lang
self.set_src_lang_special_tokens(self._src_lang )
def __getstate__( self ) -> Dict:
_a : Tuple = self.__dict__.copy()
_a : Optional[Any] = None
return state
def __setstate__( self , _a ) -> None:
_a : List[Any] = d
# for backward compatibility
if not hasattr(self , '''sp_model_kwargs''' ):
_a : Dict = {}
_a : Optional[Any] = spm.SentencePieceProcessor(**self.sp_model_kwargs )
self.sp_model.Load(self.vocab_file )
def __lowercase ( self ) -> Dict:
_a : Dict = {self.convert_ids_to_tokens(_a ): i for i in range(self.vocab_size )}
vocab.update(self.added_tokens_encoder )
return vocab
def __lowercase ( self , _a ) -> List[str]:
return self.sp_model.encode(_a , out_type=_a )
def __lowercase ( self , _a ) -> int:
if token in self.fairseq_tokens_to_ids:
return self.fairseq_tokens_to_ids[token]
_a : List[Any] = self.sp_model.PieceToId(_a )
# 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 __lowercase ( self , _a ) -> str:
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 __lowercase ( self , _a ) -> Dict:
_a : Any = []
_a : Optional[int] = ''''''
_a : Dict = False
for token in tokens:
# make sure that special tokens are not decoded using sentencepiece model
if token in self.all_special_tokens:
if not prev_is_special:
out_string += " "
out_string += self.sp_model.decode(_a ) + token
_a : int = True
_a : List[Any] = []
else:
current_sub_tokens.append(_a )
_a : Tuple = False
out_string += self.sp_model.decode(_a )
return out_string.strip()
def __lowercase ( self , _a , _a = None ) -> Tuple[str]:
if not os.path.isdir(_a ):
logger.error(F"""Vocabulary path ({save_directory}) should be a directory""" )
return
_a : Dict = os.path.join(
_a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] )
if os.path.abspath(self.vocab_file ) != os.path.abspath(_a ) and os.path.isfile(self.vocab_file ):
copyfile(self.vocab_file , _a )
elif not os.path.isfile(self.vocab_file ):
with open(_a , '''wb''' ) as fi:
_a : Optional[Any] = self.sp_model.serialized_model_proto()
fi.write(_a )
return (out_vocab_file,)
def __lowercase ( self , _a , _a = None , _a = False ) -> List[int]:
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=_a , token_ids_a=_a , already_has_special_tokens=_a )
_a : List[Any] = [1] * len(self.prefix_tokens )
_a : Dict = [1] * len(self.suffix_tokens )
if token_ids_a is None:
return prefix_ones + ([0] * len(_a )) + suffix_ones
return prefix_ones + ([0] * len(_a )) + ([0] * len(_a )) + suffix_ones
def __lowercase ( self , _a , _a = None ) -> List[int]:
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 __lowercase ( self , _a , _a , _a , _a , **_a ) -> Optional[Any]:
if src_lang is None or tgt_lang is None:
raise ValueError('''Translation requires a `src_lang` and a `tgt_lang` for this model''' )
_a : Optional[Any] = src_lang
_a : Dict = self(_a , add_special_tokens=_a , return_tensors=_a , **_a )
_a : Optional[Any] = self.convert_tokens_to_ids(_a )
_a : Union[str, Any] = tgt_lang_id
return inputs
def __lowercase ( self , _a , _a = "en_XX" , _a = None , _a = "ro_RO" , **_a , ) -> BatchEncoding:
_a : Tuple = src_lang
_a : Dict = tgt_lang
return super().prepare_seqaseq_batch(_a , _a , **_a )
def __lowercase ( self ) -> int:
return self.set_src_lang_special_tokens(self.src_lang )
def __lowercase ( self ) -> Optional[int]:
return self.set_tgt_lang_special_tokens(self.tgt_lang )
def __lowercase ( self , _a ) -> None:
_a : int = self.lang_code_to_id[src_lang]
_a : List[str] = [self.cur_lang_code_id]
_a : List[str] = [self.eos_token_id]
def __lowercase ( self , _a ) -> None:
_a : List[str] = self.lang_code_to_id[tgt_lang]
_a : str = [self.cur_lang_code_id]
_a : Any = [self.eos_token_id]
| 369
|
# 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
import os
from accelerate.test_utils import execute_subprocess_async
def __UpperCAmelCase ( __a : Dict=None ) -> str:
"""simple docstring"""
if subparsers is not None:
_a : Union[str, Any] = subparsers.add_parser('''test''' )
else:
_a : List[str] = argparse.ArgumentParser('''Accelerate test command''' )
parser.add_argument(
'''--config_file''' ,default=__a ,help=(
'''The path to use to store the config file. Will default to a file named default_config.yaml in the cache '''
'''location, which is the content of the environment `HF_HOME` suffixed with \'accelerate\', or if you don\'t have '''
'''such an environment variable, your cache directory (\'~/.cache\' or the content of `XDG_CACHE_HOME`) suffixed '''
'''with \'huggingface\'.'''
) ,)
if subparsers is not None:
parser.set_defaults(func=__a )
return parser
def __UpperCAmelCase ( __a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
_a : Dict = os.path.sep.join(__file__.split(os.path.sep )[:-2] + ['''test_utils''', '''scripts''', '''test_script.py'''] )
if args.config_file is None:
_a : List[Any] = script_name
else:
_a : Union[str, Any] = F"""--config_file={args.config_file} {script_name}"""
_a : str = ['''accelerate-launch'''] + test_args.split()
_a : str = execute_subprocess_async(__a ,env=os.environ.copy() )
if result.returncode == 0:
print('''Test is a success! You are ready for your distributed training!''' )
def __UpperCAmelCase ( ) -> List[Any]:
"""simple docstring"""
_a : Optional[int] = test_command_parser()
_a : List[Any] = parser.parse_args()
test_command(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
from dataclasses import dataclass, field
from typing import ClassVar, Dict
from ..features import Features, Sequence, Value
from .base import TaskTemplate
@dataclass(frozen=__lowercase )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : str = field(default="question-answering-extractive" , metadata={"include_in_asdict_even_if_is_default": True} )
UpperCAmelCase__ : ClassVar[Features] = Features({"question": Value("string" ), "context": Value("string" )} )
UpperCAmelCase__ : ClassVar[Features] = Features(
{
"answers": Sequence(
{
"text": Value("string" ),
"answer_start": Value("int32" ),
} )
} )
UpperCAmelCase__ : str = "question"
UpperCAmelCase__ : str = "context"
UpperCAmelCase__ : str = "answers"
@property
def __lowercase ( self ) -> Dict[str, str]:
return {self.question_column: "question", self.context_column: "context", self.answers_column: "answers"}
| 370
|
import json
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import BertTokenizerFast
from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer
from transformers.testing_utils import require_tokenizers, require_vision
from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available
if is_vision_available():
from PIL import Image
from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor
@require_tokenizers
@require_vision
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = tempfile.mkdtemp()
# fmt: off
_a : Optional[int] = ['''[UNK]''', '''[CLS]''', '''[SEP]''', '''[PAD]''', '''[MASK]''', '''want''', '''##want''', '''##ed''', '''wa''', '''un''', '''runn''', '''##ing''', ''',''', '''low''', '''lowest''']
# fmt: on
_a : Tuple = 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] ) )
_a : Any = {
'''do_resize''': True,
'''size''': {'''height''': 1_8, '''width''': 1_8},
'''do_normalize''': True,
'''image_mean''': [0.5, 0.5, 0.5],
'''image_std''': [0.5, 0.5, 0.5],
}
_a : str = os.path.join(self.tmpdirname , _a )
with open(self.image_processor_file , '''w''' , encoding='''utf-8''' ) as fp:
json.dump(_a , _a )
def __lowercase ( self , **_a ) -> Any:
return BertTokenizer.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self , **_a ) -> str:
return ViTImageProcessor.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self ) -> List[Any]:
shutil.rmtree(self.tmpdirname )
def __lowercase ( self ) -> Any:
_a : Union[str, Any] = [np.random.randint(2_5_5 , size=(3, 3_0, 4_0_0) , dtype=np.uinta )]
_a : Tuple = [Image.fromarray(np.moveaxis(_a , 0 , -1 ) ) for x in image_inputs]
return image_inputs
def __lowercase ( self ) -> str:
_a : List[str] = self.get_tokenizer()
_a : Tuple = self.get_image_processor()
_a : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
processor.save_pretrained(self.tmpdirname )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Dict:
_a : List[str] = VisionTextDualEncoderProcessor(
tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() )
processor.save_pretrained(self.tmpdirname )
_a : Any = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' )
_a : List[Any] = self.get_image_processor(do_normalize=_a , padding_value=1.0 )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(
self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=_a , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Any:
_a : Dict = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : int = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[str] = self.prepare_image_inputs()
_a : List[Any] = image_processor(_a , return_tensors='''np''' )
_a : Dict = processor(images=_a , return_tensors='''np''' )
for key in input_feat_extract.keys():
self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1e-2 )
def __lowercase ( self ) -> List[str]:
_a : Union[str, Any] = self.get_image_processor()
_a : Dict = self.get_tokenizer()
_a : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Tuple = '''lower newer'''
_a : int = processor(text=_a )
_a : str = tokenizer(_a )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def __lowercase ( self ) -> List[Any]:
_a : Any = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : Tuple = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[Any] = '''lower newer'''
_a : Union[str, Any] = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , ['''input_ids''', '''token_type_ids''', '''attention_mask''', '''pixel_values'''] )
# test if it raises when no input is passed
with self.assertRaises(_a ):
processor()
def __lowercase ( self ) -> Optional[int]:
_a : Union[str, Any] = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : Any = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Any = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
_a : int = processor.batch_decode(_a )
_a : int = tokenizer.batch_decode(_a )
self.assertListEqual(_a , _a )
def __lowercase ( self ) -> List[Any]:
_a : Tuple = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : str = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Optional[int] = '''lower newer'''
_a : Dict = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
| 15
| 0
|
from __future__ import annotations
import time
from math import sqrt
# 1 for manhattan, 0 for euclidean
a__ = 0
a__ = [
[0, 0, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0], # 0 are free path whereas 1's are obstacles
[0, 0, 0, 0, 0, 0, 0],
[0, 0, 1, 0, 0, 0, 0],
[1, 0, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0],
]
a__ = [[-1, 0], [0, -1], [1, 0], [0, 1]] # up, left, down, right
a__ = tuple[int, int]
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self , _a , _a , _a , _a , _a , _a , ) -> None:
_a : Any = pos_x
_a : List[str] = pos_y
_a : Tuple = (pos_y, pos_x)
_a : List[str] = goal_x
_a : Union[str, Any] = goal_y
_a : Optional[Any] = g_cost
_a : Union[str, Any] = parent
_a : List[Any] = self.calculate_heuristic()
_a : int = self.g_cost + self.h_cost
def __lowercase ( self ) -> float:
_a : Optional[int] = self.pos_x - self.goal_x
_a : str = self.pos_y - self.goal_y
if HEURISTIC == 1:
return abs(_a ) + abs(_a )
else:
return sqrt(dy**2 + dx**2 )
def __lt__( self , _a ) -> bool:
return self.f_cost < other.f_cost
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self , _a , _a ) -> Dict:
_a : Any = Node(start[1] , start[0] , goal[1] , goal[0] , 0 , _a )
_a : Optional[int] = Node(goal[1] , goal[0] , goal[1] , goal[0] , 9_9_9_9_9 , _a )
_a : str = [self.start]
_a : list[Node] = []
_a : int = False
def __lowercase ( self ) -> list[TPosition]:
while self.open_nodes:
# Open Nodes are sorted using __lt__
self.open_nodes.sort()
_a : Any = self.open_nodes.pop(0 )
if current_node.pos == self.target.pos:
return self.retrace_path(_a )
self.closed_nodes.append(_a )
_a : Dict = self.get_successors(_a )
for child_node in successors:
if child_node in self.closed_nodes:
continue
if child_node not in self.open_nodes:
self.open_nodes.append(_a )
else:
# retrieve the best current path
_a : List[Any] = self.open_nodes.pop(self.open_nodes.index(_a ) )
if child_node.g_cost < better_node.g_cost:
self.open_nodes.append(_a )
else:
self.open_nodes.append(_a )
return [self.start.pos]
def __lowercase ( self , _a ) -> list[Node]:
_a : Tuple = []
for action in delta:
_a : Optional[int] = parent.pos_x + action[1]
_a : List[str] = parent.pos_y + action[0]
if not (0 <= pos_x <= len(grid[0] ) - 1 and 0 <= pos_y <= len(_a ) - 1):
continue
if grid[pos_y][pos_x] != 0:
continue
successors.append(
Node(
_a , _a , self.target.pos_y , self.target.pos_x , parent.g_cost + 1 , _a , ) )
return successors
def __lowercase ( self , _a ) -> list[TPosition]:
_a : str = node
_a : int = []
while current_node is not None:
path.append((current_node.pos_y, current_node.pos_x) )
_a : Any = current_node.parent
path.reverse()
return path
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self , _a , _a ) -> None:
_a : Union[str, Any] = AStar(_a , _a )
_a : List[Any] = AStar(_a , _a )
_a : Any = False
def __lowercase ( self ) -> list[TPosition]:
while self.fwd_astar.open_nodes or self.bwd_astar.open_nodes:
self.fwd_astar.open_nodes.sort()
self.bwd_astar.open_nodes.sort()
_a : Dict = self.fwd_astar.open_nodes.pop(0 )
_a : Any = self.bwd_astar.open_nodes.pop(0 )
if current_bwd_node.pos == current_fwd_node.pos:
return self.retrace_bidirectional_path(
_a , _a )
self.fwd_astar.closed_nodes.append(_a )
self.bwd_astar.closed_nodes.append(_a )
_a : Dict = current_bwd_node
_a : Any = current_fwd_node
_a : Tuple = {
self.fwd_astar: self.fwd_astar.get_successors(_a ),
self.bwd_astar: self.bwd_astar.get_successors(_a ),
}
for astar in [self.fwd_astar, self.bwd_astar]:
for child_node in successors[astar]:
if child_node in astar.closed_nodes:
continue
if child_node not in astar.open_nodes:
astar.open_nodes.append(_a )
else:
# retrieve the best current path
_a : Tuple = astar.open_nodes.pop(
astar.open_nodes.index(_a ) )
if child_node.g_cost < better_node.g_cost:
astar.open_nodes.append(_a )
else:
astar.open_nodes.append(_a )
return [self.fwd_astar.start.pos]
def __lowercase ( self , _a , _a ) -> list[TPosition]:
_a : List[str] = self.fwd_astar.retrace_path(_a )
_a : int = self.bwd_astar.retrace_path(_a )
bwd_path.pop()
bwd_path.reverse()
_a : Dict = fwd_path + bwd_path
return path
if __name__ == "__main__":
# all coordinates are given in format [y,x]
a__ = (0, 0)
a__ = (len(grid) - 1, len(grid[0]) - 1)
for elem in grid:
print(elem)
a__ = time.time()
a__ = AStar(init, goal)
a__ = a_star.search()
a__ = time.time() - start_time
print(f'''AStar execution time = {end_time:f} seconds''')
a__ = time.time()
a__ = BidirectionalAStar(init, goal)
a__ = time.time() - bd_start_time
print(f'''BidirectionalAStar execution time = {bd_end_time:f} seconds''')
| 371
|
import argparse
import json
import os
import fairseq
import torch
from fairseq.data import Dictionary
from transformers import (
HubertConfig,
HubertForCTC,
HubertModel,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaProcessor,
logging,
)
logging.set_verbosity_info()
a__ = logging.get_logger(__name__)
a__ = {
'''post_extract_proj''': '''feature_projection.projection''',
'''encoder.pos_conv.0''': '''encoder.pos_conv_embed.conv''',
'''self_attn.k_proj''': '''encoder.layers.*.attention.k_proj''',
'''self_attn.v_proj''': '''encoder.layers.*.attention.v_proj''',
'''self_attn.q_proj''': '''encoder.layers.*.attention.q_proj''',
'''self_attn.out_proj''': '''encoder.layers.*.attention.out_proj''',
'''self_attn_layer_norm''': '''encoder.layers.*.layer_norm''',
'''fc1''': '''encoder.layers.*.feed_forward.intermediate_dense''',
'''fc2''': '''encoder.layers.*.feed_forward.output_dense''',
'''final_layer_norm''': '''encoder.layers.*.final_layer_norm''',
'''encoder.layer_norm''': '''encoder.layer_norm''',
'''w2v_model.layer_norm''': '''feature_projection.layer_norm''',
'''w2v_encoder.proj''': '''lm_head''',
'''mask_emb''': '''masked_spec_embed''',
}
def __UpperCAmelCase ( __a : List[Any] ,__a : Optional[int] ,__a : Optional[int] ,__a : List[str] ,__a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
for attribute in key.split('''.''' ):
_a : Optional[Any] = getattr(__a ,__a )
if weight_type is not None:
_a : Dict = getattr(__a ,__a ).shape
else:
_a : Optional[int] = 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":
_a : List[Any] = value
elif weight_type == "weight_g":
_a : Any = value
elif weight_type == "weight_v":
_a : Union[str, Any] = value
elif weight_type == "bias":
_a : Optional[int] = value
else:
_a : List[Any] = value
logger.info(F"""{key + '.' + weight_type if weight_type is not None else ''} was initialized from {full_name}.""" )
def __UpperCAmelCase ( __a : Any ,__a : Union[str, Any] ,__a : Union[str, Any] ) -> int:
"""simple docstring"""
_a : Union[str, Any] = []
_a : Union[str, Any] = fairseq_model.state_dict()
_a : Union[str, Any] = hf_model.hubert.feature_extractor if is_finetuned else hf_model.feature_extractor
for name, value in fairseq_dict.items():
_a : int = False
if "conv_layers" in name:
load_conv_layer(
__a ,__a ,__a ,__a ,hf_model.config.feat_extract_norm == '''group''' ,)
_a : Optional[Any] = True
else:
for key, mapped_key in MAPPING.items():
_a : Union[str, Any] = '''hubert.''' + 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] and not is_finetuned):
_a : Any = True
if "*" in mapped_key:
_a : Optional[int] = name.split(__a )[0].split('''.''' )[-2]
_a : Any = mapped_key.replace('''*''' ,__a )
if "weight_g" in name:
_a : List[Any] = '''weight_g'''
elif "weight_v" in name:
_a : List[str] = '''weight_v'''
elif "weight" in name:
_a : Any = '''weight'''
elif "bias" in name:
_a : str = '''bias'''
else:
_a : Any = None
set_recursively(__a ,__a ,__a ,__a ,__a )
continue
if not is_used:
unused_weights.append(__a )
logger.warning(F"""Unused weights: {unused_weights}""" )
def __UpperCAmelCase ( __a : int ,__a : Optional[Any] ,__a : Dict ,__a : List[str] ,__a : Any ) -> Tuple:
"""simple docstring"""
_a : int = full_name.split('''conv_layers.''' )[-1]
_a : Any = name.split('''.''' )
_a : List[Any] = int(items[0] )
_a : Optional[int] = 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."""
)
_a : Optional[int] = 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."""
)
_a : Optional[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:
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."
)
_a : int = 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."""
)
_a : Any = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(__a )
@torch.no_grad()
def __UpperCAmelCase ( __a : Dict ,__a : List[Any] ,__a : List[str]=None ,__a : Optional[int]=None ,__a : int=True ) -> List[Any]:
"""simple docstring"""
if config_path is not None:
_a : Tuple = HubertConfig.from_pretrained(__a )
else:
_a : Any = HubertConfig()
if is_finetuned:
if dict_path:
_a : Tuple = Dictionary.load(__a )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
_a : Any = target_dict.pad_index
_a : Tuple = target_dict.bos_index
_a : Optional[int] = target_dict.eos_index
_a : Optional[Any] = len(target_dict.symbols )
_a : Tuple = os.path.join(__a ,'''vocab.json''' )
if not os.path.isdir(__a ):
logger.error('''--pytorch_dump_folder_path ({}) should be a directory'''.format(__a ) )
return
os.makedirs(__a ,exist_ok=__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ) as vocab_handle:
json.dump(target_dict.indices ,__a )
_a : Tuple = WavaVecaCTCTokenizer(
__a ,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=__a ,)
_a : Tuple = True if config.feat_extract_norm == '''layer''' else False
_a : List[Any] = WavaVecaFeatureExtractor(
feature_size=1 ,sampling_rate=16_000 ,padding_value=0 ,do_normalize=__a ,return_attention_mask=__a ,)
_a : List[Any] = WavaVecaProcessor(feature_extractor=__a ,tokenizer=__a )
processor.save_pretrained(__a )
_a : Tuple = HubertForCTC(__a )
else:
_a : Tuple = HubertModel(__a )
if is_finetuned:
_a , _a , _a : int = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
_a , _a , _a : str = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] )
_a : Any = model[0].eval()
recursively_load_weights(__a ,__a ,__a )
hf_wavavec.save_pretrained(__a )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
parser.add_argument('''--checkpoint_path''', default=None, type=str, help='''Path to fairseq checkpoint''')
parser.add_argument('''--dict_path''', default=None, type=str, help='''Path to dict of fine-tuned model''')
parser.add_argument('''--config_path''', default=None, type=str, help='''Path to hf config.json of model to convert''')
parser.add_argument(
'''--not_finetuned''', action='''store_true''', help='''Whether the model to convert is a fine-tuned model or not'''
)
a__ = parser.parse_args()
convert_hubert_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 15
| 0
|
import math
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self , _a=0 ) -> int: # a graph with Node 0,1,...,N-1
_a : List[str] = n
_a : Any = [
[math.inf for j in range(0 , _a )] for i in range(0 , _a )
] # adjacency matrix for weight
_a : Tuple = [
[math.inf for j in range(0 , _a )] for i in range(0 , _a )
] # dp[i][j] stores minimum distance from i to j
def __lowercase ( self , _a , _a , _a ) -> Optional[Any]:
_a : int = w
def __lowercase ( self ) -> str:
for k in range(0 , self.n ):
for i in range(0 , self.n ):
for j in range(0 , self.n ):
_a : Optional[int] = min(self.dp[i][j] , self.dp[i][k] + self.dp[k][j] )
def __lowercase ( self , _a , _a ) -> Union[str, Any]:
return self.dp[u][v]
if __name__ == "__main__":
a__ = Graph(5)
graph.add_edge(0, 2, 9)
graph.add_edge(0, 4, 10)
graph.add_edge(1, 3, 5)
graph.add_edge(2, 3, 7)
graph.add_edge(3, 0, 10)
graph.add_edge(3, 1, 2)
graph.add_edge(3, 2, 1)
graph.add_edge(3, 4, 6)
graph.add_edge(4, 1, 3)
graph.add_edge(4, 2, 4)
graph.add_edge(4, 3, 9)
graph.floyd_warshall()
graph.show_min(1, 4)
graph.show_min(0, 3)
| 350
|
import warnings
from typing import List, Optional, Union
from ...processing_utils import ProcessorMixin
from ...tokenization_utils_base import BatchEncoding, PaddingStrategy, PreTokenizedInput, TextInput, TruncationStrategy
from ...utils import TensorType
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[str] = ["image_processor", "tokenizer"]
UpperCAmelCase__ : str = "ViltImageProcessor"
UpperCAmelCase__ : Union[str, Any] = ("BertTokenizer", "BertTokenizerFast")
def __init__( self , _a=None , _a=None , **_a ) -> Any:
_a : Union[str, Any] = None
if "feature_extractor" in kwargs:
warnings.warn(
'''The `feature_extractor` argument is deprecated and will be removed in v5, use `image_processor`'''
''' instead.''' , _a , )
_a : Dict = kwargs.pop('''feature_extractor''' )
_a : Optional[int] = image_processor if image_processor is not None else feature_extractor
if image_processor is None:
raise ValueError('''You need to specify an `image_processor`.''' )
if tokenizer is None:
raise ValueError('''You need to specify a `tokenizer`.''' )
super().__init__(_a , _a )
_a : int = self.image_processor
def __call__( self , _a , _a = None , _a = True , _a = False , _a = None , _a = None , _a = 0 , _a = None , _a = None , _a = None , _a = False , _a = False , _a = False , _a = False , _a = True , _a = None , **_a , ) -> BatchEncoding:
_a : Tuple = self.tokenizer(
text=_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , stride=_a , pad_to_multiple_of=_a , return_token_type_ids=_a , return_attention_mask=_a , return_overflowing_tokens=_a , return_special_tokens_mask=_a , return_offsets_mapping=_a , return_length=_a , verbose=_a , return_tensors=_a , **_a , )
# add pixel_values + pixel_mask
_a : str = self.image_processor(_a , return_tensors=_a )
encoding.update(_a )
return encoding
def __lowercase ( self , *_a , **_a ) -> Optional[Any]:
return self.tokenizer.batch_decode(*_a , **_a )
def __lowercase ( self , *_a , **_a ) -> str:
return self.tokenizer.decode(*_a , **_a )
@property
def __lowercase ( self ) -> Optional[int]:
_a : str = self.tokenizer.model_input_names
_a : Optional[Any] = self.image_processor.model_input_names
return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names ) )
@property
def __lowercase ( self ) -> Optional[Any]:
warnings.warn(
'''`feature_extractor_class` is deprecated and will be removed in v5. Use `image_processor_class` instead.''' , _a , )
return self.image_processor_class
@property
def __lowercase ( self ) -> Any:
warnings.warn(
'''`feature_extractor` is deprecated and will be removed in v5. Use `image_processor` instead.''' , _a , )
return self.image_processor
| 15
| 0
|
import darl # noqa
import gym
import tqdm
from diffusers.experimental import ValueGuidedRLPipeline
a__ = {
'''n_samples''': 64,
'''horizon''': 32,
'''num_inference_steps''': 20,
'''n_guide_steps''': 2, # can set to 0 for faster sampling, does not use value network
'''scale_grad_by_std''': True,
'''scale''': 0.1,
'''eta''': 0.0,
'''t_grad_cutoff''': 2,
'''device''': '''cpu''',
}
if __name__ == "__main__":
a__ = '''hopper-medium-v2'''
a__ = gym.make(env_name)
a__ = ValueGuidedRLPipeline.from_pretrained(
'''bglick13/hopper-medium-v2-value-function-hor32''',
env=env,
)
env.seed(0)
a__ = env.reset()
a__ = 0
a__ = 0
a__ = 1000
a__ = [obs.copy()]
try:
for t in tqdm.tqdm(range(T)):
# call the policy
a__ = pipeline(obs, planning_horizon=32)
# execute action in environment
a__ , a__ , a__ , a__ = env.step(denorm_actions)
a__ = env.get_normalized_score(total_reward)
# update return
total_reward += reward
total_score += score
print(
f'''Step: {t}, Reward: {reward}, Total Reward: {total_reward}, Score: {score}, Total Score:'''
f''' {total_score}'''
)
# save observations for rendering
rollout.append(next_observation.copy())
a__ = next_observation
except KeyboardInterrupt:
pass
print(f'''Total reward: {total_reward}''')
| 351
|
from math import ceil
def __UpperCAmelCase ( __a : int = 1_001 ) -> int:
"""simple docstring"""
_a : Dict = 1
for i in range(1 ,int(ceil(n / 2.0 ) ) ):
_a : int = 2 * i + 1
_a : str = 2 * i
_a : Any = total + 4 * odd**2 - 6 * even
return total
if __name__ == "__main__":
import sys
if len(sys.argv) == 1:
print(solution())
else:
try:
a__ = int(sys.argv[1])
print(solution(n))
except ValueError:
print('''Invalid entry - please enter a number''')
| 15
| 0
|
"""simple docstring"""
import argparse
from tax import checkpoints
from transformers import AutoConfig, FlaxAutoModelForSeqaSeqLM
def __UpperCAmelCase ( __a : Tuple ,__a : Union[str, Any] ,__a : Union[str, Any] ) -> Union[str, Any]:
"""simple docstring"""
_a : Dict = AutoConfig.from_pretrained(__a )
_a : Union[str, Any] = FlaxAutoModelForSeqaSeqLM.from_config(config=__a )
_a : Optional[Any] = checkpoints.load_tax_checkpoint(__a )
_a : Tuple = '''wi_0''' in tax_model['''target''']['''encoder''']['''layers_0''']['''mlp''']
if config.model_type == "t5":
_a : Optional[int] = '''SelfAttention'''
if config.model_type == "longt5" and config.encoder_attention_type == "local":
_a : int = '''LocalSelfAttention'''
elif config.model_type == "longt5" and config.encoder_attention_type == "transient-global":
_a : Optional[Any] = '''TransientGlobalSelfAttention'''
else:
raise ValueError(
'''Given config is expected to have `model_type=\'t5\'`, or `model_type=\'longt5` with `encoder_attention_type`'''
''' attribute with a value from [\'local\', \'transient-global].''' )
# Encoder
for layer_index in range(config.num_layers ):
_a : Optional[int] = F"""layers_{str(__a )}"""
# Self-Attention
_a : Optional[Any] = tax_model['''target''']['''encoder'''][layer_name]['''attention''']['''key''']['''kernel''']
_a : Dict = tax_model['''target''']['''encoder'''][layer_name]['''attention''']['''out''']['''kernel''']
_a : Optional[Any] = tax_model['''target''']['''encoder'''][layer_name]['''attention''']['''query''']['''kernel''']
_a : Dict = tax_model['''target''']['''encoder'''][layer_name]['''attention''']['''value''']['''kernel''']
# Global input layer norm
if config.model_type == "longt5" and config.encoder_attention_type == "transient-global":
_a : Tuple = tax_model['''target''']['''encoder'''][layer_name]['''attention''']['''T5LayerNorm_0''']['''scale''']
# Layer Normalization
_a : Optional[int] = tax_model['''target''']['''encoder'''][layer_name]['''pre_attention_layer_norm''']['''scale''']
if split_mlp_wi:
_a : Optional[Any] = tax_model['''target''']['''encoder'''][layer_name]['''mlp''']['''wi_0''']['''kernel''']
_a : List[str] = tax_model['''target''']['''encoder'''][layer_name]['''mlp''']['''wi_1''']['''kernel''']
else:
_a : Optional[int] = tax_model['''target''']['''encoder'''][layer_name]['''mlp''']['''wi''']['''kernel''']
_a : List[str] = tax_model['''target''']['''encoder'''][layer_name]['''mlp''']['''wo''']['''kernel''']
# Layer Normalization
_a : str = tax_model['''target''']['''encoder'''][layer_name]['''pre_mlp_layer_norm''']['''scale''']
# Assigning
_a : List[Any] = flax_model.params['''encoder''']['''block'''][str(__a )]['''layer''']
_a : Union[str, Any] = tax_attention_key
_a : int = tax_attention_out
_a : str = tax_attention_query
_a : Optional[Any] = tax_attention_value
_a : Union[str, Any] = tax_attention_layer_norm
# Global input layer norm
if config.model_type == "longt5" and config.encoder_attention_type == "transient-global":
_a : Dict = tax_global_layer_norm
if split_mlp_wi:
_a : Any = tax_mlp_wi_a
_a : int = tax_mlp_wi_a
else:
_a : Dict = tax_mlp_wi
_a : int = tax_mlp_wo
_a : Tuple = tax_mlp_layer_norm
_a : List[Any] = flax_model_encoder_layer_block
# Only for layer 0:
_a : Tuple = tax_model['''target''']['''encoder''']['''relpos_bias''']['''rel_embedding'''].T
_a : Tuple = tax_encoder_rel_embedding
# Side/global relative position_bias + layer norm
if config.model_type == "longt5" and config.encoder_attention_type == "transient-global":
_a : int = tax_model['''target''']['''encoder''']['''side_relpos_bias''']['''rel_embedding'''].T
_a : Optional[int] = tax_encoder_global_rel_embedding
# Assigning
_a : List[Any] = tax_model['''target''']['''encoder''']['''encoder_norm''']['''scale''']
_a : Optional[Any] = tax_encoder_norm
# Decoder
for layer_index in range(config.num_layers ):
_a : str = F"""layers_{str(__a )}"""
# Self-Attention
_a : Optional[Any] = tax_model['''target''']['''decoder'''][layer_name]['''self_attention''']['''key''']['''kernel''']
_a : List[str] = tax_model['''target''']['''decoder'''][layer_name]['''self_attention''']['''out''']['''kernel''']
_a : List[Any] = tax_model['''target''']['''decoder'''][layer_name]['''self_attention''']['''query''']['''kernel''']
_a : Optional[Any] = tax_model['''target''']['''decoder'''][layer_name]['''self_attention''']['''value''']['''kernel''']
# Layer Normalization
_a : str = tax_model['''target''']['''decoder'''][layer_name]['''pre_self_attention_layer_norm'''][
'''scale'''
]
# Encoder-Decoder-Attention
_a : Tuple = tax_model['''target''']['''decoder'''][layer_name]['''encoder_decoder_attention''']
_a : Optional[int] = tax_enc_dec_attention_module['''key''']['''kernel''']
_a : str = tax_enc_dec_attention_module['''out''']['''kernel''']
_a : List[Any] = tax_enc_dec_attention_module['''query''']['''kernel''']
_a : str = tax_enc_dec_attention_module['''value''']['''kernel''']
# Layer Normalization
_a : Any = tax_model['''target''']['''decoder'''][layer_name]['''pre_cross_attention_layer_norm''']['''scale''']
# MLP
if split_mlp_wi:
_a : str = tax_model['''target''']['''decoder'''][layer_name]['''mlp''']['''wi_0''']['''kernel''']
_a : Tuple = tax_model['''target''']['''decoder'''][layer_name]['''mlp''']['''wi_1''']['''kernel''']
else:
_a : Optional[int] = tax_model['''target''']['''decoder'''][layer_name]['''mlp''']['''wi''']['''kernel''']
_a : Optional[int] = tax_model['''target''']['''decoder'''][layer_name]['''mlp''']['''wo''']['''kernel''']
# Layer Normalization
_a : Optional[Any] = tax_model['''target''']['''decoder'''][layer_name]['''pre_mlp_layer_norm''']['''scale''']
# Assigning
_a : Optional[Any] = flax_model.params['''decoder''']['''block'''][str(__a )]['''layer''']
_a : List[Any] = tax_attention_key
_a : Optional[int] = tax_attention_out
_a : str = tax_attention_query
_a : Tuple = tax_attention_value
_a : List[str] = tax_pre_attention_layer_norm
_a : Any = tax_enc_dec_attention_key
_a : Dict = tax_enc_dec_attention_out
_a : Optional[Any] = tax_enc_dec_attention_query
_a : List[str] = tax_enc_dec_attention_value
_a : Dict = tax_cross_layer_norm
if split_mlp_wi:
_a : Optional[int] = tax_mlp_wi_a
_a : Dict = tax_mlp_wi_a
else:
_a : Optional[Any] = tax_mlp_wi
_a : Any = tax_mlp_wo
_a : Optional[Any] = txa_mlp_layer_norm
_a : Union[str, Any] = flax_model_decoder_layer_block
# Decoder Normalization
_a : Tuple = tax_model['''target''']['''decoder''']['''decoder_norm''']['''scale''']
_a : str = txa_decoder_norm
# Only for layer 0:
_a : Dict = tax_model['''target''']['''decoder''']['''relpos_bias''']['''rel_embedding'''].T
_a : str = tax_decoder_rel_embedding
# Token Embeddings
_a : Optional[Any] = tax_model['''target''']['''token_embedder''']['''embedding''']
_a : Any = txa_token_embeddings
# LM Head (only in v1.1 and LongT5 checkpoints)
if "logits_dense" in tax_model["target"]["decoder"]:
_a : Any = tax_model['''target''']['''decoder''']['''logits_dense''']['''kernel''']
flax_model.save_pretrained(__a )
print('''T5X Model was sucessfully converted!''' )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--t5x_checkpoint_path''', default=None, type=str, required=True, help='''Path the T5X checkpoint.'''
)
parser.add_argument('''--config_name''', default=None, type=str, required=True, help='''Config name of LongT5/T5 model.''')
parser.add_argument(
'''--flax_dump_folder_path''', default=None, type=str, required=True, help='''Path to the output FLAX model.'''
)
a__ = parser.parse_args()
convert_tax_checkpoint_to_flax(args.tax_checkpoint_path, args.config_name, args.flax_dump_folder_path)
| 352
|
from typing import Dict, Iterable, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format, to_pil_image
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_pytesseract_available, is_vision_available, logging, requires_backends
if is_vision_available():
import PIL
# soft dependency
if is_pytesseract_available():
import pytesseract
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : str ,__a : Union[str, Any] ) -> List[str]:
"""simple docstring"""
return [
int(1_000 * (box[0] / width) ),
int(1_000 * (box[1] / height) ),
int(1_000 * (box[2] / width) ),
int(1_000 * (box[3] / height) ),
]
def __UpperCAmelCase ( __a : np.ndarray ,__a : Optional[str] ,__a : Optional[str] ) -> List[Any]:
"""simple docstring"""
_a : str = to_pil_image(__a )
_a , _a : Optional[Any] = pil_image.size
_a : Tuple = pytesseract.image_to_data(__a ,lang=__a ,output_type='''dict''' ,config=__a )
_a , _a , _a , _a , _a : List[str] = data['''text'''], data['''left'''], data['''top'''], data['''width'''], data['''height''']
# filter empty words and corresponding coordinates
_a : Dict = [idx for idx, word in enumerate(__a ) if not word.strip()]
_a : str = [word for idx, word in enumerate(__a ) if idx not in irrelevant_indices]
_a : List[str] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : str = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
# turn coordinates into (left, top, left+width, top+height) format
_a : int = []
for x, y, w, h in zip(__a ,__a ,__a ,__a ):
_a : List[str] = [x, y, x + w, y + h]
actual_boxes.append(__a )
# finally, normalize the bounding boxes
_a : Dict = []
for box in actual_boxes:
normalized_boxes.append(normalize_box(__a ,__a ,__a ) )
assert len(__a ) == len(__a ), "Not as many words as there are bounding boxes"
return words, normalized_boxes
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[int] = ["pixel_values"]
def __init__( self , _a = True , _a = None , _a = PILImageResampling.BILINEAR , _a = True , _a = 1 / 2_5_5 , _a = True , _a = None , _a = None , _a = True , _a = None , _a = "" , **_a , ) -> None:
super().__init__(**_a )
_a : List[str] = size if size is not None else {'''height''': 2_2_4, '''width''': 2_2_4}
_a : Union[str, Any] = get_size_dict(_a )
_a : int = do_resize
_a : Optional[int] = size
_a : str = resample
_a : str = do_rescale
_a : Any = rescale_value
_a : Optional[Any] = do_normalize
_a : int = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
_a : List[str] = image_std if image_std is not None else IMAGENET_STANDARD_STD
_a : List[Any] = apply_ocr
_a : Optional[int] = ocr_lang
_a : Tuple = tesseract_config
def __lowercase ( self , _a , _a , _a = PILImageResampling.BILINEAR , _a = None , **_a , ) -> np.ndarray:
_a : Any = get_size_dict(_a )
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()}""" )
_a : Optional[int] = (size['''height'''], size['''width'''])
return resize(_a , size=_a , resample=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
return rescale(_a , scale=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a , _a = None , **_a , ) -> np.ndarray:
return normalize(_a , mean=_a , std=_a , data_format=_a , **_a )
def __lowercase ( 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:
_a : Optional[int] = do_resize if do_resize is not None else self.do_resize
_a : Union[str, Any] = size if size is not None else self.size
_a : Any = get_size_dict(_a )
_a : List[str] = resample if resample is not None else self.resample
_a : int = do_rescale if do_rescale is not None else self.do_rescale
_a : Union[str, Any] = rescale_factor if rescale_factor is not None else self.rescale_factor
_a : int = do_normalize if do_normalize is not None else self.do_normalize
_a : str = image_mean if image_mean is not None else self.image_mean
_a : Tuple = image_std if image_std is not None else self.image_std
_a : Any = apply_ocr if apply_ocr is not None else self.apply_ocr
_a : int = ocr_lang if ocr_lang is not None else self.ocr_lang
_a : Optional[int] = tesseract_config if tesseract_config is not None else self.tesseract_config
_a : List[Any] = make_list_of_images(_a )
if not valid_images(_a ):
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_normalize and (image_mean is None or image_std is None):
raise ValueError('''If do_normalize is True, image_mean and image_std must be specified.''' )
# All transformations expect numpy arrays.
_a : Any = [to_numpy_array(_a ) for image in images]
# Tesseract OCR to get words + normalized bounding boxes
if apply_ocr:
requires_backends(self , '''pytesseract''' )
_a : str = []
_a : str = []
for image in images:
_a , _a : Union[str, Any] = apply_tesseract(_a , _a , _a )
words_batch.append(_a )
boxes_batch.append(_a )
if do_resize:
_a : List[str] = [self.resize(image=_a , size=_a , resample=_a ) for image in images]
if do_rescale:
_a : Optional[Any] = [self.rescale(image=_a , scale=_a ) for image in images]
if do_normalize:
_a : List[Any] = [self.normalize(image=_a , mean=_a , std=_a ) for image in images]
_a : List[str] = [to_channel_dimension_format(_a , _a ) for image in images]
_a : List[str] = BatchFeature(data={'''pixel_values''': images} , tensor_type=_a )
if apply_ocr:
_a : Optional[int] = words_batch
_a : List[Any] = boxes_batch
return data
| 15
| 0
|
"""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
a__ = [
{'''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 ( __a : Optional[int]=True ) -> Optional[int]:
"""simple docstring"""
if with_config:
return [
{
"testcase_name": d["dataset"] + "/" + d["config_name"],
"dataset": d["dataset"],
"config_name": d["config_name"],
}
for d in DATASETS_ON_HF_GCP
]
else:
return [
{"testcase_name": dataset, "dataset": dataset} for dataset in {d["dataset"] for d in DATASETS_ON_HF_GCP}
]
@parameterized.named_parameters(list_datasets_on_hf_gcp_parameters(with_config=__lowercase ) )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : int = None
UpperCAmelCase__ : int = None
def __lowercase ( self , _a , _a ) -> Optional[Any]:
with TemporaryDirectory() as tmp_dir:
_a : Union[str, Any] = dataset_module_factory(_a , cache_dir=_a )
_a : Tuple = import_main_class(dataset_module.module_path , dataset=_a )
_a : DatasetBuilder = builder_cls(
cache_dir=_a , config_name=_a , hash=dataset_module.hash , )
_a : Union[str, Any] = '''/'''.join(
[
HF_GCP_BASE_URL,
builder_instance._relative_data_dir(with_hash=_a ).replace(os.sep , '''/''' ),
config.DATASET_INFO_FILENAME,
] )
_a : Dict = cached_path(_a , cache_dir=_a )
self.assertTrue(os.path.exists(_a ) )
@pytest.mark.integration
def __UpperCAmelCase ( __a : List[str] ) -> int:
"""simple docstring"""
_a : Union[str, Any] = tmp_path_factory.mktemp('''test_hf_gcp''' ) / '''test_wikipedia_simple'''
_a : str = dataset_module_factory('''wikipedia''' ,cache_dir=__a )
_a : Dict = import_main_class(dataset_module.module_path )
_a : DatasetBuilder = builder_cls(
cache_dir=__a ,config_name='''20220301.frr''' ,hash=dataset_module.hash ,)
# use the HF cloud storage, not the original download_and_prepare that uses apache-beam
_a : Union[str, Any] = None
builder_instance.download_and_prepare()
_a : Optional[int] = builder_instance.as_dataset()
assert ds
@pytest.mark.integration
def __UpperCAmelCase ( __a : Optional[Any] ) -> Dict:
"""simple docstring"""
_a : Optional[Any] = dataset_module_factory('''wikipedia''' ,cache_dir=__a )
_a : Any = import_main_class(dataset_module.module_path ,dataset=__a )
_a : DatasetBuilder = builder_cls(
cache_dir=__a ,config_name='''20220301.frr''' ,hash=dataset_module.hash ,)
_a : Dict = builder_instance.as_streaming_dataset()
assert ds
assert isinstance(__a ,__a )
assert "train" in ds
assert isinstance(ds['''train'''] ,__a )
assert next(iter(ds['''train'''] ) )
| 353
|
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from argparse import ArgumentParser
from accelerate.commands.config import get_config_parser
from accelerate.commands.env import env_command_parser
from accelerate.commands.launch import launch_command_parser
from accelerate.commands.test import test_command_parser
from accelerate.commands.tpu import tpu_command_parser
def __UpperCAmelCase ( ) -> Optional[Any]:
"""simple docstring"""
_a : int = ArgumentParser('''Accelerate CLI tool''' ,usage='''accelerate <command> [<args>]''' ,allow_abbrev=__a )
_a : Optional[int] = parser.add_subparsers(help='''accelerate command helpers''' )
# Register commands
get_config_parser(subparsers=__a )
env_command_parser(subparsers=__a )
launch_command_parser(subparsers=__a )
tpu_command_parser(subparsers=__a )
test_command_parser(subparsers=__a )
# Let's go
_a : Dict = parser.parse_args()
if not hasattr(__a ,'''func''' ):
parser.print_help()
exit(1 )
# Run
args.func(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
"""simple docstring"""
import importlib
import json
import os
from collections import OrderedDict
from typing import Dict, Optional, Union
# Build the list of all image processors
from ...configuration_utils import PretrainedConfig
from ...dynamic_module_utils import get_class_from_dynamic_module, resolve_trust_remote_code
from ...image_processing_utils import ImageProcessingMixin
from ...utils import CONFIG_NAME, IMAGE_PROCESSOR_NAME, get_file_from_repo, logging
from .auto_factory import _LazyAutoMapping
from .configuration_auto import (
CONFIG_MAPPING_NAMES,
AutoConfig,
model_type_to_module_name,
replace_list_option_in_docstrings,
)
a__ = logging.get_logger(__name__)
a__ = OrderedDict(
[
('''align''', '''EfficientNetImageProcessor'''),
('''beit''', '''BeitImageProcessor'''),
('''bit''', '''BitImageProcessor'''),
('''blip''', '''BlipImageProcessor'''),
('''blip-2''', '''BlipImageProcessor'''),
('''bridgetower''', '''BridgeTowerImageProcessor'''),
('''chinese_clip''', '''ChineseCLIPImageProcessor'''),
('''clip''', '''CLIPImageProcessor'''),
('''clipseg''', '''ViTImageProcessor'''),
('''conditional_detr''', '''ConditionalDetrImageProcessor'''),
('''convnext''', '''ConvNextImageProcessor'''),
('''convnextv2''', '''ConvNextImageProcessor'''),
('''cvt''', '''ConvNextImageProcessor'''),
('''data2vec-vision''', '''BeitImageProcessor'''),
('''deformable_detr''', '''DeformableDetrImageProcessor'''),
('''deit''', '''DeiTImageProcessor'''),
('''deta''', '''DetaImageProcessor'''),
('''detr''', '''DetrImageProcessor'''),
('''dinat''', '''ViTImageProcessor'''),
('''donut-swin''', '''DonutImageProcessor'''),
('''dpt''', '''DPTImageProcessor'''),
('''efficientformer''', '''EfficientFormerImageProcessor'''),
('''efficientnet''', '''EfficientNetImageProcessor'''),
('''flava''', '''FlavaImageProcessor'''),
('''focalnet''', '''BitImageProcessor'''),
('''git''', '''CLIPImageProcessor'''),
('''glpn''', '''GLPNImageProcessor'''),
('''groupvit''', '''CLIPImageProcessor'''),
('''imagegpt''', '''ImageGPTImageProcessor'''),
('''instructblip''', '''BlipImageProcessor'''),
('''layoutlmv2''', '''LayoutLMv2ImageProcessor'''),
('''layoutlmv3''', '''LayoutLMv3ImageProcessor'''),
('''levit''', '''LevitImageProcessor'''),
('''mask2former''', '''Mask2FormerImageProcessor'''),
('''maskformer''', '''MaskFormerImageProcessor'''),
('''mgp-str''', '''ViTImageProcessor'''),
('''mobilenet_v1''', '''MobileNetV1ImageProcessor'''),
('''mobilenet_v2''', '''MobileNetV2ImageProcessor'''),
('''mobilevit''', '''MobileViTImageProcessor'''),
('''mobilevit''', '''MobileViTImageProcessor'''),
('''mobilevitv2''', '''MobileViTImageProcessor'''),
('''nat''', '''ViTImageProcessor'''),
('''oneformer''', '''OneFormerImageProcessor'''),
('''owlvit''', '''OwlViTImageProcessor'''),
('''perceiver''', '''PerceiverImageProcessor'''),
('''pix2struct''', '''Pix2StructImageProcessor'''),
('''poolformer''', '''PoolFormerImageProcessor'''),
('''regnet''', '''ConvNextImageProcessor'''),
('''resnet''', '''ConvNextImageProcessor'''),
('''sam''', '''SamImageProcessor'''),
('''segformer''', '''SegformerImageProcessor'''),
('''swiftformer''', '''ViTImageProcessor'''),
('''swin''', '''ViTImageProcessor'''),
('''swin2sr''', '''Swin2SRImageProcessor'''),
('''swinv2''', '''ViTImageProcessor'''),
('''table-transformer''', '''DetrImageProcessor'''),
('''timesformer''', '''VideoMAEImageProcessor'''),
('''tvlt''', '''TvltImageProcessor'''),
('''upernet''', '''SegformerImageProcessor'''),
('''van''', '''ConvNextImageProcessor'''),
('''videomae''', '''VideoMAEImageProcessor'''),
('''vilt''', '''ViltImageProcessor'''),
('''vit''', '''ViTImageProcessor'''),
('''vit_hybrid''', '''ViTHybridImageProcessor'''),
('''vit_mae''', '''ViTImageProcessor'''),
('''vit_msn''', '''ViTImageProcessor'''),
('''xclip''', '''CLIPImageProcessor'''),
('''yolos''', '''YolosImageProcessor'''),
]
)
a__ = _LazyAutoMapping(CONFIG_MAPPING_NAMES, IMAGE_PROCESSOR_MAPPING_NAMES)
def __UpperCAmelCase ( __a : str ) -> Union[str, Any]:
"""simple docstring"""
for module_name, extractors in IMAGE_PROCESSOR_MAPPING_NAMES.items():
if class_name in extractors:
_a : List[Any] = model_type_to_module_name(__a )
_a : Tuple = importlib.import_module(F""".{module_name}""" ,'''transformers.models''' )
try:
return getattr(__a ,__a )
except AttributeError:
continue
for _, extractor in IMAGE_PROCESSOR_MAPPING._extra_content.items():
if getattr(__a ,'''__name__''' ,__a ) == class_name:
return extractor
# We did not fine the class, but maybe it's because a dep is missing. In that case, the class will be in the main
# init and we return the proper dummy to get an appropriate error message.
_a : Optional[int] = importlib.import_module('''transformers''' )
if hasattr(__a ,__a ):
return getattr(__a ,__a )
return None
def __UpperCAmelCase ( __a : Union[str, os.PathLike] ,__a : Optional[Union[str, os.PathLike]] = None ,__a : bool = False ,__a : bool = False ,__a : Optional[Dict[str, str]] = None ,__a : Optional[Union[bool, str]] = None ,__a : Optional[str] = None ,__a : bool = False ,**__a : Dict ,) -> Tuple:
"""simple docstring"""
_a : Optional[int] = get_file_from_repo(
__a ,__a ,cache_dir=__a ,force_download=__a ,resume_download=__a ,proxies=__a ,use_auth_token=__a ,revision=__a ,local_files_only=__a ,)
if resolved_config_file is None:
logger.info(
'''Could not locate the image processor configuration file, will try to use the model config instead.''' )
return {}
with open(__a ,encoding='''utf-8''' ) as reader:
return json.load(__a )
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self ) -> List[Any]:
raise EnvironmentError(
'''AutoImageProcessor is designed to be instantiated '''
'''using the `AutoImageProcessor.from_pretrained(pretrained_model_name_or_path)` method.''' )
@classmethod
@replace_list_option_in_docstrings(_a )
def __lowercase ( cls , _a , **_a ) -> Optional[Any]:
_a : Optional[Any] = kwargs.pop('''config''' , _a )
_a : Optional[Any] = kwargs.pop('''trust_remote_code''' , _a )
_a : Any = True
_a : Any = ImageProcessingMixin.get_image_processor_dict(_a , **_a )
_a : Optional[int] = config_dict.get('''image_processor_type''' , _a )
_a : Union[str, Any] = None
if "AutoImageProcessor" in config_dict.get('''auto_map''' , {} ):
_a : Optional[Any] = config_dict['''auto_map''']['''AutoImageProcessor''']
# If we still don't have the image processor class, check if we're loading from a previous feature extractor config
# and if so, infer the image processor class from there.
if image_processor_class is None and image_processor_auto_map is None:
_a : int = config_dict.pop('''feature_extractor_type''' , _a )
if feature_extractor_class is not None:
logger.warning(
'''Could not find image processor class in the image processor config or the model config. Loading'''
''' based on pattern matching with the model\'s feature extractor configuration.''' )
_a : str = feature_extractor_class.replace('''FeatureExtractor''' , '''ImageProcessor''' )
if "AutoFeatureExtractor" in config_dict.get('''auto_map''' , {} ):
_a : int = config_dict['''auto_map''']['''AutoFeatureExtractor''']
_a : Optional[int] = feature_extractor_auto_map.replace('''FeatureExtractor''' , '''ImageProcessor''' )
logger.warning(
'''Could not find image processor auto map in the image processor config or the model config.'''
''' Loading based on pattern matching with the model\'s feature extractor configuration.''' )
# If we don't find the image processor class in the image processor config, let's try the model config.
if image_processor_class is None and image_processor_auto_map is None:
if not isinstance(_a , _a ):
_a : List[Any] = AutoConfig.from_pretrained(_a , **_a )
# It could be in `config.image_processor_type``
_a : Dict = getattr(_a , '''image_processor_type''' , _a )
if hasattr(_a , '''auto_map''' ) and "AutoImageProcessor" in config.auto_map:
_a : Optional[Any] = config.auto_map['''AutoImageProcessor''']
if image_processor_class is not None:
_a : List[Any] = image_processor_class_from_name(_a )
_a : int = image_processor_auto_map is not None
_a : Optional[int] = image_processor_class is not None or type(_a ) in IMAGE_PROCESSOR_MAPPING
_a : Any = resolve_trust_remote_code(
_a , _a , _a , _a )
if has_remote_code and trust_remote_code:
_a : Any = get_class_from_dynamic_module(
_a , _a , **_a )
_a : Dict = kwargs.pop('''code_revision''' , _a )
if os.path.isdir(_a ):
image_processor_class.register_for_auto_class()
return image_processor_class.from_dict(_a , **_a )
elif image_processor_class is not None:
return image_processor_class.from_dict(_a , **_a )
# Last try: we use the IMAGE_PROCESSOR_MAPPING.
elif type(_a ) in IMAGE_PROCESSOR_MAPPING:
_a : str = IMAGE_PROCESSOR_MAPPING[type(_a )]
return image_processor_class.from_dict(_a , **_a )
raise ValueError(
F"""Unrecognized image processor in {pretrained_model_name_or_path}. Should have a """
F"""`image_processor_type` key in its {IMAGE_PROCESSOR_NAME} of {CONFIG_NAME}, or one of the following """
F"""`model_type` keys in its {CONFIG_NAME}: {', '.join(c for c in IMAGE_PROCESSOR_MAPPING_NAMES.keys() )}""" )
@staticmethod
def __lowercase ( _a , _a ) -> Dict:
IMAGE_PROCESSOR_MAPPING.register(_a , _a )
| 354
|
import itertools
import os
import random
import tempfile
import unittest
import numpy as np
from transformers import TvltFeatureExtractor, is_datasets_available
from transformers.testing_utils import check_json_file_has_correct_format, require_torch, require_torchaudio
from transformers.utils.import_utils import is_torch_available
from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
if is_torch_available():
import torch
if is_datasets_available():
from datasets import load_dataset
a__ = random.Random()
def __UpperCAmelCase ( __a : Tuple ,__a : str=1.0 ,__a : Optional[int]=None ,__a : List[Any]=None ) -> Any:
"""simple docstring"""
if rng is None:
_a : Dict = global_rng
_a : Optional[Any] = []
for batch_idx in range(shape[0] ):
values.append([] )
for _ in range(shape[1] ):
values[-1].append(rng.random() * scale )
return values
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __init__( self , _a , _a=7 , _a=4_0_0 , _a=2_0_0_0 , _a=2_0_4_8 , _a=1_2_8 , _a=1 , _a=5_1_2 , _a=3_0 , _a=4_4_1_0_0 , ) -> List[Any]:
_a : Optional[Any] = parent
_a : str = batch_size
_a : List[str] = min_seq_length
_a : str = max_seq_length
_a : Dict = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
_a : List[Any] = spectrogram_length
_a : List[str] = feature_size
_a : List[Any] = num_audio_channels
_a : Tuple = hop_length
_a : Optional[int] = chunk_length
_a : int = sampling_rate
def __lowercase ( self ) -> Union[str, Any]:
return {
"spectrogram_length": self.spectrogram_length,
"feature_size": self.feature_size,
"num_audio_channels": self.num_audio_channels,
"hop_length": self.hop_length,
"chunk_length": self.chunk_length,
"sampling_rate": self.sampling_rate,
}
def __lowercase ( self , _a=False , _a=False ) -> List[Any]:
def _flatten(_a ):
return list(itertools.chain(*_a ) )
if equal_length:
_a : List[Any] = [floats_list((self.max_seq_length, self.feature_size) ) for _ in range(self.batch_size )]
else:
# make sure that inputs increase in size
_a : List[Any] = [
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 : str = [np.asarray(_a ) for x in speech_inputs]
return speech_inputs
@require_torch
@require_torchaudio
class UpperCAmelCase_ ( __lowercase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = TvltFeatureExtractor
def __lowercase ( self ) -> Dict:
_a : List[str] = TvltFeatureExtractionTester(self )
def __lowercase ( self ) -> Any:
_a : List[Any] = self.feature_extraction_class(**self.feat_extract_dict )
self.assertTrue(hasattr(_a , '''spectrogram_length''' ) )
self.assertTrue(hasattr(_a , '''feature_size''' ) )
self.assertTrue(hasattr(_a , '''num_audio_channels''' ) )
self.assertTrue(hasattr(_a , '''hop_length''' ) )
self.assertTrue(hasattr(_a , '''chunk_length''' ) )
self.assertTrue(hasattr(_a , '''sampling_rate''' ) )
def __lowercase ( self ) -> Optional[int]:
_a : Optional[Any] = self.feature_extraction_class(**self.feat_extract_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : int = feat_extract_first.save_pretrained(_a )[0]
check_json_file_has_correct_format(_a )
_a : Dict = self.feature_extraction_class.from_pretrained(_a )
_a : List[Any] = feat_extract_first.to_dict()
_a : Union[str, Any] = feat_extract_second.to_dict()
_a : Any = dict_first.pop('''mel_filters''' )
_a : int = dict_second.pop('''mel_filters''' )
self.assertTrue(np.allclose(_a , _a ) )
self.assertEqual(_a , _a )
def __lowercase ( self ) -> Optional[int]:
_a : Any = self.feature_extraction_class(**self.feat_extract_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : Optional[int] = os.path.join(_a , '''feat_extract.json''' )
feat_extract_first.to_json_file(_a )
_a : List[str] = self.feature_extraction_class.from_json_file(_a )
_a : List[Any] = feat_extract_first.to_dict()
_a : Dict = feat_extract_second.to_dict()
_a : str = dict_first.pop('''mel_filters''' )
_a : str = dict_second.pop('''mel_filters''' )
self.assertTrue(np.allclose(_a , _a ) )
self.assertEqual(_a , _a )
def __lowercase ( self ) -> Union[str, Any]:
# Initialize feature_extractor
_a : Union[str, Any] = self.feature_extraction_class(**self.feat_extract_dict )
# create three inputs of length 800, 1000, and 1200
_a : Any = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
_a : List[str] = [np.asarray(_a ) for speech_input in speech_inputs]
# Test not batched input
_a : Tuple = feature_extractor(np_speech_inputs[0] , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test batched
_a : Dict = feature_extractor(_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test audio masking
_a : Union[str, Any] = feature_extractor(
_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 , mask_audio=_a ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test 2-D numpy arrays are batched.
_a : Optional[Any] = [floats_list((1, x) )[0] for x in (8_0_0, 8_0_0, 8_0_0)]
_a : int = np.asarray(_a )
_a : Tuple = feature_extractor(_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[Any] = load_dataset('''hf-internal-testing/librispeech_asr_dummy''' , '''clean''' , split='''validation''' )
# automatic decoding with librispeech
_a : Optional[int] = ds.sort('''id''' ).select(range(_a ) )[:num_samples]['''audio''']
return [x["array"] for x in speech_samples]
def __lowercase ( self ) -> int:
_a : Union[str, Any] = self._load_datasamples(1 )
_a : int = TvltFeatureExtractor()
_a : Union[str, Any] = feature_extractor(_a , return_tensors='''pt''' ).audio_values
self.assertEquals(audio_values.shape , (1, 1, 1_9_2, 1_2_8) )
_a : Union[str, Any] = torch.tensor([[-0.3032, -0.2708], [-0.4434, -0.4007]] )
self.assertTrue(torch.allclose(audio_values[0, 0, :2, :2] , _a , atol=1e-4 ) )
| 15
| 0
|
import json
import os
import re
import sys
import urllib.request
import requests
from bsa import BeautifulSoup
a__ = {
'''User-Agent''': '''Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36'''
''' (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582'''
}
def __UpperCAmelCase ( __a : str = "dhaka" ,__a : int = 5 ) -> int:
"""simple docstring"""
_a : Optional[Any] = min(__a ,50 ) # Prevent abuse!
_a : str = {
'''q''': query,
'''tbm''': '''isch''',
'''hl''': '''en''',
'''ijn''': '''0''',
}
_a : Dict = requests.get('''https://www.google.com/search''' ,params=__a ,headers=__a )
_a : List[str] = BeautifulSoup(html.text ,'''html.parser''' )
_a : Union[str, Any] = ''''''.join(
re.findall(R'''AF_initDataCallback\(([^<]+)\);''' ,str(soup.select('''script''' ) ) ) )
_a : Optional[Any] = json.dumps(__a )
_a : Tuple = json.loads(__a )
_a : List[str] = re.findall(
R'''\[\"GRID_STATE0\",null,\[\[1,\[0,\".*?\",(.*),\"All\",''' ,__a ,)
if not matched_google_image_data:
return 0
_a : Dict = re.sub(
R'''\[\"(https\:\/\/encrypted-tbn0\.gstatic\.com\/images\?.*?)\",\d+,\d+\]''' ,'''''' ,str(__a ) ,)
_a : List[str] = re.findall(
R'''(?:\'|,),\[\"(https:|http.*?)\",\d+,\d+\]''' ,__a ,)
for index, fixed_full_res_image in enumerate(__a ):
if index >= max_images:
return index
_a : List[Any] = bytes(__a ,'''ascii''' ).decode(
'''unicode-escape''' )
_a : Optional[Any] = bytes(__a ,'''ascii''' ).decode(
'''unicode-escape''' )
_a : Any = urllib.request.build_opener()
_a : List[str] = [
(
'''User-Agent''',
'''Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36'''
''' (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36 Edge/18.19582''',
)
]
urllib.request.install_opener(__a )
_a : Any = F"""query_{query.replace(' ' ,'_' )}"""
if not os.path.exists(__a ):
os.makedirs(__a )
urllib.request.urlretrieve( # noqa: S310
__a ,F"""{path_name}/original_size_img_{index}.jpg""" )
return index
if __name__ == "__main__":
try:
a__ = download_images_from_google_query(sys.argv[1])
print(f'''{image_count} images were downloaded to disk.''')
except IndexError:
print('''Please provide a search term.''')
raise
| 355
|
from typing import Dict
import numpy as np
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, GenericTensor, Pipeline, PipelineException
if is_tf_available():
import tensorflow as tf
from ..tf_utils import stable_softmax
if is_torch_available():
import torch
a__ = logging.get_logger(__name__)
@add_end_docstrings(
__lowercase , r"\n top_k (`int`, defaults to 5):\n The number of predictions to return.\n targets (`str` or `List[str]`, *optional*):\n When passed, the model will limit the scores to the passed targets instead of looking up in the whole\n vocab. If the provided targets are not in the model vocab, they will be tokenized and the first resulting\n token will be used (with a warning, and that might be slower).\n\n " , )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self , _a ) -> np.ndarray:
if self.framework == "tf":
_a : List[str] = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()
elif self.framework == "pt":
_a : Tuple = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a )
else:
raise ValueError('''Unsupported framework''' )
return masked_index
def __lowercase ( self , _a ) -> np.ndarray:
_a : int = self.get_masked_index(_a )
_a : Tuple = np.prod(masked_index.shape )
if numel < 1:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , F"""No mask_token ({self.tokenizer.mask_token}) found on the input""" , )
def __lowercase ( self , _a ) -> Optional[int]:
if isinstance(_a , _a ):
for model_input in model_inputs:
self._ensure_exactly_one_mask_token(model_input['''input_ids'''][0] )
else:
for input_ids in model_inputs["input_ids"]:
self._ensure_exactly_one_mask_token(_a )
def __lowercase ( self , _a , _a=None , **_a ) -> Dict[str, GenericTensor]:
if return_tensors is None:
_a : Union[str, Any] = self.framework
_a : str = self.tokenizer(_a , return_tensors=_a )
self.ensure_exactly_one_mask_token(_a )
return model_inputs
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[str] = self.model(**_a )
_a : Any = model_inputs['''input_ids''']
return model_outputs
def __lowercase ( self , _a , _a=5 , _a=None ) -> str:
# Cap top_k if there are targets
if target_ids is not None and target_ids.shape[0] < top_k:
_a : List[Any] = target_ids.shape[0]
_a : Any = model_outputs['''input_ids'''][0]
_a : List[str] = model_outputs['''logits''']
if self.framework == "tf":
_a : Tuple = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()[:, 0]
_a : List[str] = outputs.numpy()
_a : Dict = outputs[0, masked_index, :]
_a : str = stable_softmax(_a , axis=-1 )
if target_ids is not None:
_a : Any = tf.gather_nd(tf.squeeze(_a , 0 ) , target_ids.reshape(-1 , 1 ) )
_a : Union[str, Any] = tf.expand_dims(_a , 0 )
_a : Optional[int] = tf.math.top_k(_a , k=_a )
_a , _a : Optional[Any] = topk.values.numpy(), topk.indices.numpy()
else:
_a : Optional[Any] = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a ).squeeze(-1 )
# Fill mask pipeline supports only one ${mask_token} per sample
_a : List[str] = outputs[0, masked_index, :]
_a : List[Any] = logits.softmax(dim=-1 )
if target_ids is not None:
_a : List[Any] = probs[..., target_ids]
_a , _a : Optional[Any] = probs.topk(_a )
_a : Dict = []
_a : List[Any] = values.shape[0] == 1
for i, (_values, _predictions) in enumerate(zip(values.tolist() , predictions.tolist() ) ):
_a : Optional[Any] = []
for v, p in zip(_values , _predictions ):
# Copy is important since we're going to modify this array in place
_a : Optional[int] = input_ids.numpy().copy()
if target_ids is not None:
_a : Tuple = target_ids[p].tolist()
_a : List[str] = p
# Filter padding out:
_a : List[Any] = tokens[np.where(tokens != self.tokenizer.pad_token_id )]
# Originally we skip special tokens to give readable output.
# For multi masks though, the other [MASK] would be removed otherwise
# making the output look odd, so we add them back
_a : List[str] = self.tokenizer.decode(_a , skip_special_tokens=_a )
_a : List[Any] = {'''score''': v, '''token''': p, '''token_str''': self.tokenizer.decode([p] ), '''sequence''': sequence}
row.append(_a )
result.append(_a )
if single_mask:
return result[0]
return result
def __lowercase ( self , _a , _a=None ) -> Dict:
if isinstance(_a , _a ):
_a : Tuple = [targets]
try:
_a : int = self.tokenizer.get_vocab()
except Exception:
_a : Any = {}
_a : List[Any] = []
for target in targets:
_a : List[Any] = vocab.get(_a , _a )
if id_ is None:
_a : Tuple = self.tokenizer(
_a , add_special_tokens=_a , return_attention_mask=_a , return_token_type_ids=_a , max_length=1 , truncation=_a , )['''input_ids''']
if len(_a ) == 0:
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
'''We cannot replace it with anything meaningful, ignoring it''' )
continue
_a : Tuple = input_ids[0]
# XXX: If users encounter this pass
# it becomes pretty slow, so let's make sure
# The warning enables them to fix the input to
# get faster performance.
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
F"""Replacing with `{self.tokenizer.convert_ids_to_tokens(id_ )}`.""" )
target_ids.append(id_ )
_a : List[str] = list(set(_a ) )
if len(_a ) == 0:
raise ValueError('''At least one target must be provided when passed.''' )
_a : int = np.array(_a )
return target_ids
def __lowercase ( self , _a=None , _a=None ) -> Tuple:
_a : str = {}
if targets is not None:
_a : List[Any] = self.get_target_ids(_a , _a )
_a : Optional[Any] = target_ids
if top_k is not None:
_a : Union[str, Any] = top_k
if self.tokenizer.mask_token_id is None:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , '''The tokenizer does not define a `mask_token`.''' )
return {}, {}, postprocess_params
def __call__( self , _a , *_a , **_a ) -> int:
_a : Optional[Any] = super().__call__(_a , **_a )
if isinstance(_a , _a ) and len(_a ) == 1:
return outputs[0]
return outputs
| 15
| 0
|
from typing import List, Optional
import numpy as np
from ...processing_utils import ProcessorMixin
from ...utils import to_numpy
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Tuple = "EncodecFeatureExtractor"
UpperCAmelCase__ : str = ("T5Tokenizer", "T5TokenizerFast")
def __init__( self , _a , _a ) -> Dict:
super().__init__(_a , _a )
_a : int = self.feature_extractor
_a : Optional[int] = False
def __lowercase ( self , _a=None , _a=None , _a=True ) -> List[str]:
return self.tokenizer.get_decoder_prompt_ids(task=_a , language=_a , no_timestamps=_a )
def __call__( self , *_a , **_a ) -> int:
# For backward compatibility
if self._in_target_context_manager:
return self.current_processor(*_a , **_a )
_a : Any = kwargs.pop('''audio''' , _a )
_a : Tuple = kwargs.pop('''sampling_rate''' , _a )
_a : List[Any] = kwargs.pop('''text''' , _a )
if len(_a ) > 0:
_a : Any = args[0]
_a : Tuple = args[1:]
if audio is None and text is None:
raise ValueError('''You need to specify either an `audio` or `text` input to process.''' )
if text is not None:
_a : Any = self.tokenizer(_a , **_a )
if audio is not None:
_a : Tuple = self.feature_extractor(_a , *_a , sampling_rate=_a , **_a )
if audio is None:
return inputs
elif text is None:
return audio_inputs
else:
_a : Any = audio_inputs['''input_values''']
if "padding_mask" in audio_inputs:
_a : List[Any] = audio_inputs['''padding_mask''']
return inputs
def __lowercase ( self , *_a , **_a ) -> Optional[int]:
_a : Tuple = kwargs.pop('''audio''' , _a )
_a : List[Any] = kwargs.pop('''padding_mask''' , _a )
if len(_a ) > 0:
_a : List[str] = args[0]
_a : Optional[Any] = args[1:]
if audio_values is not None:
return self._decode_audio(_a , padding_mask=_a )
else:
return self.tokenizer.batch_decode(*_a , **_a )
def __lowercase ( self , *_a , **_a ) -> Any:
return self.tokenizer.decode(*_a , **_a )
def __lowercase ( self , _a , _a = None ) -> List[np.ndarray]:
_a : Any = to_numpy(_a )
_a : Optional[int] = audio_values.shape
if padding_mask is None:
return list(_a )
_a : List[str] = to_numpy(_a )
# match the sequence length of the padding mask to the generated audio arrays by padding with the **non-padding**
# token (so that the generated audio values are **not** treated as padded tokens)
_a : Optional[Any] = seq_len - padding_mask.shape[-1]
_a : Dict = 1 - self.feature_extractor.padding_value
_a : Dict = np.pad(_a , ((0, 0), (0, difference)) , '''constant''' , constant_values=_a )
_a : Optional[Any] = audio_values.tolist()
for i in range(_a ):
_a : Any = np.asarray(audio_values[i] )[
padding_mask[i][None, :] != self.feature_extractor.padding_value
]
_a : Optional[int] = sliced_audio.reshape(_a , -1 )
return audio_values
| 356
|
import argparse
import json
import logging
import os
import sys
from unittest.mock import patch
from transformers.testing_utils import TestCasePlus, get_gpu_count, slow
a__ = [
os.path.join(os.path.dirname(__file__), dirname)
for dirname in [
'''text-classification''',
'''language-modeling''',
'''summarization''',
'''token-classification''',
'''question-answering''',
]
]
sys.path.extend(SRC_DIRS)
if SRC_DIRS is not None:
import run_clm_flax
import run_flax_glue
import run_flax_ner
import run_mlm_flax
import run_qa
import run_summarization_flax
import run_ta_mlm_flax
logging.basicConfig(level=logging.DEBUG)
a__ = logging.getLogger()
def __UpperCAmelCase ( ) -> Optional[int]:
"""simple docstring"""
_a : Any = argparse.ArgumentParser()
parser.add_argument('''-f''' )
_a : Dict = parser.parse_args()
return args.f
def __UpperCAmelCase ( __a : Optional[int] ,__a : List[str]="eval" ) -> Any:
"""simple docstring"""
_a : Any = os.path.join(__a ,F"""{split}_results.json""" )
if os.path.exists(__a ):
with open(__a ,'''r''' ) as f:
return json.load(__a )
raise ValueError(F"""can't find {path}""" )
a__ = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self ) -> str:
_a : Any = self.get_auto_remove_tmp_dir()
_a : Optional[Any] = F"""
run_glue.py
--model_name_or_path distilbert-base-uncased
--output_dir {tmp_dir}
--train_file ./tests/fixtures/tests_samples/MRPC/train.csv
--validation_file ./tests/fixtures/tests_samples/MRPC/dev.csv
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
--learning_rate=1e-4
--eval_steps=2
--warmup_steps=2
--seed=42
--max_seq_length=128
""".split()
with patch.object(_a , '''argv''' , _a ):
run_flax_glue.main()
_a : Any = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.75 )
@slow
def __lowercase ( self ) -> Dict:
_a : Tuple = self.get_auto_remove_tmp_dir()
_a : Tuple = F"""
run_clm_flax.py
--model_name_or_path distilgpt2
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--do_train
--do_eval
--block_size 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--num_train_epochs 2
--logging_steps 2 --eval_steps 2
--output_dir {tmp_dir}
--overwrite_output_dir
""".split()
with patch.object(_a , '''argv''' , _a ):
run_clm_flax.main()
_a : List[str] = get_results(_a )
self.assertLess(result['''eval_perplexity'''] , 1_0_0 )
@slow
def __lowercase ( self ) -> Optional[int]:
_a : str = self.get_auto_remove_tmp_dir()
_a : Optional[int] = F"""
run_summarization.py
--model_name_or_path t5-small
--train_file tests/fixtures/tests_samples/xsum/sample.json
--validation_file tests/fixtures/tests_samples/xsum/sample.json
--test_file tests/fixtures/tests_samples/xsum/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--num_train_epochs=3
--warmup_steps=8
--do_train
--do_eval
--do_predict
--learning_rate=2e-4
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
--predict_with_generate
""".split()
with patch.object(_a , '''argv''' , _a ):
run_summarization_flax.main()
_a : Optional[int] = get_results(_a , split='''test''' )
self.assertGreaterEqual(result['''test_rouge1'''] , 1_0 )
self.assertGreaterEqual(result['''test_rouge2'''] , 2 )
self.assertGreaterEqual(result['''test_rougeL'''] , 7 )
self.assertGreaterEqual(result['''test_rougeLsum'''] , 7 )
@slow
def __lowercase ( self ) -> Tuple:
_a : List[str] = self.get_auto_remove_tmp_dir()
_a : List[Any] = F"""
run_mlm.py
--model_name_or_path distilroberta-base
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--output_dir {tmp_dir}
--overwrite_output_dir
--max_seq_length 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--logging_steps 2 --eval_steps 2
--do_train
--do_eval
--num_train_epochs=1
""".split()
with patch.object(_a , '''argv''' , _a ):
run_mlm_flax.main()
_a : List[Any] = get_results(_a )
self.assertLess(result['''eval_perplexity'''] , 4_2 )
@slow
def __lowercase ( self ) -> Dict:
_a : Optional[Any] = self.get_auto_remove_tmp_dir()
_a : int = F"""
run_t5_mlm_flax.py
--model_name_or_path t5-small
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--do_train
--do_eval
--max_seq_length 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--num_train_epochs 2
--logging_steps 2 --eval_steps 2
--output_dir {tmp_dir}
--overwrite_output_dir
""".split()
with patch.object(_a , '''argv''' , _a ):
run_ta_mlm_flax.main()
_a : List[Any] = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.42 )
@slow
def __lowercase ( self ) -> Optional[Any]:
# with so little data distributed training needs more epochs to get the score on par with 0/1 gpu
_a : Any = 7 if get_gpu_count() > 1 else 2
_a : List[Any] = self.get_auto_remove_tmp_dir()
_a : List[Any] = F"""
run_flax_ner.py
--model_name_or_path bert-base-uncased
--train_file tests/fixtures/tests_samples/conll/sample.json
--validation_file tests/fixtures/tests_samples/conll/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--do_train
--do_eval
--warmup_steps=2
--learning_rate=2e-4
--logging_steps 2 --eval_steps 2
--per_device_train_batch_size=2
--per_device_eval_batch_size=2
--num_train_epochs={epochs}
--seed 7
""".split()
with patch.object(_a , '''argv''' , _a ):
run_flax_ner.main()
_a : Dict = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.75 )
self.assertGreaterEqual(result['''eval_f1'''] , 0.3 )
@slow
def __lowercase ( self ) -> Any:
_a : Optional[int] = self.get_auto_remove_tmp_dir()
_a : Union[str, Any] = F"""
run_qa.py
--model_name_or_path bert-base-uncased
--version_2_with_negative
--train_file tests/fixtures/tests_samples/SQUAD/sample.json
--validation_file tests/fixtures/tests_samples/SQUAD/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--num_train_epochs=3
--warmup_steps=2
--do_train
--do_eval
--logging_steps 2 --eval_steps 2
--learning_rate=2e-4
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
""".split()
with patch.object(_a , '''argv''' , _a ):
run_qa.main()
_a : Any = get_results(_a )
self.assertGreaterEqual(result['''eval_f1'''] , 3_0 )
self.assertGreaterEqual(result['''eval_exact'''] , 3_0 )
| 15
| 0
|
import itertools
import json
import os
import unittest
from transformers import AddedToken, LongformerTokenizer, LongformerTokenizerFast
from transformers.models.longformer.tokenization_longformer import VOCAB_FILES_NAMES
from transformers.testing_utils import require_tokenizers, slow
from ...test_tokenization_common import TokenizerTesterMixin
@require_tokenizers
class UpperCAmelCase_ ( __lowercase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = LongformerTokenizer
UpperCAmelCase__ : Optional[int] = True
UpperCAmelCase__ : List[str] = LongformerTokenizerFast
UpperCAmelCase__ : Any = True
def __lowercase ( self ) -> int:
super().setUp()
# Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
_a : Union[str, Any] = [
'''l''',
'''o''',
'''w''',
'''e''',
'''r''',
'''s''',
'''t''',
'''i''',
'''d''',
'''n''',
'''\u0120''',
'''\u0120l''',
'''\u0120n''',
'''\u0120lo''',
'''\u0120low''',
'''er''',
'''\u0120lowest''',
'''\u0120newer''',
'''\u0120wider''',
'''<unk>''',
]
_a : Optional[Any] = dict(zip(_a , range(len(_a ) ) ) )
_a : Dict = ['''#version: 0.2''', '''\u0120 l''', '''\u0120l o''', '''\u0120lo w''', '''e r''', '''''']
_a : List[Any] = {'''unk_token''': '''<unk>'''}
_a : Union[str, Any] = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['''vocab_file'''] )
_a : List[Any] = os.path.join(self.tmpdirname , VOCAB_FILES_NAMES['''merges_file'''] )
with open(self.vocab_file , '''w''' , encoding='''utf-8''' ) as fp:
fp.write(json.dumps(_a ) + '''\n''' )
with open(self.merges_file , '''w''' , encoding='''utf-8''' ) as fp:
fp.write('''\n'''.join(_a ) )
def __lowercase ( self , **_a ) -> int:
kwargs.update(self.special_tokens_map )
return self.tokenizer_class.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self , **_a ) -> Tuple:
kwargs.update(self.special_tokens_map )
return self.rust_tokenizer_class.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self , _a ) -> Union[str, Any]:
_a : Optional[Any] = '''lower newer'''
_a : Optional[int] = '''lower newer'''
return input_text, output_text
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = self.tokenizer_class(self.vocab_file , self.merges_file , **self.special_tokens_map )
_a : Any = '''lower newer'''
_a : str = ['''l''', '''o''', '''w''', '''er''', '''\u0120''', '''n''', '''e''', '''w''', '''er''']
_a : Optional[Any] = tokenizer.tokenize(_a ) # , add_prefix_space=True)
self.assertListEqual(_a , _a )
_a : Dict = tokens + [tokenizer.unk_token]
_a : Union[str, Any] = [0, 1, 2, 1_5, 1_0, 9, 3, 2, 1_5, 1_9]
self.assertListEqual(tokenizer.convert_tokens_to_ids(_a ) , _a )
def __lowercase ( self ) -> int:
_a : str = self.get_tokenizer()
self.assertListEqual(tokenizer.encode('''Hello world!''' , add_special_tokens=_a ) , [0, 3_1_4_1_4, 2_3_2, 3_2_8, 2] )
self.assertListEqual(
tokenizer.encode('''Hello world! cécé herlolip 418''' , add_special_tokens=_a ) , [0, 3_1_4_1_4, 2_3_2, 3_2_8, 7_4_0, 1_1_4_0, 1_2_6_9_5, 6_9, 4_6_0_7_8, 1_5_8_8, 2] , )
@slow
def __lowercase ( self ) -> List[str]:
_a : Optional[Any] = self.tokenizer_class.from_pretrained('''allenai/longformer-base-4096''' )
_a : Tuple = tokenizer.encode('''sequence builders''' , add_special_tokens=_a )
_a : Optional[int] = tokenizer.encode('''multi-sequence build''' , add_special_tokens=_a )
_a : List[str] = tokenizer.encode(
'''sequence builders''' , add_special_tokens=_a , add_prefix_space=_a )
_a : Optional[int] = tokenizer.encode(
'''sequence builders''' , '''multi-sequence build''' , add_special_tokens=_a , add_prefix_space=_a )
_a : Optional[Any] = tokenizer.build_inputs_with_special_tokens(_a )
_a : Optional[int] = tokenizer.build_inputs_with_special_tokens(_a , _a )
assert encoded_sentence == encoded_text_from_decode
assert encoded_pair == encoded_pair_from_decode
def __lowercase ( self ) -> Optional[int]:
_a : Dict = self.get_tokenizer()
_a : Any = '''Encode this sequence.'''
_a : Tuple = tokenizer.byte_encoder[''' '''.encode('''utf-8''' )[0]]
# Testing encoder arguments
_a : Any = tokenizer.encode(_a , add_special_tokens=_a , add_prefix_space=_a )
_a : List[str] = tokenizer.convert_ids_to_tokens(encoded[0] )[0]
self.assertNotEqual(_a , _a )
_a : str = tokenizer.encode(_a , add_special_tokens=_a , add_prefix_space=_a )
_a : str = tokenizer.convert_ids_to_tokens(encoded[0] )[0]
self.assertEqual(_a , _a )
tokenizer.add_special_tokens({'''bos_token''': '''<s>'''} )
_a : List[str] = tokenizer.encode(_a , add_special_tokens=_a )
_a : List[Any] = tokenizer.convert_ids_to_tokens(encoded[1] )[0]
self.assertNotEqual(_a , _a )
# Testing spaces after special tokens
_a : str = '''<mask>'''
tokenizer.add_special_tokens(
{'''mask_token''': AddedToken(_a , lstrip=_a , rstrip=_a )} ) # mask token has a left space
_a : Optional[int] = tokenizer.convert_tokens_to_ids(_a )
_a : Optional[Any] = '''Encode <mask> sequence'''
_a : Tuple = '''Encode <mask>sequence'''
_a : Optional[Any] = tokenizer.encode(_a )
_a : Optional[Any] = encoded.index(_a )
_a : str = tokenizer.convert_ids_to_tokens(encoded[mask_loc + 1] )[0]
self.assertEqual(_a , _a )
_a : Optional[Any] = tokenizer.encode(_a )
_a : int = encoded.index(_a )
_a : Dict = tokenizer.convert_ids_to_tokens(encoded[mask_loc + 1] )[0]
self.assertNotEqual(_a , _a )
def __lowercase ( self ) -> Any:
pass
def __lowercase ( self ) -> str:
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(F"""{tokenizer.__class__.__name__} ({pretrained_name})""" ):
_a : Tuple = self.rust_tokenizer_class.from_pretrained(_a , **_a )
_a : Union[str, Any] = self.tokenizer_class.from_pretrained(_a , **_a )
_a : List[Any] = '''A, <mask> AllenNLP sentence.'''
_a : str = tokenizer_r.encode_plus(_a , add_special_tokens=_a , return_token_type_ids=_a )
_a : List[str] = tokenizer_p.encode_plus(_a , add_special_tokens=_a , return_token_type_ids=_a )
# token_type_ids should put 0 everywhere
self.assertEqual(sum(tokens_r['''token_type_ids'''] ) , sum(tokens_p['''token_type_ids'''] ) )
# attention_mask should put 1 everywhere, so sum over length should be 1
self.assertEqual(
sum(tokens_r['''attention_mask'''] ) / len(tokens_r['''attention_mask'''] ) , sum(tokens_p['''attention_mask'''] ) / len(tokens_p['''attention_mask'''] ) , )
_a : Union[str, Any] = tokenizer_r.convert_ids_to_tokens(tokens_r['''input_ids'''] )
_a : List[str] = tokenizer_p.convert_ids_to_tokens(tokens_p['''input_ids'''] )
# Rust correctly handles the space before the mask while python doesnt
self.assertSequenceEqual(tokens_p['''input_ids'''] , [0, 2_5_0, 6, 5_0_2_6_4, 3_8_2_3, 4_8_7, 2_1_9_9_2, 3_6_4_5, 4, 2] )
self.assertSequenceEqual(tokens_r['''input_ids'''] , [0, 2_5_0, 6, 5_0_2_6_4, 3_8_2_3, 4_8_7, 2_1_9_9_2, 3_6_4_5, 4, 2] )
self.assertSequenceEqual(
_a , ['''<s>''', '''A''', ''',''', '''<mask>''', '''ĠAllen''', '''N''', '''LP''', '''Ġsentence''', '''.''', '''</s>'''] )
self.assertSequenceEqual(
_a , ['''<s>''', '''A''', ''',''', '''<mask>''', '''ĠAllen''', '''N''', '''LP''', '''Ġsentence''', '''.''', '''</s>'''] )
def __lowercase ( self ) -> Optional[int]:
for trim_offsets, add_prefix_space in itertools.product([True, False] , repeat=2 ):
_a : str = self.rust_tokenizer_class.from_pretrained(
self.tmpdirname , use_fast=_a , add_prefix_space=_a , trim_offsets=_a )
_a : int = json.loads(tokenizer_r.backend_tokenizer.pre_tokenizer.__getstate__() )
_a : Dict = json.loads(tokenizer_r.backend_tokenizer.post_processor.__getstate__() )
self.assertEqual(pre_tokenizer_state['''add_prefix_space'''] , _a )
self.assertEqual(post_processor_state['''add_prefix_space'''] , _a )
self.assertEqual(post_processor_state['''trim_offsets'''] , _a )
def __lowercase ( self ) -> Union[str, Any]:
# Test which aims to verify that the offsets are well adapted to the argument `add_prefix_space` and
# `trim_offsets`
for tokenizer, pretrained_name, kwargs in self.tokenizers_list:
with self.subTest(F"""{tokenizer.__class__.__name__} ({pretrained_name})""" ):
_a : List[str] = '''hello''' # `hello` is a token in the vocabulary of `pretrained_name`
_a : List[str] = F"""{text_of_1_token} {text_of_1_token}"""
_a : List[str] = self.rust_tokenizer_class.from_pretrained(
_a , use_fast=_a , add_prefix_space=_a , trim_offsets=_a )
_a : Optional[int] = tokenizer_r(_a , return_offsets_mapping=_a , add_special_tokens=_a )
self.assertEqual(encoding.offset_mapping[0] , (0, len(_a )) )
self.assertEqual(
encoding.offset_mapping[1] , (len(_a ) + 1, len(_a ) + 1 + len(_a )) , )
_a : Tuple = self.rust_tokenizer_class.from_pretrained(
_a , use_fast=_a , add_prefix_space=_a , trim_offsets=_a )
_a : int = tokenizer_r(_a , return_offsets_mapping=_a , add_special_tokens=_a )
self.assertEqual(encoding.offset_mapping[0] , (0, len(_a )) )
self.assertEqual(
encoding.offset_mapping[1] , (len(_a ) + 1, len(_a ) + 1 + len(_a )) , )
_a : Optional[int] = self.rust_tokenizer_class.from_pretrained(
_a , use_fast=_a , add_prefix_space=_a , trim_offsets=_a )
_a : str = tokenizer_r(_a , return_offsets_mapping=_a , add_special_tokens=_a )
self.assertEqual(encoding.offset_mapping[0] , (0, len(_a )) )
self.assertEqual(
encoding.offset_mapping[1] , (len(_a ), len(_a ) + 1 + len(_a )) , )
_a : Optional[Any] = self.rust_tokenizer_class.from_pretrained(
_a , use_fast=_a , add_prefix_space=_a , trim_offsets=_a )
_a : Union[str, Any] = tokenizer_r(_a , return_offsets_mapping=_a , add_special_tokens=_a )
self.assertEqual(encoding.offset_mapping[0] , (0, len(_a )) )
self.assertEqual(
encoding.offset_mapping[1] , (len(_a ), len(_a ) + 1 + len(_a )) , )
_a : List[str] = F""" {text}"""
# tokenizer_r = self.rust_tokenizer_class.from_pretrained(
# pretrained_name, use_fast=True, add_prefix_space=True, trim_offsets=True
# )
# encoding = tokenizer_r(text, return_offsets_mapping=True, add_special_tokens=False)
# self.assertEqual(encoding.offset_mapping[0], (1, 1 + len(text_of_1_token)))
# self.assertEqual(
# encoding.offset_mapping[1],
# (1 + len(text_of_1_token) + 1, 1 + len(text_of_1_token) + 1 + len(text_of_1_token)),
# )
_a : str = self.rust_tokenizer_class.from_pretrained(
_a , use_fast=_a , add_prefix_space=_a , trim_offsets=_a )
_a : str = tokenizer_r(_a , return_offsets_mapping=_a , add_special_tokens=_a )
self.assertEqual(encoding.offset_mapping[0] , (1, 1 + len(_a )) )
self.assertEqual(
encoding.offset_mapping[1] , (1 + len(_a ) + 1, 1 + len(_a ) + 1 + len(_a )) , )
_a : Optional[int] = self.rust_tokenizer_class.from_pretrained(
_a , use_fast=_a , add_prefix_space=_a , trim_offsets=_a )
_a : Dict = tokenizer_r(_a , return_offsets_mapping=_a , add_special_tokens=_a )
self.assertEqual(encoding.offset_mapping[0] , (0, 1 + len(_a )) )
self.assertEqual(
encoding.offset_mapping[1] , (1 + len(_a ), 1 + len(_a ) + 1 + len(_a )) , )
_a : List[str] = self.rust_tokenizer_class.from_pretrained(
_a , use_fast=_a , add_prefix_space=_a , trim_offsets=_a )
_a : str = tokenizer_r(_a , return_offsets_mapping=_a , add_special_tokens=_a )
self.assertEqual(encoding.offset_mapping[0] , (0, 1 + len(_a )) )
self.assertEqual(
encoding.offset_mapping[1] , (1 + len(_a ), 1 + len(_a ) + 1 + len(_a )) , )
| 357
|
import argparse
import os
import re
import packaging.version
a__ = '''examples/'''
a__ = {
'''examples''': (re.compile(R'''^check_min_version\("[^"]+"\)\s*$''', re.MULTILINE), '''check_min_version("VERSION")\n'''),
'''init''': (re.compile(R'''^__version__\s+=\s+"([^"]+)"\s*$''', re.MULTILINE), '''__version__ = "VERSION"\n'''),
'''setup''': (re.compile(R'''^(\s*)version\s*=\s*"[^"]+",''', re.MULTILINE), R'''\1version="VERSION",'''),
'''doc''': (re.compile(R'''^(\s*)release\s*=\s*"[^"]+"$''', re.MULTILINE), '''release = "VERSION"\n'''),
}
a__ = {
'''init''': '''src/transformers/__init__.py''',
'''setup''': '''setup.py''',
}
a__ = '''README.md'''
def __UpperCAmelCase ( __a : List[str] ,__a : int ,__a : Optional[Any] ) -> int:
"""simple docstring"""
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Tuple = f.read()
_a , _a : str = REPLACE_PATTERNS[pattern]
_a : List[str] = replace.replace('''VERSION''' ,__a )
_a : List[Any] = re_pattern.sub(__a ,__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.write(__a )
def __UpperCAmelCase ( __a : Any ) -> List[Any]:
"""simple docstring"""
for folder, directories, fnames in os.walk(__a ):
# Removing some of the folders with non-actively maintained examples from the walk
if "research_projects" in directories:
directories.remove('''research_projects''' )
if "legacy" in directories:
directories.remove('''legacy''' )
for fname in fnames:
if fname.endswith('''.py''' ):
update_version_in_file(os.path.join(__a ,__a ) ,__a ,pattern='''examples''' )
def __UpperCAmelCase ( __a : List[Any] ,__a : List[str]=False ) -> int:
"""simple docstring"""
for pattern, fname in REPLACE_FILES.items():
update_version_in_file(__a ,__a ,__a )
if not patch:
update_version_in_examples(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
_a : Optional[Any] = '''🤗 Transformers currently provides the following architectures'''
_a : str = '''1. Want to contribute a new model?'''
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Optional[int] = f.readlines()
# Find the start of the list.
_a : Optional[int] = 0
while not lines[start_index].startswith(_start_prompt ):
start_index += 1
start_index += 1
_a : List[Any] = start_index
# Update the lines in the model list.
while not lines[index].startswith(_end_prompt ):
if lines[index].startswith('''1.''' ):
_a : Tuple = lines[index].replace(
'''https://huggingface.co/docs/transformers/main/model_doc''' ,'''https://huggingface.co/docs/transformers/model_doc''' ,)
index += 1
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.writelines(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
with open(REPLACE_FILES['''init'''] ,'''r''' ) as f:
_a : Optional[Any] = f.read()
_a : Optional[Any] = REPLACE_PATTERNS['''init'''][0].search(__a ).groups()[0]
return packaging.version.parse(__a )
def __UpperCAmelCase ( __a : Dict=False ) -> str:
"""simple docstring"""
_a : Optional[Any] = get_version()
if patch and default_version.is_devrelease:
raise ValueError('''Can\'t create a patch version from the dev branch, checkout a released version!''' )
if default_version.is_devrelease:
_a : List[Any] = default_version.base_version
elif patch:
_a : str = F"""{default_version.major}.{default_version.minor}.{default_version.micro + 1}"""
else:
_a : List[str] = F"""{default_version.major}.{default_version.minor + 1}.0"""
# Now let's ask nicely if that's the right one.
_a : Dict = input(F"""Which version are you releasing? [{default_version}]""" )
if len(__a ) == 0:
_a : int = default_version
print(F"""Updating version to {version}.""" )
global_version_update(__a ,patch=__a )
if not patch:
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
def __UpperCAmelCase ( ) -> Tuple:
"""simple docstring"""
_a : str = get_version()
_a : int = F"""{current_version.major}.{current_version.minor + 1}.0.dev0"""
_a : List[Any] = current_version.base_version
# Check with the user we got that right.
_a : Union[str, Any] = input(F"""Which version are we developing now? [{dev_version}]""" )
if len(__a ) == 0:
_a : List[str] = dev_version
print(F"""Updating version to {version}.""" )
global_version_update(__a )
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--post_release''', action='''store_true''', help='''Whether this is pre or post release.''')
parser.add_argument('''--patch''', action='''store_true''', help='''Whether or not this is a patch release.''')
a__ = parser.parse_args()
if not args.post_release:
pre_release_work(patch=args.patch)
elif args.patch:
print('''Nothing to do after a patch :-)''')
else:
post_release_work()
| 15
| 0
|
import unittest
from transformers import AlbertConfig, is_torch_available
from transformers.models.auto import get_values
from transformers.testing_utils import require_torch, slow, torch_device
from ...test_configuration_common import ConfigTester
from ...test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
MODEL_FOR_PRETRAINING_MAPPING,
AlbertForMaskedLM,
AlbertForMultipleChoice,
AlbertForPreTraining,
AlbertForQuestionAnswering,
AlbertForSequenceClassification,
AlbertForTokenClassification,
AlbertModel,
)
from transformers.models.albert.modeling_albert import ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self , _a , _a=1_3 , _a=7 , _a=True , _a=True , _a=True , _a=True , _a=9_9 , _a=1_6 , _a=3_6 , _a=6 , _a=6 , _a=6 , _a=3_7 , _a="gelu" , _a=0.1 , _a=0.1 , _a=5_1_2 , _a=1_6 , _a=2 , _a=0.02 , _a=3 , _a=4 , _a=None , ) -> Optional[Any]:
_a : Union[str, Any] = parent
_a : int = batch_size
_a : Dict = seq_length
_a : Union[str, Any] = is_training
_a : Union[str, Any] = use_input_mask
_a : int = use_token_type_ids
_a : str = use_labels
_a : int = vocab_size
_a : Optional[Any] = embedding_size
_a : str = hidden_size
_a : Optional[int] = num_hidden_layers
_a : str = num_hidden_groups
_a : List[Any] = num_attention_heads
_a : List[Any] = intermediate_size
_a : List[str] = hidden_act
_a : int = hidden_dropout_prob
_a : Optional[int] = attention_probs_dropout_prob
_a : List[str] = max_position_embeddings
_a : List[Any] = type_vocab_size
_a : List[str] = type_sequence_label_size
_a : Optional[Any] = initializer_range
_a : Tuple = num_labels
_a : str = num_choices
_a : Tuple = scope
def __lowercase ( self ) -> Optional[Any]:
_a : List[str] = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
_a : Optional[Any] = None
if self.use_input_mask:
_a : Dict = random_attention_mask([self.batch_size, self.seq_length] )
_a : List[Any] = None
if self.use_token_type_ids:
_a : Optional[Any] = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
_a : List[str] = None
_a : Tuple = None
_a : Any = None
if self.use_labels:
_a : int = ids_tensor([self.batch_size] , self.type_sequence_label_size )
_a : Any = ids_tensor([self.batch_size, self.seq_length] , self.num_labels )
_a : List[str] = ids_tensor([self.batch_size] , self.num_choices )
_a : List[str] = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def __lowercase ( self ) -> Tuple:
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 __lowercase ( self , _a , _a , _a , _a , _a , _a , _a ) -> Dict:
_a : Dict = AlbertModel(config=_a )
model.to(_a )
model.eval()
_a : List[Any] = model(_a , attention_mask=_a , token_type_ids=_a )
_a : Union[str, Any] = model(_a , token_type_ids=_a )
_a : str = model(_a )
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 __lowercase ( self , _a , _a , _a , _a , _a , _a , _a ) -> List[Any]:
_a : Any = AlbertForPreTraining(config=_a )
model.to(_a )
model.eval()
_a : Dict = model(
_a , attention_mask=_a , token_type_ids=_a , labels=_a , sentence_order_label=_a , )
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 __lowercase ( self , _a , _a , _a , _a , _a , _a , _a ) -> Tuple:
_a : Dict = AlbertForMaskedLM(config=_a )
model.to(_a )
model.eval()
_a : Tuple = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a , _a ) -> Optional[int]:
_a : Optional[Any] = AlbertForQuestionAnswering(config=_a )
model.to(_a )
model.eval()
_a : Tuple = model(
_a , attention_mask=_a , token_type_ids=_a , start_positions=_a , end_positions=_a , )
self.parent.assertEqual(result.start_logits.shape , (self.batch_size, self.seq_length) )
self.parent.assertEqual(result.end_logits.shape , (self.batch_size, self.seq_length) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a , _a ) -> str:
_a : Tuple = self.num_labels
_a : Dict = AlbertForSequenceClassification(_a )
model.to(_a )
model.eval()
_a : List[Any] = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_labels) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a , _a ) -> List[Any]:
_a : Optional[int] = self.num_labels
_a : Union[str, Any] = AlbertForTokenClassification(config=_a )
model.to(_a )
model.eval()
_a : Optional[int] = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a , _a ) -> Optional[int]:
_a : Dict = self.num_choices
_a : int = AlbertForMultipleChoice(config=_a )
model.to(_a )
model.eval()
_a : Union[str, Any] = input_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
_a : str = token_type_ids.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
_a : Optional[Any] = input_mask.unsqueeze(1 ).expand(-1 , self.num_choices , -1 ).contiguous()
_a : Tuple = model(
_a , attention_mask=_a , token_type_ids=_a , labels=_a , )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.num_choices) )
def __lowercase ( self ) -> Union[str, Any]:
_a : List[str] = self.prepare_config_and_inputs()
(
_a
) : Optional[int] = config_and_inputs
_a : List[Any] = {'''input_ids''': input_ids, '''token_type_ids''': token_type_ids, '''attention_mask''': input_mask}
return config, inputs_dict
@require_torch
class UpperCAmelCase_ ( __lowercase , __lowercase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : Dict = (
(
AlbertModel,
AlbertForPreTraining,
AlbertForMaskedLM,
AlbertForMultipleChoice,
AlbertForSequenceClassification,
AlbertForTokenClassification,
AlbertForQuestionAnswering,
)
if is_torch_available()
else ()
)
UpperCAmelCase__ : Optional[int] = (
{
"feature-extraction": AlbertModel,
"fill-mask": AlbertForMaskedLM,
"question-answering": AlbertForQuestionAnswering,
"text-classification": AlbertForSequenceClassification,
"token-classification": AlbertForTokenClassification,
"zero-shot": AlbertForSequenceClassification,
}
if is_torch_available()
else {}
)
UpperCAmelCase__ : Dict = True
def __lowercase ( self , _a , _a , _a=False ) -> List[str]:
_a : int = super()._prepare_for_class(_a , _a , return_labels=_a )
if return_labels:
if model_class in get_values(_a ):
_a : Any = torch.zeros(
(self.model_tester.batch_size, self.model_tester.seq_length) , dtype=torch.long , device=_a )
_a : str = torch.zeros(
self.model_tester.batch_size , dtype=torch.long , device=_a )
return inputs_dict
def __lowercase ( self ) -> str:
_a : Tuple = AlbertModelTester(self )
_a : int = ConfigTester(self , config_class=_a , hidden_size=3_7 )
def __lowercase ( self ) -> Dict:
self.config_tester.run_common_tests()
def __lowercase ( self ) -> Any:
_a : Any = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*_a )
def __lowercase ( self ) -> str:
_a : Union[str, Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_pretraining(*_a )
def __lowercase ( self ) -> int:
_a : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_masked_lm(*_a )
def __lowercase ( self ) -> Optional[int]:
_a : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_multiple_choice(*_a )
def __lowercase ( self ) -> Union[str, Any]:
_a : int = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_question_answering(*_a )
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_for_sequence_classification(*_a )
def __lowercase ( self ) -> Any:
_a : Optional[int] = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
_a : Any = type
self.model_tester.create_and_check_model(*_a )
@slow
def __lowercase ( self ) -> Dict:
for model_name in ALBERT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
_a : List[str] = AlbertModel.from_pretrained(_a )
self.assertIsNotNone(_a )
@require_torch
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
@slow
def __lowercase ( self ) -> int:
_a : Optional[int] = AlbertModel.from_pretrained('''albert-base-v2''' )
_a : int = torch.tensor([[0, 3_4_5, 2_3_2, 3_2_8, 7_4_0, 1_4_0, 1_6_9_5, 6_9, 6_0_7_8, 1_5_8_8, 2]] )
_a : int = torch.tensor([[0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]] )
with torch.no_grad():
_a : List[str] = model(_a , attention_mask=_a )[0]
_a : List[str] = torch.Size((1, 1_1, 7_6_8) )
self.assertEqual(output.shape , _a )
_a : List[Any] = 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] , _a , atol=1e-4 ) )
| 358
|
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
if n == 1 or not isinstance(__a ,__a ):
return 0
elif n == 2:
return 1
else:
_a : Any = [0, 1]
for i in range(2 ,n + 1 ):
sequence.append(sequence[i - 1] + sequence[i - 2] )
return sequence[n]
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
_a : Any = 0
_a : Dict = 2
while digits < n:
index += 1
_a : Dict = len(str(fibonacci(__a ) ) )
return index
def __UpperCAmelCase ( __a : int = 1_000 ) -> int:
"""simple docstring"""
return fibonacci_digits_index(__a )
if __name__ == "__main__":
print(solution(int(str(input()).strip())))
| 15
| 0
|
import os
import re
import unicodedata
from shutil import copyfile
from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union
import sentencepiece as spm
from ...tokenization_utils import PreTrainedTokenizer
from ...utils import is_torch_available, logging
if is_torch_available():
import torch
if TYPE_CHECKING:
from transformers.pipelines.conversational import Conversation
a__ = logging.get_logger(__name__)
a__ = {'''vocab_file''': '''spiece.model'''}
a__ = {
'''vocab_file''': {
'''AI-Sweden/gpt-sw3-126m''': '''https://huggingface.co/AI-Sweden/gpt-sw3-126m/resolve/main/spiece.model''',
'''AI-Sweden/gpt-sw3-350m''': '''https://huggingface.co/AI-Sweden/gpt-sw3-350m/resolve/main/spiece.model''',
'''AI-Sweden/gpt-sw3-1.6b''': '''https://huggingface.co/AI-Sweden/gpt-sw3-1.6b/resolve/main/spiece.model''',
'''AI-Sweden/gpt-sw3-6.7b''': '''https://huggingface.co/AI-Sweden/gpt-sw3-6.7b/resolve/main/spiece.model''',
'''AI-Sweden/gpt-sw3-20b''': '''https://huggingface.co/AI-Sweden/gpt-sw3-20b/resolve/main/spiece.model''',
}
}
a__ = {
'''AI-Sweden/gpt-sw3-126m''': 2048,
'''AI-Sweden/gpt-sw3-350m''': 2048,
'''AI-Sweden/gpt-sw3-1.6b''': 2048,
'''AI-Sweden/gpt-sw3-6.7b''': 2048,
'''AI-Sweden/gpt-sw3-20b''': 2048,
}
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : str = VOCAB_FILES_NAMES
UpperCAmelCase__ : Union[str, Any] = PRETRAINED_VOCAB_FILES_MAP
UpperCAmelCase__ : Optional[int] = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
UpperCAmelCase__ : Optional[Any] = ["input_ids", "attention_mask"]
def __init__( self , _a , _a=False , _a=False , _a=False , _a=None , _a=None , _a=None , _a=None , _a = None , **_a , ) -> None:
_a : Optional[int] = {} if sp_model_kwargs is None else sp_model_kwargs
_a : Union[str, Any] = kwargs.get('''name_or_path''' )
if name_or_path is None:
logger.warning(
'''name_or_path not provided, will work for all GPTSw3 models except gpt-sw3-7b,'''
''' you are testing the model, this can safely be ignored''' )
_a : Union[str, Any] = '''None'''
# Default definitions for our 2 tokenizer versions, with None-checks to enable proper testing
_a : Any = '''<|endoftext|>''' if eos_token is None else eos_token
_a : Any = '''<unk>''' if unk_token is None else unk_token
if "gpt-sw3-7b" in name_or_path:
_a : Union[str, Any] = unk_token if pad_token is None else pad_token
_a : Tuple = eos_token if bos_token is None else bos_token
else:
_a : Optional[Any] = '''<pad>''' if pad_token is None else pad_token
_a : Union[str, Any] = '''<s>''' if bos_token is None else bos_token
super().__init__(
do_lower_case=_a , remove_space=_a , keep_accents=_a , bos_token=_a , eos_token=_a , unk_token=_a , pad_token=_a , sp_model_kwargs=self.sp_model_kwargs , **_a , )
_a : Optional[Any] = do_lower_case
_a : Any = remove_space
_a : List[Any] = keep_accents
_a : Tuple = vocab_file
_a : Tuple = spm.SentencePieceProcessor(**self.sp_model_kwargs )
self.sp_model.Load(_a )
# Used for whitespace normalization in input texts
# fmt : off
_a : Union[str, Any] = {''' ''', ''' ''', ''' ''', ''' ''', ''' ''', ''' ''', ''' ''', ''' ''', ''' ''', ''' ''', '''''', ''''''}
# fmt : on
# Regular expression to remove non-printing characters (e.g. some unicode control chars) in preprocessing
_a : Any = re.compile(
F"""[{''.join(map(_a , list(range(0 , 9 ) ) + list(range(1_1 , 3_2 ) ) + list(range(1_2_7 , 1_6_0 ) ) + [1_6_0, 1_7_3, 8_2_0_3] ) )}]""" )
def __getstate__( self ) -> List[str]:
_a : Optional[Any] = self.__dict__.copy()
_a : Any = None
return state
def __setstate__( self , _a ) -> Dict:
_a : int = d
# for backward compatibility
if not hasattr(self , '''sp_model_kwargs''' ):
_a : Tuple = {}
_a : Any = spm.SentencePieceProcessor(**self.sp_model_kwargs )
self.sp_model.Load(self.vocab_file )
@property
# Copied from transformers.models.albert.tokenization_albert.AlbertTokenizer.vocab_size
def __lowercase ( self ) -> int:
return len(self.sp_model )
def __lowercase ( self , _a ) -> str:
_a : Optional[int] = self.non_printing_characters_re.sub('''''' , _a )
# Normalize whitespaces
_a : Dict = ''''''.join([char if char not in self.whitespaces else ''' ''' for char in text] )
# NFC Unicode normalization
_a : Optional[Any] = unicodedata.normalize('''NFC''' , _a )
return text
def __lowercase ( self , _a , **_a ) -> List[str]:
_a : Optional[Any] = self.preprocess_text(_a )
return self.sp_model.encode(_a , out_type=_a )
def __lowercase ( self , _a ) -> int:
return self.sp_model.PieceToId(_a )
def __lowercase ( self , _a ) -> str:
return self.sp_model.IdToPiece(_a )
@staticmethod
def __lowercase ( _a ) -> str:
return out_string
def __lowercase ( self , _a ) -> str:
_a : int = []
_a : Any = ''''''
_a : Tuple = False
for token in tokens:
# make sure that special tokens are not decoded using sentencepiece model
if token in self.all_special_tokens:
# TODO: Check if this is needed, as it ensures that decode(encode(doc)) != doc by adding extra whitespace in the decoded document
if not prev_is_special:
out_string += " "
out_string += self.sp_model.decode(_a ) + token
_a : Tuple = True
_a : List[str] = []
else:
current_sub_tokens.append(_a )
_a : Any = False
out_string += self.sp_model.decode(_a )
return out_string
def __lowercase ( self ) -> Dict[str, int]:
_a : List[str] = {self.convert_ids_to_tokens(_a ): i for i in range(self.vocab_size )}
vocab.update(self.added_tokens_encoder )
return vocab
def __lowercase ( self , _a , _a = None ) -> Tuple[str]:
if not os.path.isdir(_a ):
logger.error(F"""Vocabulary path ({save_directory}) should be a directory""" )
return
_a : Any = os.path.join(
_a , (filename_prefix + '''-''' if filename_prefix else '''''') + VOCAB_FILES_NAMES['''vocab_file'''] )
if os.path.abspath(self.vocab_file ) != os.path.abspath(_a ) and os.path.isfile(self.vocab_file ):
copyfile(self.vocab_file , _a )
elif not os.path.isfile(self.vocab_file ):
with open(_a , '''wb''' ) as fi:
_a : Optional[int] = self.sp_model.serialized_model_proto()
fi.write(_a )
return (out_vocab_file,)
def __lowercase ( self , _a , _a = False ) -> Union[List[int], List[List[int]], "torch.Tensor"]:
if isinstance(_a , _a ):
_a : Tuple = self.preprocess_text(_a )
_a : int = self.sp_model.encode(_a )
else:
_a : Dict = [self.preprocess_text(_a ) for t in text]
_a : Optional[int] = self.sp_model.encode(_a )
if return_tensors is True or return_tensors == "pt":
_a : List[str] = torch.tensor(_a )
return token_ids
def __lowercase ( self , _a ) -> str:
return self.sp_model.decode(_a )
def __lowercase ( self , _a ) -> List[int]:
_a : List[Any] = [F"""User: {text}""" if is_user else F"""Bot: {text}""" for is_user, text in conversation.iter_texts()]
_a : Optional[int] = (
F"""{self.eos_token}{self.bos_token}""" + F"""{self.bos_token}""".join(_a ) + F"""{self.bos_token}Bot:"""
)
return self.encode(text=_a )
| 359
|
from sklearn.metrics import fa_score, matthews_corrcoef
import datasets
from .record_evaluation import evaluate as evaluate_record
a__ = '''\
@article{wang2019superglue,
title={SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems},
author={Wang, Alex and Pruksachatkun, Yada and Nangia, Nikita and Singh, Amanpreet and Michael, Julian and Hill, Felix and Levy, Omer and Bowman, Samuel R},
journal={arXiv preprint arXiv:1905.00537},
year={2019}
}
'''
a__ = '''\
SuperGLUE (https://super.gluebenchmark.com/) is a new benchmark styled after
GLUE with a new set of more difficult language understanding tasks, improved
resources, and a new public leaderboard.
'''
a__ = '''
Compute SuperGLUE evaluation metric associated to each SuperGLUE dataset.
Args:
predictions: list of predictions to score. Depending on the SuperGlUE subset:
- for \'record\': list of question-answer dictionaries with the following keys:
- \'idx\': index of the question as specified by the dataset
- \'prediction_text\': the predicted answer text
- for \'multirc\': list of question-answer dictionaries with the following keys:
- \'idx\': index of the question-answer pair as specified by the dataset
- \'prediction\': the predicted answer label
- otherwise: list of predicted labels
references: list of reference labels. Depending on the SuperGLUE subset:
- for \'record\': list of question-answers dictionaries with the following keys:
- \'idx\': index of the question as specified by the dataset
- \'answers\': list of possible answers
- otherwise: list of reference labels
Returns: depending on the SuperGLUE subset:
- for \'record\':
- \'exact_match\': Exact match between answer and gold answer
- \'f1\': F1 score
- for \'multirc\':
- \'exact_match\': Exact match between answer and gold answer
- \'f1_m\': Per-question macro-F1 score
- \'f1_a\': Average F1 score over all answers
- for \'axb\':
\'matthews_correlation\': Matthew Correlation
- for \'cb\':
- \'accuracy\': Accuracy
- \'f1\': F1 score
- for all others:
- \'accuracy\': Accuracy
Examples:
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'copa\') # any of ["copa", "rte", "wic", "wsc", "wsc.fixed", "boolq", "axg"]
>>> predictions = [0, 1]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'accuracy\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'cb\')
>>> predictions = [0, 1]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'accuracy\': 1.0, \'f1\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'record\')
>>> predictions = [{\'idx\': {\'passage\': 0, \'query\': 0}, \'prediction_text\': \'answer\'}]
>>> references = [{\'idx\': {\'passage\': 0, \'query\': 0}, \'answers\': [\'answer\', \'another_answer\']}]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 1.0, \'f1\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'multirc\')
>>> predictions = [{\'idx\': {\'answer\': 0, \'paragraph\': 0, \'question\': 0}, \'prediction\': 0}, {\'idx\': {\'answer\': 1, \'paragraph\': 2, \'question\': 3}, \'prediction\': 1}]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 1.0, \'f1_m\': 1.0, \'f1_a\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'axb\')
>>> references = [0, 1]
>>> predictions = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'matthews_correlation\': 1.0}
'''
def __UpperCAmelCase ( __a : int ,__a : List[str] ) -> Optional[Any]:
"""simple docstring"""
return float((preds == labels).mean() )
def __UpperCAmelCase ( __a : List[Any] ,__a : Union[str, Any] ,__a : List[str]="binary" ) -> Optional[int]:
"""simple docstring"""
_a : List[str] = simple_accuracy(__a ,__a )
_a : Any = float(fa_score(y_true=__a ,y_pred=__a ,average=__a ) )
return {
"accuracy": acc,
"f1": fa,
}
def __UpperCAmelCase ( __a : Optional[Any] ,__a : str ) -> List[Any]:
"""simple docstring"""
_a : Union[str, Any] = {}
for id_pred, label in zip(__a ,__a ):
_a : Optional[int] = F"""{id_pred['idx']['paragraph']}-{id_pred['idx']['question']}"""
_a : Optional[Any] = id_pred['''prediction''']
if question_id in question_map:
question_map[question_id].append((pred, label) )
else:
_a : str = [(pred, label)]
_a , _a : Any = [], []
for question, preds_labels in question_map.items():
_a , _a : Any = zip(*__a )
_a : List[Any] = fa_score(y_true=__a ,y_pred=__a ,average='''macro''' )
fas.append(__a )
_a : List[str] = int(sum(pred == label for pred, label in preds_labels ) == len(__a ) )
ems.append(__a )
_a : List[str] = float(sum(__a ) / len(__a ) )
_a : str = sum(__a ) / len(__a )
_a : Optional[int] = float(fa_score(y_true=__a ,y_pred=[id_pred['''prediction'''] for id_pred in ids_preds] ) )
return {"exact_match": em, "f1_m": fa_m, "f1_a": fa_a}
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> List[Any]:
if self.config_name not in [
"boolq",
"cb",
"copa",
"multirc",
"record",
"rte",
"wic",
"wsc",
"wsc.fixed",
"axb",
"axg",
]:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["boolq", "cb", "copa", "multirc", "record", "rte", "wic", "wsc", "wsc.fixed", "axb", "axg",]''' )
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(self._get_feature_types() ) , codebase_urls=[] , reference_urls=[] , format='''numpy''' if not self.config_name == '''record''' and not self.config_name == '''multirc''' else None , )
def __lowercase ( self ) -> Any:
if self.config_name == "record":
return {
"predictions": {
"idx": {
"passage": datasets.Value('''int64''' ),
"query": datasets.Value('''int64''' ),
},
"prediction_text": datasets.Value('''string''' ),
},
"references": {
"idx": {
"passage": datasets.Value('''int64''' ),
"query": datasets.Value('''int64''' ),
},
"answers": datasets.Sequence(datasets.Value('''string''' ) ),
},
}
elif self.config_name == "multirc":
return {
"predictions": {
"idx": {
"answer": datasets.Value('''int64''' ),
"paragraph": datasets.Value('''int64''' ),
"question": datasets.Value('''int64''' ),
},
"prediction": datasets.Value('''int64''' ),
},
"references": datasets.Value('''int64''' ),
}
else:
return {
"predictions": datasets.Value('''int64''' ),
"references": datasets.Value('''int64''' ),
}
def __lowercase ( self , _a , _a ) -> Optional[Any]:
if self.config_name == "axb":
return {"matthews_correlation": matthews_corrcoef(_a , _a )}
elif self.config_name == "cb":
return acc_and_fa(_a , _a , fa_avg='''macro''' )
elif self.config_name == "record":
_a : Any = [
{
'''qas''': [
{'''id''': ref['''idx''']['''query'''], '''answers''': [{'''text''': ans} for ans in ref['''answers''']]}
for ref in references
]
}
]
_a : Any = {pred['''idx''']['''query''']: pred['''prediction_text'''] for pred in predictions}
return evaluate_record(_a , _a )[0]
elif self.config_name == "multirc":
return evaluate_multirc(_a , _a )
elif self.config_name in ["copa", "rte", "wic", "wsc", "wsc.fixed", "boolq", "axg"]:
return {"accuracy": simple_accuracy(_a , _a )}
else:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["boolq", "cb", "copa", "multirc", "record", "rte", "wic", "wsc", "wsc.fixed", "axb", "axg",]''' )
| 15
| 0
|
from typing import List
import numpy as np
def __UpperCAmelCase ( __a : dict ) -> int:
"""simple docstring"""
_a : int = {key: len(__a ) for key, value in gen_kwargs.items() if isinstance(__a ,__a )}
if len(set(lists_lengths.values() ) ) > 1:
raise RuntimeError(
(
'''Sharding is ambiguous for this dataset: '''
+ '''we found several data sources lists of different lengths, and we don\'t know over which list we should parallelize:\n'''
+ '''\n'''.join(F"""\t- key {key} has length {length}""" for key, length in lists_lengths.items() )
+ '''\nTo fix this, check the \'gen_kwargs\' and make sure to use lists only for data sources, '''
+ '''and use tuples otherwise. In the end there should only be one single list, or several lists with the same length.'''
) )
_a : Union[str, Any] = max(lists_lengths.values() ,default=0 )
return max(1 ,__a )
def __UpperCAmelCase ( __a : int ,__a : int ) -> List[range]:
"""simple docstring"""
_a : List[str] = []
for group_idx in range(__a ):
_a : str = num_shards // max_num_jobs + (group_idx < (num_shards % max_num_jobs))
if num_shards_to_add == 0:
break
_a : int = shards_indices_per_group[-1].stop if shards_indices_per_group else 0
_a : Dict = range(__a ,start + num_shards_to_add )
shards_indices_per_group.append(__a )
return shards_indices_per_group
def __UpperCAmelCase ( __a : dict ,__a : int ) -> List[dict]:
"""simple docstring"""
_a : Dict = _number_of_shards_in_gen_kwargs(__a )
if num_shards == 1:
return [dict(__a )]
else:
_a : Optional[int] = _distribute_shards(num_shards=__a ,max_num_jobs=__a )
return [
{
key: [value[shard_idx] for shard_idx in shard_indices_per_group[group_idx]]
if isinstance(__a ,__a )
else value
for key, value in gen_kwargs.items()
}
for group_idx in range(len(__a ) )
]
def __UpperCAmelCase ( __a : List[dict] ) -> dict:
"""simple docstring"""
return {
key: [value for gen_kwargs in gen_kwargs_list for value in gen_kwargs[key]]
if isinstance(gen_kwargs_list[0][key] ,__a )
else gen_kwargs_list[0][key]
for key in gen_kwargs_list[0]
}
def __UpperCAmelCase ( __a : np.random.Generator ,__a : dict ) -> dict:
"""simple docstring"""
_a : Dict = {len(__a ) for value in gen_kwargs.values() if isinstance(__a ,__a )}
_a : List[Any] = {}
for size in list_sizes:
_a : Dict = list(range(__a ) )
rng.shuffle(indices_per_size[size] )
# Now let's copy the gen_kwargs and shuffle the lists based on their sizes
_a : List[Any] = dict(__a )
for key, value in shuffled_kwargs.items():
if isinstance(__a ,__a ):
_a : Tuple = [value[i] for i in indices_per_size[len(__a )]]
return shuffled_kwargs
| 360
|
import numpy as np
def __UpperCAmelCase ( __a : np.ndarray ,__a : np.ndarray ,__a : float = 1E-12 ,__a : int = 100 ,) -> tuple[float, np.ndarray]:
"""simple docstring"""
assert np.shape(__a )[0] == np.shape(__a )[1]
# Ensure proper dimensionality.
assert np.shape(__a )[0] == np.shape(__a )[0]
# Ensure inputs are either both complex or both real
assert np.iscomplexobj(__a ) == np.iscomplexobj(__a )
_a : List[str] = np.iscomplexobj(__a )
if is_complex:
# Ensure complex input_matrix is Hermitian
assert np.array_equal(__a ,input_matrix.conj().T )
# Set convergence to False. Will define convergence when we exceed max_iterations
# or when we have small changes from one iteration to next.
_a : List[str] = False
_a : List[str] = 0
_a : Tuple = 0
_a : str = 1E12
while not convergence:
# Multiple matrix by the vector.
_a : str = np.dot(__a ,__a )
# Normalize the resulting output vector.
_a : List[Any] = w / np.linalg.norm(__a )
# Find rayleigh quotient
# (faster than usual b/c we know vector is normalized already)
_a : Dict = vector.conj().T if is_complex else vector.T
_a : Tuple = np.dot(__a ,np.dot(__a ,__a ) )
# Check convergence.
_a : List[str] = np.abs(lambda_ - lambda_previous ) / lambda_
iterations += 1
if error <= error_tol or iterations >= max_iterations:
_a : Dict = True
_a : str = lambda_
if is_complex:
_a : Tuple = np.real(lambda_ )
return lambda_, vector
def __UpperCAmelCase ( ) -> None:
"""simple docstring"""
_a : List[str] = np.array([[41, 4, 20], [4, 26, 30], [20, 30, 50]] )
_a : int = np.array([41, 4, 20] )
_a : Optional[Any] = real_input_matrix.astype(np.complexaaa )
_a : int = np.triu(1j * complex_input_matrix ,1 )
complex_input_matrix += imag_matrix
complex_input_matrix += -1 * imag_matrix.T
_a : Union[str, Any] = np.array([41, 4, 20] ).astype(np.complexaaa )
for problem_type in ["real", "complex"]:
if problem_type == "real":
_a : Optional[int] = real_input_matrix
_a : Union[str, Any] = real_vector
elif problem_type == "complex":
_a : str = complex_input_matrix
_a : str = complex_vector
# Our implementation.
_a , _a : Optional[Any] = power_iteration(__a ,__a )
# Numpy implementation.
# Get eigenvalues and eigenvectors using built-in numpy
# eigh (eigh used for symmetric or hermetian matrices).
_a , _a : List[str] = np.linalg.eigh(__a )
# Last eigenvalue is the maximum one.
_a : Tuple = eigen_values[-1]
# Last column in this matrix is eigenvector corresponding to largest eigenvalue.
_a : List[Any] = eigen_vectors[:, -1]
# Check our implementation and numpy gives close answers.
assert np.abs(eigen_value - eigen_value_max ) <= 1E-6
# Take absolute values element wise of each eigenvector.
# as they are only unique to a minus sign.
assert np.linalg.norm(np.abs(__a ) - np.abs(__a ) ) <= 1E-6
if __name__ == "__main__":
import doctest
doctest.testmod()
test_power_iteration()
| 15
| 0
|
def __UpperCAmelCase ( __a : List[Any] ) -> Tuple:
"""simple docstring"""
_a : str = []
_a : Union[str, Any] = set({'''(''', '''[''', '''{'''} )
_a : List[Any] = set({''')''', ''']''', '''}'''} )
_a : str = {'''{''': '''}''', '''[''': ''']''', '''(''': ''')'''}
for i in range(len(__a ) ):
if s[i] in open_brackets:
stack.append(s[i] )
elif s[i] in closed_brackets and (
len(__a ) == 0 or (len(__a ) > 0 and open_to_closed[stack.pop()] != s[i])
):
return False
return len(__a ) == 0
def __UpperCAmelCase ( ) -> Union[str, Any]:
"""simple docstring"""
_a : Any = input('''Enter sequence of brackets: ''' )
if is_balanced(__a ):
print(__a ,'''is balanced''' )
else:
print(__a ,'''is not balanced''' )
if __name__ == "__main__":
main()
| 361
|
import itertools
from dataclasses import dataclass
from typing import Optional
import pandas as pd
import pyarrow as pa
import datasets
from datasets.table import table_cast
@dataclass
class UpperCAmelCase_ ( datasets.BuilderConfig ):
"""simple docstring"""
UpperCAmelCase__ : Optional[datasets.Features] = None
class UpperCAmelCase_ ( datasets.ArrowBasedBuilder ):
"""simple docstring"""
UpperCAmelCase__ : Any = PandasConfig
def __lowercase ( self ) -> Any:
return datasets.DatasetInfo(features=self.config.features )
def __lowercase ( self , _a ) -> List[Any]:
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}""" )
_a : str = dl_manager.download_and_extract(self.config.data_files )
if isinstance(_a , (str, list, tuple) ):
_a : Dict = data_files
if isinstance(_a , _a ):
_a : Dict = [files]
# Use `dl_manager.iter_files` to skip hidden files in an extracted archive
_a : int = [dl_manager.iter_files(_a ) for file in files]
return [datasets.SplitGenerator(name=datasets.Split.TRAIN , gen_kwargs={'''files''': files} )]
_a : Optional[Any] = []
for split_name, files in data_files.items():
if isinstance(_a , _a ):
_a : List[str] = [files]
# Use `dl_manager.iter_files` to skip hidden files in an extracted archive
_a : Any = [dl_manager.iter_files(_a ) for file in files]
splits.append(datasets.SplitGenerator(name=_a , gen_kwargs={'''files''': files} ) )
return splits
def __lowercase ( self , _a ) -> pa.Table:
if self.config.features is not None:
# more expensive cast to support nested features with keys in a different order
# allows str <-> int/float or str to Audio for example
_a : Optional[Any] = table_cast(_a , self.config.features.arrow_schema )
return pa_table
def __lowercase ( self , _a ) -> List[str]:
for i, file in enumerate(itertools.chain.from_iterable(_a ) ):
with open(_a , '''rb''' ) as f:
_a : str = pa.Table.from_pandas(pd.read_pickle(_a ) )
yield i, self._cast_table(_a )
| 15
| 0
|
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a__ = {'''configuration_vit_msn''': ['''VIT_MSN_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''ViTMSNConfig''']}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''VIT_MSN_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''ViTMSNModel''',
'''ViTMSNForImageClassification''',
'''ViTMSNPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_vit_msn import VIT_MSN_PRETRAINED_CONFIG_ARCHIVE_MAP, ViTMSNConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_vit_msn import (
VIT_MSN_PRETRAINED_MODEL_ARCHIVE_LIST,
ViTMSNForImageClassification,
ViTMSNModel,
ViTMSNPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 362
|
def __UpperCAmelCase ( __a : int ,__a : int ,__a : int ) -> int:
"""simple docstring"""
if exponent == 1:
return base
if exponent % 2 == 0:
_a : List[Any] = _modexpt(__a ,exponent // 2 ,__a ) % modulo_value
return (x * x) % modulo_value
else:
return (base * _modexpt(__a ,exponent - 1 ,__a )) % modulo_value
def __UpperCAmelCase ( __a : int = 1_777 ,__a : int = 1_855 ,__a : int = 8 ) -> int:
"""simple docstring"""
_a : List[Any] = base
for _ in range(1 ,__a ):
_a : Any = _modexpt(__a ,__a ,10**digits )
return result
if __name__ == "__main__":
print(f'''{solution() = }''')
| 15
| 0
|
def __UpperCAmelCase ( __a : int ,__a : int ) -> int:
"""simple docstring"""
return int((input_a, input_a).count(0 ) == 0 )
def __UpperCAmelCase ( ) -> None:
"""simple docstring"""
assert and_gate(0 ,0 ) == 0
assert and_gate(0 ,1 ) == 0
assert and_gate(1 ,0 ) == 0
assert and_gate(1 ,1 ) == 1
if __name__ == "__main__":
test_and_gate()
print(and_gate(1, 0))
print(and_gate(0, 0))
print(and_gate(0, 1))
print(and_gate(1, 1))
| 363
|
import numpy as np
import torch
from torch.nn import CrossEntropyLoss
from transformers import AutoModelForCausalLM, AutoTokenizer
import datasets
from datasets import logging
a__ = '''\
'''
a__ = '''
Perplexity (PPL) is one of the most common metrics for evaluating language models.
It is defined as the exponentiated average negative log-likelihood of a sequence.
For more information, see https://huggingface.co/docs/transformers/perplexity
'''
a__ = '''
Args:
model_id (str): model used for calculating Perplexity
NOTE: Perplexity can only be calculated for causal language models.
This includes models such as gpt2, causal variations of bert,
causal versions of t5, and more (the full list can be found
in the AutoModelForCausalLM documentation here:
https://huggingface.co/docs/transformers/master/en/model_doc/auto#transformers.AutoModelForCausalLM )
input_texts (list of str): input text, each separate text snippet
is one list entry.
batch_size (int): the batch size to run texts through the model. Defaults to 16.
add_start_token (bool): whether to add the start token to the texts,
so the perplexity can include the probability of the first word. Defaults to True.
device (str): device to run on, defaults to \'cuda\' when available
Returns:
perplexity: dictionary containing the perplexity scores for the texts
in the input list, as well as the mean perplexity. If one of the input texts is
longer than the max input length of the model, then it is truncated to the
max length for the perplexity computation.
Examples:
Example 1:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = ["lorem ipsum", "Happy Birthday!", "Bienvenue"]
>>> results = perplexity.compute(model_id=\'gpt2\',
... add_start_token=False,
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
78.22
>>> print(round(results["perplexities"][0], 2))
11.11
Example 2:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = datasets.load_dataset("wikitext",
... "wikitext-2-raw-v1",
... split="test")["text"][:50] # doctest:+ELLIPSIS
[...]
>>> input_texts = [s for s in input_texts if s!=\'\']
>>> results = perplexity.compute(model_id=\'gpt2\',
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
60.35
>>> print(round(results["perplexities"][0], 2))
81.12
'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> Any:
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(
{
'''input_texts''': datasets.Value('''string''' ),
} ) , reference_urls=['''https://huggingface.co/docs/transformers/perplexity'''] , )
def __lowercase ( self , _a , _a , _a = 1_6 , _a = True , _a=None ) -> List[Any]:
if device is not None:
assert device in ["gpu", "cpu", "cuda"], "device should be either gpu or cpu."
if device == "gpu":
_a : List[str] = '''cuda'''
else:
_a : Optional[Any] = '''cuda''' if torch.cuda.is_available() else '''cpu'''
_a : Dict = AutoModelForCausalLM.from_pretrained(_a )
_a : List[Any] = model.to(_a )
_a : List[str] = AutoTokenizer.from_pretrained(_a )
# if batch_size > 1 (which generally leads to padding being required), and
# if there is not an already assigned pad_token, assign an existing
# special token to also be the padding token
if tokenizer.pad_token is None and batch_size > 1:
_a : str = list(tokenizer.special_tokens_map_extended.values() )
# check that the model already has at least one special token defined
assert (
len(_a ) > 0
), "If batch_size > 1, model must have at least one special token to use for padding. Please use a different model or set batch_size=1."
# assign one of the special tokens to also be the pad token
tokenizer.add_special_tokens({'''pad_token''': existing_special_tokens[0]} )
if add_start_token:
# leave room for <BOS> token to be added:
assert (
tokenizer.bos_token is not None
), "Input model must already have a BOS token if using add_start_token=True. Please use a different model, or set add_start_token=False"
_a : List[Any] = model.config.max_length - 1
else:
_a : List[str] = model.config.max_length
_a : Union[str, Any] = tokenizer(
_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , return_tensors='''pt''' , return_attention_mask=_a , ).to(_a )
_a : List[Any] = encodings['''input_ids''']
_a : int = encodings['''attention_mask''']
# check that each input is long enough:
if add_start_token:
assert torch.all(torch.ge(attn_masks.sum(1 ) , 1 ) ), "Each input text must be at least one token long."
else:
assert torch.all(
torch.ge(attn_masks.sum(1 ) , 2 ) ), "When add_start_token=False, each input text must be at least two tokens long. Run with add_start_token=True if inputting strings of only one token, and remove all empty input strings."
_a : Optional[int] = []
_a : Dict = CrossEntropyLoss(reduction='''none''' )
for start_index in logging.tqdm(range(0 , len(_a ) , _a ) ):
_a : Dict = min(start_index + batch_size , len(_a ) )
_a : Union[str, Any] = encoded_texts[start_index:end_index]
_a : int = attn_masks[start_index:end_index]
if add_start_token:
_a : Dict = torch.tensor([[tokenizer.bos_token_id]] * encoded_batch.size(dim=0 ) ).to(_a )
_a : List[str] = torch.cat([bos_tokens_tensor, encoded_batch] , dim=1 )
_a : Dict = torch.cat(
[torch.ones(bos_tokens_tensor.size() , dtype=torch.intaa ).to(_a ), attn_mask] , dim=1 )
_a : Dict = encoded_batch
with torch.no_grad():
_a : Any = model(_a , attention_mask=_a ).logits
_a : List[str] = out_logits[..., :-1, :].contiguous()
_a : Union[str, Any] = labels[..., 1:].contiguous()
_a : Optional[int] = attn_mask[..., 1:].contiguous()
_a : Union[str, Any] = torch.expa(
(loss_fct(shift_logits.transpose(1 , 2 ) , _a ) * shift_attention_mask_batch).sum(1 )
/ shift_attention_mask_batch.sum(1 ) )
ppls += perplexity_batch.tolist()
return {"perplexities": ppls, "mean_perplexity": np.mean(_a )}
| 15
| 0
|
"""simple docstring"""
import torch
from diffusers import DDIMParallelScheduler
from .test_schedulers import SchedulerCommonTest
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = (DDIMParallelScheduler,)
UpperCAmelCase__ : int = (("eta", 0.0), ("num_inference_steps", 50))
def __lowercase ( self , **_a ) -> Dict:
_a : int = {
'''num_train_timesteps''': 1_0_0_0,
'''beta_start''': 0.0001,
'''beta_end''': 0.02,
'''beta_schedule''': '''linear''',
'''clip_sample''': True,
}
config.update(**_a )
return config
def __lowercase ( self , **_a ) -> Union[str, Any]:
_a : List[Any] = self.scheduler_classes[0]
_a : Dict = self.get_scheduler_config(**_a )
_a : Dict = scheduler_class(**_a )
_a : Any = 1_0, 0.0
_a : Optional[int] = self.dummy_model()
_a : Tuple = self.dummy_sample_deter
scheduler.set_timesteps(_a )
for t in scheduler.timesteps:
_a : Tuple = model(_a , _a )
_a : str = scheduler.step(_a , _a , _a , _a ).prev_sample
return sample
def __lowercase ( self ) -> Tuple:
for timesteps in [1_0_0, 5_0_0, 1_0_0_0]:
self.check_over_configs(num_train_timesteps=_a )
def __lowercase ( self ) -> Dict:
for steps_offset in [0, 1]:
self.check_over_configs(steps_offset=_a )
_a : int = self.scheduler_classes[0]
_a : Any = self.get_scheduler_config(steps_offset=1 )
_a : List[str] = scheduler_class(**_a )
scheduler.set_timesteps(5 )
assert torch.equal(scheduler.timesteps , torch.LongTensor([8_0_1, 6_0_1, 4_0_1, 2_0_1, 1] ) )
def __lowercase ( self ) -> int:
for beta_start, beta_end in zip([0.0001, 0.001, 0.01, 0.1] , [0.002, 0.02, 0.2, 2] ):
self.check_over_configs(beta_start=_a , beta_end=_a )
def __lowercase ( self ) -> List[Any]:
for schedule in ["linear", "squaredcos_cap_v2"]:
self.check_over_configs(beta_schedule=_a )
def __lowercase ( self ) -> int:
for prediction_type in ["epsilon", "v_prediction"]:
self.check_over_configs(prediction_type=_a )
def __lowercase ( self ) -> Dict:
for clip_sample in [True, False]:
self.check_over_configs(clip_sample=_a )
def __lowercase ( self ) -> Dict:
for timestep_spacing in ["trailing", "leading"]:
self.check_over_configs(timestep_spacing=_a )
def __lowercase ( self ) -> Optional[int]:
for rescale_betas_zero_snr in [True, False]:
self.check_over_configs(rescale_betas_zero_snr=_a )
def __lowercase ( self ) -> Optional[Any]:
self.check_over_configs(thresholding=_a )
for threshold in [0.5, 1.0, 2.0]:
for prediction_type in ["epsilon", "v_prediction"]:
self.check_over_configs(
thresholding=_a , prediction_type=_a , sample_max_value=_a , )
def __lowercase ( self ) -> Optional[int]:
for t in [1, 1_0, 4_9]:
self.check_over_forward(time_step=_a )
def __lowercase ( self ) -> int:
for t, num_inference_steps in zip([1, 1_0, 5_0] , [1_0, 5_0, 5_0_0] ):
self.check_over_forward(time_step=_a , num_inference_steps=_a )
def __lowercase ( self ) -> Tuple:
for t, eta in zip([1, 1_0, 4_9] , [0.0, 0.5, 1.0] ):
self.check_over_forward(time_step=_a , eta=_a )
def __lowercase ( self ) -> Optional[int]:
_a : Any = self.scheduler_classes[0]
_a : List[str] = self.get_scheduler_config()
_a : int = scheduler_class(**_a )
assert torch.sum(torch.abs(scheduler._get_variance(0 , 0 ) - 0.0 ) ) < 1e-5
assert torch.sum(torch.abs(scheduler._get_variance(4_2_0 , 4_0_0 ) - 0.1_4771 ) ) < 1e-5
assert torch.sum(torch.abs(scheduler._get_variance(9_8_0 , 9_6_0 ) - 0.3_2460 ) ) < 1e-5
assert torch.sum(torch.abs(scheduler._get_variance(0 , 0 ) - 0.0 ) ) < 1e-5
assert torch.sum(torch.abs(scheduler._get_variance(4_8_7 , 4_8_6 ) - 0.0_0979 ) ) < 1e-5
assert torch.sum(torch.abs(scheduler._get_variance(9_9_9 , 9_9_8 ) - 0.02 ) ) < 1e-5
def __lowercase ( self ) -> List[str]:
_a : List[Any] = self.scheduler_classes[0]
_a : Union[str, Any] = self.get_scheduler_config()
_a : Tuple = scheduler_class(**_a )
_a : Union[str, Any] = 1_0, 0.0
scheduler.set_timesteps(_a )
_a : int = self.dummy_model()
_a : int = self.dummy_sample_deter
_a : List[str] = self.dummy_sample_deter + 0.1
_a : List[str] = self.dummy_sample_deter - 0.1
_a : int = samplea.shape[0]
_a : int = torch.stack([samplea, samplea, samplea] , dim=0 )
_a : Optional[int] = torch.arange(_a )[0:3, None].repeat(1 , _a )
_a : List[Any] = model(samples.flatten(0 , 1 ) , timesteps.flatten(0 , 1 ) )
_a : Dict = scheduler.batch_step_no_noise(_a , timesteps.flatten(0 , 1 ) , samples.flatten(0 , 1 ) , _a )
_a : List[Any] = torch.sum(torch.abs(_a ) )
_a : str = torch.mean(torch.abs(_a ) )
assert abs(result_sum.item() - 1147.7904 ) < 1e-2
assert abs(result_mean.item() - 0.4982 ) < 1e-3
def __lowercase ( self ) -> List[str]:
_a : Optional[int] = self.full_loop()
_a : Optional[Any] = torch.sum(torch.abs(_a ) )
_a : Optional[Any] = torch.mean(torch.abs(_a ) )
assert abs(result_sum.item() - 172.0067 ) < 1e-2
assert abs(result_mean.item() - 0.22_3967 ) < 1e-3
def __lowercase ( self ) -> Optional[int]:
_a : Optional[int] = self.full_loop(prediction_type='''v_prediction''' )
_a : Union[str, Any] = torch.sum(torch.abs(_a ) )
_a : Optional[Any] = torch.mean(torch.abs(_a ) )
assert abs(result_sum.item() - 52.5302 ) < 1e-2
assert abs(result_mean.item() - 0.0684 ) < 1e-3
def __lowercase ( self ) -> Dict:
# We specify different beta, so that the first alpha is 0.99
_a : Any = self.full_loop(set_alpha_to_one=_a , beta_start=0.01 )
_a : str = torch.sum(torch.abs(_a ) )
_a : Union[str, Any] = torch.mean(torch.abs(_a ) )
assert abs(result_sum.item() - 149.8295 ) < 1e-2
assert abs(result_mean.item() - 0.1951 ) < 1e-3
def __lowercase ( self ) -> Union[str, Any]:
# We specify different beta, so that the first alpha is 0.99
_a : List[Any] = self.full_loop(set_alpha_to_one=_a , beta_start=0.01 )
_a : str = torch.sum(torch.abs(_a ) )
_a : List[Any] = torch.mean(torch.abs(_a ) )
assert abs(result_sum.item() - 149.0784 ) < 1e-2
assert abs(result_mean.item() - 0.1941 ) < 1e-3
| 364
|
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a__ = {
'''configuration_xmod''': [
'''XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP''',
'''XmodConfig''',
'''XmodOnnxConfig''',
],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''XMOD_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''XmodForCausalLM''',
'''XmodForMaskedLM''',
'''XmodForMultipleChoice''',
'''XmodForQuestionAnswering''',
'''XmodForSequenceClassification''',
'''XmodForTokenClassification''',
'''XmodModel''',
'''XmodPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_xmod import XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP, XmodConfig, XmodOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_xmod import (
XMOD_PRETRAINED_MODEL_ARCHIVE_LIST,
XmodForCausalLM,
XmodForMaskedLM,
XmodForMultipleChoice,
XmodForQuestionAnswering,
XmodForSequenceClassification,
XmodForTokenClassification,
XmodModel,
XmodPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 15
| 0
|
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self ) -> None:
_a : dict[str, TrieNode] = {} # Mapping from char to TrieNode
_a : Optional[int] = False
def __lowercase ( self , _a ) -> None:
for word in words:
self.insert(_a )
def __lowercase ( self , _a ) -> None:
_a : Tuple = self
for char in word:
if char not in curr.nodes:
_a : Union[str, Any] = TrieNode()
_a : str = curr.nodes[char]
_a : Any = True
def __lowercase ( self , _a ) -> bool:
_a : Tuple = self
for char in word:
if char not in curr.nodes:
return False
_a : Dict = curr.nodes[char]
return curr.is_leaf
def __lowercase ( self , _a ) -> None:
def _delete(_a , _a , _a ) -> bool:
if index == len(_a ):
# If word does not exist
if not curr.is_leaf:
return False
_a : List[str] = False
return len(curr.nodes ) == 0
_a : Union[str, Any] = word[index]
_a : Dict = curr.nodes.get(_a )
# If char not in current trie node
if not char_node:
return False
# Flag to check if node can be deleted
_a : Tuple = _delete(_a , _a , index + 1 )
if delete_curr:
del curr.nodes[char]
return len(curr.nodes ) == 0
return delete_curr
_delete(self , _a , 0 )
def __UpperCAmelCase ( __a : TrieNode ,__a : str ) -> None:
"""simple docstring"""
if node.is_leaf:
print(__a ,end=''' ''' )
for key, value in node.nodes.items():
print_words(__a ,word + key )
def __UpperCAmelCase ( ) -> bool:
"""simple docstring"""
_a : int = '''banana bananas bandana band apple all beast'''.split()
_a : Union[str, Any] = TrieNode()
root.insert_many(__a )
# print_words(root, "")
assert all(root.find(__a ) for word in words )
assert root.find('''banana''' )
assert not root.find('''bandanas''' )
assert not root.find('''apps''' )
assert root.find('''apple''' )
assert root.find('''all''' )
root.delete('''all''' )
assert not root.find('''all''' )
root.delete('''banana''' )
assert not root.find('''banana''' )
assert root.find('''bananas''' )
return True
def __UpperCAmelCase ( __a : str ,__a : bool ) -> None:
"""simple docstring"""
print(str(__a ) ,'''works!''' if passes else '''doesn\'t work :(''' )
def __UpperCAmelCase ( ) -> None:
"""simple docstring"""
assert test_trie()
def __UpperCAmelCase ( ) -> None:
"""simple docstring"""
print_results('''Testing trie functionality''' ,test_trie() )
if __name__ == "__main__":
main()
| 365
|
import re
import tempfile
from pathlib import Path
import pytest
import yaml
from datasets.utils.readme import ReadMe
# @pytest.fixture
# def example_yaml_structure():
a__ = yaml.safe_load(
'''\
name: ""
allow_empty: false
allow_empty_text: true
subsections:
- name: "Dataset Card for X" # First-level markdown heading
allow_empty: false
allow_empty_text: true
subsections:
- name: "Table of Contents"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Dataset Description"
allow_empty: false
allow_empty_text: false
subsections:
- name: "Dataset Summary"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Supported Tasks and Leaderboards"
allow_empty: true
allow_empty_text: true
subsections: null
- name: Languages
allow_empty: false
allow_empty_text: true
subsections: null
'''
)
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
#### Extra Ignored Subsection
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Extra Ignored Subsection''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
}
],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tEmpty YAML markers are present in the README.'''
)
a__ = '''\
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tNo YAML markers are present in the README.'''
)
a__ = '''\
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tOnly the start of YAML tags present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Summary` but it is empty.\n-\tExpected some text in section `Dataset Summary` but it is empty (text in subsections are ignored).'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Card for My Dataset` but it is empty.\n-\tSection `Dataset Card for My Dataset` expected the following subsections: `Table of Contents`, `Dataset Description`. Found \'None\'.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tSection `Dataset Description` is missing subsection: `Supported Tasks and Leaderboards`.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Languages` but it is empty.'''
a__ = '''\
---
language:
- zh
- en
---
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
# Dataset Card My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has several first-level headings: `Dataset Card for My Dataset`, `Dataset Card My Dataset`. Only one heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tNo first-level heading starting with `Dataset Card for` found in README. Skipping further validation for this README.'''
a__ = ''''''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.\n-\tNo YAML markers are present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found while parsing the README at `{path}`:\n-\tMultiple sections with the same heading `Dataset Card for My Dataset` have been found. Please keep only one of these sections.'''
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : List[str] ) -> Optional[int]:
"""simple docstring"""
assert ReadMe.from_string(__a ,__a ).to_dict() == expected_dict
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[str] ,__a : Optional[Any] ) -> Union[str, Any]:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
_a : List[Any] = ReadMe.from_string(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Dict ,__a : Dict ) -> Tuple:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
ReadMe.from_string(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> Tuple:
"""simple docstring"""
ReadMe.from_string(__a ,__a ,suppress_parsing_errors=__a )
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : Any ) -> Optional[int]:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Tuple = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[Any] = ReadMe.from_readme(__a ,__a ).to_dict()
assert out["name"] == path
assert out["text"] == ""
assert out["is_empty_text"]
assert out["subsections"] == expected_dict["subsections"]
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[Any] ,__a : List[Any] ) -> int:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[int] = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
_a : Any = ReadMe.from_readme(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : str ,__a : Union[str, Any] ) -> Dict:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Optional[Any] = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : str = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
ReadMe.from_readme(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> str:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
ReadMe.from_readme(__a ,__a ,suppress_parsing_errors=__a )
| 15
| 0
|
from typing import Dict, Iterable, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format, to_pil_image
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_pytesseract_available, is_vision_available, logging, requires_backends
if is_vision_available():
import PIL
# soft dependency
if is_pytesseract_available():
import pytesseract
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : str ,__a : Union[str, Any] ) -> List[str]:
"""simple docstring"""
return [
int(1_000 * (box[0] / width) ),
int(1_000 * (box[1] / height) ),
int(1_000 * (box[2] / width) ),
int(1_000 * (box[3] / height) ),
]
def __UpperCAmelCase ( __a : np.ndarray ,__a : Optional[str] ,__a : Optional[str] ) -> List[Any]:
"""simple docstring"""
_a : str = to_pil_image(__a )
_a : Optional[Any] = pil_image.size
_a : Tuple = pytesseract.image_to_data(__a ,lang=__a ,output_type='''dict''' ,config=__a )
_a : List[str] = data['''text'''], data['''left'''], data['''top'''], data['''width'''], data['''height''']
# filter empty words and corresponding coordinates
_a : Dict = [idx for idx, word in enumerate(__a ) if not word.strip()]
_a : str = [word for idx, word in enumerate(__a ) if idx not in irrelevant_indices]
_a : List[str] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : str = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
# turn coordinates into (left, top, left+width, top+height) format
_a : int = []
for x, y, w, h in zip(__a ,__a ,__a ,__a ):
_a : List[str] = [x, y, x + w, y + h]
actual_boxes.append(__a )
# finally, normalize the bounding boxes
_a : Dict = []
for box in actual_boxes:
normalized_boxes.append(normalize_box(__a ,__a ,__a ) )
assert len(__a ) == len(__a ), "Not as many words as there are bounding boxes"
return words, normalized_boxes
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[int] = ["pixel_values"]
def __init__( self , _a = True , _a = None , _a = PILImageResampling.BILINEAR , _a = True , _a = 1 / 2_5_5 , _a = True , _a = None , _a = None , _a = True , _a = None , _a = "" , **_a , ) -> None:
super().__init__(**_a )
_a : List[str] = size if size is not None else {'''height''': 2_2_4, '''width''': 2_2_4}
_a : Union[str, Any] = get_size_dict(_a )
_a : int = do_resize
_a : Optional[int] = size
_a : str = resample
_a : str = do_rescale
_a : Any = rescale_value
_a : Optional[Any] = do_normalize
_a : int = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
_a : List[str] = image_std if image_std is not None else IMAGENET_STANDARD_STD
_a : List[Any] = apply_ocr
_a : Optional[int] = ocr_lang
_a : Tuple = tesseract_config
def __lowercase ( self , _a , _a , _a = PILImageResampling.BILINEAR , _a = None , **_a , ) -> np.ndarray:
_a : Any = get_size_dict(_a )
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()}""" )
_a : Optional[int] = (size['''height'''], size['''width'''])
return resize(_a , size=_a , resample=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
return rescale(_a , scale=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a , _a = None , **_a , ) -> np.ndarray:
return normalize(_a , mean=_a , std=_a , data_format=_a , **_a )
def __lowercase ( 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:
_a : Optional[int] = do_resize if do_resize is not None else self.do_resize
_a : Union[str, Any] = size if size is not None else self.size
_a : Any = get_size_dict(_a )
_a : List[str] = resample if resample is not None else self.resample
_a : int = do_rescale if do_rescale is not None else self.do_rescale
_a : Union[str, Any] = rescale_factor if rescale_factor is not None else self.rescale_factor
_a : int = do_normalize if do_normalize is not None else self.do_normalize
_a : str = image_mean if image_mean is not None else self.image_mean
_a : Tuple = image_std if image_std is not None else self.image_std
_a : Any = apply_ocr if apply_ocr is not None else self.apply_ocr
_a : int = ocr_lang if ocr_lang is not None else self.ocr_lang
_a : Optional[int] = tesseract_config if tesseract_config is not None else self.tesseract_config
_a : List[Any] = make_list_of_images(_a )
if not valid_images(_a ):
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_normalize and (image_mean is None or image_std is None):
raise ValueError('''If do_normalize is True, image_mean and image_std must be specified.''' )
# All transformations expect numpy arrays.
_a : Any = [to_numpy_array(_a ) for image in images]
# Tesseract OCR to get words + normalized bounding boxes
if apply_ocr:
requires_backends(self , '''pytesseract''' )
_a : str = []
_a : str = []
for image in images:
_a : Union[str, Any] = apply_tesseract(_a , _a , _a )
words_batch.append(_a )
boxes_batch.append(_a )
if do_resize:
_a : List[str] = [self.resize(image=_a , size=_a , resample=_a ) for image in images]
if do_rescale:
_a : Optional[Any] = [self.rescale(image=_a , scale=_a ) for image in images]
if do_normalize:
_a : List[Any] = [self.normalize(image=_a , mean=_a , std=_a ) for image in images]
_a : List[str] = [to_channel_dimension_format(_a , _a ) for image in images]
_a : List[str] = BatchFeature(data={'''pixel_values''': images} , tensor_type=_a )
if apply_ocr:
_a : Optional[int] = words_batch
_a : List[Any] = boxes_batch
return data
| 366
|
from __future__ import annotations
def __UpperCAmelCase ( __a : list ) -> float:
"""simple docstring"""
if not nums:
raise ValueError('''List is empty''' )
return sum(__a ) / len(__a )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 15
| 0
|
import json
import os
from pathlib import Path
from shutil import copyfile
from typing import Any, Dict, List, Optional, Tuple, Union
import sentencepiece
from ...tokenization_utils import PreTrainedTokenizer
from ...utils import logging
a__ = logging.get_logger(__name__)
a__ = '''▁'''
a__ = {
'''vocab_file''': '''vocab.json''',
'''spm_file''': '''sentencepiece.bpe.model''',
}
a__ = {
'''vocab_file''': {
'''facebook/s2t-small-librispeech-asr''': (
'''https://huggingface.co/facebook/s2t-small-librispeech-asr/resolve/main/vocab.json'''
),
},
'''spm_file''': {
'''facebook/s2t-small-librispeech-asr''': (
'''https://huggingface.co/facebook/s2t-small-librispeech-asr/resolve/main/sentencepiece.bpe.model'''
)
},
}
a__ = {
'''facebook/s2t-small-librispeech-asr''': 1024,
}
a__ = ['''pt''', '''fr''', '''ru''', '''nl''', '''ro''', '''it''', '''es''', '''de''']
a__ = {'''mustc''': MUSTC_LANGS}
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = VOCAB_FILES_NAMES
UpperCAmelCase__ : List[str] = PRETRAINED_VOCAB_FILES_MAP
UpperCAmelCase__ : Optional[Any] = MAX_MODEL_INPUT_SIZES
UpperCAmelCase__ : List[str] = ["input_ids", "attention_mask"]
UpperCAmelCase__ : List[int] = []
def __init__( self , _a , _a , _a="<s>" , _a="</s>" , _a="<pad>" , _a="<unk>" , _a=False , _a=False , _a=None , _a=None , _a = None , **_a , ) -> None:
_a : Any = {} if sp_model_kwargs is None else sp_model_kwargs
super().__init__(
bos_token=_a , eos_token=_a , unk_token=_a , pad_token=_a , do_upper_case=_a , do_lower_case=_a , tgt_lang=_a , lang_codes=_a , sp_model_kwargs=self.sp_model_kwargs , **_a , )
_a : Any = do_upper_case
_a : Tuple = do_lower_case
_a : Union[str, Any] = load_json(_a )
_a : List[Any] = {v: k for k, v in self.encoder.items()}
_a : List[str] = spm_file
_a : Union[str, Any] = load_spm(_a , self.sp_model_kwargs )
if lang_codes is not None:
_a : Optional[Any] = lang_codes
_a : str = LANGUAGES[lang_codes]
_a : str = [F"""<lang:{lang}>""" for lang in self.langs]
_a : str = {lang: self.sp_model.PieceToId(F"""<lang:{lang}>""" ) for lang in self.langs}
_a : Tuple = self.lang_tokens
_a : Any = tgt_lang if tgt_lang is not None else self.langs[0]
self.set_tgt_lang_special_tokens(self._tgt_lang )
else:
_a : Union[str, Any] = {}
@property
def __lowercase ( self ) -> int:
return len(self.encoder )
@property
def __lowercase ( self ) -> str:
return self._tgt_lang
@tgt_lang.setter
def __lowercase ( self , _a ) -> None:
_a : Tuple = new_tgt_lang
self.set_tgt_lang_special_tokens(_a )
def __lowercase ( self , _a ) -> None:
_a : Tuple = self.lang_code_to_id[tgt_lang]
_a : str = [lang_code_id]
def __lowercase ( self , _a ) -> List[str]:
return self.sp_model.encode(_a , out_type=_a )
def __lowercase ( self , _a ) -> int:
return self.encoder.get(_a , self.encoder[self.unk_token] )
def __lowercase ( self , _a ) -> str:
return self.decoder.get(_a , self.unk_token )
def __lowercase ( self , _a ) -> str:
_a : Any = []
_a : Any = ''''''
for token in tokens:
# make sure that special tokens are not decoded using sentencepiece model
if token in self.all_special_tokens:
_a : Tuple = self.sp_model.decode(_a )
out_string += (decoded.upper() if self.do_upper_case else decoded) + token + " "
_a : Optional[int] = []
else:
current_sub_tokens.append(_a )
_a : Union[str, Any] = self.sp_model.decode(_a )
out_string += decoded.upper() if self.do_upper_case else decoded
return out_string.strip()
def __lowercase ( self , _a , _a=None ) -> List[int]:
if token_ids_a is None:
return self.prefix_tokens + token_ids_a + [self.eos_token_id]
# 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.eos_token_id]
def __lowercase ( self , _a , _a = None , _a = False ) -> List[int]:
if already_has_special_tokens:
return super().get_special_tokens_mask(
token_ids_a=_a , token_ids_a=_a , already_has_special_tokens=_a )
_a : Any = [1] * len(self.prefix_tokens )
_a : Any = [1]
if token_ids_a is None:
return prefix_ones + ([0] * len(_a )) + suffix_ones
return prefix_ones + ([0] * len(_a )) + ([0] * len(_a )) + suffix_ones
def __lowercase ( self ) -> Dict:
_a : Union[str, Any] = self.encoder.copy()
vocab.update(self.added_tokens_encoder )
return vocab
def __getstate__( self ) -> Dict:
_a : str = self.__dict__.copy()
_a : List[Any] = None
return state
def __setstate__( self , _a ) -> None:
_a : Optional[int] = d
# for backward compatibility
if not hasattr(self , '''sp_model_kwargs''' ):
_a : Optional[int] = {}
_a : Union[str, Any] = load_spm(self.spm_file , self.sp_model_kwargs )
def __lowercase ( self , _a , _a = None ) -> Tuple[str]:
_a : Any = Path(_a )
assert save_dir.is_dir(), F"""{save_directory} should be a directory"""
_a : Any = save_dir / (
(filename_prefix + '''-''' if filename_prefix else '''''') + self.vocab_files_names['''vocab_file''']
)
_a : Tuple = save_dir / (
(filename_prefix + '''-''' if filename_prefix else '''''') + self.vocab_files_names['''spm_file''']
)
save_json(self.encoder , _a )
if os.path.abspath(self.spm_file ) != os.path.abspath(_a ) and os.path.isfile(self.spm_file ):
copyfile(self.spm_file , _a )
elif not os.path.isfile(self.spm_file ):
with open(_a , '''wb''' ) as fi:
_a : Tuple = self.sp_model.serialized_model_proto()
fi.write(_a )
return (str(_a ), str(_a ))
def __UpperCAmelCase ( __a : str ,__a : Dict[str, Any] ) -> sentencepiece.SentencePieceProcessor:
"""simple docstring"""
_a : List[Any] = sentencepiece.SentencePieceProcessor(**__a )
spm.Load(str(__a ) )
return spm
def __UpperCAmelCase ( __a : str ) -> Union[Dict, List]:
"""simple docstring"""
with open(__a ,'''r''' ) as f:
return json.load(__a )
def __UpperCAmelCase ( __a : Dict ,__a : str ) -> None:
"""simple docstring"""
with open(__a ,'''w''' ) as f:
json.dump(__a ,__a ,indent=2 )
| 367
|
import argparse
import os
import torch
from transformers.utils import WEIGHTS_NAME
a__ = ['''small''', '''medium''', '''large''']
a__ = '''lm_head.decoder.weight'''
a__ = '''lm_head.weight'''
def __UpperCAmelCase ( __a : str ,__a : str ) -> List[str]:
"""simple docstring"""
_a : Any = torch.load(__a )
_a : List[str] = d.pop(__a )
os.makedirs(__a ,exist_ok=__a )
torch.save(__a ,os.path.join(__a ,__a ) )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--dialogpt_path''', default='''.''', type=str)
a__ = parser.parse_args()
for MODEL in DIALOGPT_MODELS:
a__ = os.path.join(args.dialogpt_path, f'''{MODEL}_ft.pkl''')
a__ = f'''./DialoGPT-{MODEL}'''
convert_dialogpt_checkpoint(
checkpoint_path,
pytorch_dump_folder_path,
)
| 15
| 0
|
import dataclasses
import re
import string
from typing import Any, Dict, Iterator, List, Mapping, Optional, Sequence, Tuple
import numpy as np
from . import residue_constants
a__ = Mapping[str, np.ndarray]
a__ = Mapping[str, Any] # Is a nested dict.
a__ = 0.01
@dataclasses.dataclass(frozen=__lowercase )
class UpperCAmelCase_ :
"""simple docstring"""
UpperCAmelCase__ : np.ndarray # [num_res, num_atom_type, 3]
# Amino-acid type for each residue represented as an integer between 0 and
# 20, where 20 is 'X'.
UpperCAmelCase__ : np.ndarray # [num_res]
# Binary float mask to indicate presence of a particular atom. 1.0 if an atom
# is present and 0.0 if not. This should be used for loss masking.
UpperCAmelCase__ : np.ndarray # [num_res, num_atom_type]
# Residue index as used in PDB. It is not necessarily continuous or 0-indexed.
UpperCAmelCase__ : np.ndarray # [num_res]
# B-factors, or temperature factors, of each residue (in sq. angstroms units),
# representing the displacement of the residue from its ground truth mean
# value.
UpperCAmelCase__ : np.ndarray # [num_res, num_atom_type]
# Chain indices for multi-chain predictions
UpperCAmelCase__ : Optional[np.ndarray] = None
# Optional remark about the protein. Included as a comment in output PDB
# files
UpperCAmelCase__ : Optional[str] = None
# Templates used to generate this protein (prediction-only)
UpperCAmelCase__ : Optional[Sequence[str]] = None
# Chain corresponding to each parent
UpperCAmelCase__ : Optional[Sequence[int]] = None
def __UpperCAmelCase ( __a : str ) -> Protein:
"""simple docstring"""
_a : int = R'''(\[[A-Z]+\]\n)'''
_a : List[str] = [tag.strip() for tag in re.split(__a ,__a ) if len(__a ) > 0]
_a : Iterator[Tuple[str, List[str]]] = zip(tags[0::2] ,[l.split('''\n''' ) for l in tags[1::2]] )
_a : List[str] = ["N", "CA", "C"]
_a : Tuple = None
_a : int = None
_a : int = None
for g in groups:
if "[PRIMARY]" == g[0]:
_a : Any = g[1][0].strip()
for i in range(len(__a ) ):
if seq[i] not in residue_constants.restypes:
_a : Any = '''X''' # FIXME: strings are immutable
_a : Dict = np.array(
[residue_constants.restype_order.get(__a ,residue_constants.restype_num ) for res_symbol in seq] )
elif "[TERTIARY]" == g[0]:
_a : List[List[float]] = []
for axis in range(3 ):
tertiary.append(list(map(__a ,g[1][axis].split() ) ) )
_a : Union[str, Any] = np.array(__a )
_a : Any = np.zeros((len(tertiary[0] ) // 3, residue_constants.atom_type_num, 3) ).astype(np.floataa )
for i, atom in enumerate(__a ):
_a : Optional[int] = np.transpose(tertiary_np[:, i::3] )
atom_positions *= PICO_TO_ANGSTROM
elif "[MASK]" == g[0]:
_a : Any = np.array(list(map({'''-''': 0, '''+''': 1}.get ,g[1][0].strip() ) ) )
_a : Tuple = np.zeros(
(
len(__a ),
residue_constants.atom_type_num,
) ).astype(np.floataa )
for i, atom in enumerate(__a ):
_a : Union[str, Any] = 1
atom_mask *= mask[..., None]
assert aatype is not None
return Protein(
atom_positions=__a ,atom_mask=__a ,aatype=__a ,residue_index=np.arange(len(__a ) ) ,b_factors=__a ,)
def __UpperCAmelCase ( __a : Protein ,__a : int = 0 ) -> List[str]:
"""simple docstring"""
_a : List[str] = []
_a : str = prot.remark
if remark is not None:
pdb_headers.append(F"""REMARK {remark}""" )
_a : Optional[int] = prot.parents
_a : Tuple = prot.parents_chain_index
if parents is not None and parents_chain_index is not None:
_a : int = [p for i, p in zip(__a ,__a ) if i == chain_id]
if parents is None or len(__a ) == 0:
_a : List[Any] = ['''N/A''']
pdb_headers.append(F"""PARENT {' '.join(__a )}""" )
return pdb_headers
def __UpperCAmelCase ( __a : Protein ,__a : str ) -> str:
"""simple docstring"""
_a : List[str] = []
_a : Union[str, Any] = pdb_str.split('''\n''' )
_a : Any = prot.remark
if remark is not None:
out_pdb_lines.append(F"""REMARK {remark}""" )
_a : List[List[str]]
if prot.parents is not None and len(prot.parents ) > 0:
_a : Tuple = []
if prot.parents_chain_index is not None:
_a : Dict[str, List[str]] = {}
for p, i in zip(prot.parents ,prot.parents_chain_index ):
parent_dict.setdefault(str(__a ) ,[] )
parent_dict[str(__a )].append(__a )
_a : List[Any] = max([int(__a ) for chain_idx in parent_dict] )
for i in range(max_idx + 1 ):
_a : int = parent_dict.get(str(__a ) ,['''N/A'''] )
parents_per_chain.append(__a )
else:
parents_per_chain.append(list(prot.parents ) )
else:
_a : Tuple = [['''N/A''']]
def make_parent_line(__a : Sequence[str] ) -> str:
return F"""PARENT {' '.join(__a )}"""
out_pdb_lines.append(make_parent_line(parents_per_chain[0] ) )
_a : List[Any] = 0
for i, l in enumerate(__a ):
if "PARENT" not in l and "REMARK" not in l:
out_pdb_lines.append(__a )
if "TER" in l and "END" not in lines[i + 1]:
chain_counter += 1
if not chain_counter >= len(__a ):
_a : str = parents_per_chain[chain_counter]
else:
_a : Optional[int] = ['''N/A''']
out_pdb_lines.append(make_parent_line(__a ) )
return "\n".join(__a )
def __UpperCAmelCase ( __a : Protein ) -> str:
"""simple docstring"""
_a : str = residue_constants.restypes + ['''X''']
def res_atoa(__a : int ) -> str:
return residue_constants.restype_atoa.get(restypes[r] ,'''UNK''' )
_a : Dict = residue_constants.atom_types
_a : List[str] = []
_a : List[Any] = prot.atom_mask
_a : Union[str, Any] = prot.aatype
_a : Optional[Any] = prot.atom_positions
_a : Dict = prot.residue_index.astype(np.intaa )
_a : Tuple = prot.b_factors
_a : str = prot.chain_index
if np.any(aatype > residue_constants.restype_num ):
raise ValueError('''Invalid aatypes.''' )
_a : List[Any] = get_pdb_headers(__a )
if len(__a ) > 0:
pdb_lines.extend(__a )
_a : List[str] = aatype.shape[0]
_a : Optional[Any] = 1
_a : List[str] = 0
_a : Dict = string.ascii_uppercase
_a : Optional[Any] = None
# Add all atom sites.
for i in range(__a ):
_a : Any = res_atoa(aatype[i] )
for atom_name, pos, mask, b_factor in zip(__a ,atom_positions[i] ,atom_mask[i] ,b_factors[i] ):
if mask < 0.5:
continue
_a : Optional[Any] = '''ATOM'''
_a : List[Any] = atom_name if len(__a ) == 4 else F""" {atom_name}"""
_a : Union[str, Any] = ''''''
_a : Dict = ''''''
_a : List[Any] = 1.00
_a : int = atom_name[0] # Protein supports only C, N, O, S, this works.
_a : int = ''''''
_a : Union[str, Any] = '''A'''
if chain_index is not None:
_a : int = chain_tags[chain_index[i]]
# PDB is a columnar format, every space matters here!
_a : Tuple = (
F"""{record_type:<6}{atom_index:>5} {name:<4}{alt_loc:>1}"""
F"""{res_name_a:>3} {chain_tag:>1}"""
F"""{residue_index[i]:>4}{insertion_code:>1} """
F"""{pos[0]:>8.3f}{pos[1]:>8.3f}{pos[2]:>8.3f}"""
F"""{occupancy:>6.2f}{b_factor:>6.2f} """
F"""{element:>2}{charge:>2}"""
)
pdb_lines.append(__a )
atom_index += 1
_a : List[Any] = i == n - 1
if chain_index is not None:
if i != n - 1 and chain_index[i + 1] != prev_chain_index:
_a : Optional[int] = True
_a : Union[str, Any] = chain_index[i + 1]
if should_terminate:
# Close the chain.
_a : str = '''TER'''
_a : Any = (
F"""{chain_end:<6}{atom_index:>5} {res_atoa(aatype[i] ):>3} {chain_tag:>1}{residue_index[i]:>4}"""
)
pdb_lines.append(__a )
atom_index += 1
if i != n - 1:
# "prev" is a misnomer here. This happens at the beginning of
# each new chain.
pdb_lines.extend(get_pdb_headers(__a ,__a ) )
pdb_lines.append('''END''' )
pdb_lines.append('''''' )
return "\n".join(__a )
def __UpperCAmelCase ( __a : Protein ) -> np.ndarray:
"""simple docstring"""
return residue_constants.STANDARD_ATOM_MASK[prot.aatype]
def __UpperCAmelCase ( __a : FeatureDict ,__a : ModelOutput ,__a : Optional[np.ndarray] = None ,__a : Optional[np.ndarray] = None ,__a : Optional[str] = None ,__a : Optional[Sequence[str]] = None ,__a : Optional[Sequence[int]] = None ,) -> Protein:
"""simple docstring"""
return Protein(
aatype=features['''aatype'''] ,atom_positions=result['''final_atom_positions'''] ,atom_mask=result['''final_atom_mask'''] ,residue_index=features['''residue_index'''] + 1 ,b_factors=b_factors if b_factors is not None else np.zeros_like(result['''final_atom_mask'''] ) ,chain_index=__a ,remark=__a ,parents=__a ,parents_chain_index=__a ,)
| 368
|
import enum
import warnings
from .. import MODEL_FOR_CAUSAL_LM_MAPPING, TF_MODEL_FOR_CAUSAL_LM_MAPPING
from ..utils import add_end_docstrings, is_tf_available
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
class UpperCAmelCase_ ( enum.Enum ):
"""simple docstring"""
UpperCAmelCase__ : int = 0
UpperCAmelCase__ : Union[str, Any] = 1
UpperCAmelCase__ : Optional[Any] = 2
@add_end_docstrings(__lowercase )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = "\n In 1991, the remains of Russian Tsar Nicholas II and his family (except for Alexei and Maria) are discovered. The\n voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the remainder of the story. 1883 Western\n Siberia, a young Grigori Rasputin is asked by his father and a group of men to perform magic. Rasputin has a vision\n and denounces one of the men as a horse thief. Although his father initially slaps him for making such an\n accusation, Rasputin watches as the man is chased outside and beaten. Twenty years later, Rasputin sees a vision of\n the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous, with people, even a bishop,\n begging for his blessing. <eod> </s> <eos>\n "
def __init__( self , *_a , **_a ) -> List[str]:
super().__init__(*_a , **_a )
self.check_model_type(
TF_MODEL_FOR_CAUSAL_LM_MAPPING if self.framework == '''tf''' else MODEL_FOR_CAUSAL_LM_MAPPING )
if "prefix" not in self._preprocess_params:
# This is very specific. The logic is quite complex and needs to be done
# as a "default".
# It also defines both some preprocess_kwargs and generate_kwargs
# which is why we cannot put them in their respective methods.
_a : Dict = None
if self.model.config.prefix is not None:
_a : List[Any] = self.model.config.prefix
if prefix is None and self.model.__class__.__name__ in [
"XLNetLMHeadModel",
"TransfoXLLMHeadModel",
"TFXLNetLMHeadModel",
"TFTransfoXLLMHeadModel",
]:
# For XLNet and TransformerXL we add an article to the prompt to give more state to the model.
_a : Optional[Any] = self.XL_PREFIX
if prefix is not None:
# Recalculate some generate_kwargs linked to prefix.
_a , _a , _a : str = self._sanitize_parameters(prefix=_a , **self._forward_params )
_a : Optional[Any] = {**self._preprocess_params, **preprocess_params}
_a : List[Any] = {**self._forward_params, **forward_params}
def __lowercase ( self , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , **_a , ) -> Optional[int]:
_a : List[Any] = {}
if prefix is not None:
_a : Optional[Any] = prefix
if prefix:
_a : Dict = self.tokenizer(
_a , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Tuple = prefix_inputs['''input_ids'''].shape[-1]
if handle_long_generation is not None:
if handle_long_generation not in {"hole"}:
raise ValueError(
F"""{handle_long_generation} is not a valid value for `handle_long_generation` parameter expected"""
''' [None, \'hole\']''' )
_a : Dict = handle_long_generation
preprocess_params.update(_a )
_a : Tuple = generate_kwargs
_a : Any = {}
if return_full_text is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_full_text`''' )
if return_tensors is not None:
raise ValueError('''`return_full_text` is mutually exclusive with `return_tensors`''' )
_a : List[str] = ReturnType.FULL_TEXT if return_full_text else ReturnType.NEW_TEXT
if return_tensors is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_tensors`''' )
_a : Any = ReturnType.TENSORS
if return_type is not None:
_a : Any = return_type
if clean_up_tokenization_spaces is not None:
_a : List[Any] = clean_up_tokenization_spaces
if stop_sequence is not None:
_a : Tuple = self.tokenizer.encode(_a , add_special_tokens=_a )
if len(_a ) > 1:
warnings.warn(
'''Stopping on a multiple token sequence is not yet supported on transformers. The first token of'''
''' the stop sequence will be used as the stop sequence string in the interim.''' )
_a : List[Any] = stop_sequence_ids[0]
return preprocess_params, forward_params, postprocess_params
def __lowercase ( self , *_a , **_a ) -> Union[str, Any]:
# Parse arguments
if self.model.__class__.__name__ in ["TransfoXLLMHeadModel"]:
kwargs.update({'''add_space_before_punct_symbol''': True} )
return super()._parse_and_tokenize(*_a , **_a )
def __call__( self , _a , **_a ) -> List[str]:
return super().__call__(_a , **_a )
def __lowercase ( self , _a , _a="" , _a=None , **_a ) -> List[Any]:
_a : Optional[int] = self.tokenizer(
prefix + prompt_text , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Union[str, Any] = prompt_text
if handle_long_generation == "hole":
_a : List[str] = inputs['''input_ids'''].shape[-1]
if "max_new_tokens" in generate_kwargs:
_a : int = generate_kwargs['''max_new_tokens''']
else:
_a : List[Any] = generate_kwargs.get('''max_length''' , self.model.config.max_length ) - cur_len
if new_tokens < 0:
raise ValueError('''We cannot infer how many new tokens are expected''' )
if cur_len + new_tokens > self.tokenizer.model_max_length:
_a : List[str] = self.tokenizer.model_max_length - new_tokens
if keep_length <= 0:
raise ValueError(
'''We cannot use `hole` to handle this generation the number of desired tokens exceeds the'''
''' models max length''' )
_a : List[Any] = inputs['''input_ids'''][:, -keep_length:]
if "attention_mask" in inputs:
_a : List[str] = inputs['''attention_mask'''][:, -keep_length:]
return inputs
def __lowercase ( self , _a , **_a ) -> Optional[int]:
_a : Any = model_inputs['''input_ids''']
_a : Optional[Any] = model_inputs.get('''attention_mask''' , _a )
# Allow empty prompts
if input_ids.shape[1] == 0:
_a : int = None
_a : int = None
_a : List[str] = 1
else:
_a : List[Any] = input_ids.shape[0]
_a : Union[str, Any] = model_inputs.pop('''prompt_text''' )
# If there is a prefix, we may need to adjust the generation length. Do so without permanently modifying
# generate_kwargs, as some of the parameterization may come from the initialization of the pipeline.
_a : int = generate_kwargs.pop('''prefix_length''' , 0 )
if prefix_length > 0:
_a : Tuple = '''max_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].max_new_tokens is not None
)
if not has_max_new_tokens:
_a : int = generate_kwargs.get('''max_length''' ) or self.model.config.max_length
generate_kwargs["max_length"] += prefix_length
_a : Dict = '''min_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].min_new_tokens is not None
)
if not has_min_new_tokens and "min_length" in generate_kwargs:
generate_kwargs["min_length"] += prefix_length
# BS x SL
_a : Optional[Any] = self.model.generate(input_ids=_a , attention_mask=_a , **_a )
_a : int = generated_sequence.shape[0]
if self.framework == "pt":
_a : Tuple = generated_sequence.reshape(_a , out_b // in_b , *generated_sequence.shape[1:] )
elif self.framework == "tf":
_a : List[Any] = tf.reshape(_a , (in_b, out_b // in_b, *generated_sequence.shape[1:]) )
return {"generated_sequence": generated_sequence, "input_ids": input_ids, "prompt_text": prompt_text}
def __lowercase ( self , _a , _a=ReturnType.FULL_TEXT , _a=True ) -> int:
_a : Tuple = model_outputs['''generated_sequence'''][0]
_a : int = model_outputs['''input_ids''']
_a : Any = model_outputs['''prompt_text''']
_a : Any = generated_sequence.numpy().tolist()
_a : Any = []
for sequence in generated_sequence:
if return_type == ReturnType.TENSORS:
_a : Optional[int] = {'''generated_token_ids''': sequence}
elif return_type in {ReturnType.NEW_TEXT, ReturnType.FULL_TEXT}:
# Decode text
_a : str = self.tokenizer.decode(
_a , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , )
# Remove PADDING prompt of the sequence if XLNet or Transfo-XL model is used
if input_ids is None:
_a : Union[str, Any] = 0
else:
_a : str = len(
self.tokenizer.decode(
input_ids[0] , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , ) )
if return_type == ReturnType.FULL_TEXT:
_a : str = prompt_text + text[prompt_length:]
else:
_a : List[str] = text[prompt_length:]
_a : Union[str, Any] = {'''generated_text''': all_text}
records.append(_a )
return records
| 15
| 0
|
import sys
import turtle
def __UpperCAmelCase ( __a : tuple[float, float] ,__a : tuple[float, float] ) -> tuple[float, float]:
"""simple docstring"""
return (pa[0] + pa[0]) / 2, (pa[1] + pa[1]) / 2
def __UpperCAmelCase ( __a : tuple[float, float] ,__a : tuple[float, float] ,__a : tuple[float, float] ,__a : int ,) -> None:
"""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(__a ,get_mid(__a ,__a ) ,get_mid(__a ,__a ) ,depth - 1 )
triangle(__a ,get_mid(__a ,__a ) ,get_mid(__a ,__a ) ,depth - 1 )
triangle(__a ,get_mid(__a ,__a ) ,get_mid(__a ,__a ) ,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]))
| 369
|
# 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
import os
from accelerate.test_utils import execute_subprocess_async
def __UpperCAmelCase ( __a : Dict=None ) -> str:
"""simple docstring"""
if subparsers is not None:
_a : Union[str, Any] = subparsers.add_parser('''test''' )
else:
_a : List[str] = argparse.ArgumentParser('''Accelerate test command''' )
parser.add_argument(
'''--config_file''' ,default=__a ,help=(
'''The path to use to store the config file. Will default to a file named default_config.yaml in the cache '''
'''location, which is the content of the environment `HF_HOME` suffixed with \'accelerate\', or if you don\'t have '''
'''such an environment variable, your cache directory (\'~/.cache\' or the content of `XDG_CACHE_HOME`) suffixed '''
'''with \'huggingface\'.'''
) ,)
if subparsers is not None:
parser.set_defaults(func=__a )
return parser
def __UpperCAmelCase ( __a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
_a : Dict = os.path.sep.join(__file__.split(os.path.sep )[:-2] + ['''test_utils''', '''scripts''', '''test_script.py'''] )
if args.config_file is None:
_a : List[Any] = script_name
else:
_a : Union[str, Any] = F"""--config_file={args.config_file} {script_name}"""
_a : str = ['''accelerate-launch'''] + test_args.split()
_a : str = execute_subprocess_async(__a ,env=os.environ.copy() )
if result.returncode == 0:
print('''Test is a success! You are ready for your distributed training!''' )
def __UpperCAmelCase ( ) -> List[Any]:
"""simple docstring"""
_a : Optional[int] = test_command_parser()
_a : List[Any] = parser.parse_args()
test_command(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
def __UpperCAmelCase ( __a : int ,__a : int ) -> int:
"""simple docstring"""
return x if y == 0 else greatest_common_divisor(__a ,x % y )
def __UpperCAmelCase ( __a : int ,__a : int ) -> int:
"""simple docstring"""
return (x * y) // greatest_common_divisor(__a ,__a )
def __UpperCAmelCase ( __a : int = 20 ) -> int:
"""simple docstring"""
_a : str = 1
for i in range(1 ,n + 1 ):
_a : Tuple = lcm(__a ,__a )
return g
if __name__ == "__main__":
print(f'''{solution() = }''')
| 370
|
import json
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import BertTokenizerFast
from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer
from transformers.testing_utils import require_tokenizers, require_vision
from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available
if is_vision_available():
from PIL import Image
from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor
@require_tokenizers
@require_vision
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = tempfile.mkdtemp()
# fmt: off
_a : Optional[int] = ['''[UNK]''', '''[CLS]''', '''[SEP]''', '''[PAD]''', '''[MASK]''', '''want''', '''##want''', '''##ed''', '''wa''', '''un''', '''runn''', '''##ing''', ''',''', '''low''', '''lowest''']
# fmt: on
_a : Tuple = 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] ) )
_a : Any = {
'''do_resize''': True,
'''size''': {'''height''': 1_8, '''width''': 1_8},
'''do_normalize''': True,
'''image_mean''': [0.5, 0.5, 0.5],
'''image_std''': [0.5, 0.5, 0.5],
}
_a : str = os.path.join(self.tmpdirname , _a )
with open(self.image_processor_file , '''w''' , encoding='''utf-8''' ) as fp:
json.dump(_a , _a )
def __lowercase ( self , **_a ) -> Any:
return BertTokenizer.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self , **_a ) -> str:
return ViTImageProcessor.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self ) -> List[Any]:
shutil.rmtree(self.tmpdirname )
def __lowercase ( self ) -> Any:
_a : Union[str, Any] = [np.random.randint(2_5_5 , size=(3, 3_0, 4_0_0) , dtype=np.uinta )]
_a : Tuple = [Image.fromarray(np.moveaxis(_a , 0 , -1 ) ) for x in image_inputs]
return image_inputs
def __lowercase ( self ) -> str:
_a : List[str] = self.get_tokenizer()
_a : Tuple = self.get_image_processor()
_a : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
processor.save_pretrained(self.tmpdirname )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Dict:
_a : List[str] = VisionTextDualEncoderProcessor(
tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() )
processor.save_pretrained(self.tmpdirname )
_a : Any = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' )
_a : List[Any] = self.get_image_processor(do_normalize=_a , padding_value=1.0 )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(
self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=_a , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Any:
_a : Dict = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : int = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[str] = self.prepare_image_inputs()
_a : List[Any] = image_processor(_a , return_tensors='''np''' )
_a : Dict = processor(images=_a , return_tensors='''np''' )
for key in input_feat_extract.keys():
self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1e-2 )
def __lowercase ( self ) -> List[str]:
_a : Union[str, Any] = self.get_image_processor()
_a : Dict = self.get_tokenizer()
_a : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Tuple = '''lower newer'''
_a : int = processor(text=_a )
_a : str = tokenizer(_a )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def __lowercase ( self ) -> List[Any]:
_a : Any = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : Tuple = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[Any] = '''lower newer'''
_a : Union[str, Any] = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , ['''input_ids''', '''token_type_ids''', '''attention_mask''', '''pixel_values'''] )
# test if it raises when no input is passed
with self.assertRaises(_a ):
processor()
def __lowercase ( self ) -> Optional[int]:
_a : Union[str, Any] = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : Any = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Any = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
_a : int = processor.batch_decode(_a )
_a : int = tokenizer.batch_decode(_a )
self.assertListEqual(_a , _a )
def __lowercase ( self ) -> List[Any]:
_a : Tuple = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : str = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Optional[int] = '''lower newer'''
_a : Dict = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
| 15
| 0
|
from typing import Dict, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import center_crop, normalize, rescale, resize, to_channel_dimension_format
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_vision_available, logging
if is_vision_available():
import PIL
a__ = logging.get_logger(__name__)
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Union[str, Any] = ["pixel_values"]
def __init__( self , _a = True , _a = None , _a = PIL.Image.BICUBIC , _a = True , _a = None , _a = 1 / 2_5_5 , _a = True , _a = True , _a = None , _a = None , **_a , ) -> None:
super().__init__(**_a )
_a : Dict = size if size is not None else {'''height''': 2_5_6, '''width''': 2_5_6}
_a : int = get_size_dict(_a )
_a : Tuple = crop_size if crop_size is not None else {'''height''': 2_2_4, '''width''': 2_2_4}
_a : Dict = get_size_dict(_a , param_name='''crop_size''' )
_a : Union[str, Any] = do_resize
_a : Any = size
_a : List[Any] = resample
_a : List[Any] = do_center_crop
_a : Union[str, Any] = crop_size
_a : str = do_rescale
_a : Union[str, Any] = rescale_factor
_a : int = do_normalize
_a : List[Any] = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
_a : Optional[Any] = image_std if image_std is not None else IMAGENET_STANDARD_STD
def __lowercase ( self , _a , _a , _a = PIL.Image.BICUBIC , _a = None , **_a , ) -> np.ndarray:
_a : List[Any] = get_size_dict(_a )
if "height" not in size or "width" not in size:
raise ValueError(F"""The size dictionary must have keys 'height' and 'width'. Got {size.keys()}""" )
return resize(
_a , size=(size['''height'''], size['''width''']) , resample=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
_a : Union[str, Any] = get_size_dict(_a )
if "height" not in size or "width" not in size:
raise ValueError(F"""The size dictionary must have keys 'height' and 'width'. Got {size.keys()}""" )
return center_crop(_a , size=(size['''height'''], size['''width''']) , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> List[Any]:
return rescale(_a , scale=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a , _a = None , **_a , ) -> np.ndarray:
return normalize(_a , mean=_a , std=_a , data_format=_a , **_a )
def __lowercase ( 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 , **_a , ) -> PIL.Image.Image:
_a : int = do_resize if do_resize is not None else self.do_resize
_a : int = resample if resample is not None else self.resample
_a : Optional[Any] = do_center_crop if do_center_crop is not None else self.do_center_crop
_a : Any = do_rescale if do_rescale is not None else self.do_rescale
_a : Optional[int] = rescale_factor if rescale_factor is not None else self.rescale_factor
_a : List[Any] = do_normalize if do_normalize is not None else self.do_normalize
_a : Optional[int] = image_mean if image_mean is not None else self.image_mean
_a : Tuple = image_std if image_std is not None else self.image_std
_a : Any = size if size is not None else self.size
_a : str = get_size_dict(_a )
_a : List[Any] = crop_size if crop_size is not None else self.crop_size
_a : int = get_size_dict(_a , param_name='''crop_size''' )
_a : Union[str, Any] = make_list_of_images(_a )
if not valid_images(_a ):
raise ValueError(
'''Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '''
'''torch.Tensor, tf.Tensor or jax.ndarray.''' )
if do_resize and size is None or resample is None:
raise ValueError('''Size and resample must be specified if do_resize is True.''' )
if do_center_crop and crop_size is None:
raise ValueError('''Crop size must be specified if do_center_crop is True.''' )
if do_rescale and rescale_factor is None:
raise ValueError('''Rescale factor must be specified if do_rescale is True.''' )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError('''Image mean and std must be specified if do_normalize is True.''' )
# All transformations expect numpy arrays.
_a : Dict = [to_numpy_array(_a ) for image in images]
if do_resize:
_a : Optional[int] = [self.resize(image=_a , size=_a , resample=_a ) for image in images]
if do_center_crop:
_a : Optional[Any] = [self.center_crop(image=_a , size=_a ) for image in images]
if do_rescale:
_a : Optional[Any] = [self.rescale(image=_a , scale=_a ) for image in images]
if do_normalize:
_a : Dict = [self.normalize(image=_a , mean=_a , std=_a ) for image in images]
_a : Tuple = [to_channel_dimension_format(_a , _a ) for image in images]
_a : Dict = {'''pixel_values''': images}
return BatchFeature(data=_a , tensor_type=_a )
| 371
|
import argparse
import json
import os
import fairseq
import torch
from fairseq.data import Dictionary
from transformers import (
HubertConfig,
HubertForCTC,
HubertModel,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaProcessor,
logging,
)
logging.set_verbosity_info()
a__ = logging.get_logger(__name__)
a__ = {
'''post_extract_proj''': '''feature_projection.projection''',
'''encoder.pos_conv.0''': '''encoder.pos_conv_embed.conv''',
'''self_attn.k_proj''': '''encoder.layers.*.attention.k_proj''',
'''self_attn.v_proj''': '''encoder.layers.*.attention.v_proj''',
'''self_attn.q_proj''': '''encoder.layers.*.attention.q_proj''',
'''self_attn.out_proj''': '''encoder.layers.*.attention.out_proj''',
'''self_attn_layer_norm''': '''encoder.layers.*.layer_norm''',
'''fc1''': '''encoder.layers.*.feed_forward.intermediate_dense''',
'''fc2''': '''encoder.layers.*.feed_forward.output_dense''',
'''final_layer_norm''': '''encoder.layers.*.final_layer_norm''',
'''encoder.layer_norm''': '''encoder.layer_norm''',
'''w2v_model.layer_norm''': '''feature_projection.layer_norm''',
'''w2v_encoder.proj''': '''lm_head''',
'''mask_emb''': '''masked_spec_embed''',
}
def __UpperCAmelCase ( __a : List[Any] ,__a : Optional[int] ,__a : Optional[int] ,__a : List[str] ,__a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
for attribute in key.split('''.''' ):
_a : Optional[Any] = getattr(__a ,__a )
if weight_type is not None:
_a : Dict = getattr(__a ,__a ).shape
else:
_a : Optional[int] = 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":
_a : List[Any] = value
elif weight_type == "weight_g":
_a : Any = value
elif weight_type == "weight_v":
_a : Union[str, Any] = value
elif weight_type == "bias":
_a : Optional[int] = value
else:
_a : List[Any] = value
logger.info(F"""{key + '.' + weight_type if weight_type is not None else ''} was initialized from {full_name}.""" )
def __UpperCAmelCase ( __a : Any ,__a : Union[str, Any] ,__a : Union[str, Any] ) -> int:
"""simple docstring"""
_a : Union[str, Any] = []
_a : Union[str, Any] = fairseq_model.state_dict()
_a : Union[str, Any] = hf_model.hubert.feature_extractor if is_finetuned else hf_model.feature_extractor
for name, value in fairseq_dict.items():
_a : int = False
if "conv_layers" in name:
load_conv_layer(
__a ,__a ,__a ,__a ,hf_model.config.feat_extract_norm == '''group''' ,)
_a : Optional[Any] = True
else:
for key, mapped_key in MAPPING.items():
_a : Union[str, Any] = '''hubert.''' + 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] and not is_finetuned):
_a : Any = True
if "*" in mapped_key:
_a : Optional[int] = name.split(__a )[0].split('''.''' )[-2]
_a : Any = mapped_key.replace('''*''' ,__a )
if "weight_g" in name:
_a : List[Any] = '''weight_g'''
elif "weight_v" in name:
_a : List[str] = '''weight_v'''
elif "weight" in name:
_a : Any = '''weight'''
elif "bias" in name:
_a : str = '''bias'''
else:
_a : Any = None
set_recursively(__a ,__a ,__a ,__a ,__a )
continue
if not is_used:
unused_weights.append(__a )
logger.warning(F"""Unused weights: {unused_weights}""" )
def __UpperCAmelCase ( __a : int ,__a : Optional[Any] ,__a : Dict ,__a : List[str] ,__a : Any ) -> Tuple:
"""simple docstring"""
_a : int = full_name.split('''conv_layers.''' )[-1]
_a : Any = name.split('''.''' )
_a : List[Any] = int(items[0] )
_a : Optional[int] = 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."""
)
_a : Optional[int] = 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."""
)
_a : Optional[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:
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."
)
_a : int = 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."""
)
_a : Any = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(__a )
@torch.no_grad()
def __UpperCAmelCase ( __a : Dict ,__a : List[Any] ,__a : List[str]=None ,__a : Optional[int]=None ,__a : int=True ) -> List[Any]:
"""simple docstring"""
if config_path is not None:
_a : Tuple = HubertConfig.from_pretrained(__a )
else:
_a : Any = HubertConfig()
if is_finetuned:
if dict_path:
_a : Tuple = Dictionary.load(__a )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
_a : Any = target_dict.pad_index
_a : Tuple = target_dict.bos_index
_a : Optional[int] = target_dict.eos_index
_a : Optional[Any] = len(target_dict.symbols )
_a : Tuple = os.path.join(__a ,'''vocab.json''' )
if not os.path.isdir(__a ):
logger.error('''--pytorch_dump_folder_path ({}) should be a directory'''.format(__a ) )
return
os.makedirs(__a ,exist_ok=__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ) as vocab_handle:
json.dump(target_dict.indices ,__a )
_a : Tuple = WavaVecaCTCTokenizer(
__a ,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=__a ,)
_a : Tuple = True if config.feat_extract_norm == '''layer''' else False
_a : List[Any] = WavaVecaFeatureExtractor(
feature_size=1 ,sampling_rate=16_000 ,padding_value=0 ,do_normalize=__a ,return_attention_mask=__a ,)
_a : List[Any] = WavaVecaProcessor(feature_extractor=__a ,tokenizer=__a )
processor.save_pretrained(__a )
_a : Tuple = HubertForCTC(__a )
else:
_a : Tuple = HubertModel(__a )
if is_finetuned:
_a , _a , _a : int = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
_a , _a , _a : str = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] )
_a : Any = model[0].eval()
recursively_load_weights(__a ,__a ,__a )
hf_wavavec.save_pretrained(__a )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
parser.add_argument('''--checkpoint_path''', default=None, type=str, help='''Path to fairseq checkpoint''')
parser.add_argument('''--dict_path''', default=None, type=str, help='''Path to dict of fine-tuned model''')
parser.add_argument('''--config_path''', default=None, type=str, help='''Path to hf config.json of model to convert''')
parser.add_argument(
'''--not_finetuned''', action='''store_true''', help='''Whether the model to convert is a fine-tuned model or not'''
)
a__ = parser.parse_args()
convert_hubert_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 15
| 0
|
def __UpperCAmelCase ( __a : int = 1_000_000 ) -> int:
"""simple docstring"""
_a : List[Any] = set(range(3 ,__a ,2 ) )
primes.add(2 )
for p in range(3 ,__a ,2 ):
if p not in primes:
continue
primes.difference_update(set(range(p * p ,__a ,__a ) ) )
_a : Any = [float(__a ) for n in range(limit + 1 )]
for p in primes:
for n in range(__a ,limit + 1 ,__a ):
phi[n] *= 1 - 1 / p
return int(sum(phi[2:] ) )
if __name__ == "__main__":
print(f'''{solution() = }''')
| 350
|
import warnings
from typing import List, Optional, Union
from ...processing_utils import ProcessorMixin
from ...tokenization_utils_base import BatchEncoding, PaddingStrategy, PreTokenizedInput, TextInput, TruncationStrategy
from ...utils import TensorType
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[str] = ["image_processor", "tokenizer"]
UpperCAmelCase__ : str = "ViltImageProcessor"
UpperCAmelCase__ : Union[str, Any] = ("BertTokenizer", "BertTokenizerFast")
def __init__( self , _a=None , _a=None , **_a ) -> Any:
_a : Union[str, Any] = None
if "feature_extractor" in kwargs:
warnings.warn(
'''The `feature_extractor` argument is deprecated and will be removed in v5, use `image_processor`'''
''' instead.''' , _a , )
_a : Dict = kwargs.pop('''feature_extractor''' )
_a : Optional[int] = image_processor if image_processor is not None else feature_extractor
if image_processor is None:
raise ValueError('''You need to specify an `image_processor`.''' )
if tokenizer is None:
raise ValueError('''You need to specify a `tokenizer`.''' )
super().__init__(_a , _a )
_a : int = self.image_processor
def __call__( self , _a , _a = None , _a = True , _a = False , _a = None , _a = None , _a = 0 , _a = None , _a = None , _a = None , _a = False , _a = False , _a = False , _a = False , _a = True , _a = None , **_a , ) -> BatchEncoding:
_a : Tuple = self.tokenizer(
text=_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , stride=_a , pad_to_multiple_of=_a , return_token_type_ids=_a , return_attention_mask=_a , return_overflowing_tokens=_a , return_special_tokens_mask=_a , return_offsets_mapping=_a , return_length=_a , verbose=_a , return_tensors=_a , **_a , )
# add pixel_values + pixel_mask
_a : str = self.image_processor(_a , return_tensors=_a )
encoding.update(_a )
return encoding
def __lowercase ( self , *_a , **_a ) -> Optional[Any]:
return self.tokenizer.batch_decode(*_a , **_a )
def __lowercase ( self , *_a , **_a ) -> str:
return self.tokenizer.decode(*_a , **_a )
@property
def __lowercase ( self ) -> Optional[int]:
_a : str = self.tokenizer.model_input_names
_a : Optional[Any] = self.image_processor.model_input_names
return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names ) )
@property
def __lowercase ( self ) -> Optional[Any]:
warnings.warn(
'''`feature_extractor_class` is deprecated and will be removed in v5. Use `image_processor_class` instead.''' , _a , )
return self.image_processor_class
@property
def __lowercase ( self ) -> Any:
warnings.warn(
'''`feature_extractor` is deprecated and will be removed in v5. Use `image_processor` instead.''' , _a , )
return self.image_processor
| 15
| 0
|
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_flax_available,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
a__ = {
'''configuration_distilbert''': [
'''DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP''',
'''DistilBertConfig''',
'''DistilBertOnnxConfig''',
],
'''tokenization_distilbert''': ['''DistilBertTokenizer'''],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = ['''DistilBertTokenizerFast''']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''DistilBertForMaskedLM''',
'''DistilBertForMultipleChoice''',
'''DistilBertForQuestionAnswering''',
'''DistilBertForSequenceClassification''',
'''DistilBertForTokenClassification''',
'''DistilBertModel''',
'''DistilBertPreTrainedModel''',
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''TF_DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''TFDistilBertForMaskedLM''',
'''TFDistilBertForMultipleChoice''',
'''TFDistilBertForQuestionAnswering''',
'''TFDistilBertForSequenceClassification''',
'''TFDistilBertForTokenClassification''',
'''TFDistilBertMainLayer''',
'''TFDistilBertModel''',
'''TFDistilBertPreTrainedModel''',
]
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''FlaxDistilBertForMaskedLM''',
'''FlaxDistilBertForMultipleChoice''',
'''FlaxDistilBertForQuestionAnswering''',
'''FlaxDistilBertForSequenceClassification''',
'''FlaxDistilBertForTokenClassification''',
'''FlaxDistilBertModel''',
'''FlaxDistilBertPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_distilbert import (
DISTILBERT_PRETRAINED_CONFIG_ARCHIVE_MAP,
DistilBertConfig,
DistilBertOnnxConfig,
)
from .tokenization_distilbert import DistilBertTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_distilbert_fast import DistilBertTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_distilbert import (
DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
DistilBertForMaskedLM,
DistilBertForMultipleChoice,
DistilBertForQuestionAnswering,
DistilBertForSequenceClassification,
DistilBertForTokenClassification,
DistilBertModel,
DistilBertPreTrainedModel,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_distilbert import (
TF_DISTILBERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFDistilBertForMaskedLM,
TFDistilBertForMultipleChoice,
TFDistilBertForQuestionAnswering,
TFDistilBertForSequenceClassification,
TFDistilBertForTokenClassification,
TFDistilBertMainLayer,
TFDistilBertModel,
TFDistilBertPreTrainedModel,
)
try:
if not is_flax_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_flax_distilbert import (
FlaxDistilBertForMaskedLM,
FlaxDistilBertForMultipleChoice,
FlaxDistilBertForQuestionAnswering,
FlaxDistilBertForSequenceClassification,
FlaxDistilBertForTokenClassification,
FlaxDistilBertModel,
FlaxDistilBertPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 351
|
from math import ceil
def __UpperCAmelCase ( __a : int = 1_001 ) -> int:
"""simple docstring"""
_a : Dict = 1
for i in range(1 ,int(ceil(n / 2.0 ) ) ):
_a : int = 2 * i + 1
_a : str = 2 * i
_a : Any = total + 4 * odd**2 - 6 * even
return total
if __name__ == "__main__":
import sys
if len(sys.argv) == 1:
print(solution())
else:
try:
a__ = int(sys.argv[1])
print(solution(n))
except ValueError:
print('''Invalid entry - please enter a number''')
| 15
| 0
|
"""simple docstring"""
import itertools
import os
from collections import Counter, defaultdict
from concurrent.futures import ThreadPoolExecutor, as_completed
import numpy as np
import datasets
from .execute import check_correctness
a__ = '''\
@misc{chen2021evaluating,
title={Evaluating Large Language Models Trained on Code},
author={Mark Chen and Jerry Tworek and Heewoo Jun and Qiming Yuan \
and Henrique Ponde de Oliveira Pinto and Jared Kaplan and Harri Edwards \
and Yuri Burda and Nicholas Joseph and Greg Brockman and Alex Ray \
and Raul Puri and Gretchen Krueger and Michael Petrov and Heidy Khlaaf \
and Girish Sastry and Pamela Mishkin and Brooke Chan and Scott Gray \
and Nick Ryder and Mikhail Pavlov and Alethea Power and Lukasz Kaiser \
and Mohammad Bavarian and Clemens Winter and Philippe Tillet \
and Felipe Petroski Such and Dave Cummings and Matthias Plappert \
and Fotios Chantzis and Elizabeth Barnes and Ariel Herbert-Voss \
and William Hebgen Guss and Alex Nichol and Alex Paino and Nikolas Tezak \
and Jie Tang and Igor Babuschkin and Suchir Balaji and Shantanu Jain \
and William Saunders and Christopher Hesse and Andrew N. Carr \
and Jan Leike and Josh Achiam and Vedant Misra and Evan Morikawa \
and Alec Radford and Matthew Knight and Miles Brundage and Mira Murati \
and Katie Mayer and Peter Welinder and Bob McGrew and Dario Amodei \
and Sam McCandlish and Ilya Sutskever and Wojciech Zaremba},
year={2021},
eprint={2107.03374},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
'''
a__ = '''\
This metric implements the evaluation harness for the HumanEval problem solving dataset
described in the paper "Evaluating Large Language Models Trained on Code"
(https://arxiv.org/abs/2107.03374).
'''
a__ = '''
Calculates how good are predictions given some references, using certain scores
Args:
predictions: list of candidates to evaluate. Each candidates should be a list
of strings with several code candidates to solve the problem.
references: a list with a test for each prediction. Each test should evaluate the
correctness of a code candidate.
k: number of code candidates to consider in the evaluation (Default: [1, 10, 100])
num_workers: number of workers used to evaluate the canidate programs (Default: 4).
timeout:
Returns:
pass_at_k: dict with pass rates for each k
results: dict with granular results of each unittest
Examples:
>>> code_eval = datasets.load_metric("code_eval")
>>> test_cases = ["assert add(2,3)==5"]
>>> candidates = [["def add(a,b): return a*b", "def add(a, b): return a+b"]]
>>> pass_at_k, results = code_eval.compute(references=test_cases, predictions=candidates, k=[1, 2])
>>> print(pass_at_k)
{\'pass@1\': 0.5, \'pass@2\': 1.0}
'''
a__ = '''
################################################################################
!!!WARNING!!!
################################################################################
The "code_eval" metric executes untrusted model-generated code in Python.
Although it is highly unlikely that model-generated code will do something
overtly malicious in response to this test suite, model-generated code may act
destructively due to a lack of model capability or alignment.
Users are strongly encouraged to sandbox this evaluation suite so that it
does not perform destructive actions on their host or network. For more
information on how OpenAI sandboxes its code, see the paper "Evaluating Large
Language Models Trained on Code" (https://arxiv.org/abs/2107.03374).
Once you have read this disclaimer and taken appropriate precautions,
set the environment variable HF_ALLOW_CODE_EVAL="1". Within Python you can to this
with:
>>> import os
>>> os.environ["HF_ALLOW_CODE_EVAL"] = "1"
################################################################################\
'''
a__ = '''The MIT License
Copyright (c) OpenAI (https://openai.com)
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> Optional[Any]:
return datasets.MetricInfo(
# This is the description that will appear on the metrics page.
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(
{
'''predictions''': datasets.Sequence(datasets.Value('''string''' ) ),
'''references''': datasets.Value('''string''' ),
} ) , homepage='''https://github.com/openai/human-eval''' , codebase_urls=['''https://github.com/openai/human-eval'''] , reference_urls=['''https://github.com/openai/human-eval'''] , license=_LICENSE , )
def __lowercase ( self , _a , _a , _a=[1, 1_0, 1_0_0] , _a=4 , _a=3.0 ) -> Tuple:
if os.getenv('''HF_ALLOW_CODE_EVAL''' , 0 ) != "1":
raise ValueError(_WARNING )
if os.name == "nt":
raise NotImplementedError('''This metric is currently not supported on Windows.''' )
with ThreadPoolExecutor(max_workers=_a ) as executor:
_a : Dict = []
_a : str = Counter()
_a : Any = 0
_a : List[str] = defaultdict(_a )
for task_id, (candidates, test_case) in enumerate(zip(_a , _a ) ):
for candidate in candidates:
_a : Any = candidate + '''\n''' + test_case
_a : Optional[Any] = (test_program, timeout, task_id, completion_id[task_id])
_a : Dict = executor.submit(_a , *_a )
futures.append(_a )
completion_id[task_id] += 1
n_samples += 1
for future in as_completed(_a ):
_a : Any = future.result()
results[result["task_id"]].append((result['''completion_id'''], result) )
_a : Dict = [], []
for result in results.values():
result.sort()
_a : str = [r[1]['''passed'''] for r in result]
total.append(len(_a ) )
correct.append(sum(_a ) )
_a : str = np.array(_a )
_a : int = np.array(_a )
_a : List[str] = k
_a : Dict = {F"""pass@{k}""": estimate_pass_at_k(_a , _a , _a ).mean() for k in ks if (total >= k).all()}
return pass_at_k, results
def __UpperCAmelCase ( __a : Tuple ,__a : List[str] ,__a : Any ) -> Tuple:
"""simple docstring"""
def estimator(__a : int ,__a : int ,__a : int ) -> float:
if n - c < k:
return 1.0
return 1.0 - np.prod(1.0 - k / np.arange(n - c + 1 ,n + 1 ) )
if isinstance(__a ,__a ):
_a : List[str] = itertools.repeat(__a ,len(__a ) )
else:
assert len(__a ) == len(__a )
_a : Optional[Any] = iter(__a )
return np.array([estimator(int(__a ) ,int(__a ) ,__a ) for n, c in zip(__a ,__a )] )
| 352
|
from typing import Dict, Iterable, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format, to_pil_image
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_pytesseract_available, is_vision_available, logging, requires_backends
if is_vision_available():
import PIL
# soft dependency
if is_pytesseract_available():
import pytesseract
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : str ,__a : Union[str, Any] ) -> List[str]:
"""simple docstring"""
return [
int(1_000 * (box[0] / width) ),
int(1_000 * (box[1] / height) ),
int(1_000 * (box[2] / width) ),
int(1_000 * (box[3] / height) ),
]
def __UpperCAmelCase ( __a : np.ndarray ,__a : Optional[str] ,__a : Optional[str] ) -> List[Any]:
"""simple docstring"""
_a : str = to_pil_image(__a )
_a , _a : Optional[Any] = pil_image.size
_a : Tuple = pytesseract.image_to_data(__a ,lang=__a ,output_type='''dict''' ,config=__a )
_a , _a , _a , _a , _a : List[str] = data['''text'''], data['''left'''], data['''top'''], data['''width'''], data['''height''']
# filter empty words and corresponding coordinates
_a : Dict = [idx for idx, word in enumerate(__a ) if not word.strip()]
_a : str = [word for idx, word in enumerate(__a ) if idx not in irrelevant_indices]
_a : List[str] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : str = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
# turn coordinates into (left, top, left+width, top+height) format
_a : int = []
for x, y, w, h in zip(__a ,__a ,__a ,__a ):
_a : List[str] = [x, y, x + w, y + h]
actual_boxes.append(__a )
# finally, normalize the bounding boxes
_a : Dict = []
for box in actual_boxes:
normalized_boxes.append(normalize_box(__a ,__a ,__a ) )
assert len(__a ) == len(__a ), "Not as many words as there are bounding boxes"
return words, normalized_boxes
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[int] = ["pixel_values"]
def __init__( self , _a = True , _a = None , _a = PILImageResampling.BILINEAR , _a = True , _a = 1 / 2_5_5 , _a = True , _a = None , _a = None , _a = True , _a = None , _a = "" , **_a , ) -> None:
super().__init__(**_a )
_a : List[str] = size if size is not None else {'''height''': 2_2_4, '''width''': 2_2_4}
_a : Union[str, Any] = get_size_dict(_a )
_a : int = do_resize
_a : Optional[int] = size
_a : str = resample
_a : str = do_rescale
_a : Any = rescale_value
_a : Optional[Any] = do_normalize
_a : int = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
_a : List[str] = image_std if image_std is not None else IMAGENET_STANDARD_STD
_a : List[Any] = apply_ocr
_a : Optional[int] = ocr_lang
_a : Tuple = tesseract_config
def __lowercase ( self , _a , _a , _a = PILImageResampling.BILINEAR , _a = None , **_a , ) -> np.ndarray:
_a : Any = get_size_dict(_a )
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()}""" )
_a : Optional[int] = (size['''height'''], size['''width'''])
return resize(_a , size=_a , resample=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
return rescale(_a , scale=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a , _a = None , **_a , ) -> np.ndarray:
return normalize(_a , mean=_a , std=_a , data_format=_a , **_a )
def __lowercase ( 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:
_a : Optional[int] = do_resize if do_resize is not None else self.do_resize
_a : Union[str, Any] = size if size is not None else self.size
_a : Any = get_size_dict(_a )
_a : List[str] = resample if resample is not None else self.resample
_a : int = do_rescale if do_rescale is not None else self.do_rescale
_a : Union[str, Any] = rescale_factor if rescale_factor is not None else self.rescale_factor
_a : int = do_normalize if do_normalize is not None else self.do_normalize
_a : str = image_mean if image_mean is not None else self.image_mean
_a : Tuple = image_std if image_std is not None else self.image_std
_a : Any = apply_ocr if apply_ocr is not None else self.apply_ocr
_a : int = ocr_lang if ocr_lang is not None else self.ocr_lang
_a : Optional[int] = tesseract_config if tesseract_config is not None else self.tesseract_config
_a : List[Any] = make_list_of_images(_a )
if not valid_images(_a ):
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_normalize and (image_mean is None or image_std is None):
raise ValueError('''If do_normalize is True, image_mean and image_std must be specified.''' )
# All transformations expect numpy arrays.
_a : Any = [to_numpy_array(_a ) for image in images]
# Tesseract OCR to get words + normalized bounding boxes
if apply_ocr:
requires_backends(self , '''pytesseract''' )
_a : str = []
_a : str = []
for image in images:
_a , _a : Union[str, Any] = apply_tesseract(_a , _a , _a )
words_batch.append(_a )
boxes_batch.append(_a )
if do_resize:
_a : List[str] = [self.resize(image=_a , size=_a , resample=_a ) for image in images]
if do_rescale:
_a : Optional[Any] = [self.rescale(image=_a , scale=_a ) for image in images]
if do_normalize:
_a : List[Any] = [self.normalize(image=_a , mean=_a , std=_a ) for image in images]
_a : List[str] = [to_channel_dimension_format(_a , _a ) for image in images]
_a : List[str] = BatchFeature(data={'''pixel_values''': images} , tensor_type=_a )
if apply_ocr:
_a : Optional[int] = words_batch
_a : List[Any] = boxes_batch
return data
| 15
| 0
|
"""simple docstring"""
import dataclasses
import json
import warnings
from dataclasses import dataclass, field
from time import time
from typing import List
from ..utils import logging
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : Union[str, Any]=None ,__a : Dict=None ) -> List[Any]:
"""simple docstring"""
return field(default_factory=lambda: default ,metadata=__a )
@dataclass
class UpperCAmelCase_ :
"""simple docstring"""
UpperCAmelCase__ : List[str] = list_field(
default=[] , metadata={
"help": (
"Model checkpoints to be provided to the AutoModel classes. Leave blank to benchmark the base version"
" of all available models"
)
} , )
UpperCAmelCase__ : List[int] = list_field(
default=[8] , metadata={"help": "List of batch sizes for which memory and time performance will be evaluated"} )
UpperCAmelCase__ : List[int] = list_field(
default=[8, 32, 128, 512] , metadata={"help": "List of sequence lengths for which memory and time performance will be evaluated"} , )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={"help": "Whether to benchmark inference of model. Inference can be disabled via --no-inference."} , )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={"help": "Whether to run on available cuda devices. Cuda can be disabled via --no-cuda."} , )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={"help": "Whether to run on available tpu devices. TPU can be disabled via --no-tpu."} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Use FP16 to accelerate inference."} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Benchmark training of model"} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Verbose memory tracing"} )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={"help": "Whether to perform speed measurements. Speed measurements can be disabled via --no-speed."} , )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={
"help": "Whether to perform memory measurements. Memory measurements can be disabled via --no-memory"
} , )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Trace memory line by line"} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Save result to a CSV file"} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Save all print statements in a log file"} )
UpperCAmelCase__ : bool = field(default=__lowercase , metadata={"help": "Whether to print environment information"} )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={
"help": (
"Whether to use multiprocessing for memory and speed measurement. It is highly recommended to use"
" multiprocessing for accurate CPU and GPU memory measurements. This option should only be disabled"
" for debugging / testing and on TPU."
)
} , )
UpperCAmelCase__ : str = field(
default=F'''inference_time_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving time results to csv."} , )
UpperCAmelCase__ : str = field(
default=F'''inference_memory_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving memory results to csv."} , )
UpperCAmelCase__ : str = field(
default=F'''train_time_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving time results to csv for training."} , )
UpperCAmelCase__ : str = field(
default=F'''train_memory_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving memory results to csv for training."} , )
UpperCAmelCase__ : str = field(
default=F'''env_info_{round(time() )}.csv''' , metadata={"help": "CSV filename used if saving environment information."} , )
UpperCAmelCase__ : str = field(
default=F'''log_{round(time() )}.csv''' , metadata={"help": "Log filename used if print statements are saved in log."} , )
UpperCAmelCase__ : int = field(default=3 , metadata={"help": "Times an experiment will be run."} )
UpperCAmelCase__ : bool = field(
default=__lowercase , metadata={
"help": (
"Instead of loading the model as defined in `config.architectures` if exists, just load the pretrain"
" model weights."
)
} , )
def __lowercase ( self ) -> Union[str, Any]:
warnings.warn(
F"""The class {self.__class__} is deprecated. Hugging Face Benchmarking utils"""
''' are deprecated in general and it is advised to use external Benchmarking libraries '''
''' to benchmark Transformer models.''' , _a , )
def __lowercase ( self ) -> List[Any]:
return json.dumps(dataclasses.asdict(self ) , indent=2 )
@property
def __lowercase ( self ) -> List[str]:
if len(self.models ) <= 0:
raise ValueError(
'''Please make sure you provide at least one model name / model identifier, *e.g.* `--models'''
''' bert-base-cased` or `args.models = [\'bert-base-cased\'].''' )
return self.models
@property
def __lowercase ( self ) -> List[str]:
if not self.multi_process:
return False
elif self.is_tpu:
logger.info('''Multiprocessing is currently not possible on TPU.''' )
return False
else:
return True
| 353
|
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from argparse import ArgumentParser
from accelerate.commands.config import get_config_parser
from accelerate.commands.env import env_command_parser
from accelerate.commands.launch import launch_command_parser
from accelerate.commands.test import test_command_parser
from accelerate.commands.tpu import tpu_command_parser
def __UpperCAmelCase ( ) -> Optional[Any]:
"""simple docstring"""
_a : int = ArgumentParser('''Accelerate CLI tool''' ,usage='''accelerate <command> [<args>]''' ,allow_abbrev=__a )
_a : Optional[int] = parser.add_subparsers(help='''accelerate command helpers''' )
# Register commands
get_config_parser(subparsers=__a )
env_command_parser(subparsers=__a )
launch_command_parser(subparsers=__a )
tpu_command_parser(subparsers=__a )
test_command_parser(subparsers=__a )
# Let's go
_a : Dict = parser.parse_args()
if not hasattr(__a ,'''func''' ):
parser.print_help()
exit(1 )
# Run
args.func(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
"""simple docstring"""
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_tf_available,
is_tokenizers_available,
is_torch_available,
)
a__ = {
'''configuration_lxmert''': ['''LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''LxmertConfig'''],
'''tokenization_lxmert''': ['''LxmertTokenizer'''],
}
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = ['''LxmertTokenizerFast''']
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''LxmertEncoder''',
'''LxmertForPreTraining''',
'''LxmertForQuestionAnswering''',
'''LxmertModel''',
'''LxmertPreTrainedModel''',
'''LxmertVisualFeatureEncoder''',
'''LxmertXLayer''',
]
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''TF_LXMERT_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''TFLxmertForPreTraining''',
'''TFLxmertMainLayer''',
'''TFLxmertModel''',
'''TFLxmertPreTrainedModel''',
'''TFLxmertVisualFeatureEncoder''',
]
if TYPE_CHECKING:
from .configuration_lxmert import LXMERT_PRETRAINED_CONFIG_ARCHIVE_MAP, LxmertConfig
from .tokenization_lxmert import LxmertTokenizer
try:
if not is_tokenizers_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .tokenization_lxmert_fast import LxmertTokenizerFast
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_lxmert import (
LxmertEncoder,
LxmertForPreTraining,
LxmertForQuestionAnswering,
LxmertModel,
LxmertPreTrainedModel,
LxmertVisualFeatureEncoder,
LxmertXLayer,
)
try:
if not is_tf_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_tf_lxmert import (
TF_LXMERT_PRETRAINED_MODEL_ARCHIVE_LIST,
TFLxmertForPreTraining,
TFLxmertMainLayer,
TFLxmertModel,
TFLxmertPreTrainedModel,
TFLxmertVisualFeatureEncoder,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 354
|
import itertools
import os
import random
import tempfile
import unittest
import numpy as np
from transformers import TvltFeatureExtractor, is_datasets_available
from transformers.testing_utils import check_json_file_has_correct_format, require_torch, require_torchaudio
from transformers.utils.import_utils import is_torch_available
from ...test_sequence_feature_extraction_common import SequenceFeatureExtractionTestMixin
if is_torch_available():
import torch
if is_datasets_available():
from datasets import load_dataset
a__ = random.Random()
def __UpperCAmelCase ( __a : Tuple ,__a : str=1.0 ,__a : Optional[int]=None ,__a : List[Any]=None ) -> Any:
"""simple docstring"""
if rng is None:
_a : Dict = global_rng
_a : Optional[Any] = []
for batch_idx in range(shape[0] ):
values.append([] )
for _ in range(shape[1] ):
values[-1].append(rng.random() * scale )
return values
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __init__( self , _a , _a=7 , _a=4_0_0 , _a=2_0_0_0 , _a=2_0_4_8 , _a=1_2_8 , _a=1 , _a=5_1_2 , _a=3_0 , _a=4_4_1_0_0 , ) -> List[Any]:
_a : Optional[Any] = parent
_a : str = batch_size
_a : List[str] = min_seq_length
_a : str = max_seq_length
_a : Dict = (self.max_seq_length - self.min_seq_length) // (self.batch_size - 1)
_a : List[Any] = spectrogram_length
_a : List[str] = feature_size
_a : List[Any] = num_audio_channels
_a : Tuple = hop_length
_a : Optional[int] = chunk_length
_a : int = sampling_rate
def __lowercase ( self ) -> Union[str, Any]:
return {
"spectrogram_length": self.spectrogram_length,
"feature_size": self.feature_size,
"num_audio_channels": self.num_audio_channels,
"hop_length": self.hop_length,
"chunk_length": self.chunk_length,
"sampling_rate": self.sampling_rate,
}
def __lowercase ( self , _a=False , _a=False ) -> List[Any]:
def _flatten(_a ):
return list(itertools.chain(*_a ) )
if equal_length:
_a : List[Any] = [floats_list((self.max_seq_length, self.feature_size) ) for _ in range(self.batch_size )]
else:
# make sure that inputs increase in size
_a : List[Any] = [
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 : str = [np.asarray(_a ) for x in speech_inputs]
return speech_inputs
@require_torch
@require_torchaudio
class UpperCAmelCase_ ( __lowercase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = TvltFeatureExtractor
def __lowercase ( self ) -> Dict:
_a : List[str] = TvltFeatureExtractionTester(self )
def __lowercase ( self ) -> Any:
_a : List[Any] = self.feature_extraction_class(**self.feat_extract_dict )
self.assertTrue(hasattr(_a , '''spectrogram_length''' ) )
self.assertTrue(hasattr(_a , '''feature_size''' ) )
self.assertTrue(hasattr(_a , '''num_audio_channels''' ) )
self.assertTrue(hasattr(_a , '''hop_length''' ) )
self.assertTrue(hasattr(_a , '''chunk_length''' ) )
self.assertTrue(hasattr(_a , '''sampling_rate''' ) )
def __lowercase ( self ) -> Optional[int]:
_a : Optional[Any] = self.feature_extraction_class(**self.feat_extract_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : int = feat_extract_first.save_pretrained(_a )[0]
check_json_file_has_correct_format(_a )
_a : Dict = self.feature_extraction_class.from_pretrained(_a )
_a : List[Any] = feat_extract_first.to_dict()
_a : Union[str, Any] = feat_extract_second.to_dict()
_a : Any = dict_first.pop('''mel_filters''' )
_a : int = dict_second.pop('''mel_filters''' )
self.assertTrue(np.allclose(_a , _a ) )
self.assertEqual(_a , _a )
def __lowercase ( self ) -> Optional[int]:
_a : Any = self.feature_extraction_class(**self.feat_extract_dict )
with tempfile.TemporaryDirectory() as tmpdirname:
_a : Optional[int] = os.path.join(_a , '''feat_extract.json''' )
feat_extract_first.to_json_file(_a )
_a : List[str] = self.feature_extraction_class.from_json_file(_a )
_a : List[Any] = feat_extract_first.to_dict()
_a : Dict = feat_extract_second.to_dict()
_a : str = dict_first.pop('''mel_filters''' )
_a : str = dict_second.pop('''mel_filters''' )
self.assertTrue(np.allclose(_a , _a ) )
self.assertEqual(_a , _a )
def __lowercase ( self ) -> Union[str, Any]:
# Initialize feature_extractor
_a : Union[str, Any] = self.feature_extraction_class(**self.feat_extract_dict )
# create three inputs of length 800, 1000, and 1200
_a : Any = [floats_list((1, x) )[0] for x in range(8_0_0 , 1_4_0_0 , 2_0_0 )]
_a : List[str] = [np.asarray(_a ) for speech_input in speech_inputs]
# Test not batched input
_a : Tuple = feature_extractor(np_speech_inputs[0] , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test batched
_a : Dict = feature_extractor(_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test audio masking
_a : Union[str, Any] = feature_extractor(
_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 , mask_audio=_a ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
# Test 2-D numpy arrays are batched.
_a : Optional[Any] = [floats_list((1, x) )[0] for x in (8_0_0, 8_0_0, 8_0_0)]
_a : int = np.asarray(_a )
_a : Tuple = feature_extractor(_a , return_tensors='''np''' , sampling_rate=4_4_1_0_0 ).audio_values
self.assertTrue(encoded_audios.ndim == 4 )
self.assertTrue(encoded_audios.shape[-1] == feature_extractor.feature_size )
self.assertTrue(encoded_audios.shape[-2] <= feature_extractor.spectrogram_length )
self.assertTrue(encoded_audios.shape[-3] == feature_extractor.num_channels )
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[Any] = load_dataset('''hf-internal-testing/librispeech_asr_dummy''' , '''clean''' , split='''validation''' )
# automatic decoding with librispeech
_a : Optional[int] = ds.sort('''id''' ).select(range(_a ) )[:num_samples]['''audio''']
return [x["array"] for x in speech_samples]
def __lowercase ( self ) -> int:
_a : Union[str, Any] = self._load_datasamples(1 )
_a : int = TvltFeatureExtractor()
_a : Union[str, Any] = feature_extractor(_a , return_tensors='''pt''' ).audio_values
self.assertEquals(audio_values.shape , (1, 1, 1_9_2, 1_2_8) )
_a : Union[str, Any] = torch.tensor([[-0.3032, -0.2708], [-0.4434, -0.4007]] )
self.assertTrue(torch.allclose(audio_values[0, 0, :2, :2] , _a , atol=1e-4 ) )
| 15
| 0
|
from typing import Dict, Iterable, List, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import (
center_crop,
get_resize_output_image_size,
normalize,
rescale,
resize,
to_channel_dimension_format,
)
from ...image_utils import (
IMAGENET_DEFAULT_MEAN,
IMAGENET_DEFAULT_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, logging
a__ = logging.get_logger(__name__)
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Tuple = ["pixel_values"]
def __init__( self , _a = True , _a = None , _a = PILImageResampling.BICUBIC , _a = True , _a = None , _a = True , _a = 1 / 2_5_5 , _a = True , _a = IMAGENET_DEFAULT_MEAN , _a = IMAGENET_DEFAULT_STD , **_a , ) -> None:
super().__init__(**_a )
_a : int = size if size is not None else {'''shortest_edge''': 2_2_4}
_a : List[str] = get_size_dict(_a , default_to_square=_a )
_a : Tuple = crop_size if crop_size is not None else {'''height''': 2_2_4, '''width''': 2_2_4}
_a : Optional[int] = get_size_dict(_a , param_name='''crop_size''' )
_a : Optional[int] = do_resize
_a : str = size
_a : List[str] = resample
_a : str = do_center_crop
_a : Optional[Any] = crop_size
_a : Union[str, Any] = do_rescale
_a : str = rescale_factor
_a : List[Any] = do_normalize
_a : Dict = image_mean if image_mean is not None else IMAGENET_DEFAULT_MEAN
_a : Optional[Any] = image_std if image_std is not None else IMAGENET_DEFAULT_STD
def __lowercase ( self , _a , _a , _a = PILImageResampling.BICUBIC , _a = None , **_a , ) -> np.ndarray:
_a : Optional[int] = get_size_dict(_a , default_to_square=_a )
# size_dict is a dict with either keys "height" and "width" or "shortest_edge"
if "shortest_edge" in size:
_a : Optional[int] = int((2_5_6 / 2_2_4) * size['''shortest_edge'''] )
_a : Any = get_resize_output_image_size(_a , size=_a , default_to_square=_a )
_a : Dict = {'''height''': output_size[0], '''width''': output_size[1]}
if "height" not in size_dict or "width" not in size_dict:
raise ValueError(
F"""Size dict must have keys 'height' and 'width' or 'shortest_edge'. Got {size_dict.keys()}""" )
return resize(
_a , size=(size_dict['''height'''], size_dict['''width''']) , resample=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
_a : int = get_size_dict(_a )
if "height" not in size or "width" not in size:
raise ValueError(F"""Size dict must have keys 'height' and 'width'. Got {size.keys()}""" )
return center_crop(_a , size=(size['''height'''], size['''width''']) , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
return rescale(_a , scale=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a , _a = None , **_a , ) -> np.ndarray:
return normalize(_a , mean=_a , std=_a , data_format=_a , **_a )
def __lowercase ( 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 , **_a , ) -> BatchFeature:
_a : int = do_resize if do_resize is not None else self.do_resize
_a : Optional[Any] = resample if resample is not None else self.resample
_a : Optional[int] = do_center_crop if do_center_crop is not None else self.do_center_crop
_a : Any = do_rescale if do_rescale is not None else self.do_rescale
_a : List[str] = rescale_factor if rescale_factor is not None else self.rescale_factor
_a : Dict = do_normalize if do_normalize is not None else self.do_normalize
_a : Union[str, Any] = image_mean if image_mean is not None else self.image_mean
_a : Union[str, Any] = image_std if image_std is not None else self.image_std
_a : Union[str, Any] = size if size is not None else self.size
_a : str = get_size_dict(_a , default_to_square=_a )
_a : Optional[int] = crop_size if crop_size is not None else self.crop_size
_a : Union[str, Any] = get_size_dict(_a , param_name='''crop_size''' )
_a : int = make_list_of_images(_a )
if not valid_images(_a ):
raise ValueError(
'''Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, '''
'''torch.Tensor, tf.Tensor or jax.ndarray.''' )
if do_resize and size is None:
raise ValueError('''Size must be specified if do_resize is True.''' )
if do_center_crop and crop_size is None:
raise ValueError('''Crop size must be specified if do_center_crop is True.''' )
if do_rescale and rescale_factor is None:
raise ValueError('''Rescale factor must be specified if do_rescale is True.''' )
if do_normalize and (image_mean is None or image_std is None):
raise ValueError('''Image mean and std must be specified if do_normalize is True.''' )
# All transformations expect numpy arrays.
_a : Tuple = [to_numpy_array(_a ) for image in images]
if do_resize:
_a : int = [self.resize(_a , _a , _a ) for image in images]
if do_center_crop:
_a : Any = [self.center_crop(_a , _a ) for image in images]
if do_rescale:
_a : Any = [self.rescale(_a , _a ) for image in images]
if do_normalize:
_a : Optional[int] = [self.normalize(_a , _a , _a ) for image in images]
_a : Optional[int] = [to_channel_dimension_format(_a , _a ) for image in images]
_a : Dict = {'''pixel_values''': images}
return BatchFeature(data=_a , tensor_type=_a )
| 355
|
from typing import Dict
import numpy as np
from ..utils import add_end_docstrings, is_tf_available, is_torch_available, logging
from .base import PIPELINE_INIT_ARGS, GenericTensor, Pipeline, PipelineException
if is_tf_available():
import tensorflow as tf
from ..tf_utils import stable_softmax
if is_torch_available():
import torch
a__ = logging.get_logger(__name__)
@add_end_docstrings(
__lowercase , r"\n top_k (`int`, defaults to 5):\n The number of predictions to return.\n targets (`str` or `List[str]`, *optional*):\n When passed, the model will limit the scores to the passed targets instead of looking up in the whole\n vocab. If the provided targets are not in the model vocab, they will be tokenized and the first resulting\n token will be used (with a warning, and that might be slower).\n\n " , )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self , _a ) -> np.ndarray:
if self.framework == "tf":
_a : List[str] = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()
elif self.framework == "pt":
_a : Tuple = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a )
else:
raise ValueError('''Unsupported framework''' )
return masked_index
def __lowercase ( self , _a ) -> np.ndarray:
_a : int = self.get_masked_index(_a )
_a : Tuple = np.prod(masked_index.shape )
if numel < 1:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , F"""No mask_token ({self.tokenizer.mask_token}) found on the input""" , )
def __lowercase ( self , _a ) -> Optional[int]:
if isinstance(_a , _a ):
for model_input in model_inputs:
self._ensure_exactly_one_mask_token(model_input['''input_ids'''][0] )
else:
for input_ids in model_inputs["input_ids"]:
self._ensure_exactly_one_mask_token(_a )
def __lowercase ( self , _a , _a=None , **_a ) -> Dict[str, GenericTensor]:
if return_tensors is None:
_a : Union[str, Any] = self.framework
_a : str = self.tokenizer(_a , return_tensors=_a )
self.ensure_exactly_one_mask_token(_a )
return model_inputs
def __lowercase ( self , _a ) -> Optional[Any]:
_a : List[str] = self.model(**_a )
_a : Any = model_inputs['''input_ids''']
return model_outputs
def __lowercase ( self , _a , _a=5 , _a=None ) -> str:
# Cap top_k if there are targets
if target_ids is not None and target_ids.shape[0] < top_k:
_a : List[Any] = target_ids.shape[0]
_a : Any = model_outputs['''input_ids'''][0]
_a : List[str] = model_outputs['''logits''']
if self.framework == "tf":
_a : Tuple = tf.where(input_ids == self.tokenizer.mask_token_id ).numpy()[:, 0]
_a : List[str] = outputs.numpy()
_a : Dict = outputs[0, masked_index, :]
_a : str = stable_softmax(_a , axis=-1 )
if target_ids is not None:
_a : Any = tf.gather_nd(tf.squeeze(_a , 0 ) , target_ids.reshape(-1 , 1 ) )
_a : Union[str, Any] = tf.expand_dims(_a , 0 )
_a : Optional[int] = tf.math.top_k(_a , k=_a )
_a , _a : Optional[Any] = topk.values.numpy(), topk.indices.numpy()
else:
_a : Optional[Any] = torch.nonzero(input_ids == self.tokenizer.mask_token_id , as_tuple=_a ).squeeze(-1 )
# Fill mask pipeline supports only one ${mask_token} per sample
_a : List[str] = outputs[0, masked_index, :]
_a : List[Any] = logits.softmax(dim=-1 )
if target_ids is not None:
_a : List[Any] = probs[..., target_ids]
_a , _a : Optional[Any] = probs.topk(_a )
_a : Dict = []
_a : List[Any] = values.shape[0] == 1
for i, (_values, _predictions) in enumerate(zip(values.tolist() , predictions.tolist() ) ):
_a : Optional[Any] = []
for v, p in zip(_values , _predictions ):
# Copy is important since we're going to modify this array in place
_a : Optional[int] = input_ids.numpy().copy()
if target_ids is not None:
_a : Tuple = target_ids[p].tolist()
_a : List[str] = p
# Filter padding out:
_a : List[Any] = tokens[np.where(tokens != self.tokenizer.pad_token_id )]
# Originally we skip special tokens to give readable output.
# For multi masks though, the other [MASK] would be removed otherwise
# making the output look odd, so we add them back
_a : List[str] = self.tokenizer.decode(_a , skip_special_tokens=_a )
_a : List[Any] = {'''score''': v, '''token''': p, '''token_str''': self.tokenizer.decode([p] ), '''sequence''': sequence}
row.append(_a )
result.append(_a )
if single_mask:
return result[0]
return result
def __lowercase ( self , _a , _a=None ) -> Dict:
if isinstance(_a , _a ):
_a : Tuple = [targets]
try:
_a : int = self.tokenizer.get_vocab()
except Exception:
_a : Any = {}
_a : List[Any] = []
for target in targets:
_a : List[Any] = vocab.get(_a , _a )
if id_ is None:
_a : Tuple = self.tokenizer(
_a , add_special_tokens=_a , return_attention_mask=_a , return_token_type_ids=_a , max_length=1 , truncation=_a , )['''input_ids''']
if len(_a ) == 0:
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
'''We cannot replace it with anything meaningful, ignoring it''' )
continue
_a : Tuple = input_ids[0]
# XXX: If users encounter this pass
# it becomes pretty slow, so let's make sure
# The warning enables them to fix the input to
# get faster performance.
logger.warning(
F"""The specified target token `{target}` does not exist in the model vocabulary. """
F"""Replacing with `{self.tokenizer.convert_ids_to_tokens(id_ )}`.""" )
target_ids.append(id_ )
_a : List[str] = list(set(_a ) )
if len(_a ) == 0:
raise ValueError('''At least one target must be provided when passed.''' )
_a : int = np.array(_a )
return target_ids
def __lowercase ( self , _a=None , _a=None ) -> Tuple:
_a : str = {}
if targets is not None:
_a : List[Any] = self.get_target_ids(_a , _a )
_a : Optional[Any] = target_ids
if top_k is not None:
_a : Union[str, Any] = top_k
if self.tokenizer.mask_token_id is None:
raise PipelineException(
'''fill-mask''' , self.model.base_model_prefix , '''The tokenizer does not define a `mask_token`.''' )
return {}, {}, postprocess_params
def __call__( self , _a , *_a , **_a ) -> int:
_a : Optional[Any] = super().__call__(_a , **_a )
if isinstance(_a , _a ) and len(_a ) == 1:
return outputs[0]
return outputs
| 15
| 0
|
import argparse
import json
import os
import fairseq
import torch
from fairseq.data import Dictionary
from transformers import (
HubertConfig,
HubertForCTC,
HubertModel,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaProcessor,
logging,
)
logging.set_verbosity_info()
a__ = logging.get_logger(__name__)
a__ = {
'''post_extract_proj''': '''feature_projection.projection''',
'''encoder.pos_conv.0''': '''encoder.pos_conv_embed.conv''',
'''self_attn.k_proj''': '''encoder.layers.*.attention.k_proj''',
'''self_attn.v_proj''': '''encoder.layers.*.attention.v_proj''',
'''self_attn.q_proj''': '''encoder.layers.*.attention.q_proj''',
'''self_attn.out_proj''': '''encoder.layers.*.attention.out_proj''',
'''self_attn_layer_norm''': '''encoder.layers.*.layer_norm''',
'''fc1''': '''encoder.layers.*.feed_forward.intermediate_dense''',
'''fc2''': '''encoder.layers.*.feed_forward.output_dense''',
'''final_layer_norm''': '''encoder.layers.*.final_layer_norm''',
'''encoder.layer_norm''': '''encoder.layer_norm''',
'''w2v_model.layer_norm''': '''feature_projection.layer_norm''',
'''w2v_encoder.proj''': '''lm_head''',
'''mask_emb''': '''masked_spec_embed''',
}
def __UpperCAmelCase ( __a : List[Any] ,__a : Optional[int] ,__a : Optional[int] ,__a : List[str] ,__a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
for attribute in key.split('''.''' ):
_a : Optional[Any] = getattr(__a ,__a )
if weight_type is not None:
_a : Dict = getattr(__a ,__a ).shape
else:
_a : Optional[int] = 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":
_a : List[Any] = value
elif weight_type == "weight_g":
_a : Any = value
elif weight_type == "weight_v":
_a : Union[str, Any] = value
elif weight_type == "bias":
_a : Optional[int] = value
else:
_a : List[Any] = value
logger.info(F"""{key + '.' + weight_type if weight_type is not None else ''} was initialized from {full_name}.""" )
def __UpperCAmelCase ( __a : Any ,__a : Union[str, Any] ,__a : Union[str, Any] ) -> int:
"""simple docstring"""
_a : Union[str, Any] = []
_a : Union[str, Any] = fairseq_model.state_dict()
_a : Union[str, Any] = hf_model.hubert.feature_extractor if is_finetuned else hf_model.feature_extractor
for name, value in fairseq_dict.items():
_a : int = False
if "conv_layers" in name:
load_conv_layer(
__a ,__a ,__a ,__a ,hf_model.config.feat_extract_norm == '''group''' ,)
_a : Optional[Any] = True
else:
for key, mapped_key in MAPPING.items():
_a : Union[str, Any] = '''hubert.''' + 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] and not is_finetuned):
_a : Any = True
if "*" in mapped_key:
_a : Optional[int] = name.split(__a )[0].split('''.''' )[-2]
_a : Any = mapped_key.replace('''*''' ,__a )
if "weight_g" in name:
_a : List[Any] = '''weight_g'''
elif "weight_v" in name:
_a : List[str] = '''weight_v'''
elif "weight" in name:
_a : Any = '''weight'''
elif "bias" in name:
_a : str = '''bias'''
else:
_a : Any = None
set_recursively(__a ,__a ,__a ,__a ,__a )
continue
if not is_used:
unused_weights.append(__a )
logger.warning(F"""Unused weights: {unused_weights}""" )
def __UpperCAmelCase ( __a : int ,__a : Optional[Any] ,__a : Dict ,__a : List[str] ,__a : Any ) -> Tuple:
"""simple docstring"""
_a : int = full_name.split('''conv_layers.''' )[-1]
_a : Any = name.split('''.''' )
_a : List[Any] = int(items[0] )
_a : Optional[int] = 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."""
)
_a : Optional[int] = 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."""
)
_a : Optional[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:
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."
)
_a : int = 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."""
)
_a : Any = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(__a )
@torch.no_grad()
def __UpperCAmelCase ( __a : Dict ,__a : List[Any] ,__a : List[str]=None ,__a : Optional[int]=None ,__a : int=True ) -> List[Any]:
"""simple docstring"""
if config_path is not None:
_a : Tuple = HubertConfig.from_pretrained(__a )
else:
_a : Any = HubertConfig()
if is_finetuned:
if dict_path:
_a : Tuple = Dictionary.load(__a )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
_a : Any = target_dict.pad_index
_a : Tuple = target_dict.bos_index
_a : Optional[int] = target_dict.eos_index
_a : Optional[Any] = len(target_dict.symbols )
_a : Tuple = os.path.join(__a ,'''vocab.json''' )
if not os.path.isdir(__a ):
logger.error('''--pytorch_dump_folder_path ({}) should be a directory'''.format(__a ) )
return
os.makedirs(__a ,exist_ok=__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ) as vocab_handle:
json.dump(target_dict.indices ,__a )
_a : Tuple = WavaVecaCTCTokenizer(
__a ,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=__a ,)
_a : Tuple = True if config.feat_extract_norm == '''layer''' else False
_a : List[Any] = WavaVecaFeatureExtractor(
feature_size=1 ,sampling_rate=16_000 ,padding_value=0 ,do_normalize=__a ,return_attention_mask=__a ,)
_a : List[Any] = WavaVecaProcessor(feature_extractor=__a ,tokenizer=__a )
processor.save_pretrained(__a )
_a : Tuple = HubertForCTC(__a )
else:
_a : Tuple = HubertModel(__a )
if is_finetuned:
_a : int = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
_a : str = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] )
_a : Any = model[0].eval()
recursively_load_weights(__a ,__a ,__a )
hf_wavavec.save_pretrained(__a )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
parser.add_argument('''--checkpoint_path''', default=None, type=str, help='''Path to fairseq checkpoint''')
parser.add_argument('''--dict_path''', default=None, type=str, help='''Path to dict of fine-tuned model''')
parser.add_argument('''--config_path''', default=None, type=str, help='''Path to hf config.json of model to convert''')
parser.add_argument(
'''--not_finetuned''', action='''store_true''', help='''Whether the model to convert is a fine-tuned model or not'''
)
a__ = parser.parse_args()
convert_hubert_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 356
|
import argparse
import json
import logging
import os
import sys
from unittest.mock import patch
from transformers.testing_utils import TestCasePlus, get_gpu_count, slow
a__ = [
os.path.join(os.path.dirname(__file__), dirname)
for dirname in [
'''text-classification''',
'''language-modeling''',
'''summarization''',
'''token-classification''',
'''question-answering''',
]
]
sys.path.extend(SRC_DIRS)
if SRC_DIRS is not None:
import run_clm_flax
import run_flax_glue
import run_flax_ner
import run_mlm_flax
import run_qa
import run_summarization_flax
import run_ta_mlm_flax
logging.basicConfig(level=logging.DEBUG)
a__ = logging.getLogger()
def __UpperCAmelCase ( ) -> Optional[int]:
"""simple docstring"""
_a : Any = argparse.ArgumentParser()
parser.add_argument('''-f''' )
_a : Dict = parser.parse_args()
return args.f
def __UpperCAmelCase ( __a : Optional[int] ,__a : List[str]="eval" ) -> Any:
"""simple docstring"""
_a : Any = os.path.join(__a ,F"""{split}_results.json""" )
if os.path.exists(__a ):
with open(__a ,'''r''' ) as f:
return json.load(__a )
raise ValueError(F"""can't find {path}""" )
a__ = logging.StreamHandler(sys.stdout)
logger.addHandler(stream_handler)
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self ) -> str:
_a : Any = self.get_auto_remove_tmp_dir()
_a : Optional[Any] = F"""
run_glue.py
--model_name_or_path distilbert-base-uncased
--output_dir {tmp_dir}
--train_file ./tests/fixtures/tests_samples/MRPC/train.csv
--validation_file ./tests/fixtures/tests_samples/MRPC/dev.csv
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
--learning_rate=1e-4
--eval_steps=2
--warmup_steps=2
--seed=42
--max_seq_length=128
""".split()
with patch.object(_a , '''argv''' , _a ):
run_flax_glue.main()
_a : Any = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.75 )
@slow
def __lowercase ( self ) -> Dict:
_a : Tuple = self.get_auto_remove_tmp_dir()
_a : Tuple = F"""
run_clm_flax.py
--model_name_or_path distilgpt2
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--do_train
--do_eval
--block_size 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--num_train_epochs 2
--logging_steps 2 --eval_steps 2
--output_dir {tmp_dir}
--overwrite_output_dir
""".split()
with patch.object(_a , '''argv''' , _a ):
run_clm_flax.main()
_a : List[str] = get_results(_a )
self.assertLess(result['''eval_perplexity'''] , 1_0_0 )
@slow
def __lowercase ( self ) -> Optional[int]:
_a : str = self.get_auto_remove_tmp_dir()
_a : Optional[int] = F"""
run_summarization.py
--model_name_or_path t5-small
--train_file tests/fixtures/tests_samples/xsum/sample.json
--validation_file tests/fixtures/tests_samples/xsum/sample.json
--test_file tests/fixtures/tests_samples/xsum/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--num_train_epochs=3
--warmup_steps=8
--do_train
--do_eval
--do_predict
--learning_rate=2e-4
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
--predict_with_generate
""".split()
with patch.object(_a , '''argv''' , _a ):
run_summarization_flax.main()
_a : Optional[int] = get_results(_a , split='''test''' )
self.assertGreaterEqual(result['''test_rouge1'''] , 1_0 )
self.assertGreaterEqual(result['''test_rouge2'''] , 2 )
self.assertGreaterEqual(result['''test_rougeL'''] , 7 )
self.assertGreaterEqual(result['''test_rougeLsum'''] , 7 )
@slow
def __lowercase ( self ) -> Tuple:
_a : List[str] = self.get_auto_remove_tmp_dir()
_a : List[Any] = F"""
run_mlm.py
--model_name_or_path distilroberta-base
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--output_dir {tmp_dir}
--overwrite_output_dir
--max_seq_length 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--logging_steps 2 --eval_steps 2
--do_train
--do_eval
--num_train_epochs=1
""".split()
with patch.object(_a , '''argv''' , _a ):
run_mlm_flax.main()
_a : List[Any] = get_results(_a )
self.assertLess(result['''eval_perplexity'''] , 4_2 )
@slow
def __lowercase ( self ) -> Dict:
_a : Optional[Any] = self.get_auto_remove_tmp_dir()
_a : int = F"""
run_t5_mlm_flax.py
--model_name_or_path t5-small
--train_file ./tests/fixtures/sample_text.txt
--validation_file ./tests/fixtures/sample_text.txt
--do_train
--do_eval
--max_seq_length 128
--per_device_train_batch_size 4
--per_device_eval_batch_size 4
--num_train_epochs 2
--logging_steps 2 --eval_steps 2
--output_dir {tmp_dir}
--overwrite_output_dir
""".split()
with patch.object(_a , '''argv''' , _a ):
run_ta_mlm_flax.main()
_a : List[Any] = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.42 )
@slow
def __lowercase ( self ) -> Optional[Any]:
# with so little data distributed training needs more epochs to get the score on par with 0/1 gpu
_a : Any = 7 if get_gpu_count() > 1 else 2
_a : List[Any] = self.get_auto_remove_tmp_dir()
_a : List[Any] = F"""
run_flax_ner.py
--model_name_or_path bert-base-uncased
--train_file tests/fixtures/tests_samples/conll/sample.json
--validation_file tests/fixtures/tests_samples/conll/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--do_train
--do_eval
--warmup_steps=2
--learning_rate=2e-4
--logging_steps 2 --eval_steps 2
--per_device_train_batch_size=2
--per_device_eval_batch_size=2
--num_train_epochs={epochs}
--seed 7
""".split()
with patch.object(_a , '''argv''' , _a ):
run_flax_ner.main()
_a : Dict = get_results(_a )
self.assertGreaterEqual(result['''eval_accuracy'''] , 0.75 )
self.assertGreaterEqual(result['''eval_f1'''] , 0.3 )
@slow
def __lowercase ( self ) -> Any:
_a : Optional[int] = self.get_auto_remove_tmp_dir()
_a : Union[str, Any] = F"""
run_qa.py
--model_name_or_path bert-base-uncased
--version_2_with_negative
--train_file tests/fixtures/tests_samples/SQUAD/sample.json
--validation_file tests/fixtures/tests_samples/SQUAD/sample.json
--output_dir {tmp_dir}
--overwrite_output_dir
--num_train_epochs=3
--warmup_steps=2
--do_train
--do_eval
--logging_steps 2 --eval_steps 2
--learning_rate=2e-4
--per_device_train_batch_size=2
--per_device_eval_batch_size=1
""".split()
with patch.object(_a , '''argv''' , _a ):
run_qa.main()
_a : Any = get_results(_a )
self.assertGreaterEqual(result['''eval_f1'''] , 3_0 )
self.assertGreaterEqual(result['''eval_exact'''] , 3_0 )
| 15
| 0
|
from typing import Optional
import pyspark
from .. import Features, NamedSplit
from ..download import DownloadMode
from ..packaged_modules.spark.spark import Spark
from .abc import AbstractDatasetReader
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __init__( self , _a , _a = None , _a = None , _a = True , _a = None , _a = False , _a = None , _a = True , _a = "arrow" , **_a , ) -> Union[str, Any]:
super().__init__(
split=_a , features=_a , cache_dir=_a , keep_in_memory=_a , streaming=_a , **_a , )
_a : Optional[int] = load_from_cache_file
_a : List[Any] = file_format
_a : Optional[int] = Spark(
df=_a , features=_a , cache_dir=_a , working_dir=_a , **_a , )
def __lowercase ( self ) -> Optional[Any]:
if self.streaming:
return self.builder.as_streaming_dataset(split=self.split )
_a : Union[str, Any] = None if self._load_from_cache_file else DownloadMode.FORCE_REDOWNLOAD
self.builder.download_and_prepare(
download_mode=_a , file_format=self._file_format , )
return self.builder.as_dataset(split=self.split )
| 357
|
import argparse
import os
import re
import packaging.version
a__ = '''examples/'''
a__ = {
'''examples''': (re.compile(R'''^check_min_version\("[^"]+"\)\s*$''', re.MULTILINE), '''check_min_version("VERSION")\n'''),
'''init''': (re.compile(R'''^__version__\s+=\s+"([^"]+)"\s*$''', re.MULTILINE), '''__version__ = "VERSION"\n'''),
'''setup''': (re.compile(R'''^(\s*)version\s*=\s*"[^"]+",''', re.MULTILINE), R'''\1version="VERSION",'''),
'''doc''': (re.compile(R'''^(\s*)release\s*=\s*"[^"]+"$''', re.MULTILINE), '''release = "VERSION"\n'''),
}
a__ = {
'''init''': '''src/transformers/__init__.py''',
'''setup''': '''setup.py''',
}
a__ = '''README.md'''
def __UpperCAmelCase ( __a : List[str] ,__a : int ,__a : Optional[Any] ) -> int:
"""simple docstring"""
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Tuple = f.read()
_a , _a : str = REPLACE_PATTERNS[pattern]
_a : List[str] = replace.replace('''VERSION''' ,__a )
_a : List[Any] = re_pattern.sub(__a ,__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.write(__a )
def __UpperCAmelCase ( __a : Any ) -> List[Any]:
"""simple docstring"""
for folder, directories, fnames in os.walk(__a ):
# Removing some of the folders with non-actively maintained examples from the walk
if "research_projects" in directories:
directories.remove('''research_projects''' )
if "legacy" in directories:
directories.remove('''legacy''' )
for fname in fnames:
if fname.endswith('''.py''' ):
update_version_in_file(os.path.join(__a ,__a ) ,__a ,pattern='''examples''' )
def __UpperCAmelCase ( __a : List[Any] ,__a : List[str]=False ) -> int:
"""simple docstring"""
for pattern, fname in REPLACE_FILES.items():
update_version_in_file(__a ,__a ,__a )
if not patch:
update_version_in_examples(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
_a : Optional[Any] = '''🤗 Transformers currently provides the following architectures'''
_a : str = '''1. Want to contribute a new model?'''
with open(__a ,'''r''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
_a : Optional[int] = f.readlines()
# Find the start of the list.
_a : Optional[int] = 0
while not lines[start_index].startswith(_start_prompt ):
start_index += 1
start_index += 1
_a : List[Any] = start_index
# Update the lines in the model list.
while not lines[index].startswith(_end_prompt ):
if lines[index].startswith('''1.''' ):
_a : Tuple = lines[index].replace(
'''https://huggingface.co/docs/transformers/main/model_doc''' ,'''https://huggingface.co/docs/transformers/model_doc''' ,)
index += 1
with open(__a ,'''w''' ,encoding='''utf-8''' ,newline='''\n''' ) as f:
f.writelines(__a )
def __UpperCAmelCase ( ) -> List[str]:
"""simple docstring"""
with open(REPLACE_FILES['''init'''] ,'''r''' ) as f:
_a : Optional[Any] = f.read()
_a : Optional[Any] = REPLACE_PATTERNS['''init'''][0].search(__a ).groups()[0]
return packaging.version.parse(__a )
def __UpperCAmelCase ( __a : Dict=False ) -> str:
"""simple docstring"""
_a : Optional[Any] = get_version()
if patch and default_version.is_devrelease:
raise ValueError('''Can\'t create a patch version from the dev branch, checkout a released version!''' )
if default_version.is_devrelease:
_a : List[Any] = default_version.base_version
elif patch:
_a : str = F"""{default_version.major}.{default_version.minor}.{default_version.micro + 1}"""
else:
_a : List[str] = F"""{default_version.major}.{default_version.minor + 1}.0"""
# Now let's ask nicely if that's the right one.
_a : Dict = input(F"""Which version are you releasing? [{default_version}]""" )
if len(__a ) == 0:
_a : int = default_version
print(F"""Updating version to {version}.""" )
global_version_update(__a ,patch=__a )
if not patch:
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
def __UpperCAmelCase ( ) -> Tuple:
"""simple docstring"""
_a : str = get_version()
_a : int = F"""{current_version.major}.{current_version.minor + 1}.0.dev0"""
_a : List[Any] = current_version.base_version
# Check with the user we got that right.
_a : Union[str, Any] = input(F"""Which version are we developing now? [{dev_version}]""" )
if len(__a ) == 0:
_a : List[str] = dev_version
print(F"""Updating version to {version}.""" )
global_version_update(__a )
print('''Cleaning main README, don\'t forget to run `make fix-copies`.''' )
clean_main_ref_in_model_list()
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--post_release''', action='''store_true''', help='''Whether this is pre or post release.''')
parser.add_argument('''--patch''', action='''store_true''', help='''Whether or not this is a patch release.''')
a__ = parser.parse_args()
if not args.post_release:
pre_release_work(patch=args.patch)
elif args.patch:
print('''Nothing to do after a patch :-)''')
else:
post_release_work()
| 15
| 0
|
import argparse
import json
from pathlib import Path
import requests
import timm
import torch
from huggingface_hub import hf_hub_download
from PIL import Image
from transformers import AutoImageProcessor, SwinvaConfig, SwinvaForImageClassification
def __UpperCAmelCase ( __a : Tuple ) -> Optional[int]:
"""simple docstring"""
_a : List[Any] = SwinvaConfig()
_a : Optional[Any] = swinva_name.split('''_''' )
_a : Union[str, Any] = name_split[1]
if "to" in name_split[3]:
_a : List[str] = int(name_split[3][-3:] )
else:
_a : Dict = int(name_split[3] )
if "to" in name_split[2]:
_a : List[Any] = int(name_split[2][-2:] )
else:
_a : Optional[int] = int(name_split[2][6:] )
if model_size == "tiny":
_a : Tuple = 96
_a : Optional[Any] = (2, 2, 6, 2)
_a : Any = (3, 6, 12, 24)
elif model_size == "small":
_a : Dict = 96
_a : Dict = (2, 2, 18, 2)
_a : Optional[Any] = (3, 6, 12, 24)
elif model_size == "base":
_a : Dict = 128
_a : str = (2, 2, 18, 2)
_a : Union[str, Any] = (4, 8, 16, 32)
else:
_a : Dict = 192
_a : List[str] = (2, 2, 18, 2)
_a : Any = (6, 12, 24, 48)
if "to" in swinva_name:
_a : Optional[Any] = (12, 12, 12, 6)
if ("22k" in swinva_name) and ("to" not in swinva_name):
_a : int = 21_841
_a : str = '''huggingface/label-files'''
_a : int = '''imagenet-22k-id2label.json'''
_a : Any = json.load(open(hf_hub_download(__a ,__a ,repo_type='''dataset''' ) ,'''r''' ) )
_a : Any = {int(__a ): v for k, v in idalabel.items()}
_a : str = idalabel
_a : Any = {v: k for k, v in idalabel.items()}
else:
_a : Optional[Any] = 1_000
_a : str = '''huggingface/label-files'''
_a : Optional[int] = '''imagenet-1k-id2label.json'''
_a : Optional[int] = json.load(open(hf_hub_download(__a ,__a ,repo_type='''dataset''' ) ,'''r''' ) )
_a : Any = {int(__a ): v for k, v in idalabel.items()}
_a : int = idalabel
_a : Dict = {v: k for k, v in idalabel.items()}
_a : Dict = img_size
_a : Any = num_classes
_a : str = embed_dim
_a : Dict = depths
_a : Optional[Any] = num_heads
_a : Optional[Any] = window_size
return config
def __UpperCAmelCase ( __a : Any ) -> str:
"""simple docstring"""
if "patch_embed.proj" in name:
_a : Dict = name.replace('''patch_embed.proj''' ,'''embeddings.patch_embeddings.projection''' )
if "patch_embed.norm" in name:
_a : Tuple = name.replace('''patch_embed.norm''' ,'''embeddings.norm''' )
if "layers" in name:
_a : Union[str, Any] = '''encoder.''' + name
if "attn.proj" in name:
_a : int = name.replace('''attn.proj''' ,'''attention.output.dense''' )
if "attn" in name:
_a : int = name.replace('''attn''' ,'''attention.self''' )
if "norm1" in name:
_a : Any = name.replace('''norm1''' ,'''layernorm_before''' )
if "norm2" in name:
_a : Optional[Any] = name.replace('''norm2''' ,'''layernorm_after''' )
if "mlp.fc1" in name:
_a : Any = name.replace('''mlp.fc1''' ,'''intermediate.dense''' )
if "mlp.fc2" in name:
_a : str = name.replace('''mlp.fc2''' ,'''output.dense''' )
if "q_bias" in name:
_a : Dict = name.replace('''q_bias''' ,'''query.bias''' )
if "k_bias" in name:
_a : List[Any] = name.replace('''k_bias''' ,'''key.bias''' )
if "v_bias" in name:
_a : Dict = name.replace('''v_bias''' ,'''value.bias''' )
if "cpb_mlp" in name:
_a : List[Any] = name.replace('''cpb_mlp''' ,'''continuous_position_bias_mlp''' )
if name == "norm.weight":
_a : str = '''layernorm.weight'''
if name == "norm.bias":
_a : Optional[Any] = '''layernorm.bias'''
if "head" in name:
_a : Any = name.replace('''head''' ,'''classifier''' )
else:
_a : Optional[int] = '''swinv2.''' + name
return name
def __UpperCAmelCase ( __a : Tuple ,__a : Optional[int] ) -> Any:
"""simple docstring"""
for key in orig_state_dict.copy().keys():
_a : Any = orig_state_dict.pop(__a )
if "mask" in key:
continue
elif "qkv" in key:
_a : List[Any] = key.split('''.''' )
_a : Union[str, Any] = int(key_split[1] )
_a : Optional[int] = int(key_split[3] )
_a : List[str] = model.swinva.encoder.layers[layer_num].blocks[block_num].attention.self.all_head_size
if "weight" in key:
_a : Any = val[:dim, :]
_a : Tuple = val[dim : dim * 2, :]
_a : Optional[Any] = val[-dim:, :]
else:
_a : Optional[int] = val[:dim]
_a : List[str] = val[
dim : dim * 2
]
_a : Any = val[-dim:]
else:
_a : List[str] = val
return orig_state_dict
def __UpperCAmelCase ( __a : List[Any] ,__a : Dict ) -> Any:
"""simple docstring"""
_a : Any = timm.create_model(__a ,pretrained=__a )
timm_model.eval()
_a : Optional[int] = get_swinva_config(__a )
_a : str = SwinvaForImageClassification(__a )
model.eval()
_a : Union[str, Any] = convert_state_dict(timm_model.state_dict() ,__a )
model.load_state_dict(__a )
_a : Union[str, Any] = '''http://images.cocodataset.org/val2017/000000039769.jpg'''
_a : Dict = AutoImageProcessor.from_pretrained('''microsoft/{}'''.format(swinva_name.replace('''_''' ,'''-''' ) ) )
_a : Tuple = Image.open(requests.get(__a ,stream=__a ).raw )
_a : Dict = image_processor(images=__a ,return_tensors='''pt''' )
_a : Any = timm_model(inputs['''pixel_values'''] )
_a : List[str] = model(**__a ).logits
assert torch.allclose(__a ,__a ,atol=1E-3 )
print(F"""Saving model {swinva_name} to {pytorch_dump_folder_path}""" )
model.save_pretrained(__a )
print(F"""Saving image processor to {pytorch_dump_folder_path}""" )
image_processor.save_pretrained(__a )
model.push_to_hub(
repo_path_or_name=Path(__a ,__a ) ,organization='''nandwalritik''' ,commit_message='''Add model''' ,)
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
'''--swinv2_name''',
default='''swinv2_tiny_patch4_window8_256''',
type=str,
help='''Name of the Swinv2 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__ = parser.parse_args()
convert_swinva_checkpoint(args.swinva_name, args.pytorch_dump_folder_path)
| 358
|
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
if n == 1 or not isinstance(__a ,__a ):
return 0
elif n == 2:
return 1
else:
_a : Any = [0, 1]
for i in range(2 ,n + 1 ):
sequence.append(sequence[i - 1] + sequence[i - 2] )
return sequence[n]
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
_a : Any = 0
_a : Dict = 2
while digits < n:
index += 1
_a : Dict = len(str(fibonacci(__a ) ) )
return index
def __UpperCAmelCase ( __a : int = 1_000 ) -> int:
"""simple docstring"""
return fibonacci_digits_index(__a )
if __name__ == "__main__":
print(solution(int(str(input()).strip())))
| 15
| 0
|
from ...utils import (
OptionalDependencyNotAvailable,
is_flax_available,
is_torch_available,
is_transformers_available,
)
try:
if not (is_transformers_available() and is_torch_available()):
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
from ...utils.dummy_torch_and_transformers_objects import * # noqa F403
else:
from .multicontrolnet import MultiControlNetModel
from .pipeline_controlnet import StableDiffusionControlNetPipeline
from .pipeline_controlnet_imgaimg import StableDiffusionControlNetImgaImgPipeline
from .pipeline_controlnet_inpaint import StableDiffusionControlNetInpaintPipeline
if is_transformers_available() and is_flax_available():
from .pipeline_flax_controlnet import FlaxStableDiffusionControlNetPipeline
| 359
|
from sklearn.metrics import fa_score, matthews_corrcoef
import datasets
from .record_evaluation import evaluate as evaluate_record
a__ = '''\
@article{wang2019superglue,
title={SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems},
author={Wang, Alex and Pruksachatkun, Yada and Nangia, Nikita and Singh, Amanpreet and Michael, Julian and Hill, Felix and Levy, Omer and Bowman, Samuel R},
journal={arXiv preprint arXiv:1905.00537},
year={2019}
}
'''
a__ = '''\
SuperGLUE (https://super.gluebenchmark.com/) is a new benchmark styled after
GLUE with a new set of more difficult language understanding tasks, improved
resources, and a new public leaderboard.
'''
a__ = '''
Compute SuperGLUE evaluation metric associated to each SuperGLUE dataset.
Args:
predictions: list of predictions to score. Depending on the SuperGlUE subset:
- for \'record\': list of question-answer dictionaries with the following keys:
- \'idx\': index of the question as specified by the dataset
- \'prediction_text\': the predicted answer text
- for \'multirc\': list of question-answer dictionaries with the following keys:
- \'idx\': index of the question-answer pair as specified by the dataset
- \'prediction\': the predicted answer label
- otherwise: list of predicted labels
references: list of reference labels. Depending on the SuperGLUE subset:
- for \'record\': list of question-answers dictionaries with the following keys:
- \'idx\': index of the question as specified by the dataset
- \'answers\': list of possible answers
- otherwise: list of reference labels
Returns: depending on the SuperGLUE subset:
- for \'record\':
- \'exact_match\': Exact match between answer and gold answer
- \'f1\': F1 score
- for \'multirc\':
- \'exact_match\': Exact match between answer and gold answer
- \'f1_m\': Per-question macro-F1 score
- \'f1_a\': Average F1 score over all answers
- for \'axb\':
\'matthews_correlation\': Matthew Correlation
- for \'cb\':
- \'accuracy\': Accuracy
- \'f1\': F1 score
- for all others:
- \'accuracy\': Accuracy
Examples:
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'copa\') # any of ["copa", "rte", "wic", "wsc", "wsc.fixed", "boolq", "axg"]
>>> predictions = [0, 1]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'accuracy\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'cb\')
>>> predictions = [0, 1]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'accuracy\': 1.0, \'f1\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'record\')
>>> predictions = [{\'idx\': {\'passage\': 0, \'query\': 0}, \'prediction_text\': \'answer\'}]
>>> references = [{\'idx\': {\'passage\': 0, \'query\': 0}, \'answers\': [\'answer\', \'another_answer\']}]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 1.0, \'f1\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'multirc\')
>>> predictions = [{\'idx\': {\'answer\': 0, \'paragraph\': 0, \'question\': 0}, \'prediction\': 0}, {\'idx\': {\'answer\': 1, \'paragraph\': 2, \'question\': 3}, \'prediction\': 1}]
>>> references = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'exact_match\': 1.0, \'f1_m\': 1.0, \'f1_a\': 1.0}
>>> super_glue_metric = datasets.load_metric(\'super_glue\', \'axb\')
>>> references = [0, 1]
>>> predictions = [0, 1]
>>> results = super_glue_metric.compute(predictions=predictions, references=references)
>>> print(results)
{\'matthews_correlation\': 1.0}
'''
def __UpperCAmelCase ( __a : int ,__a : List[str] ) -> Optional[Any]:
"""simple docstring"""
return float((preds == labels).mean() )
def __UpperCAmelCase ( __a : List[Any] ,__a : Union[str, Any] ,__a : List[str]="binary" ) -> Optional[int]:
"""simple docstring"""
_a : List[str] = simple_accuracy(__a ,__a )
_a : Any = float(fa_score(y_true=__a ,y_pred=__a ,average=__a ) )
return {
"accuracy": acc,
"f1": fa,
}
def __UpperCAmelCase ( __a : Optional[Any] ,__a : str ) -> List[Any]:
"""simple docstring"""
_a : Union[str, Any] = {}
for id_pred, label in zip(__a ,__a ):
_a : Optional[int] = F"""{id_pred['idx']['paragraph']}-{id_pred['idx']['question']}"""
_a : Optional[Any] = id_pred['''prediction''']
if question_id in question_map:
question_map[question_id].append((pred, label) )
else:
_a : str = [(pred, label)]
_a , _a : Any = [], []
for question, preds_labels in question_map.items():
_a , _a : Any = zip(*__a )
_a : List[Any] = fa_score(y_true=__a ,y_pred=__a ,average='''macro''' )
fas.append(__a )
_a : List[str] = int(sum(pred == label for pred, label in preds_labels ) == len(__a ) )
ems.append(__a )
_a : List[str] = float(sum(__a ) / len(__a ) )
_a : str = sum(__a ) / len(__a )
_a : Optional[int] = float(fa_score(y_true=__a ,y_pred=[id_pred['''prediction'''] for id_pred in ids_preds] ) )
return {"exact_match": em, "f1_m": fa_m, "f1_a": fa_a}
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> List[Any]:
if self.config_name not in [
"boolq",
"cb",
"copa",
"multirc",
"record",
"rte",
"wic",
"wsc",
"wsc.fixed",
"axb",
"axg",
]:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["boolq", "cb", "copa", "multirc", "record", "rte", "wic", "wsc", "wsc.fixed", "axb", "axg",]''' )
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(self._get_feature_types() ) , codebase_urls=[] , reference_urls=[] , format='''numpy''' if not self.config_name == '''record''' and not self.config_name == '''multirc''' else None , )
def __lowercase ( self ) -> Any:
if self.config_name == "record":
return {
"predictions": {
"idx": {
"passage": datasets.Value('''int64''' ),
"query": datasets.Value('''int64''' ),
},
"prediction_text": datasets.Value('''string''' ),
},
"references": {
"idx": {
"passage": datasets.Value('''int64''' ),
"query": datasets.Value('''int64''' ),
},
"answers": datasets.Sequence(datasets.Value('''string''' ) ),
},
}
elif self.config_name == "multirc":
return {
"predictions": {
"idx": {
"answer": datasets.Value('''int64''' ),
"paragraph": datasets.Value('''int64''' ),
"question": datasets.Value('''int64''' ),
},
"prediction": datasets.Value('''int64''' ),
},
"references": datasets.Value('''int64''' ),
}
else:
return {
"predictions": datasets.Value('''int64''' ),
"references": datasets.Value('''int64''' ),
}
def __lowercase ( self , _a , _a ) -> Optional[Any]:
if self.config_name == "axb":
return {"matthews_correlation": matthews_corrcoef(_a , _a )}
elif self.config_name == "cb":
return acc_and_fa(_a , _a , fa_avg='''macro''' )
elif self.config_name == "record":
_a : Any = [
{
'''qas''': [
{'''id''': ref['''idx''']['''query'''], '''answers''': [{'''text''': ans} for ans in ref['''answers''']]}
for ref in references
]
}
]
_a : Any = {pred['''idx''']['''query''']: pred['''prediction_text'''] for pred in predictions}
return evaluate_record(_a , _a )[0]
elif self.config_name == "multirc":
return evaluate_multirc(_a , _a )
elif self.config_name in ["copa", "rte", "wic", "wsc", "wsc.fixed", "boolq", "axg"]:
return {"accuracy": simple_accuracy(_a , _a )}
else:
raise KeyError(
'''You should supply a configuration name selected in '''
'''["boolq", "cb", "copa", "multirc", "record", "rte", "wic", "wsc", "wsc.fixed", "axb", "axg",]''' )
| 15
| 0
|
from typing import List, Optional, Union
import torch
from ...models import UNetaDConditionModel, VQModel
from ...pipelines import DiffusionPipeline
from ...pipelines.pipeline_utils import ImagePipelineOutput
from ...schedulers import DDPMScheduler
from ...utils import (
is_accelerate_available,
is_accelerate_version,
logging,
randn_tensor,
replace_example_docstring,
)
a__ = logging.get_logger(__name__) # pylint: disable=invalid-name
a__ = '''
Examples:
```py
>>> import torch
>>> import numpy as np
>>> from diffusers import KandinskyV22PriorPipeline, KandinskyV22ControlnetPipeline
>>> from transformers import pipeline
>>> from diffusers.utils import load_image
>>> def make_hint(image, depth_estimator):
... image = depth_estimator(image)["depth"]
... image = np.array(image)
... image = image[:, :, None]
... image = np.concatenate([image, image, image], axis=2)
... detected_map = torch.from_numpy(image).float() / 255.0
... hint = detected_map.permute(2, 0, 1)
... return hint
>>> depth_estimator = pipeline("depth-estimation")
>>> pipe_prior = KandinskyV22PriorPipeline.from_pretrained(
... "kandinsky-community/kandinsky-2-2-prior", torch_dtype=torch.float16
... )
>>> pipe_prior = pipe_prior.to("cuda")
>>> pipe = KandinskyV22ControlnetPipeline.from_pretrained(
... "kandinsky-community/kandinsky-2-2-controlnet-depth", torch_dtype=torch.float16
... )
>>> pipe = pipe.to("cuda")
>>> img = load_image(
... "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"
... "/kandinsky/cat.png"
... ).resize((768, 768))
>>> hint = make_hint(img, depth_estimator).unsqueeze(0).half().to("cuda")
>>> prompt = "A robot, 4k photo"
>>> negative_prior_prompt = "lowres, text, error, cropped, worst quality, low quality, jpeg artifacts, ugly, duplicate, morbid, mutilated, out of frame, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, mutation, deformed, blurry, dehydrated, bad anatomy, bad proportions, extra limbs, cloned face, disfigured, gross proportions, malformed limbs, missing arms, missing legs, extra arms, extra legs, fused fingers, too many fingers, long neck, username, watermark, signature"
>>> generator = torch.Generator(device="cuda").manual_seed(43)
>>> image_emb, zero_image_emb = pipe_prior(
... prompt=prompt, negative_prompt=negative_prior_prompt, generator=generator
... ).to_tuple()
>>> images = pipe(
... image_embeds=image_emb,
... negative_image_embeds=zero_image_emb,
... hint=hint,
... num_inference_steps=50,
... generator=generator,
... height=768,
... width=768,
... ).images
>>> images[0].save("robot_cat.png")
```
'''
def __UpperCAmelCase ( __a : Optional[int] ,__a : List[Any] ,__a : Union[str, Any]=8 ) -> Tuple:
"""simple docstring"""
_a : Union[str, Any] = height // scale_factor**2
if height % scale_factor**2 != 0:
new_height += 1
_a : List[Any] = width // scale_factor**2
if width % scale_factor**2 != 0:
new_width += 1
return new_height * scale_factor, new_width * scale_factor
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __init__( self , _a , _a , _a , ) -> List[str]:
"""simple docstring"""
super().__init__()
self.register_modules(
unet=_a , scheduler=_a , movq=_a , )
_a : List[Any] = 2 ** (len(self.movq.config.block_out_channels ) - 1)
def __lowercase ( self , _a , _a , _a , _a , _a , _a ) -> List[str]:
"""simple docstring"""
if latents is None:
_a : List[Any] = randn_tensor(_a , generator=_a , device=_a , dtype=_a )
else:
if latents.shape != shape:
raise ValueError(F"""Unexpected latents shape, got {latents.shape}, expected {shape}""" )
_a : Union[str, Any] = latents.to(_a )
_a : Dict = latents * scheduler.init_noise_sigma
return latents
def __lowercase ( self , _a=0 ) -> List[str]:
"""simple docstring"""
if is_accelerate_available():
from accelerate import cpu_offload
else:
raise ImportError('''Please install accelerate via `pip install accelerate`''' )
_a : List[str] = torch.device(F"""cuda:{gpu_id}""" )
_a : Tuple = [
self.unet,
self.movq,
]
for cpu_offloaded_model in models:
if cpu_offloaded_model is not None:
cpu_offload(_a , _a )
def __lowercase ( self , _a=0 ) -> List[Any]:
"""simple docstring"""
if is_accelerate_available() and is_accelerate_version('''>=''' , '''0.17.0.dev0''' ):
from accelerate import cpu_offload_with_hook
else:
raise ImportError('''`enable_model_cpu_offload` requires `accelerate v0.17.0` or higher.''' )
_a : int = torch.device(F"""cuda:{gpu_id}""" )
if self.device.type != "cpu":
self.to('''cpu''' , silence_dtype_warnings=_a )
torch.cuda.empty_cache() # otherwise we don't see the memory savings (but they probably exist)
_a : List[Any] = None
for cpu_offloaded_model in [self.unet, self.movq]:
_a : int = cpu_offload_with_hook(_a , _a , prev_module_hook=_a )
# We'll offload the last model manually.
_a : Optional[Any] = hook
@property
# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline._execution_device
def __lowercase ( self ) -> Union[str, Any]:
"""simple docstring"""
if not hasattr(self.unet , '''_hf_hook''' ):
return self.device
for module in self.unet.modules():
if (
hasattr(_a , '''_hf_hook''' )
and hasattr(module._hf_hook , '''execution_device''' )
and module._hf_hook.execution_device is not None
):
return torch.device(module._hf_hook.execution_device )
return self.device
@torch.no_grad()
@replace_example_docstring(_a )
def __call__( self , _a , _a , _a , _a = 5_1_2 , _a = 5_1_2 , _a = 1_0_0 , _a = 4.0 , _a = 1 , _a = None , _a = None , _a = "pil" , _a = True , ) -> Optional[int]:
"""simple docstring"""
_a : List[Any] = self._execution_device
_a : int = guidance_scale > 1.0
if isinstance(_a , _a ):
_a : Optional[int] = torch.cat(_a , dim=0 )
if isinstance(_a , _a ):
_a : Union[str, Any] = torch.cat(_a , dim=0 )
if isinstance(_a , _a ):
_a : Optional[Any] = torch.cat(_a , dim=0 )
_a : Union[str, Any] = image_embeds.shape[0] * num_images_per_prompt
if do_classifier_free_guidance:
_a : Optional[Any] = image_embeds.repeat_interleave(_a , dim=0 )
_a : Any = negative_image_embeds.repeat_interleave(_a , dim=0 )
_a : Union[str, Any] = hint.repeat_interleave(_a , dim=0 )
_a : Optional[Any] = torch.cat([negative_image_embeds, image_embeds] , dim=0 ).to(dtype=self.unet.dtype , device=_a )
_a : Tuple = torch.cat([hint, hint] , dim=0 ).to(dtype=self.unet.dtype , device=_a )
self.scheduler.set_timesteps(_a , device=_a )
_a : Optional[Any] = self.scheduler.timesteps
_a : str = self.movq.config.latent_channels
_a : Union[str, Any] = downscale_height_and_width(_a , _a , self.movq_scale_factor )
# create initial latent
_a : Optional[int] = self.prepare_latents(
(batch_size, num_channels_latents, height, width) , image_embeds.dtype , _a , _a , _a , self.scheduler , )
for i, t in enumerate(self.progress_bar(_a ) ):
# expand the latents if we are doing classifier free guidance
_a : Dict = torch.cat([latents] * 2 ) if do_classifier_free_guidance else latents
_a : Union[str, Any] = {'''image_embeds''': image_embeds, '''hint''': hint}
_a : str = self.unet(
sample=_a , timestep=_a , encoder_hidden_states=_a , added_cond_kwargs=_a , return_dict=_a , )[0]
if do_classifier_free_guidance:
_a : List[str] = noise_pred.split(latents.shape[1] , dim=1 )
_a : Any = noise_pred.chunk(2 )
_a : Optional[int] = variance_pred.chunk(2 )
_a : Union[str, Any] = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
_a : Tuple = torch.cat([noise_pred, variance_pred_text] , dim=1 )
if not (
hasattr(self.scheduler.config , '''variance_type''' )
and self.scheduler.config.variance_type in ["learned", "learned_range"]
):
_a : str = noise_pred.split(latents.shape[1] , dim=1 )
# compute the previous noisy sample x_t -> x_t-1
_a : str = self.scheduler.step(
_a , _a , _a , generator=_a , )[0]
# post-processing
_a : Optional[Any] = self.movq.decode(_a , force_not_quantize=_a )['''sample''']
if output_type not in ["pt", "np", "pil"]:
raise ValueError(F"""Only the output types `pt`, `pil` and `np` are supported not output_type={output_type}""" )
if output_type in ["np", "pil"]:
_a : Optional[int] = image * 0.5 + 0.5
_a : Tuple = image.clamp(0 , 1 )
_a : Optional[Any] = image.cpu().permute(0 , 2 , 3 , 1 ).float().numpy()
if output_type == "pil":
_a : Tuple = self.numpy_to_pil(_a )
if not return_dict:
return (image,)
return ImagePipelineOutput(images=_a )
| 360
|
import numpy as np
def __UpperCAmelCase ( __a : np.ndarray ,__a : np.ndarray ,__a : float = 1E-12 ,__a : int = 100 ,) -> tuple[float, np.ndarray]:
"""simple docstring"""
assert np.shape(__a )[0] == np.shape(__a )[1]
# Ensure proper dimensionality.
assert np.shape(__a )[0] == np.shape(__a )[0]
# Ensure inputs are either both complex or both real
assert np.iscomplexobj(__a ) == np.iscomplexobj(__a )
_a : List[str] = np.iscomplexobj(__a )
if is_complex:
# Ensure complex input_matrix is Hermitian
assert np.array_equal(__a ,input_matrix.conj().T )
# Set convergence to False. Will define convergence when we exceed max_iterations
# or when we have small changes from one iteration to next.
_a : List[str] = False
_a : List[str] = 0
_a : Tuple = 0
_a : str = 1E12
while not convergence:
# Multiple matrix by the vector.
_a : str = np.dot(__a ,__a )
# Normalize the resulting output vector.
_a : List[Any] = w / np.linalg.norm(__a )
# Find rayleigh quotient
# (faster than usual b/c we know vector is normalized already)
_a : Dict = vector.conj().T if is_complex else vector.T
_a : Tuple = np.dot(__a ,np.dot(__a ,__a ) )
# Check convergence.
_a : List[str] = np.abs(lambda_ - lambda_previous ) / lambda_
iterations += 1
if error <= error_tol or iterations >= max_iterations:
_a : Dict = True
_a : str = lambda_
if is_complex:
_a : Tuple = np.real(lambda_ )
return lambda_, vector
def __UpperCAmelCase ( ) -> None:
"""simple docstring"""
_a : List[str] = np.array([[41, 4, 20], [4, 26, 30], [20, 30, 50]] )
_a : int = np.array([41, 4, 20] )
_a : Optional[Any] = real_input_matrix.astype(np.complexaaa )
_a : int = np.triu(1j * complex_input_matrix ,1 )
complex_input_matrix += imag_matrix
complex_input_matrix += -1 * imag_matrix.T
_a : Union[str, Any] = np.array([41, 4, 20] ).astype(np.complexaaa )
for problem_type in ["real", "complex"]:
if problem_type == "real":
_a : Optional[int] = real_input_matrix
_a : Union[str, Any] = real_vector
elif problem_type == "complex":
_a : str = complex_input_matrix
_a : str = complex_vector
# Our implementation.
_a , _a : Optional[Any] = power_iteration(__a ,__a )
# Numpy implementation.
# Get eigenvalues and eigenvectors using built-in numpy
# eigh (eigh used for symmetric or hermetian matrices).
_a , _a : List[str] = np.linalg.eigh(__a )
# Last eigenvalue is the maximum one.
_a : Tuple = eigen_values[-1]
# Last column in this matrix is eigenvector corresponding to largest eigenvalue.
_a : List[Any] = eigen_vectors[:, -1]
# Check our implementation and numpy gives close answers.
assert np.abs(eigen_value - eigen_value_max ) <= 1E-6
# Take absolute values element wise of each eigenvector.
# as they are only unique to a minus sign.
assert np.linalg.norm(np.abs(__a ) - np.abs(__a ) ) <= 1E-6
if __name__ == "__main__":
import doctest
doctest.testmod()
test_power_iteration()
| 15
| 0
|
a__ = {str(digit): digit**5 for digit in range(10)}
def __UpperCAmelCase ( __a : int ) -> int:
"""simple docstring"""
return sum(DIGITS_FIFTH_POWER[digit] for digit in str(__a ) )
def __UpperCAmelCase ( ) -> int:
"""simple docstring"""
return sum(
number
for number in range(1_000 ,1_000_000 )
if number == digits_fifth_powers_sum(__a ) )
if __name__ == "__main__":
print(solution())
| 361
|
import itertools
from dataclasses import dataclass
from typing import Optional
import pandas as pd
import pyarrow as pa
import datasets
from datasets.table import table_cast
@dataclass
class UpperCAmelCase_ ( datasets.BuilderConfig ):
"""simple docstring"""
UpperCAmelCase__ : Optional[datasets.Features] = None
class UpperCAmelCase_ ( datasets.ArrowBasedBuilder ):
"""simple docstring"""
UpperCAmelCase__ : Any = PandasConfig
def __lowercase ( self ) -> Any:
return datasets.DatasetInfo(features=self.config.features )
def __lowercase ( self , _a ) -> List[Any]:
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}""" )
_a : str = dl_manager.download_and_extract(self.config.data_files )
if isinstance(_a , (str, list, tuple) ):
_a : Dict = data_files
if isinstance(_a , _a ):
_a : Dict = [files]
# Use `dl_manager.iter_files` to skip hidden files in an extracted archive
_a : int = [dl_manager.iter_files(_a ) for file in files]
return [datasets.SplitGenerator(name=datasets.Split.TRAIN , gen_kwargs={'''files''': files} )]
_a : Optional[Any] = []
for split_name, files in data_files.items():
if isinstance(_a , _a ):
_a : List[str] = [files]
# Use `dl_manager.iter_files` to skip hidden files in an extracted archive
_a : Any = [dl_manager.iter_files(_a ) for file in files]
splits.append(datasets.SplitGenerator(name=_a , gen_kwargs={'''files''': files} ) )
return splits
def __lowercase ( self , _a ) -> pa.Table:
if self.config.features is not None:
# more expensive cast to support nested features with keys in a different order
# allows str <-> int/float or str to Audio for example
_a : Optional[Any] = table_cast(_a , self.config.features.arrow_schema )
return pa_table
def __lowercase ( self , _a ) -> List[str]:
for i, file in enumerate(itertools.chain.from_iterable(_a ) ):
with open(_a , '''rb''' ) as f:
_a : str = pa.Table.from_pandas(pd.read_pickle(_a ) )
yield i, self._cast_table(_a )
| 15
| 0
|
from typing import Optional, Tuple, Union
import tensorflow as tf
from ...activations_tf import ACTaFN
from ...file_utils import add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward
from ...modeling_tf_outputs import (
TFBaseModelOutputWithNoAttention,
TFBaseModelOutputWithPoolingAndNoAttention,
TFSequenceClassifierOutput,
)
from ...modeling_tf_utils import TFPreTrainedModel, TFSequenceClassificationLoss, keras_serializable, unpack_inputs
from ...tf_utils import shape_list
from ...utils import logging
from .configuration_regnet import RegNetConfig
a__ = logging.get_logger(__name__)
# General docstring
a__ = '''RegNetConfig'''
# Base docstring
a__ = '''facebook/regnet-y-040'''
a__ = [1, 1088, 7, 7]
# Image classification docstring
a__ = '''facebook/regnet-y-040'''
a__ = '''tabby, tabby cat'''
a__ = [
'''facebook/regnet-y-040''',
# See all regnet models at https://huggingface.co/models?filter=regnet
]
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _a , _a = 3 , _a = 1 , _a = 1 , _a = "relu" , **_a , ) -> str:
super().__init__(**_a )
# The padding and conv has been verified in
# https://colab.research.google.com/gist/sayakpaul/854bc10eeaf21c9ee2119e0b9f3841a7/scratchpad.ipynb
_a : Optional[Any] = tf.keras.layers.ZeroPaddingaD(padding=kernel_size // 2 )
_a : Tuple = tf.keras.layers.ConvaD(
filters=_a , kernel_size=_a , strides=_a , padding='''VALID''' , groups=_a , use_bias=_a , name='''convolution''' , )
_a : Optional[Any] = tf.keras.layers.BatchNormalization(epsilon=1e-5 , momentum=0.9 , name='''normalization''' )
_a : List[str] = ACTaFN[activation] if activation is not None else tf.identity
def __lowercase ( self , _a ) -> List[Any]:
_a : List[Any] = self.convolution(self.padding(_a ) )
_a : Tuple = self.normalization(_a )
_a : Union[str, Any] = self.activation(_a )
return hidden_state
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _a , **_a ) -> Tuple:
super().__init__(**_a )
_a : List[str] = config.num_channels
_a : Dict = TFRegNetConvLayer(
out_channels=config.embedding_size , kernel_size=3 , stride=2 , activation=config.hidden_act , name='''embedder''' , )
def __lowercase ( self , _a ) -> Union[str, Any]:
_a : List[str] = shape_list(_a )[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 : Dict = tf.transpose(_a , perm=(0, 2, 3, 1) )
_a : List[str] = self.embedder(_a )
return hidden_state
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _a , _a = 2 , **_a ) -> int:
super().__init__(**_a )
_a : Optional[Any] = tf.keras.layers.ConvaD(
filters=_a , kernel_size=1 , strides=_a , use_bias=_a , name='''convolution''' )
_a : str = tf.keras.layers.BatchNormalization(epsilon=1e-5 , momentum=0.9 , name='''normalization''' )
def __lowercase ( self , _a , _a = False ) -> tf.Tensor:
return self.normalization(self.convolution(_a ) , training=_a )
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _a , _a , **_a ) -> int:
super().__init__(**_a )
_a : List[Any] = tf.keras.layers.GlobalAveragePoolingaD(keepdims=_a , name='''pooler''' )
_a : Optional[int] = [
tf.keras.layers.ConvaD(filters=_a , kernel_size=1 , activation='''relu''' , name='''attention.0''' ),
tf.keras.layers.ConvaD(filters=_a , kernel_size=1 , activation='''sigmoid''' , name='''attention.2''' ),
]
def __lowercase ( self , _a ) -> List[Any]:
# [batch_size, h, w, num_channels] -> [batch_size, 1, 1, num_channels]
_a : str = self.pooler(_a )
for layer_module in self.attention:
_a : Dict = layer_module(_a )
_a : Optional[Any] = hidden_state * pooled
return hidden_state
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _a , _a , _a , _a = 1 , **_a ) -> Union[str, Any]:
super().__init__(**_a )
_a : Union[str, Any] = in_channels != out_channels or stride != 1
_a : str = max(1 , out_channels // config.groups_width )
_a : Tuple = (
TFRegNetShortCut(_a , stride=_a , 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 : Tuple = [
TFRegNetConvLayer(_a , kernel_size=1 , activation=config.hidden_act , name='''layer.0''' ),
TFRegNetConvLayer(
_a , stride=_a , groups=_a , activation=config.hidden_act , name='''layer.1''' ),
TFRegNetConvLayer(_a , kernel_size=1 , activation=_a , name='''layer.2''' ),
]
_a : int = ACTaFN[config.hidden_act]
def __lowercase ( self , _a ) -> Any:
_a : Tuple = hidden_state
for layer_module in self.layers:
_a : List[str] = layer_module(_a )
_a : Optional[int] = self.shortcut(_a )
hidden_state += residual
_a : Union[str, Any] = self.activation(_a )
return hidden_state
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _a , _a , _a , _a = 1 , **_a ) -> List[str]:
super().__init__(**_a )
_a : List[str] = in_channels != out_channels or stride != 1
_a : Union[str, Any] = max(1 , out_channels // config.groups_width )
_a : int = (
TFRegNetShortCut(_a , stride=_a , name='''shortcut''' )
if should_apply_shortcut
else tf.keras.layers.Activation('''linear''' , name='''shortcut''' )
)
_a : Any = [
TFRegNetConvLayer(_a , kernel_size=1 , activation=config.hidden_act , name='''layer.0''' ),
TFRegNetConvLayer(
_a , stride=_a , groups=_a , activation=config.hidden_act , name='''layer.1''' ),
TFRegNetSELayer(_a , reduced_channels=int(round(in_channels / 4 ) ) , name='''layer.2''' ),
TFRegNetConvLayer(_a , kernel_size=1 , activation=_a , name='''layer.3''' ),
]
_a : Union[str, Any] = ACTaFN[config.hidden_act]
def __lowercase ( self , _a ) -> List[str]:
_a : Tuple = hidden_state
for layer_module in self.layers:
_a : str = layer_module(_a )
_a : int = self.shortcut(_a )
hidden_state += residual
_a : List[Any] = self.activation(_a )
return hidden_state
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _a , _a , _a , _a = 2 , _a = 2 , **_a ) -> Union[str, Any]:
super().__init__(**_a )
_a : Union[str, Any] = TFRegNetXLayer if config.layer_type == '''x''' else TFRegNetYLayer
_a : Any = [
# downsampling is done in the first layer with stride of 2
layer(_a , _a , _a , stride=_a , name='''layers.0''' ),
*[layer(_a , _a , _a , name=F"""layers.{i+1}""" ) for i in range(depth - 1 )],
]
def __lowercase ( self , _a ) -> Any:
for layer_module in self.layers:
_a : Tuple = layer_module(_a )
return hidden_state
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
def __init__( self , _a , **_a ) -> List[str]:
super().__init__(**_a )
_a : int = []
# based on `downsample_in_first_stage`, the first layer of the first stage may or may not downsample the input
self.stages.append(
TFRegNetStage(
_a , 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 : Any = zip(config.hidden_sizes , config.hidden_sizes[1:] )
for i, ((in_channels, out_channels), depth) in enumerate(zip(_a , config.depths[1:] ) ):
self.stages.append(TFRegNetStage(_a , _a , _a , depth=_a , name=F"""stages.{i+1}""" ) )
def __lowercase ( self , _a , _a = False , _a = True ) -> TFBaseModelOutputWithNoAttention:
_a : List[Any] = () if output_hidden_states else None
for stage_module in self.stages:
if output_hidden_states:
_a : Optional[Any] = hidden_states + (hidden_state,)
_a : Union[str, Any] = stage_module(_a )
if output_hidden_states:
_a : Tuple = 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=_a , hidden_states=_a )
@keras_serializable
class UpperCAmelCase_ ( tf.keras.layers.Layer ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = RegNetConfig
def __init__( self , _a , **_a ) -> Optional[int]:
super().__init__(**_a )
_a : List[Any] = config
_a : Union[str, Any] = TFRegNetEmbeddings(_a , name='''embedder''' )
_a : List[str] = TFRegNetEncoder(_a , name='''encoder''' )
_a : Optional[int] = tf.keras.layers.GlobalAveragePoolingaD(keepdims=_a , name='''pooler''' )
@unpack_inputs
def __lowercase ( self , _a , _a = None , _a = None , _a = False , ) -> TFBaseModelOutputWithPoolingAndNoAttention:
_a : int = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
_a : Optional[Any] = return_dict if return_dict is not None else self.config.use_return_dict
_a : Union[str, Any] = self.embedder(_a , training=_a )
_a : Any = self.encoder(
_a , output_hidden_states=_a , return_dict=_a , training=_a )
_a : Tuple = encoder_outputs[0]
_a : int = self.pooler(_a )
# Change to NCHW output format have uniformity in the modules
_a : Union[str, Any] = tf.transpose(_a , perm=(0, 3, 1, 2) )
_a : Optional[int] = tf.transpose(_a , perm=(0, 3, 1, 2) )
# Change the other hidden state outputs to NCHW as well
if output_hidden_states:
_a : int = tuple([tf.transpose(_a , 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=_a , pooler_output=_a , hidden_states=hidden_states if output_hidden_states else encoder_outputs.hidden_states , )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Union[str, Any] = RegNetConfig
UpperCAmelCase__ : int = "regnet"
UpperCAmelCase__ : Dict = "pixel_values"
@property
def __lowercase ( self ) -> Tuple:
return {"pixel_values": tf.TensorSpec(shape=(None, self.config.num_channels, 2_2_4, 2_2_4) , dtype=tf.floataa )}
a__ = R'''
Parameters:
This model is a Tensorflow
[tf.keras.layers.Layer](https://www.tensorflow.org/api_docs/python/tf/keras/layers/Layer) sub-class. Use it as a
regular Tensorflow Module and refer to the Tensorflow documentation for all matter related to general usage and
behavior.
config ([`RegNetConfig`]): Model configuration class with all the parameters of the model.
Initializing with a config file does not load the weights associated with the model, only the
configuration. Check out the [`~TFPreTrainedModel.from_pretrained`] method to load the model weights.
'''
a__ = R'''
Args:
pixel_values (`tf.Tensor` of shape `(batch_size, num_channels, height, width)`):
Pixel values. Pixel values can be obtained using [`AutoImageProcessor`]. See
[`ConveNextImageProcessor.__call__`] for details.
output_hidden_states (`bool`, *optional*):
Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for
more detail.
return_dict (`bool`, *optional*):
Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
'''
@add_start_docstrings(
"The bare RegNet model outputting raw features without any specific head on top." , __lowercase , )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __init__( self , _a , *_a , **_a ) -> Any:
super().__init__(_a , *_a , **_a )
_a : Optional[Any] = TFRegNetMainLayer(_a , name='''regnet''' )
@unpack_inputs
@add_start_docstrings_to_model_forward(_a )
@add_code_sample_docstrings(
checkpoint=_CHECKPOINT_FOR_DOC , output_type=_a , config_class=_CONFIG_FOR_DOC , modality='''vision''' , expected_output=_EXPECTED_OUTPUT_SHAPE , )
def __lowercase ( self , _a , _a = None , _a = None , _a=False , ) -> Union[TFBaseModelOutputWithPoolingAndNoAttention, Tuple[tf.Tensor]]:
_a : str = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
_a : Optional[int] = return_dict if return_dict is not None else self.config.use_return_dict
_a : Tuple = self.regnet(
pixel_values=_a , output_hidden_states=_a , return_dict=_a , training=_a , )
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 " , __lowercase , )
class UpperCAmelCase_ ( __lowercase , __lowercase ):
"""simple docstring"""
def __init__( self , _a , *_a , **_a ) -> Tuple:
super().__init__(_a , *_a , **_a )
_a : str = config.num_labels
_a : int = TFRegNetMainLayer(_a , name='''regnet''' )
# classification head
_a : Optional[int] = [
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(_a )
@add_code_sample_docstrings(
checkpoint=_IMAGE_CLASS_CHECKPOINT , output_type=_a , config_class=_CONFIG_FOR_DOC , expected_output=_IMAGE_CLASS_EXPECTED_OUTPUT , )
def __lowercase ( self , _a = None , _a = None , _a = None , _a = None , _a=False , ) -> Union[TFSequenceClassifierOutput, Tuple[tf.Tensor]]:
_a : int = (
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
)
_a : List[str] = return_dict if return_dict is not None else self.config.use_return_dict
_a : List[str] = self.regnet(
_a , output_hidden_states=_a , return_dict=_a , training=_a )
_a : Dict = outputs.pooler_output if return_dict else outputs[1]
_a : Dict = self.classifier[0](_a )
_a : List[str] = self.classifier[1](_a )
_a : List[Any] = None if labels is None else self.hf_compute_loss(labels=_a , logits=_a )
if not return_dict:
_a : int = (logits,) + outputs[2:]
return ((loss,) + output) if loss is not None else output
return TFSequenceClassifierOutput(loss=_a , logits=_a , hidden_states=outputs.hidden_states )
| 362
|
def __UpperCAmelCase ( __a : int ,__a : int ,__a : int ) -> int:
"""simple docstring"""
if exponent == 1:
return base
if exponent % 2 == 0:
_a : List[Any] = _modexpt(__a ,exponent // 2 ,__a ) % modulo_value
return (x * x) % modulo_value
else:
return (base * _modexpt(__a ,exponent - 1 ,__a )) % modulo_value
def __UpperCAmelCase ( __a : int = 1_777 ,__a : int = 1_855 ,__a : int = 8 ) -> int:
"""simple docstring"""
_a : List[Any] = base
for _ in range(1 ,__a ):
_a : Any = _modexpt(__a ,__a ,10**digits )
return result
if __name__ == "__main__":
print(f'''{solution() = }''')
| 15
| 0
|
def __UpperCAmelCase ( __a : int ,__a : list ) -> List[str]:
"""simple docstring"""
_enforce_args(__a ,__a )
if n == 0:
return 0
_a : Optional[Any] = float('''-inf''' )
for i in range(1 ,n + 1 ):
_a : List[Any] = max(
__a ,prices[i - 1] + naive_cut_rod_recursive(n - i ,__a ) )
return max_revue
def __UpperCAmelCase ( __a : int ,__a : list ) -> Dict:
"""simple docstring"""
_enforce_args(__a ,__a )
_a : Optional[Any] = [float('''-inf''' ) for _ in range(n + 1 )]
return _top_down_cut_rod_recursive(__a ,__a ,__a )
def __UpperCAmelCase ( __a : int ,__a : list ,__a : list ) -> Union[str, Any]:
"""simple docstring"""
if max_rev[n] >= 0:
return max_rev[n]
elif n == 0:
return 0
else:
_a : Optional[int] = float('''-inf''' )
for i in range(1 ,n + 1 ):
_a : Union[str, Any] = max(
__a ,prices[i - 1] + _top_down_cut_rod_recursive(n - i ,__a ,__a ) ,)
_a : Dict = max_revenue
return max_rev[n]
def __UpperCAmelCase ( __a : int ,__a : list ) -> str:
"""simple docstring"""
_enforce_args(__a ,__a )
# length(max_rev) = n + 1, to accommodate for the revenue obtainable from a rod of
# length 0.
_a : Any = [float('''-inf''' ) for _ in range(n + 1 )]
_a : str = 0
for i in range(1 ,n + 1 ):
_a : List[Any] = max_rev[i]
for j in range(1 ,i + 1 ):
_a : Tuple = max(__a ,prices[j - 1] + max_rev[i - j] )
_a : Optional[int] = max_revenue_i
return max_rev[n]
def __UpperCAmelCase ( __a : int ,__a : list ) -> Optional[Any]:
"""simple docstring"""
if n < 0:
_a : Union[str, Any] = F"""n must be greater than or equal to 0. Got n = {n}"""
raise ValueError(__a )
if n > len(__a ):
_a : Optional[Any] = (
'''Each integral piece of rod must have a corresponding price. '''
F"""Got n = {n} but length of prices = {len(__a )}"""
)
raise ValueError(__a )
def __UpperCAmelCase ( ) -> Dict:
"""simple docstring"""
_a : Dict = [6, 10, 12, 15, 20, 23]
_a : Tuple = len(__a )
# the best revenue comes from cutting the rod into 6 pieces, each
# of length 1 resulting in a revenue of 6 * 6 = 36.
_a : str = 36
_a : List[Any] = top_down_cut_rod(__a ,__a )
_a : Union[str, Any] = bottom_up_cut_rod(__a ,__a )
_a : Dict = naive_cut_rod_recursive(__a ,__a )
assert expected_max_revenue == max_rev_top_down
assert max_rev_top_down == max_rev_bottom_up
assert max_rev_bottom_up == max_rev_naive
if __name__ == "__main__":
main()
| 363
|
import numpy as np
import torch
from torch.nn import CrossEntropyLoss
from transformers import AutoModelForCausalLM, AutoTokenizer
import datasets
from datasets import logging
a__ = '''\
'''
a__ = '''
Perplexity (PPL) is one of the most common metrics for evaluating language models.
It is defined as the exponentiated average negative log-likelihood of a sequence.
For more information, see https://huggingface.co/docs/transformers/perplexity
'''
a__ = '''
Args:
model_id (str): model used for calculating Perplexity
NOTE: Perplexity can only be calculated for causal language models.
This includes models such as gpt2, causal variations of bert,
causal versions of t5, and more (the full list can be found
in the AutoModelForCausalLM documentation here:
https://huggingface.co/docs/transformers/master/en/model_doc/auto#transformers.AutoModelForCausalLM )
input_texts (list of str): input text, each separate text snippet
is one list entry.
batch_size (int): the batch size to run texts through the model. Defaults to 16.
add_start_token (bool): whether to add the start token to the texts,
so the perplexity can include the probability of the first word. Defaults to True.
device (str): device to run on, defaults to \'cuda\' when available
Returns:
perplexity: dictionary containing the perplexity scores for the texts
in the input list, as well as the mean perplexity. If one of the input texts is
longer than the max input length of the model, then it is truncated to the
max length for the perplexity computation.
Examples:
Example 1:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = ["lorem ipsum", "Happy Birthday!", "Bienvenue"]
>>> results = perplexity.compute(model_id=\'gpt2\',
... add_start_token=False,
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
78.22
>>> print(round(results["perplexities"][0], 2))
11.11
Example 2:
>>> perplexity = datasets.load_metric("perplexity")
>>> input_texts = datasets.load_dataset("wikitext",
... "wikitext-2-raw-v1",
... split="test")["text"][:50] # doctest:+ELLIPSIS
[...]
>>> input_texts = [s for s in input_texts if s!=\'\']
>>> results = perplexity.compute(model_id=\'gpt2\',
... input_texts=input_texts) # doctest:+ELLIPSIS
>>> print(list(results.keys()))
[\'perplexities\', \'mean_perplexity\']
>>> print(round(results["mean_perplexity"], 2))
60.35
>>> print(round(results["perplexities"][0], 2))
81.12
'''
@datasets.utils.file_utils.add_start_docstrings(_DESCRIPTION , _KWARGS_DESCRIPTION )
class UpperCAmelCase_ ( datasets.Metric ):
"""simple docstring"""
def __lowercase ( self ) -> Any:
return datasets.MetricInfo(
description=_DESCRIPTION , citation=_CITATION , inputs_description=_KWARGS_DESCRIPTION , features=datasets.Features(
{
'''input_texts''': datasets.Value('''string''' ),
} ) , reference_urls=['''https://huggingface.co/docs/transformers/perplexity'''] , )
def __lowercase ( self , _a , _a , _a = 1_6 , _a = True , _a=None ) -> List[Any]:
if device is not None:
assert device in ["gpu", "cpu", "cuda"], "device should be either gpu or cpu."
if device == "gpu":
_a : List[str] = '''cuda'''
else:
_a : Optional[Any] = '''cuda''' if torch.cuda.is_available() else '''cpu'''
_a : Dict = AutoModelForCausalLM.from_pretrained(_a )
_a : List[Any] = model.to(_a )
_a : List[str] = AutoTokenizer.from_pretrained(_a )
# if batch_size > 1 (which generally leads to padding being required), and
# if there is not an already assigned pad_token, assign an existing
# special token to also be the padding token
if tokenizer.pad_token is None and batch_size > 1:
_a : str = list(tokenizer.special_tokens_map_extended.values() )
# check that the model already has at least one special token defined
assert (
len(_a ) > 0
), "If batch_size > 1, model must have at least one special token to use for padding. Please use a different model or set batch_size=1."
# assign one of the special tokens to also be the pad token
tokenizer.add_special_tokens({'''pad_token''': existing_special_tokens[0]} )
if add_start_token:
# leave room for <BOS> token to be added:
assert (
tokenizer.bos_token is not None
), "Input model must already have a BOS token if using add_start_token=True. Please use a different model, or set add_start_token=False"
_a : List[Any] = model.config.max_length - 1
else:
_a : List[str] = model.config.max_length
_a : Union[str, Any] = tokenizer(
_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , return_tensors='''pt''' , return_attention_mask=_a , ).to(_a )
_a : List[Any] = encodings['''input_ids''']
_a : int = encodings['''attention_mask''']
# check that each input is long enough:
if add_start_token:
assert torch.all(torch.ge(attn_masks.sum(1 ) , 1 ) ), "Each input text must be at least one token long."
else:
assert torch.all(
torch.ge(attn_masks.sum(1 ) , 2 ) ), "When add_start_token=False, each input text must be at least two tokens long. Run with add_start_token=True if inputting strings of only one token, and remove all empty input strings."
_a : Optional[int] = []
_a : Dict = CrossEntropyLoss(reduction='''none''' )
for start_index in logging.tqdm(range(0 , len(_a ) , _a ) ):
_a : Dict = min(start_index + batch_size , len(_a ) )
_a : Union[str, Any] = encoded_texts[start_index:end_index]
_a : int = attn_masks[start_index:end_index]
if add_start_token:
_a : Dict = torch.tensor([[tokenizer.bos_token_id]] * encoded_batch.size(dim=0 ) ).to(_a )
_a : List[str] = torch.cat([bos_tokens_tensor, encoded_batch] , dim=1 )
_a : Dict = torch.cat(
[torch.ones(bos_tokens_tensor.size() , dtype=torch.intaa ).to(_a ), attn_mask] , dim=1 )
_a : Dict = encoded_batch
with torch.no_grad():
_a : Any = model(_a , attention_mask=_a ).logits
_a : List[str] = out_logits[..., :-1, :].contiguous()
_a : Union[str, Any] = labels[..., 1:].contiguous()
_a : Optional[int] = attn_mask[..., 1:].contiguous()
_a : Union[str, Any] = torch.expa(
(loss_fct(shift_logits.transpose(1 , 2 ) , _a ) * shift_attention_mask_batch).sum(1 )
/ shift_attention_mask_batch.sum(1 ) )
ppls += perplexity_batch.tolist()
return {"perplexities": ppls, "mean_perplexity": np.mean(_a )}
| 15
| 0
|
"""simple docstring"""
from typing import List, Optional, Union
from ...configuration_utils import PretrainedConfig
from ...utils import logging
a__ = logging.get_logger(__name__)
a__ = {
'''huggingface/time-series-transformer-tourism-monthly''': (
'''https://huggingface.co/huggingface/time-series-transformer-tourism-monthly/resolve/main/config.json'''
),
# See all TimeSeriesTransformer models at https://huggingface.co/models?filter=time_series_transformer
}
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Tuple = "time_series_transformer"
UpperCAmelCase__ : Union[str, Any] = {
"hidden_size": "d_model",
"num_attention_heads": "encoder_attention_heads",
"num_hidden_layers": "encoder_layers",
}
def __init__( self , _a = None , _a = None , _a = "student_t" , _a = "nll" , _a = 1 , _a = [1, 2, 3, 4, 5, 6, 7] , _a = "mean" , _a = 0 , _a = 0 , _a = 0 , _a = 0 , _a = None , _a = None , _a = 3_2 , _a = 3_2 , _a = 2 , _a = 2 , _a = 2 , _a = 2 , _a = True , _a = "gelu" , _a = 6_4 , _a = 0.1 , _a = 0.1 , _a = 0.1 , _a = 0.1 , _a = 0.1 , _a = 1_0_0 , _a = 0.02 , _a=True , **_a , ) -> Optional[int]:
# time series specific configuration
_a : Optional[Any] = prediction_length
_a : Dict = context_length or prediction_length
_a : Optional[Any] = distribution_output
_a : Dict = loss
_a : Dict = input_size
_a : Dict = num_time_features
_a : Any = lags_sequence
_a : Optional[Any] = scaling
_a : Any = num_dynamic_real_features
_a : Any = num_static_real_features
_a : Optional[int] = num_static_categorical_features
if cardinality and num_static_categorical_features > 0:
if len(_a ) != num_static_categorical_features:
raise ValueError(
'''The cardinality should be a list of the same length as `num_static_categorical_features`''' )
_a : List[Any] = cardinality
else:
_a : Optional[int] = [0]
if embedding_dimension and num_static_categorical_features > 0:
if len(_a ) != num_static_categorical_features:
raise ValueError(
'''The embedding dimension should be a list of the same length as `num_static_categorical_features`''' )
_a : str = embedding_dimension
else:
_a : Optional[int] = [min(5_0 , (cat + 1) // 2 ) for cat in self.cardinality]
_a : Union[str, Any] = num_parallel_samples
# Transformer architecture configuration
_a : Any = input_size * len(_a ) + self._number_of_features
_a : int = d_model
_a : int = encoder_attention_heads
_a : Dict = decoder_attention_heads
_a : Optional[int] = encoder_ffn_dim
_a : Union[str, Any] = decoder_ffn_dim
_a : Dict = encoder_layers
_a : Tuple = decoder_layers
_a : List[Any] = dropout
_a : Tuple = attention_dropout
_a : Any = activation_dropout
_a : int = encoder_layerdrop
_a : int = decoder_layerdrop
_a : Tuple = activation_function
_a : Optional[Any] = init_std
_a : int = use_cache
super().__init__(is_encoder_decoder=_a , **_a )
@property
def __lowercase ( self ) -> int:
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
)
| 364
|
# Copyright 2023 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a__ = {
'''configuration_xmod''': [
'''XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP''',
'''XmodConfig''',
'''XmodOnnxConfig''',
],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''XMOD_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''XmodForCausalLM''',
'''XmodForMaskedLM''',
'''XmodForMultipleChoice''',
'''XmodForQuestionAnswering''',
'''XmodForSequenceClassification''',
'''XmodForTokenClassification''',
'''XmodModel''',
'''XmodPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_xmod import XMOD_PRETRAINED_CONFIG_ARCHIVE_MAP, XmodConfig, XmodOnnxConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_xmod import (
XMOD_PRETRAINED_MODEL_ARCHIVE_LIST,
XmodForCausalLM,
XmodForMaskedLM,
XmodForMultipleChoice,
XmodForQuestionAnswering,
XmodForSequenceClassification,
XmodForTokenClassification,
XmodModel,
XmodPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 15
| 0
|
import gc
import unittest
import numpy as np
import torch
from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
from diffusers import (
AutoencoderKL,
DDIMScheduler,
PNDMScheduler,
StableDiffusionLDMaDPipeline,
UNetaDConditionModel,
)
from diffusers.utils import nightly, slow, torch_device
from diffusers.utils.testing_utils import enable_full_determinism, require_torch_gpu
from ..pipeline_params import TEXT_TO_IMAGE_BATCH_PARAMS, TEXT_TO_IMAGE_IMAGE_PARAMS, TEXT_TO_IMAGE_PARAMS
enable_full_determinism()
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : int = StableDiffusionLDMaDPipeline
UpperCAmelCase__ : List[Any] = TEXT_TO_IMAGE_PARAMS
UpperCAmelCase__ : Optional[Any] = TEXT_TO_IMAGE_BATCH_PARAMS
UpperCAmelCase__ : str = TEXT_TO_IMAGE_IMAGE_PARAMS
def __lowercase ( self ) -> List[str]:
torch.manual_seed(0 )
_a : Optional[Any] = UNetaDConditionModel(
block_out_channels=(3_2, 6_4) , layers_per_block=2 , sample_size=3_2 , in_channels=4 , out_channels=4 , down_block_types=('''DownBlock2D''', '''CrossAttnDownBlock2D''') , up_block_types=('''CrossAttnUpBlock2D''', '''UpBlock2D''') , cross_attention_dim=3_2 , )
_a : Optional[Any] = DDIMScheduler(
beta_start=0.0_0085 , beta_end=0.012 , beta_schedule='''scaled_linear''' , clip_sample=_a , set_alpha_to_one=_a , )
torch.manual_seed(0 )
_a : str = AutoencoderKL(
block_out_channels=[3_2, 6_4] , in_channels=6 , out_channels=6 , down_block_types=['''DownEncoderBlock2D''', '''DownEncoderBlock2D'''] , up_block_types=['''UpDecoderBlock2D''', '''UpDecoderBlock2D'''] , latent_channels=4 , )
torch.manual_seed(0 )
_a : Any = 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 , )
_a : List[str] = CLIPTextModel(_a )
_a : List[str] = CLIPTokenizer.from_pretrained('''hf-internal-testing/tiny-random-clip''' )
_a : Tuple = {
'''unet''': unet,
'''scheduler''': scheduler,
'''vae''': vae,
'''text_encoder''': text_encoder,
'''tokenizer''': tokenizer,
'''safety_checker''': None,
'''feature_extractor''': None,
}
return components
def __lowercase ( self , _a , _a=0 ) -> Optional[Any]:
if str(_a ).startswith('''mps''' ):
_a : Tuple = torch.manual_seed(_a )
else:
_a : str = torch.Generator(device=_a ).manual_seed(_a )
_a : Optional[int] = {
'''prompt''': '''A painting of a squirrel eating a burger''',
'''generator''': generator,
'''num_inference_steps''': 2,
'''guidance_scale''': 6.0,
'''output_type''': '''numpy''',
}
return inputs
def __lowercase ( self ) -> int:
_a : Any = '''cpu''' # ensure determinism for the device-dependent torch.Generator
_a : Union[str, Any] = self.get_dummy_components()
_a : str = StableDiffusionLDMaDPipeline(**_a )
_a : int = ldmad_pipe.to(_a )
ldmad_pipe.set_progress_bar_config(disable=_a )
_a : Dict = self.get_dummy_inputs(_a )
_a : Any = ldmad_pipe(**_a )
_a : Union[str, Any] = output.rgb, output.depth
_a : Tuple = rgb[0, -3:, -3:, -1]
_a : int = depth[0, -3:, -1]
assert rgb.shape == (1, 6_4, 6_4, 3)
assert depth.shape == (1, 6_4, 6_4)
_a : int = np.array(
[0.3733_8176, 0.7_0247, 0.7420_3193, 0.5164_3604, 0.5825_6793, 0.6093_2136, 0.418_1095, 0.4835_5877, 0.4653_5262] )
_a : Any = np.array([103.4_6727, 85.81_2004, 87.84_9236] )
assert np.abs(image_slice_rgb.flatten() - expected_slice_rgb ).max() < 1e-2
assert np.abs(image_slice_depth.flatten() - expected_slice_depth ).max() < 1e-2
def __lowercase ( self ) -> List[str]:
_a : Optional[int] = self.get_dummy_components()
_a : Optional[Any] = StableDiffusionLDMaDPipeline(**_a )
_a : int = ldmad_pipe.to(_a )
ldmad_pipe.set_progress_bar_config(disable=_a )
_a : Union[str, Any] = self.get_dummy_inputs(_a )
_a : Union[str, Any] = 3 * [inputs['''prompt''']]
# forward
_a : List[Any] = ldmad_pipe(**_a )
_a : str = output.rgb, output.depth
_a : str = rgb_slice_a[0, -3:, -3:, -1]
_a : Optional[Any] = depth_slice_a[0, -3:, -1]
_a : Any = self.get_dummy_inputs(_a )
_a : List[Any] = 3 * [inputs.pop('''prompt''' )]
_a : int = ldmad_pipe.tokenizer(
_a , padding='''max_length''' , max_length=ldmad_pipe.tokenizer.model_max_length , truncation=_a , return_tensors='''pt''' , )
_a : Tuple = text_inputs['''input_ids'''].to(_a )
_a : Union[str, Any] = ldmad_pipe.text_encoder(_a )[0]
_a : Optional[int] = prompt_embeds
# forward
_a : Dict = ldmad_pipe(**_a )
_a : List[str] = output.rgb, output.depth
_a : Union[str, Any] = rgb_slice_a[0, -3:, -3:, -1]
_a : int = depth_slice_a[0, -3:, -1]
assert np.abs(rgb_slice_a.flatten() - rgb_slice_a.flatten() ).max() < 1e-4
assert np.abs(depth_slice_a.flatten() - depth_slice_a.flatten() ).max() < 1e-4
def __lowercase ( self ) -> Dict:
_a : Optional[Any] = '''cpu''' # ensure determinism for the device-dependent torch.Generator
_a : List[str] = self.get_dummy_components()
_a : Any = PNDMScheduler(skip_prk_steps=_a )
_a : Any = StableDiffusionLDMaDPipeline(**_a )
_a : Any = ldmad_pipe.to(_a )
ldmad_pipe.set_progress_bar_config(disable=_a )
_a : Tuple = self.get_dummy_inputs(_a )
_a : int = '''french fries'''
_a : Tuple = ldmad_pipe(**_a , negative_prompt=_a )
_a : Optional[Any] = output.rgb, output.depth
_a : List[Any] = rgb[0, -3:, -3:, -1]
_a : Tuple = depth[0, -3:, -1]
assert rgb.shape == (1, 6_4, 6_4, 3)
assert depth.shape == (1, 6_4, 6_4)
_a : Any = np.array(
[0.3_7044, 0.7181_1503, 0.722_3251, 0.4860_3675, 0.563_8391, 0.636_4948, 0.4283_3704, 0.490_1315, 0.4792_6217] )
_a : int = np.array([107.8_4738, 84.6_2802, 89.96_2135] )
assert np.abs(rgb_slice.flatten() - expected_slice_rgb ).max() < 1e-2
assert np.abs(depth_slice.flatten() - expected_slice_depth ).max() < 1e-2
@slow
@require_torch_gpu
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Dict:
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def __lowercase ( self , _a , _a="cpu" , _a=torch.floataa , _a=0 ) -> Tuple:
_a : int = torch.Generator(device=_a ).manual_seed(_a )
_a : Optional[int] = np.random.RandomState(_a ).standard_normal((1, 4, 6_4, 6_4) )
_a : Dict = torch.from_numpy(_a ).to(device=_a , dtype=_a )
_a : int = {
'''prompt''': '''a photograph of an astronaut riding a horse''',
'''latents''': latents,
'''generator''': generator,
'''num_inference_steps''': 3,
'''guidance_scale''': 7.5,
'''output_type''': '''numpy''',
}
return inputs
def __lowercase ( self ) -> Optional[Any]:
_a : Tuple = StableDiffusionLDMaDPipeline.from_pretrained('''Intel/ldm3d''' )
_a : Optional[Any] = ldmad_pipe.to(_a )
ldmad_pipe.set_progress_bar_config(disable=_a )
_a : int = self.get_inputs(_a )
_a : Optional[int] = ldmad_pipe(**_a )
_a : Any = output.rgb, output.depth
_a : Tuple = rgb[0, -3:, -3:, -1].flatten()
_a : int = rgb[0, -3:, -1].flatten()
assert rgb.shape == (1, 5_1_2, 5_1_2, 3)
assert depth.shape == (1, 5_1_2, 5_1_2)
_a : int = np.array(
[0.5380_5465, 0.5670_7305, 0.548_6515, 0.5701_2236, 0.581_4511, 0.5625_3487, 0.5484_3014, 0.5509_2263, 0.645_9706] )
_a : Dict = np.array(
[0.926_3781, 0.667_8672, 0.548_6515, 0.9220_2145, 0.6783_1135, 0.5625_3487, 0.924_1694, 0.755_1478, 0.645_9706] )
assert np.abs(rgb_slice - expected_slice_rgb ).max() < 3e-3
assert np.abs(depth_slice - expected_slice_depth ).max() < 3e-3
@nightly
@require_torch_gpu
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Tuple:
super().tearDown()
gc.collect()
torch.cuda.empty_cache()
def __lowercase ( self , _a , _a="cpu" , _a=torch.floataa , _a=0 ) -> Optional[int]:
_a : Optional[Any] = torch.Generator(device=_a ).manual_seed(_a )
_a : Dict = np.random.RandomState(_a ).standard_normal((1, 4, 6_4, 6_4) )
_a : Optional[Any] = torch.from_numpy(_a ).to(device=_a , dtype=_a )
_a : str = {
'''prompt''': '''a photograph of an astronaut riding a horse''',
'''latents''': latents,
'''generator''': generator,
'''num_inference_steps''': 5_0,
'''guidance_scale''': 7.5,
'''output_type''': '''numpy''',
}
return inputs
def __lowercase ( self ) -> Tuple:
_a : Optional[Any] = StableDiffusionLDMaDPipeline.from_pretrained('''Intel/ldm3d''' ).to(_a )
ldmad_pipe.set_progress_bar_config(disable=_a )
_a : Any = self.get_inputs(_a )
_a : List[str] = ldmad_pipe(**_a )
_a : Any = output.rgb, output.depth
_a : str = 0.49_5586
_a : Union[str, Any] = 0.3379_5515
_a : Any = 112.4_8518
_a : List[Any] = 98.48_9746
assert np.abs(expected_rgb_mean - rgb.mean() ) < 1e-3
assert np.abs(expected_rgb_std - rgb.std() ) < 1e-3
assert np.abs(expected_depth_mean - depth.mean() ) < 1e-3
assert np.abs(expected_depth_std - depth.std() ) < 1e-3
def __lowercase ( self ) -> int:
_a : Optional[Any] = StableDiffusionLDMaDPipeline.from_pretrained('''Intel/ldm3d-4c''' ).to(_a )
ldmad_pipe.set_progress_bar_config(disable=_a )
_a : List[str] = self.get_inputs(_a )
_a : List[Any] = ldmad_pipe(**_a )
_a : Union[str, Any] = output.rgb, output.depth
_a : int = 0.419_4127
_a : str = 0.3537_5586
_a : str = 0.563_8502
_a : Tuple = 0.3468_6103
assert rgb.shape == (1, 5_1_2, 5_1_2, 3)
assert depth.shape == (1, 5_1_2, 5_1_2, 1)
assert np.abs(expected_rgb_mean - rgb.mean() ) < 1e-3
assert np.abs(expected_rgb_std - rgb.std() ) < 1e-3
assert np.abs(expected_depth_mean - depth.mean() ) < 1e-3
assert np.abs(expected_depth_std - depth.std() ) < 1e-3
| 365
|
import re
import tempfile
from pathlib import Path
import pytest
import yaml
from datasets.utils.readme import ReadMe
# @pytest.fixture
# def example_yaml_structure():
a__ = yaml.safe_load(
'''\
name: ""
allow_empty: false
allow_empty_text: true
subsections:
- name: "Dataset Card for X" # First-level markdown heading
allow_empty: false
allow_empty_text: true
subsections:
- name: "Table of Contents"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Dataset Description"
allow_empty: false
allow_empty_text: false
subsections:
- name: "Dataset Summary"
allow_empty: false
allow_empty_text: false
subsections: null
- name: "Supported Tasks and Leaderboards"
allow_empty: true
allow_empty_text: true
subsections: null
- name: Languages
allow_empty: false
allow_empty_text: true
subsections: null
'''
)
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
#### Extra Ignored Subsection
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = {
'''name''': '''root''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{
'''name''': '''Dataset Card for My Dataset''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [
{'''name''': '''Table of Contents''', '''text''': '''Some text here.''', '''is_empty_text''': False, '''subsections''': []},
{
'''name''': '''Dataset Description''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Dataset Summary''',
'''text''': '''Some text here.''',
'''is_empty_text''': False,
'''subsections''': [
{
'''name''': '''Extra Ignored Subsection''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
}
],
},
{
'''name''': '''Supported Tasks and Leaderboards''',
'''text''': '''''',
'''is_empty_text''': True,
'''subsections''': [],
},
{'''name''': '''Languages''', '''text''': '''Language Text''', '''is_empty_text''': False, '''subsections''': []},
],
},
],
}
],
}
a__ = '''\
---
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tEmpty YAML markers are present in the README.'''
)
a__ = '''\
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = (
'''The following issues were found for the README at `{path}`:\n-\tNo YAML markers are present in the README.'''
)
a__ = '''\
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tOnly the start of YAML tags present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Summary` but it is empty.\n-\tExpected some text in section `Dataset Summary` but it is empty (text in subsections are ignored).'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Dataset Card for My Dataset` but it is empty.\n-\tSection `Dataset Card for My Dataset` expected the following subsections: `Table of Contents`, `Dataset Description`. Found \'None\'.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tSection `Dataset Description` is missing subsection: `Supported Tasks and Leaderboards`.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tExpected some content in section `Languages` but it is empty.'''
a__ = '''\
---
language:
- zh
- en
---
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
# Dataset Card My Dataset
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has several first-level headings: `Dataset Card for My Dataset`, `Dataset Card My Dataset`. Only one heading is expected. Skipping further validation for this README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found for the README at `{path}`:\n-\tNo first-level heading starting with `Dataset Card for` found in README. Skipping further validation for this README.'''
a__ = ''''''
a__ = '''The following issues were found for the README at `{path}`:\n-\tThe README has no first-level headings. One heading is expected. Skipping further validation for this README.\n-\tNo YAML markers are present in the README.'''
a__ = '''\
---
language:
- zh
- en
---
# Dataset Card for My Dataset
# Dataset Card for My Dataset
## Table of Contents
Some text here.
## Dataset Description
Some text here.
### Dataset Summary
Some text here.
### Supported Tasks and Leaderboards
### Languages
Language Text
'''
a__ = '''The following issues were found while parsing the README at `{path}`:\n-\tMultiple sections with the same heading `Dataset Card for My Dataset` have been found. Please keep only one of these sections.'''
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : List[str] ) -> Optional[int]:
"""simple docstring"""
assert ReadMe.from_string(__a ,__a ).to_dict() == expected_dict
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[str] ,__a : Optional[Any] ) -> Union[str, Any]:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
_a : List[Any] = ReadMe.from_string(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Dict ,__a : Dict ) -> Tuple:
"""simple docstring"""
with pytest.raises(__a ,match=re.escape(expected_error.format(path='''root''' ) ) ):
ReadMe.from_string(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> Tuple:
"""simple docstring"""
ReadMe.from_string(__a ,__a ,suppress_parsing_errors=__a )
@pytest.mark.parametrize(
'''readme_md, expected_dict''' ,[
(README_CORRECT, CORRECT_DICT),
(README_CORRECT_FOUR_LEVEL, CORRECT_DICT_FOUR_LEVEL),
] ,)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : Any ) -> Optional[int]:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Tuple = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[Any] = ReadMe.from_readme(__a ,__a ).to_dict()
assert out["name"] == path
assert out["text"] == ""
assert out["is_empty_text"]
assert out["subsections"] == expected_dict["subsections"]
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_NO_YAML, EXPECTED_ERROR_README_NO_YAML),
(README_EMPTY_YAML, EXPECTED_ERROR_README_EMPTY_YAML),
(README_INCORRECT_YAML, EXPECTED_ERROR_README_INCORRECT_YAML),
(README_EMPTY, EXPECTED_ERROR_README_EMPTY),
(README_NONE_SUBSECTION, EXPECTED_ERROR_README_NONE_SUBSECTION),
(README_MISSING_FIRST_LEVEL, EXPECTED_ERROR_README_MISSING_FIRST_LEVEL),
(README_MISSING_SUBSECTION, EXPECTED_ERROR_README_MISSING_SUBSECTION),
(README_MISSING_TEXT, EXPECTED_ERROR_README_MISSING_TEXT),
(README_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_WRONG_FIRST_LEVEL),
(README_MULTIPLE_WRONG_FIRST_LEVEL, EXPECTED_ERROR_README_MULTIPLE_WRONG_FIRST_LEVEL),
(README_MISSING_CONTENT, EXPECTED_ERROR_README_MISSING_CONTENT),
] ,)
def __UpperCAmelCase ( __a : List[Any] ,__a : List[Any] ) -> int:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : Optional[int] = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
_a : Any = ReadMe.from_readme(__a ,__a )
readme.validate()
@pytest.mark.parametrize(
'''readme_md, expected_error''' ,[
(README_MULTIPLE_SAME_HEADING_1, EXPECTED_ERROR_README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : str ,__a : Union[str, Any] ) -> Dict:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : Optional[Any] = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
_a : str = expected_error.format(path=__a )
with pytest.raises(__a ,match=re.escape(__a ) ):
ReadMe.from_readme(__a ,__a )
@pytest.mark.parametrize(
'''readme_md,''' ,[
(README_MULTIPLE_SAME_HEADING_1),
] ,)
def __UpperCAmelCase ( __a : Optional[Any] ) -> str:
"""simple docstring"""
with tempfile.TemporaryDirectory() as tmp_dir:
_a : int = Path(__a ) / '''README.md'''
with open(__a ,'''w+''' ) as readme_file:
readme_file.write(__a )
ReadMe.from_readme(__a ,__a ,suppress_parsing_errors=__a )
| 15
| 0
|
from __future__ import annotations
from fractions import Fraction
from math import gcd, sqrt
def __UpperCAmelCase ( __a : int ) -> bool:
"""simple docstring"""
_a : int = int(number**0.5 )
return number == sq * sq
def __UpperCAmelCase ( __a : int ,__a : int ,__a : int ,__a : int ,__a : int ,__a : int ) -> tuple[int, int]:
"""simple docstring"""
_a : int = x_num * y_den * z_den + y_num * x_den * z_den + z_num * x_den * y_den
_a : int = x_den * y_den * z_den
_a : int = gcd(__a ,__a )
top //= hcf
bottom //= hcf
return top, bottom
def __UpperCAmelCase ( __a : int = 35 ) -> int:
"""simple docstring"""
_a : set = set()
_a : int
_a : Fraction = Fraction(0 )
_a : tuple[int, int]
for x_num in range(1 ,order + 1 ):
for x_den in range(x_num + 1 ,order + 1 ):
for y_num in range(1 ,order + 1 ):
for y_den in range(y_num + 1 ,order + 1 ):
# n=1
_a : Optional[Any] = x_num * y_den + x_den * y_num
_a : Any = x_den * y_den
_a : List[str] = gcd(__a ,__a )
z_num //= hcf
z_den //= hcf
if 0 < z_num < z_den <= order:
_a : List[Any] = add_three(
__a ,__a ,__a ,__a ,__a ,__a )
unique_s.add(__a )
# n=2
_a : Dict = (
x_num * x_num * y_den * y_den + x_den * x_den * y_num * y_num
)
_a : Tuple = x_den * x_den * y_den * y_den
if is_sq(__a ) and is_sq(__a ):
_a : List[str] = int(sqrt(__a ) )
_a : List[str] = int(sqrt(__a ) )
_a : List[Any] = gcd(__a ,__a )
z_num //= hcf
z_den //= hcf
if 0 < z_num < z_den <= order:
_a : List[Any] = add_three(
__a ,__a ,__a ,__a ,__a ,__a )
unique_s.add(__a )
# n=-1
_a : Optional[Any] = x_num * y_num
_a : List[Any] = x_den * y_num + x_num * y_den
_a : List[Any] = gcd(__a ,__a )
z_num //= hcf
z_den //= hcf
if 0 < z_num < z_den <= order:
_a : Any = add_three(
__a ,__a ,__a ,__a ,__a ,__a )
unique_s.add(__a )
# n=2
_a : int = x_num * x_num * y_num * y_num
_a : Tuple = (
x_den * x_den * y_num * y_num + x_num * x_num * y_den * y_den
)
if is_sq(__a ) and is_sq(__a ):
_a : str = int(sqrt(__a ) )
_a : Optional[int] = int(sqrt(__a ) )
_a : Optional[int] = gcd(__a ,__a )
z_num //= hcf
z_den //= hcf
if 0 < z_num < z_den <= order:
_a : Optional[int] = add_three(
__a ,__a ,__a ,__a ,__a ,__a )
unique_s.add(__a )
for num, den in unique_s:
total += Fraction(__a ,__a )
return total.denominator + total.numerator
if __name__ == "__main__":
print(f'''{solution() = }''')
| 366
|
from __future__ import annotations
def __UpperCAmelCase ( __a : list ) -> float:
"""simple docstring"""
if not nums:
raise ValueError('''List is empty''' )
return sum(__a ) / len(__a )
if __name__ == "__main__":
import doctest
doctest.testmod()
| 15
| 0
|
from transformers import BertTokenizer, EncoderDecoderModel, SeqaSeqTrainer, SeqaSeqTrainingArguments
from transformers.testing_utils import TestCasePlus, require_torch, slow
from transformers.utils import is_datasets_available
if is_datasets_available():
import datasets
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
@slow
@require_torch
def __lowercase ( self ) -> Dict:
_a : Optional[int] = EncoderDecoderModel.from_encoder_decoder_pretrained('''prajjwal1/bert-tiny''' , '''prajjwal1/bert-tiny''' )
_a : str = BertTokenizer.from_pretrained('''bert-base-uncased''' )
_a : str = bertabert.config.encoder.vocab_size
_a : int = tokenizer.sep_token_id
_a : List[Any] = tokenizer.cls_token_id
_a : Tuple = 1_2_8
_a : Union[str, Any] = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''train[:1%]''' )
_a : Optional[int] = datasets.load_dataset('''cnn_dailymail''' , '''3.0.0''' , split='''validation[:1%]''' )
_a : Union[str, Any] = train_dataset.select(range(3_2 ) )
_a : List[Any] = val_dataset.select(range(1_6 ) )
_a : Tuple = 4
def _map_to_encoder_decoder_inputs(_a ):
# Tokenizer will automatically set [BOS] <text> [EOS]
_a : List[Any] = tokenizer(batch['''article'''] , padding='''max_length''' , truncation=_a , max_length=5_1_2 )
_a : List[Any] = tokenizer(batch['''highlights'''] , padding='''max_length''' , truncation=_a , max_length=1_2_8 )
_a : Any = inputs.input_ids
_a : List[str] = inputs.attention_mask
_a : List[Any] = outputs.input_ids
_a : Any = outputs.input_ids.copy()
_a : Tuple = [
[-1_0_0 if token == tokenizer.pad_token_id else token for token in labels] for labels in batch['''labels''']
]
_a : Optional[int] = outputs.attention_mask
assert all(len(_a ) == 5_1_2 for x in inputs.input_ids )
assert all(len(_a ) == 1_2_8 for x in outputs.input_ids )
return batch
def _compute_metrics(_a ):
_a : int = pred.label_ids
_a : int = pred.predictions
# all unnecessary tokens are removed
_a : int = tokenizer.batch_decode(_a , skip_special_tokens=_a )
_a : Union[str, Any] = tokenizer.batch_decode(_a , skip_special_tokens=_a )
_a : str = sum([int(pred_str[i] == label_str[i] ) for i in range(len(_a ) )] ) / len(_a )
return {"accuracy": accuracy}
# map train dataset
_a : List[Any] = train_dataset.map(
_map_to_encoder_decoder_inputs , batched=_a , batch_size=_a , remove_columns=['''article''', '''highlights'''] , )
train_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
# same for validation dataset
_a : List[str] = val_dataset.map(
_map_to_encoder_decoder_inputs , batched=_a , batch_size=_a , remove_columns=['''article''', '''highlights'''] , )
val_dataset.set_format(
type='''torch''' , columns=['''input_ids''', '''attention_mask''', '''decoder_input_ids''', '''decoder_attention_mask''', '''labels'''] , )
_a : Union[str, Any] = self.get_auto_remove_tmp_dir()
_a : List[Any] = SeqaSeqTrainingArguments(
output_dir=_a , per_device_train_batch_size=_a , per_device_eval_batch_size=_a , predict_with_generate=_a , evaluation_strategy='''steps''' , do_train=_a , do_eval=_a , warmup_steps=0 , eval_steps=2 , logging_steps=2 , )
# instantiate trainer
_a : str = SeqaSeqTrainer(
model=_a , args=_a , compute_metrics=_compute_metrics , train_dataset=_a , eval_dataset=_a , tokenizer=_a , )
# start training
trainer.train()
| 367
|
import argparse
import os
import torch
from transformers.utils import WEIGHTS_NAME
a__ = ['''small''', '''medium''', '''large''']
a__ = '''lm_head.decoder.weight'''
a__ = '''lm_head.weight'''
def __UpperCAmelCase ( __a : str ,__a : str ) -> List[str]:
"""simple docstring"""
_a : Any = torch.load(__a )
_a : List[str] = d.pop(__a )
os.makedirs(__a ,exist_ok=__a )
torch.save(__a ,os.path.join(__a ,__a ) )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--dialogpt_path''', default='''.''', type=str)
a__ = parser.parse_args()
for MODEL in DIALOGPT_MODELS:
a__ = os.path.join(args.dialogpt_path, f'''{MODEL}_ft.pkl''')
a__ = f'''./DialoGPT-{MODEL}'''
convert_dialogpt_checkpoint(
checkpoint_path,
pytorch_dump_folder_path,
)
| 15
| 0
|
import copy
import os
from collections import OrderedDict
from typing import TYPE_CHECKING, Any, Dict, Mapping, Optional, Union
if TYPE_CHECKING:
from ...processing_utils import ProcessorMixin
from ...utils import TensorType
from ...configuration_utils import PretrainedConfig
from ...onnx import OnnxConfig
from ...utils import logging
a__ = logging.get_logger(__name__)
a__ = {
'''google/owlvit-base-patch32''': '''https://huggingface.co/google/owlvit-base-patch32/resolve/main/config.json''',
'''google/owlvit-base-patch16''': '''https://huggingface.co/google/owlvit-base-patch16/resolve/main/config.json''',
'''google/owlvit-large-patch14''': '''https://huggingface.co/google/owlvit-large-patch14/resolve/main/config.json''',
}
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : str = "owlvit_text_model"
def __init__( self , _a=4_9_4_0_8 , _a=5_1_2 , _a=2_0_4_8 , _a=1_2 , _a=8 , _a=1_6 , _a="quick_gelu" , _a=1e-5 , _a=0.0 , _a=0.02 , _a=1.0 , _a=0 , _a=4_9_4_0_6 , _a=4_9_4_0_7 , **_a , ) -> Dict:
super().__init__(pad_token_id=_a , bos_token_id=_a , eos_token_id=_a , **_a )
_a : int = vocab_size
_a : Any = hidden_size
_a : Optional[Any] = intermediate_size
_a : Optional[Any] = num_hidden_layers
_a : Optional[int] = num_attention_heads
_a : Tuple = max_position_embeddings
_a : Any = hidden_act
_a : str = layer_norm_eps
_a : int = attention_dropout
_a : Tuple = initializer_range
_a : Optional[Any] = initializer_factor
@classmethod
def __lowercase ( cls , _a , **_a ) -> "PretrainedConfig":
cls._set_token_in_kwargs(_a )
_a : Tuple = cls.get_config_dict(_a , **_a )
# get the text config dict if we are loading from OwlViTConfig
if config_dict.get('''model_type''' ) == "owlvit":
_a : List[Any] = config_dict['''text_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(_a , **_a )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[Any] = "owlvit_vision_model"
def __init__( self , _a=7_6_8 , _a=3_0_7_2 , _a=1_2 , _a=1_2 , _a=3 , _a=7_6_8 , _a=3_2 , _a="quick_gelu" , _a=1e-5 , _a=0.0 , _a=0.02 , _a=1.0 , **_a , ) -> str:
super().__init__(**_a )
_a : Optional[int] = hidden_size
_a : Union[str, Any] = intermediate_size
_a : Dict = num_hidden_layers
_a : Optional[Any] = num_attention_heads
_a : str = num_channels
_a : Optional[int] = image_size
_a : Optional[Any] = patch_size
_a : Optional[int] = hidden_act
_a : Optional[int] = layer_norm_eps
_a : int = attention_dropout
_a : Optional[Any] = initializer_range
_a : Any = initializer_factor
@classmethod
def __lowercase ( cls , _a , **_a ) -> "PretrainedConfig":
cls._set_token_in_kwargs(_a )
_a : Tuple = cls.get_config_dict(_a , **_a )
# get the vision config dict if we are loading from OwlViTConfig
if config_dict.get('''model_type''' ) == "owlvit":
_a : Optional[int] = 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(_a , **_a )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[str] = "owlvit"
UpperCAmelCase__ : Optional[int] = True
def __init__( self , _a=None , _a=None , _a=5_1_2 , _a=2.6592 , _a=True , **_a , ) -> Tuple:
super().__init__(**_a )
if text_config is None:
_a : str = {}
logger.info('''text_config is None. Initializing the OwlViTTextConfig with default values.''' )
if vision_config is None:
_a : List[Any] = {}
logger.info('''vision_config is None. initializing the OwlViTVisionConfig with default values.''' )
_a : int = OwlViTTextConfig(**_a )
_a : Union[str, Any] = OwlViTVisionConfig(**_a )
_a : List[Any] = projection_dim
_a : Optional[Any] = logit_scale_init_value
_a : List[str] = return_dict
_a : Union[str, Any] = 1.0
@classmethod
def __lowercase ( cls , _a , **_a ) -> "PretrainedConfig":
cls._set_token_in_kwargs(_a )
_a : Optional[Any] = cls.get_config_dict(_a , **_a )
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(_a , **_a )
@classmethod
def __lowercase ( cls , _a , _a , **_a ) -> Union[str, Any]:
_a : Dict = {}
_a : Tuple = text_config
_a : Optional[Any] = vision_config
return cls.from_dict(_a , **_a )
def __lowercase ( self ) -> List[str]:
_a : Tuple = copy.deepcopy(self.__dict__ )
_a : Tuple = self.text_config.to_dict()
_a : Any = self.vision_config.to_dict()
_a : Tuple = self.__class__.model_type
return output
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
@property
def __lowercase ( self ) -> Mapping[str, Mapping[int, str]]:
return OrderedDict(
[
('''input_ids''', {0: '''batch''', 1: '''sequence'''}),
('''pixel_values''', {0: '''batch''', 1: '''num_channels''', 2: '''height''', 3: '''width'''}),
('''attention_mask''', {0: '''batch''', 1: '''sequence'''}),
] )
@property
def __lowercase ( self ) -> Mapping[str, Mapping[int, str]]:
return OrderedDict(
[
('''logits_per_image''', {0: '''batch'''}),
('''logits_per_text''', {0: '''batch'''}),
('''text_embeds''', {0: '''batch'''}),
('''image_embeds''', {0: '''batch'''}),
] )
@property
def __lowercase ( self ) -> float:
return 1e-4
def __lowercase ( self , _a , _a = -1 , _a = -1 , _a = None , ) -> Mapping[str, Any]:
_a : int = super().generate_dummy_inputs(
processor.tokenizer , batch_size=_a , seq_length=_a , framework=_a )
_a : int = super().generate_dummy_inputs(
processor.image_processor , batch_size=_a , framework=_a )
return {**text_input_dict, **image_input_dict}
@property
def __lowercase ( self ) -> int:
return 1_4
| 368
|
import enum
import warnings
from .. import MODEL_FOR_CAUSAL_LM_MAPPING, TF_MODEL_FOR_CAUSAL_LM_MAPPING
from ..utils import add_end_docstrings, is_tf_available
from .base import PIPELINE_INIT_ARGS, Pipeline
if is_tf_available():
import tensorflow as tf
class UpperCAmelCase_ ( enum.Enum ):
"""simple docstring"""
UpperCAmelCase__ : int = 0
UpperCAmelCase__ : Union[str, Any] = 1
UpperCAmelCase__ : Optional[Any] = 2
@add_end_docstrings(__lowercase )
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[Any] = "\n In 1991, the remains of Russian Tsar Nicholas II and his family (except for Alexei and Maria) are discovered. The\n voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the remainder of the story. 1883 Western\n Siberia, a young Grigori Rasputin is asked by his father and a group of men to perform magic. Rasputin has a vision\n and denounces one of the men as a horse thief. Although his father initially slaps him for making such an\n accusation, Rasputin watches as the man is chased outside and beaten. Twenty years later, Rasputin sees a vision of\n the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous, with people, even a bishop,\n begging for his blessing. <eod> </s> <eos>\n "
def __init__( self , *_a , **_a ) -> List[str]:
super().__init__(*_a , **_a )
self.check_model_type(
TF_MODEL_FOR_CAUSAL_LM_MAPPING if self.framework == '''tf''' else MODEL_FOR_CAUSAL_LM_MAPPING )
if "prefix" not in self._preprocess_params:
# This is very specific. The logic is quite complex and needs to be done
# as a "default".
# It also defines both some preprocess_kwargs and generate_kwargs
# which is why we cannot put them in their respective methods.
_a : Dict = None
if self.model.config.prefix is not None:
_a : List[Any] = self.model.config.prefix
if prefix is None and self.model.__class__.__name__ in [
"XLNetLMHeadModel",
"TransfoXLLMHeadModel",
"TFXLNetLMHeadModel",
"TFTransfoXLLMHeadModel",
]:
# For XLNet and TransformerXL we add an article to the prompt to give more state to the model.
_a : Optional[Any] = self.XL_PREFIX
if prefix is not None:
# Recalculate some generate_kwargs linked to prefix.
_a , _a , _a : str = self._sanitize_parameters(prefix=_a , **self._forward_params )
_a : Optional[Any] = {**self._preprocess_params, **preprocess_params}
_a : List[Any] = {**self._forward_params, **forward_params}
def __lowercase ( self , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , _a=None , **_a , ) -> Optional[int]:
_a : List[Any] = {}
if prefix is not None:
_a : Optional[Any] = prefix
if prefix:
_a : Dict = self.tokenizer(
_a , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Tuple = prefix_inputs['''input_ids'''].shape[-1]
if handle_long_generation is not None:
if handle_long_generation not in {"hole"}:
raise ValueError(
F"""{handle_long_generation} is not a valid value for `handle_long_generation` parameter expected"""
''' [None, \'hole\']''' )
_a : Dict = handle_long_generation
preprocess_params.update(_a )
_a : Tuple = generate_kwargs
_a : Any = {}
if return_full_text is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_full_text`''' )
if return_tensors is not None:
raise ValueError('''`return_full_text` is mutually exclusive with `return_tensors`''' )
_a : List[str] = ReturnType.FULL_TEXT if return_full_text else ReturnType.NEW_TEXT
if return_tensors is not None and return_type is None:
if return_text is not None:
raise ValueError('''`return_text` is mutually exclusive with `return_tensors`''' )
_a : Any = ReturnType.TENSORS
if return_type is not None:
_a : Any = return_type
if clean_up_tokenization_spaces is not None:
_a : List[Any] = clean_up_tokenization_spaces
if stop_sequence is not None:
_a : Tuple = self.tokenizer.encode(_a , add_special_tokens=_a )
if len(_a ) > 1:
warnings.warn(
'''Stopping on a multiple token sequence is not yet supported on transformers. The first token of'''
''' the stop sequence will be used as the stop sequence string in the interim.''' )
_a : List[Any] = stop_sequence_ids[0]
return preprocess_params, forward_params, postprocess_params
def __lowercase ( self , *_a , **_a ) -> Union[str, Any]:
# Parse arguments
if self.model.__class__.__name__ in ["TransfoXLLMHeadModel"]:
kwargs.update({'''add_space_before_punct_symbol''': True} )
return super()._parse_and_tokenize(*_a , **_a )
def __call__( self , _a , **_a ) -> List[str]:
return super().__call__(_a , **_a )
def __lowercase ( self , _a , _a="" , _a=None , **_a ) -> List[Any]:
_a : Optional[int] = self.tokenizer(
prefix + prompt_text , padding=_a , add_special_tokens=_a , return_tensors=self.framework )
_a : Union[str, Any] = prompt_text
if handle_long_generation == "hole":
_a : List[str] = inputs['''input_ids'''].shape[-1]
if "max_new_tokens" in generate_kwargs:
_a : int = generate_kwargs['''max_new_tokens''']
else:
_a : List[Any] = generate_kwargs.get('''max_length''' , self.model.config.max_length ) - cur_len
if new_tokens < 0:
raise ValueError('''We cannot infer how many new tokens are expected''' )
if cur_len + new_tokens > self.tokenizer.model_max_length:
_a : List[str] = self.tokenizer.model_max_length - new_tokens
if keep_length <= 0:
raise ValueError(
'''We cannot use `hole` to handle this generation the number of desired tokens exceeds the'''
''' models max length''' )
_a : List[Any] = inputs['''input_ids'''][:, -keep_length:]
if "attention_mask" in inputs:
_a : List[str] = inputs['''attention_mask'''][:, -keep_length:]
return inputs
def __lowercase ( self , _a , **_a ) -> Optional[int]:
_a : Any = model_inputs['''input_ids''']
_a : Optional[Any] = model_inputs.get('''attention_mask''' , _a )
# Allow empty prompts
if input_ids.shape[1] == 0:
_a : int = None
_a : int = None
_a : List[str] = 1
else:
_a : List[Any] = input_ids.shape[0]
_a : Union[str, Any] = model_inputs.pop('''prompt_text''' )
# If there is a prefix, we may need to adjust the generation length. Do so without permanently modifying
# generate_kwargs, as some of the parameterization may come from the initialization of the pipeline.
_a : int = generate_kwargs.pop('''prefix_length''' , 0 )
if prefix_length > 0:
_a : Tuple = '''max_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].max_new_tokens is not None
)
if not has_max_new_tokens:
_a : int = generate_kwargs.get('''max_length''' ) or self.model.config.max_length
generate_kwargs["max_length"] += prefix_length
_a : Dict = '''min_new_tokens''' in generate_kwargs or (
'''generation_config''' in generate_kwargs
and generate_kwargs['''generation_config'''].min_new_tokens is not None
)
if not has_min_new_tokens and "min_length" in generate_kwargs:
generate_kwargs["min_length"] += prefix_length
# BS x SL
_a : Optional[Any] = self.model.generate(input_ids=_a , attention_mask=_a , **_a )
_a : int = generated_sequence.shape[0]
if self.framework == "pt":
_a : Tuple = generated_sequence.reshape(_a , out_b // in_b , *generated_sequence.shape[1:] )
elif self.framework == "tf":
_a : List[Any] = tf.reshape(_a , (in_b, out_b // in_b, *generated_sequence.shape[1:]) )
return {"generated_sequence": generated_sequence, "input_ids": input_ids, "prompt_text": prompt_text}
def __lowercase ( self , _a , _a=ReturnType.FULL_TEXT , _a=True ) -> int:
_a : Tuple = model_outputs['''generated_sequence'''][0]
_a : int = model_outputs['''input_ids''']
_a : Any = model_outputs['''prompt_text''']
_a : Any = generated_sequence.numpy().tolist()
_a : Any = []
for sequence in generated_sequence:
if return_type == ReturnType.TENSORS:
_a : Optional[int] = {'''generated_token_ids''': sequence}
elif return_type in {ReturnType.NEW_TEXT, ReturnType.FULL_TEXT}:
# Decode text
_a : str = self.tokenizer.decode(
_a , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , )
# Remove PADDING prompt of the sequence if XLNet or Transfo-XL model is used
if input_ids is None:
_a : Union[str, Any] = 0
else:
_a : str = len(
self.tokenizer.decode(
input_ids[0] , skip_special_tokens=_a , clean_up_tokenization_spaces=_a , ) )
if return_type == ReturnType.FULL_TEXT:
_a : str = prompt_text + text[prompt_length:]
else:
_a : List[str] = text[prompt_length:]
_a : Union[str, Any] = {'''generated_text''': all_text}
records.append(_a )
return records
| 15
| 0
|
from math import sqrt
import numpy as np
from sympy import symbols
# Coefficient
# Speed of light (m/s)
a__ = 299792458
# Symbols
a__ , a__ , a__ , a__ = symbols('''ct x y z''')
def __UpperCAmelCase ( __a : float ) -> float:
"""simple docstring"""
if velocity > c:
raise ValueError('''Speed must not exceed light speed 299,792,458 [m/s]!''' )
elif velocity < 1:
# Usually the speed should be much higher than 1 (c order of magnitude)
raise ValueError('''Speed must be greater than or equal to 1!''' )
return velocity / c
def __UpperCAmelCase ( __a : float ) -> float:
"""simple docstring"""
return 1 / sqrt(1 - beta(__a ) ** 2 )
def __UpperCAmelCase ( __a : float ) -> np.ndarray:
"""simple docstring"""
return np.array(
[
[gamma(__a ), -gamma(__a ) * beta(__a ), 0, 0],
[-gamma(__a ) * beta(__a ), gamma(__a ), 0, 0],
[0, 0, 1, 0],
[0, 0, 0, 1],
] )
def __UpperCAmelCase ( __a : float ,__a : np.ndarray | None = None ) -> np.ndarray:
"""simple docstring"""
if event is None:
_a : List[Any] = np.array([ct, x, y, z] ) # Symbolic four vector
else:
event[0] *= c # x0 is ct (speed of light * time)
return transformation_matrix(__a ) @ event
if __name__ == "__main__":
import doctest
doctest.testmod()
# Example of symbolic vector:
a__ = transform(29979245)
print('''Example of four vector: ''')
print(f'''ct\' = {four_vector[0]}''')
print(f'''x\' = {four_vector[1]}''')
print(f'''y\' = {four_vector[2]}''')
print(f'''z\' = {four_vector[3]}''')
# Substitute symbols with numerical values
a__ = {ct: c, x: 1, y: 1, z: 1}
a__ = [four_vector[i].subs(sub_dict) for i in range(4)]
print(f'''\n{numerical_vector}''')
| 369
|
# 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
import os
from accelerate.test_utils import execute_subprocess_async
def __UpperCAmelCase ( __a : Dict=None ) -> str:
"""simple docstring"""
if subparsers is not None:
_a : Union[str, Any] = subparsers.add_parser('''test''' )
else:
_a : List[str] = argparse.ArgumentParser('''Accelerate test command''' )
parser.add_argument(
'''--config_file''' ,default=__a ,help=(
'''The path to use to store the config file. Will default to a file named default_config.yaml in the cache '''
'''location, which is the content of the environment `HF_HOME` suffixed with \'accelerate\', or if you don\'t have '''
'''such an environment variable, your cache directory (\'~/.cache\' or the content of `XDG_CACHE_HOME`) suffixed '''
'''with \'huggingface\'.'''
) ,)
if subparsers is not None:
parser.set_defaults(func=__a )
return parser
def __UpperCAmelCase ( __a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
_a : Dict = os.path.sep.join(__file__.split(os.path.sep )[:-2] + ['''test_utils''', '''scripts''', '''test_script.py'''] )
if args.config_file is None:
_a : List[Any] = script_name
else:
_a : Union[str, Any] = F"""--config_file={args.config_file} {script_name}"""
_a : str = ['''accelerate-launch'''] + test_args.split()
_a : str = execute_subprocess_async(__a ,env=os.environ.copy() )
if result.returncode == 0:
print('''Test is a success! You are ready for your distributed training!''' )
def __UpperCAmelCase ( ) -> List[Any]:
"""simple docstring"""
_a : Optional[int] = test_command_parser()
_a : List[Any] = parser.parse_args()
test_command(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
from typing import TYPE_CHECKING
from ...utils import (
OptionalDependencyNotAvailable,
_LazyModule,
is_torch_available,
)
a__ = {
'''configuration_gpt_bigcode''': ['''GPT_BIGCODE_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''GPTBigCodeConfig'''],
}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''GPT_BIGCODE_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''GPTBigCodeForSequenceClassification''',
'''GPTBigCodeForTokenClassification''',
'''GPTBigCodeForCausalLM''',
'''GPTBigCodeModel''',
'''GPTBigCodePreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_gpt_bigcode import GPT_BIGCODE_PRETRAINED_CONFIG_ARCHIVE_MAP, GPTBigCodeConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_gpt_bigcode import (
GPT_BIGCODE_PRETRAINED_MODEL_ARCHIVE_LIST,
GPTBigCodeForCausalLM,
GPTBigCodeForSequenceClassification,
GPTBigCodeForTokenClassification,
GPTBigCodeModel,
GPTBigCodePreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 370
|
import json
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import BertTokenizerFast
from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer
from transformers.testing_utils import require_tokenizers, require_vision
from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available
if is_vision_available():
from PIL import Image
from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor
@require_tokenizers
@require_vision
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = tempfile.mkdtemp()
# fmt: off
_a : Optional[int] = ['''[UNK]''', '''[CLS]''', '''[SEP]''', '''[PAD]''', '''[MASK]''', '''want''', '''##want''', '''##ed''', '''wa''', '''un''', '''runn''', '''##ing''', ''',''', '''low''', '''lowest''']
# fmt: on
_a : Tuple = 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] ) )
_a : Any = {
'''do_resize''': True,
'''size''': {'''height''': 1_8, '''width''': 1_8},
'''do_normalize''': True,
'''image_mean''': [0.5, 0.5, 0.5],
'''image_std''': [0.5, 0.5, 0.5],
}
_a : str = os.path.join(self.tmpdirname , _a )
with open(self.image_processor_file , '''w''' , encoding='''utf-8''' ) as fp:
json.dump(_a , _a )
def __lowercase ( self , **_a ) -> Any:
return BertTokenizer.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self , **_a ) -> str:
return ViTImageProcessor.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self ) -> List[Any]:
shutil.rmtree(self.tmpdirname )
def __lowercase ( self ) -> Any:
_a : Union[str, Any] = [np.random.randint(2_5_5 , size=(3, 3_0, 4_0_0) , dtype=np.uinta )]
_a : Tuple = [Image.fromarray(np.moveaxis(_a , 0 , -1 ) ) for x in image_inputs]
return image_inputs
def __lowercase ( self ) -> str:
_a : List[str] = self.get_tokenizer()
_a : Tuple = self.get_image_processor()
_a : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
processor.save_pretrained(self.tmpdirname )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Dict:
_a : List[str] = VisionTextDualEncoderProcessor(
tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() )
processor.save_pretrained(self.tmpdirname )
_a : Any = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' )
_a : List[Any] = self.get_image_processor(do_normalize=_a , padding_value=1.0 )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(
self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=_a , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Any:
_a : Dict = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : int = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[str] = self.prepare_image_inputs()
_a : List[Any] = image_processor(_a , return_tensors='''np''' )
_a : Dict = processor(images=_a , return_tensors='''np''' )
for key in input_feat_extract.keys():
self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1e-2 )
def __lowercase ( self ) -> List[str]:
_a : Union[str, Any] = self.get_image_processor()
_a : Dict = self.get_tokenizer()
_a : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Tuple = '''lower newer'''
_a : int = processor(text=_a )
_a : str = tokenizer(_a )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def __lowercase ( self ) -> List[Any]:
_a : Any = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : Tuple = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[Any] = '''lower newer'''
_a : Union[str, Any] = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , ['''input_ids''', '''token_type_ids''', '''attention_mask''', '''pixel_values'''] )
# test if it raises when no input is passed
with self.assertRaises(_a ):
processor()
def __lowercase ( self ) -> Optional[int]:
_a : Union[str, Any] = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : Any = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Any = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
_a : int = processor.batch_decode(_a )
_a : int = tokenizer.batch_decode(_a )
self.assertListEqual(_a , _a )
def __lowercase ( self ) -> List[Any]:
_a : Tuple = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : str = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Optional[int] = '''lower newer'''
_a : Dict = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
| 15
| 0
|
import math
import unittest
from transformers import BioGptConfig, 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, random_attention_mask
from ...test_pipeline_mixin import PipelineTesterMixin
if is_torch_available():
import torch
from transformers import (
BioGptForCausalLM,
BioGptForSequenceClassification,
BioGptForTokenClassification,
BioGptModel,
BioGptTokenizer,
)
from transformers.models.biogpt.modeling_biogpt import BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST
class UpperCAmelCase_ :
"""simple docstring"""
def __init__( self , _a , _a=1_3 , _a=7 , _a=True , _a=True , _a=False , _a=True , _a=9_9 , _a=3_2 , _a=5 , _a=4 , _a=3_7 , _a="gelu" , _a=0.1 , _a=0.1 , _a=5_1_2 , _a=1_6 , _a=2 , _a=0.02 , _a=3 , _a=4 , _a=None , ) -> Optional[Any]:
_a : str = parent
_a : Optional[Any] = batch_size
_a : Tuple = seq_length
_a : Union[str, Any] = is_training
_a : Optional[Any] = use_input_mask
_a : List[Any] = use_token_type_ids
_a : Any = use_labels
_a : Tuple = vocab_size
_a : Tuple = hidden_size
_a : Optional[int] = num_hidden_layers
_a : List[Any] = num_attention_heads
_a : List[Any] = intermediate_size
_a : int = hidden_act
_a : List[Any] = hidden_dropout_prob
_a : str = attention_probs_dropout_prob
_a : int = max_position_embeddings
_a : List[Any] = type_vocab_size
_a : int = type_sequence_label_size
_a : str = initializer_range
_a : Dict = num_labels
_a : Union[str, Any] = num_choices
_a : Any = scope
def __lowercase ( self ) -> str:
_a : Union[str, Any] = ids_tensor([self.batch_size, self.seq_length] , self.vocab_size )
_a : Any = None
if self.use_input_mask:
_a : Dict = random_attention_mask([self.batch_size, self.seq_length] )
_a : int = None
if self.use_token_type_ids:
_a : List[str] = ids_tensor([self.batch_size, self.seq_length] , self.type_vocab_size )
_a : Tuple = None
_a : Union[str, Any] = None
_a : Any = None
if self.use_labels:
_a : Tuple = 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 : Dict = ids_tensor([self.batch_size] , self.num_choices )
_a : Any = self.get_config()
return config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels, choice_labels
def __lowercase ( self ) -> str:
return BioGptConfig(
vocab_size=self.vocab_size , hidden_size=self.hidden_size , num_hidden_layers=self.num_hidden_layers , num_attention_heads=self.num_attention_heads , intermediate_size=self.intermediate_size , hidden_act=self.hidden_act , hidden_dropout_prob=self.hidden_dropout_prob , attention_probs_dropout_prob=self.attention_probs_dropout_prob , max_position_embeddings=self.max_position_embeddings , type_vocab_size=self.type_vocab_size , is_decoder=_a , initializer_range=self.initializer_range , )
def __lowercase ( self , _a , _a , _a , _a , _a , _a , _a ) -> Optional[int]:
_a : Union[str, Any] = BioGptModel(config=_a )
model.to(_a )
model.eval()
_a : Optional[Any] = model(_a , attention_mask=_a )
_a : List[Any] = model(_a )
self.parent.assertEqual(result.last_hidden_state.shape , (self.batch_size, self.seq_length, self.hidden_size) )
def __lowercase ( self , _a , _a , _a , _a , _a , _a , _a , _a , _a , ) -> Tuple:
_a : int = BioGptForCausalLM(config=_a )
model.to(_a )
model.eval()
_a : Dict = model(_a , attention_mask=_a , token_type_ids=_a , labels=_a )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
def __lowercase ( self , _a , _a , _a , _a , _a , *_a ) -> Union[str, Any]:
_a : List[Any] = BioGptModel(config=_a )
model.to(_a )
model.eval()
# create attention mask
_a : List[str] = torch.ones(input_ids.shape , dtype=torch.long , device=_a )
_a : List[Any] = self.seq_length // 2
_a : Union[str, Any] = 0
# first forward pass
_a : Optional[Any] = model(_a , attention_mask=_a ).to_tuple()
# create hypothetical next token and extent to next_input_ids
_a : Tuple = ids_tensor((self.batch_size, 1) , config.vocab_size )
# change a random masked slice from input_ids
_a : Tuple = ids_tensor((1,) , _a ).item() + 1
_a : List[Any] = ids_tensor((self.batch_size, 1) , config.vocab_size ).squeeze(-1 )
_a : Dict = random_other_next_tokens
# append to next input_ids and attn_mask
_a : Tuple = torch.cat([input_ids, next_tokens] , dim=-1 )
_a : List[str] = torch.cat(
[attn_mask, torch.ones((attn_mask.shape[0], 1) , dtype=torch.long , device=_a )] , dim=1 , )
# get two different outputs
_a : Optional[int] = model(_a , attention_mask=_a )['''last_hidden_state''']
_a : Tuple = model(_a , past_key_values=_a , attention_mask=_a )['''last_hidden_state''']
# select random slice
_a : Optional[int] = ids_tensor((1,) , output_from_past.shape[-1] ).item()
_a : Any = output_from_no_past[:, -1, random_slice_idx].detach()
_a : int = output_from_past[:, 0, random_slice_idx].detach()
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(_a , _a , atol=1e-3 ) )
def __lowercase ( self , _a , _a , _a , _a , _a , *_a ) -> Dict:
_a : Optional[Any] = BioGptModel(config=_a ).to(_a ).eval()
_a : Optional[Any] = torch.ones(input_ids.shape , dtype=torch.long , device=_a )
# first forward pass
_a : Union[str, Any] = model(_a , attention_mask=_a , use_cache=_a )
_a : str = outputs.to_tuple()
# create hypothetical multiple next token and extent to next_input_ids
_a : Optional[Any] = ids_tensor((self.batch_size, 3) , config.vocab_size )
_a : Optional[Any] = ids_tensor((self.batch_size, 3) , 2 )
# append to next input_ids and
_a : List[str] = torch.cat([input_ids, next_tokens] , dim=-1 )
_a : int = torch.cat([attention_mask, next_attn_mask] , dim=-1 )
_a : Optional[Any] = model(_a , attention_mask=_a )['''last_hidden_state''']
_a : int = model(_a , attention_mask=_a , past_key_values=_a )[
'''last_hidden_state'''
]
# select random slice
_a : Optional[Any] = ids_tensor((1,) , output_from_past.shape[-1] ).item()
_a : str = output_from_no_past[:, -3:, random_slice_idx].detach()
_a : Optional[int] = output_from_past[:, :, random_slice_idx].detach()
self.parent.assertTrue(output_from_past_slice.shape[1] == next_tokens.shape[1] )
# test that outputs are equal for slice
self.parent.assertTrue(torch.allclose(_a , _a , atol=1e-3 ) )
def __lowercase ( self , _a , _a , _a , _a , _a , *_a , _a=False ) -> Any:
_a : List[Any] = BioGptForCausalLM(_a )
model.to(_a )
if gradient_checkpointing:
model.gradient_checkpointing_enable()
_a : Dict = model(_a , labels=_a )
self.parent.assertEqual(result.loss.shape , () )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.vocab_size) )
result.loss.backward()
def __lowercase ( self , _a , *_a ) -> str:
_a : Union[str, Any] = BioGptModel(_a )
_a : Union[str, Any] = model.config.initializer_range / math.sqrt(2 * model.config.num_hidden_layers )
for key in model.state_dict().keys():
if "c_proj" in key and "weight" in key:
self.parent.assertLessEqual(abs(torch.std(model.state_dict()[key] ) - model_std ) , 0.001 )
self.parent.assertLessEqual(abs(torch.mean(model.state_dict()[key] ) - 0.0 ) , 0.01 )
def __lowercase ( self , _a , _a , _a , _a , _a , *_a ) -> Union[str, Any]:
_a : Optional[Any] = self.num_labels
_a : str = BioGptForTokenClassification(_a )
model.to(_a )
model.eval()
_a : int = model(_a , attention_mask=_a , token_type_ids=_a )
self.parent.assertEqual(result.logits.shape , (self.batch_size, self.seq_length, self.num_labels) )
def __lowercase ( self ) -> List[str]:
_a : List[Any] = self.prepare_config_and_inputs()
(
_a
) : Union[str, Any] = config_and_inputs
_a : Tuple = {'''input_ids''': input_ids, '''attention_mask''': input_mask}
return config, inputs_dict
@require_torch
class UpperCAmelCase_ ( __lowercase , __lowercase , __lowercase , unittest.TestCase ):
"""simple docstring"""
UpperCAmelCase__ : int = (
(BioGptModel, BioGptForCausalLM, BioGptForSequenceClassification, BioGptForTokenClassification)
if is_torch_available()
else ()
)
UpperCAmelCase__ : Optional[Any] = (BioGptForCausalLM,) if is_torch_available() else ()
UpperCAmelCase__ : int = (
{
"feature-extraction": BioGptModel,
"text-classification": BioGptForSequenceClassification,
"text-generation": BioGptForCausalLM,
"token-classification": BioGptForTokenClassification,
"zero-shot": BioGptForSequenceClassification,
}
if is_torch_available()
else {}
)
UpperCAmelCase__ : Optional[Any] = False
def __lowercase ( self ) -> str:
_a : str = BioGptModelTester(self )
_a : Any = ConfigTester(self , config_class=_a , hidden_size=3_7 )
def __lowercase ( self ) -> List[str]:
self.config_tester.run_common_tests()
def __lowercase ( self ) -> Any:
_a : Optional[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_model(*_a )
def __lowercase ( self ) -> Any:
_a : Optional[int] = self.model_tester.prepare_config_and_inputs()
for type in ["absolute", "relative_key", "relative_key_query"]:
_a : Any = type
self.model_tester.create_and_check_model(*_a )
def __lowercase ( self ) -> List[Any]:
_a : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_biogpt_model_attention_mask_past(*_a )
def __lowercase ( self ) -> Optional[Any]:
_a : List[Any] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_forward_and_backwards(*_a , gradient_checkpointing=_a )
def __lowercase ( self ) -> Optional[Any]:
_a : List[str] = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_biogpt_model_past_large_inputs(*_a )
def __lowercase ( self ) -> List[str]:
_a : Dict = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_biogpt_weight_initialization(*_a )
def __lowercase ( self ) -> Optional[int]:
_a : int = self.model_tester.prepare_config_and_inputs()
self.model_tester.create_and_check_biogpt_for_token_classification(*_a )
@slow
def __lowercase ( self ) -> List[Any]:
_a : List[str] = BioGptForCausalLM.from_pretrained('''microsoft/biogpt''' )
model.to(_a )
_a : Tuple = BioGptTokenizer.from_pretrained('''microsoft/biogpt''' )
_a : Optional[Any] = '''left'''
# Define PAD Token = EOS Token = 50256
_a : Dict = tokenizer.eos_token
_a : Tuple = model.config.eos_token_id
# use different length sentences to test batching
_a : str = [
'''Hello, my dog is a little''',
'''Today, I''',
]
_a : Tuple = tokenizer(_a , return_tensors='''pt''' , padding=_a )
_a : Optional[int] = inputs['''input_ids'''].to(_a )
_a : int = model.generate(
input_ids=_a , attention_mask=inputs['''attention_mask'''].to(_a ) , )
_a : str = tokenizer(sentences[0] , return_tensors='''pt''' ).input_ids.to(_a )
_a : Optional[int] = model.generate(input_ids=_a )
_a : Any = inputs_non_padded.shape[-1] - inputs['''attention_mask'''][-1].long().sum().cpu().item()
_a : Any = tokenizer(sentences[1] , return_tensors='''pt''' ).input_ids.to(_a )
_a : List[str] = model.generate(input_ids=_a , max_length=model.config.max_length - num_paddings )
_a : int = tokenizer.batch_decode(_a , skip_special_tokens=_a )
_a : List[str] = tokenizer.decode(output_non_padded[0] , skip_special_tokens=_a )
_a : Dict = tokenizer.decode(output_padded[0] , skip_special_tokens=_a )
_a : Optional[int] = [
'''Hello, my dog is a little bit bigger than a little bit.''',
'''Today, I have a good idea of how to use the information''',
]
self.assertListEqual(_a , _a )
self.assertListEqual(_a , [non_padded_sentence, padded_sentence] )
@slow
def __lowercase ( self ) -> Any:
for model_name in BIOGPT_PRETRAINED_MODEL_ARCHIVE_LIST[:1]:
_a : int = BioGptModel.from_pretrained(_a )
self.assertIsNotNone(_a )
def __lowercase ( self ) -> Union[str, Any]:
_a : Any = self.model_tester.prepare_config_and_inputs_for_common()
_a : List[Any] = 3
_a : str = input_dict['''input_ids''']
_a : Any = input_ids.ne(1 ).to(_a )
_a : Union[str, Any] = ids_tensor([self.model_tester.batch_size] , self.model_tester.type_sequence_label_size )
_a : Any = BioGptForSequenceClassification(_a )
model.to(_a )
model.eval()
_a : Tuple = model(_a , attention_mask=_a , labels=_a )
self.assertEqual(result.logits.shape , (self.model_tester.batch_size, self.model_tester.num_labels) )
def __lowercase ( self ) -> Any:
_a : Dict = self.model_tester.prepare_config_and_inputs_for_common()
_a : Tuple = 3
_a : Union[str, Any] = '''multi_label_classification'''
_a : Optional[int] = input_dict['''input_ids''']
_a : Optional[int] = input_ids.ne(1 ).to(_a )
_a : Union[str, Any] = ids_tensor(
[self.model_tester.batch_size, config.num_labels] , self.model_tester.type_sequence_label_size ).to(torch.float )
_a : str = BioGptForSequenceClassification(_a )
model.to(_a )
model.eval()
_a : Optional[int] = model(_a , attention_mask=_a , labels=_a )
self.assertEqual(result.logits.shape , (self.model_tester.batch_size, self.model_tester.num_labels) )
@require_torch
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
@slow
def __lowercase ( self ) -> Dict:
_a : Optional[Any] = BioGptForCausalLM.from_pretrained('''microsoft/biogpt''' )
_a : List[str] = torch.tensor([[2, 4_8_0_5, 9, 6_5_6, 2_1]] )
_a : List[str] = model(_a )[0]
_a : List[Any] = 4_2_3_8_4
_a : Tuple = torch.Size((1, 5, vocab_size) )
self.assertEqual(output.shape , _a )
_a : Dict = torch.tensor(
[[[-9.5236, -9.8918, 10.4557], [-11.0469, -9.6423, 8.1022], [-8.8664, -7.8826, 5.5325]]] )
self.assertTrue(torch.allclose(output[:, :3, :3] , _a , atol=1e-4 ) )
@slow
def __lowercase ( self ) -> Dict:
_a : Optional[int] = BioGptTokenizer.from_pretrained('''microsoft/biogpt''' )
_a : Any = BioGptForCausalLM.from_pretrained('''microsoft/biogpt''' )
model.to(_a )
torch.manual_seed(0 )
_a : Tuple = tokenizer('''COVID-19 is''' , return_tensors='''pt''' ).to(_a )
_a : Optional[int] = model.generate(
**_a , min_length=1_0_0 , max_length=1_0_2_4 , num_beams=5 , early_stopping=_a , )
_a : Optional[int] = tokenizer.decode(output_ids[0] , skip_special_tokens=_a )
_a : Optional[int] = (
'''COVID-19 is a global pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the'''
''' causative agent of coronavirus disease 2019 (COVID-19), which has spread to more than 200 countries and'''
''' territories, including the United States (US), Canada, Australia, New Zealand, the United Kingdom (UK),'''
''' and the United States of America (USA), as of March 11, 2020, with more than 800,000 confirmed cases and'''
''' more than 800,000 deaths.'''
)
self.assertEqual(_a , _a )
| 371
|
import argparse
import json
import os
import fairseq
import torch
from fairseq.data import Dictionary
from transformers import (
HubertConfig,
HubertForCTC,
HubertModel,
WavaVecaCTCTokenizer,
WavaVecaFeatureExtractor,
WavaVecaProcessor,
logging,
)
logging.set_verbosity_info()
a__ = logging.get_logger(__name__)
a__ = {
'''post_extract_proj''': '''feature_projection.projection''',
'''encoder.pos_conv.0''': '''encoder.pos_conv_embed.conv''',
'''self_attn.k_proj''': '''encoder.layers.*.attention.k_proj''',
'''self_attn.v_proj''': '''encoder.layers.*.attention.v_proj''',
'''self_attn.q_proj''': '''encoder.layers.*.attention.q_proj''',
'''self_attn.out_proj''': '''encoder.layers.*.attention.out_proj''',
'''self_attn_layer_norm''': '''encoder.layers.*.layer_norm''',
'''fc1''': '''encoder.layers.*.feed_forward.intermediate_dense''',
'''fc2''': '''encoder.layers.*.feed_forward.output_dense''',
'''final_layer_norm''': '''encoder.layers.*.final_layer_norm''',
'''encoder.layer_norm''': '''encoder.layer_norm''',
'''w2v_model.layer_norm''': '''feature_projection.layer_norm''',
'''w2v_encoder.proj''': '''lm_head''',
'''mask_emb''': '''masked_spec_embed''',
}
def __UpperCAmelCase ( __a : List[Any] ,__a : Optional[int] ,__a : Optional[int] ,__a : List[str] ,__a : List[Any] ) -> Union[str, Any]:
"""simple docstring"""
for attribute in key.split('''.''' ):
_a : Optional[Any] = getattr(__a ,__a )
if weight_type is not None:
_a : Dict = getattr(__a ,__a ).shape
else:
_a : Optional[int] = 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":
_a : List[Any] = value
elif weight_type == "weight_g":
_a : Any = value
elif weight_type == "weight_v":
_a : Union[str, Any] = value
elif weight_type == "bias":
_a : Optional[int] = value
else:
_a : List[Any] = value
logger.info(F"""{key + '.' + weight_type if weight_type is not None else ''} was initialized from {full_name}.""" )
def __UpperCAmelCase ( __a : Any ,__a : Union[str, Any] ,__a : Union[str, Any] ) -> int:
"""simple docstring"""
_a : Union[str, Any] = []
_a : Union[str, Any] = fairseq_model.state_dict()
_a : Union[str, Any] = hf_model.hubert.feature_extractor if is_finetuned else hf_model.feature_extractor
for name, value in fairseq_dict.items():
_a : int = False
if "conv_layers" in name:
load_conv_layer(
__a ,__a ,__a ,__a ,hf_model.config.feat_extract_norm == '''group''' ,)
_a : Optional[Any] = True
else:
for key, mapped_key in MAPPING.items():
_a : Union[str, Any] = '''hubert.''' + 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] and not is_finetuned):
_a : Any = True
if "*" in mapped_key:
_a : Optional[int] = name.split(__a )[0].split('''.''' )[-2]
_a : Any = mapped_key.replace('''*''' ,__a )
if "weight_g" in name:
_a : List[Any] = '''weight_g'''
elif "weight_v" in name:
_a : List[str] = '''weight_v'''
elif "weight" in name:
_a : Any = '''weight'''
elif "bias" in name:
_a : str = '''bias'''
else:
_a : Any = None
set_recursively(__a ,__a ,__a ,__a ,__a )
continue
if not is_used:
unused_weights.append(__a )
logger.warning(F"""Unused weights: {unused_weights}""" )
def __UpperCAmelCase ( __a : int ,__a : Optional[Any] ,__a : Dict ,__a : List[str] ,__a : Any ) -> Tuple:
"""simple docstring"""
_a : int = full_name.split('''conv_layers.''' )[-1]
_a : Any = name.split('''.''' )
_a : List[Any] = int(items[0] )
_a : Optional[int] = 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."""
)
_a : Optional[int] = 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."""
)
_a : Optional[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:
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."
)
_a : int = 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."""
)
_a : Any = value
logger.info(F"""Feat extract layer norm weight of layer {layer_id} was initialized from {full_name}.""" )
else:
unused_weights.append(__a )
@torch.no_grad()
def __UpperCAmelCase ( __a : Dict ,__a : List[Any] ,__a : List[str]=None ,__a : Optional[int]=None ,__a : int=True ) -> List[Any]:
"""simple docstring"""
if config_path is not None:
_a : Tuple = HubertConfig.from_pretrained(__a )
else:
_a : Any = HubertConfig()
if is_finetuned:
if dict_path:
_a : Tuple = Dictionary.load(__a )
# important change bos & pad token id since CTC symbol is <pad> and
# not <s> as in fairseq
_a : Any = target_dict.pad_index
_a : Tuple = target_dict.bos_index
_a : Optional[int] = target_dict.eos_index
_a : Optional[Any] = len(target_dict.symbols )
_a : Tuple = os.path.join(__a ,'''vocab.json''' )
if not os.path.isdir(__a ):
logger.error('''--pytorch_dump_folder_path ({}) should be a directory'''.format(__a ) )
return
os.makedirs(__a ,exist_ok=__a )
with open(__a ,'''w''' ,encoding='''utf-8''' ) as vocab_handle:
json.dump(target_dict.indices ,__a )
_a : Tuple = WavaVecaCTCTokenizer(
__a ,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=__a ,)
_a : Tuple = True if config.feat_extract_norm == '''layer''' else False
_a : List[Any] = WavaVecaFeatureExtractor(
feature_size=1 ,sampling_rate=16_000 ,padding_value=0 ,do_normalize=__a ,return_attention_mask=__a ,)
_a : List[Any] = WavaVecaProcessor(feature_extractor=__a ,tokenizer=__a )
processor.save_pretrained(__a )
_a : Tuple = HubertForCTC(__a )
else:
_a : Tuple = HubertModel(__a )
if is_finetuned:
_a , _a , _a : int = fairseq.checkpoint_utils.load_model_ensemble_and_task(
[checkpoint_path] ,arg_overrides={'''data''': '''/'''.join(dict_path.split('''/''' )[:-1] )} )
else:
_a , _a , _a : str = fairseq.checkpoint_utils.load_model_ensemble_and_task([checkpoint_path] )
_a : Any = model[0].eval()
recursively_load_weights(__a ,__a ,__a )
hf_wavavec.save_pretrained(__a )
if __name__ == "__main__":
a__ = argparse.ArgumentParser()
parser.add_argument('''--pytorch_dump_folder_path''', default=None, type=str, help='''Path to the output PyTorch model.''')
parser.add_argument('''--checkpoint_path''', default=None, type=str, help='''Path to fairseq checkpoint''')
parser.add_argument('''--dict_path''', default=None, type=str, help='''Path to dict of fine-tuned model''')
parser.add_argument('''--config_path''', default=None, type=str, help='''Path to hf config.json of model to convert''')
parser.add_argument(
'''--not_finetuned''', action='''store_true''', help='''Whether the model to convert is a fine-tuned model or not'''
)
a__ = parser.parse_args()
convert_hubert_checkpoint(
args.checkpoint_path, args.pytorch_dump_folder_path, args.config_path, args.dict_path, not args.not_finetuned
)
| 15
| 0
|
from typing import TYPE_CHECKING
from ...utils import OptionalDependencyNotAvailable, _LazyModule, is_torch_available
a__ = {'''configuration_wavlm''': ['''WAVLM_PRETRAINED_CONFIG_ARCHIVE_MAP''', '''WavLMConfig''']}
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
a__ = [
'''WAVLM_PRETRAINED_MODEL_ARCHIVE_LIST''',
'''WavLMForAudioFrameClassification''',
'''WavLMForCTC''',
'''WavLMForSequenceClassification''',
'''WavLMForXVector''',
'''WavLMModel''',
'''WavLMPreTrainedModel''',
]
if TYPE_CHECKING:
from .configuration_wavlm import WAVLM_PRETRAINED_CONFIG_ARCHIVE_MAP, WavLMConfig
try:
if not is_torch_available():
raise OptionalDependencyNotAvailable()
except OptionalDependencyNotAvailable:
pass
else:
from .modeling_wavlm import (
WAVLM_PRETRAINED_MODEL_ARCHIVE_LIST,
WavLMForAudioFrameClassification,
WavLMForCTC,
WavLMForSequenceClassification,
WavLMForXVector,
WavLMModel,
WavLMPreTrainedModel,
)
else:
import sys
a__ = _LazyModule(__name__, globals()['''__file__'''], _import_structure, module_spec=__spec__)
| 350
|
import warnings
from typing import List, Optional, Union
from ...processing_utils import ProcessorMixin
from ...tokenization_utils_base import BatchEncoding, PaddingStrategy, PreTokenizedInput, TextInput, TruncationStrategy
from ...utils import TensorType
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : List[str] = ["image_processor", "tokenizer"]
UpperCAmelCase__ : str = "ViltImageProcessor"
UpperCAmelCase__ : Union[str, Any] = ("BertTokenizer", "BertTokenizerFast")
def __init__( self , _a=None , _a=None , **_a ) -> Any:
_a : Union[str, Any] = None
if "feature_extractor" in kwargs:
warnings.warn(
'''The `feature_extractor` argument is deprecated and will be removed in v5, use `image_processor`'''
''' instead.''' , _a , )
_a : Dict = kwargs.pop('''feature_extractor''' )
_a : Optional[int] = image_processor if image_processor is not None else feature_extractor
if image_processor is None:
raise ValueError('''You need to specify an `image_processor`.''' )
if tokenizer is None:
raise ValueError('''You need to specify a `tokenizer`.''' )
super().__init__(_a , _a )
_a : int = self.image_processor
def __call__( self , _a , _a = None , _a = True , _a = False , _a = None , _a = None , _a = 0 , _a = None , _a = None , _a = None , _a = False , _a = False , _a = False , _a = False , _a = True , _a = None , **_a , ) -> BatchEncoding:
_a : Tuple = self.tokenizer(
text=_a , add_special_tokens=_a , padding=_a , truncation=_a , max_length=_a , stride=_a , pad_to_multiple_of=_a , return_token_type_ids=_a , return_attention_mask=_a , return_overflowing_tokens=_a , return_special_tokens_mask=_a , return_offsets_mapping=_a , return_length=_a , verbose=_a , return_tensors=_a , **_a , )
# add pixel_values + pixel_mask
_a : str = self.image_processor(_a , return_tensors=_a )
encoding.update(_a )
return encoding
def __lowercase ( self , *_a , **_a ) -> Optional[Any]:
return self.tokenizer.batch_decode(*_a , **_a )
def __lowercase ( self , *_a , **_a ) -> str:
return self.tokenizer.decode(*_a , **_a )
@property
def __lowercase ( self ) -> Optional[int]:
_a : str = self.tokenizer.model_input_names
_a : Optional[Any] = self.image_processor.model_input_names
return list(dict.fromkeys(tokenizer_input_names + image_processor_input_names ) )
@property
def __lowercase ( self ) -> Optional[Any]:
warnings.warn(
'''`feature_extractor_class` is deprecated and will be removed in v5. Use `image_processor_class` instead.''' , _a , )
return self.image_processor_class
@property
def __lowercase ( self ) -> Any:
warnings.warn(
'''`feature_extractor` is deprecated and will be removed in v5. Use `image_processor` instead.''' , _a , )
return self.image_processor
| 15
| 0
|
import os
import tempfile
from functools import partial
from unittest import TestCase
from unittest.mock import patch
import numpy as np
import pytest
from datasets.arrow_dataset import Dataset
from datasets.search import ElasticSearchIndex, FaissIndex, MissingIndex
from .utils import require_elasticsearch, require_faiss
a__ = pytest.mark.integration
@require_faiss
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self ) -> List[Any]:
_a : Tuple = Dataset.from_dict({'''filename''': ['''my_name-train''' + '''_''' + str(_a ) for x in np.arange(3_0 ).tolist()]} )
return dset
def __lowercase ( self ) -> List[Any]:
import faiss
_a : Dataset = self._create_dummy_dataset()
_a : Union[str, Any] = dset.map(
lambda _a , _a : {"vecs": i * np.ones(5 , dtype=np.floataa )} , with_indices=_a , keep_in_memory=_a )
_a : List[str] = dset.add_faiss_index('''vecs''' , batch_size=1_0_0 , metric_type=faiss.METRIC_INNER_PRODUCT )
_a : Optional[Any] = dset.get_nearest_examples('''vecs''' , np.ones(5 , dtype=np.floataa ) )
self.assertEqual(examples['''filename'''][0] , '''my_name-train_29''' )
dset.drop_index('''vecs''' )
def __lowercase ( self ) -> Optional[int]:
import faiss
_a : Dataset = self._create_dummy_dataset()
dset.add_faiss_index_from_external_arrays(
external_arrays=np.ones((3_0, 5) ) * np.arange(3_0 ).reshape(-1 , 1 ) , index_name='''vecs''' , batch_size=1_0_0 , metric_type=faiss.METRIC_INNER_PRODUCT , )
_a : Union[str, Any] = dset.get_nearest_examples('''vecs''' , np.ones(5 , dtype=np.floataa ) )
self.assertEqual(examples['''filename'''][0] , '''my_name-train_29''' )
def __lowercase ( self ) -> str:
import faiss
_a : Dataset = self._create_dummy_dataset()
dset.add_faiss_index_from_external_arrays(
external_arrays=np.ones((3_0, 5) ) * np.arange(3_0 ).reshape(-1 , 1 ) , index_name='''vecs''' , metric_type=faiss.METRIC_INNER_PRODUCT , )
# Setting delete=False and unlinking manually is not pretty... but it is required on Windows to
# ensure somewhat stable behaviour. If we don't, we get PermissionErrors. This is an age-old issue.
# see https://bugs.python.org/issue14243 and
# https://stackoverflow.com/questions/23212435/permission-denied-to-write-to-my-temporary-file/23212515
with tempfile.NamedTemporaryFile(delete=_a ) as tmp_file:
dset.save_faiss_index('''vecs''' , tmp_file.name )
dset.load_faiss_index('''vecs2''' , tmp_file.name )
os.unlink(tmp_file.name )
_a : str = dset.get_nearest_examples('''vecs2''' , np.ones(5 , dtype=np.floataa ) )
self.assertEqual(examples['''filename'''][0] , '''my_name-train_29''' )
def __lowercase ( self ) -> Optional[int]:
_a : Dataset = self._create_dummy_dataset()
dset.add_faiss_index_from_external_arrays(
external_arrays=np.ones((3_0, 5) ) * np.arange(3_0 ).reshape(-1 , 1 ) , index_name='''vecs''' )
dset.drop_index('''vecs''' )
self.assertRaises(_a , partial(dset.get_nearest_examples , '''vecs2''' , np.ones(5 , dtype=np.floataa ) ) )
def __lowercase ( self ) -> Any:
from elasticsearch import Elasticsearch
_a : Dataset = self._create_dummy_dataset()
with patch('''elasticsearch.Elasticsearch.search''' ) as mocked_search, patch(
'''elasticsearch.client.IndicesClient.create''' ) as mocked_index_create, patch('''elasticsearch.helpers.streaming_bulk''' ) as mocked_bulk:
_a : Dict = {'''acknowledged''': True}
mocked_bulk.return_value([(True, None)] * 3_0 )
_a : Tuple = {'''hits''': {'''hits''': [{'''_score''': 1, '''_id''': 2_9}]}}
_a : Optional[int] = Elasticsearch()
dset.add_elasticsearch_index('''filename''' , es_client=_a )
_a : Optional[Any] = dset.get_nearest_examples('''filename''' , '''my_name-train_29''' )
self.assertEqual(examples['''filename'''][0] , '''my_name-train_29''' )
@require_faiss
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self ) -> Tuple:
import faiss
_a : Tuple = FaissIndex(metric_type=faiss.METRIC_INNER_PRODUCT )
# add vectors
index.add_vectors(np.eye(5 , dtype=np.floataa ) )
self.assertIsNotNone(index.faiss_index )
self.assertEqual(index.faiss_index.ntotal , 5 )
index.add_vectors(np.zeros((5, 5) , dtype=np.floataa ) )
self.assertEqual(index.faiss_index.ntotal , 1_0 )
# single query
_a : Optional[int] = np.zeros(5 , dtype=np.floataa )
_a : Optional[int] = 1
_a : int = index.search(_a )
self.assertRaises(_a , index.search , query.reshape(-1 , 1 ) )
self.assertGreater(scores[0] , 0 )
self.assertEqual(indices[0] , 1 )
# batched queries
_a : Dict = np.eye(5 , dtype=np.floataa )[::-1]
_a : Any = index.search_batch(_a )
self.assertRaises(_a , index.search_batch , queries[0] )
_a : Dict = [scores[0] for scores in total_scores]
_a : Any = [indices[0] for indices in total_indices]
self.assertGreater(np.min(_a ) , 0 )
self.assertListEqual([4, 3, 2, 1, 0] , _a )
def __lowercase ( self ) -> str:
import faiss
_a : int = FaissIndex(string_factory='''Flat''' )
index.add_vectors(np.eye(5 , dtype=np.floataa ) )
self.assertIsInstance(index.faiss_index , faiss.IndexFlat )
_a : Optional[Any] = FaissIndex(string_factory='''LSH''' )
index.add_vectors(np.eye(5 , dtype=np.floataa ) )
self.assertIsInstance(index.faiss_index , faiss.IndexLSH )
with self.assertRaises(_a ):
_a : List[str] = FaissIndex(string_factory='''Flat''' , custom_index=faiss.IndexFlat(5 ) )
def __lowercase ( self ) -> str:
import faiss
_a : int = faiss.IndexFlat(5 )
_a : str = FaissIndex(custom_index=_a )
index.add_vectors(np.eye(5 , dtype=np.floataa ) )
self.assertIsInstance(index.faiss_index , faiss.IndexFlat )
def __lowercase ( self ) -> str:
import faiss
_a : Tuple = FaissIndex(metric_type=faiss.METRIC_INNER_PRODUCT )
index.add_vectors(np.eye(5 , dtype=np.floataa ) )
# Setting delete=False and unlinking manually is not pretty... but it is required on Windows to
# ensure somewhat stable behaviour. If we don't, we get PermissionErrors. This is an age-old issue.
# see https://bugs.python.org/issue14243 and
# https://stackoverflow.com/questions/23212435/permission-denied-to-write-to-my-temporary-file/23212515
with tempfile.NamedTemporaryFile(delete=_a ) as tmp_file:
index.save(tmp_file.name )
_a : Union[str, Any] = FaissIndex.load(tmp_file.name )
os.unlink(tmp_file.name )
_a : List[Any] = np.zeros(5 , dtype=np.floataa )
_a : int = 1
_a : Union[str, Any] = index.search(_a )
self.assertGreater(scores[0] , 0 )
self.assertEqual(indices[0] , 1 )
@require_faiss
def __UpperCAmelCase ( __a : Optional[int] ) -> List[Any]:
"""simple docstring"""
import faiss
_a : Optional[Any] = FaissIndex(metric_type=faiss.METRIC_INNER_PRODUCT )
index.add_vectors(np.eye(5 ,dtype=np.floataa ) )
_a : str = '''index.faiss'''
_a : int = F"""mock://{index_name}"""
index.save(__a ,storage_options=mockfs.storage_options )
_a : int = FaissIndex.load(__a ,storage_options=mockfs.storage_options )
_a : Tuple = np.zeros(5 ,dtype=np.floataa )
_a : str = 1
_a : Union[str, Any] = index.search(__a )
assert scores[0] > 0
assert indices[0] == 1
@require_elasticsearch
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self ) -> List[str]:
from elasticsearch import Elasticsearch
with patch('''elasticsearch.Elasticsearch.search''' ) as mocked_search, patch(
'''elasticsearch.client.IndicesClient.create''' ) as mocked_index_create, patch('''elasticsearch.helpers.streaming_bulk''' ) as mocked_bulk:
_a : int = Elasticsearch()
_a : List[Any] = {'''acknowledged''': True}
_a : List[str] = ElasticSearchIndex(es_client=_a )
mocked_bulk.return_value([(True, None)] * 3 )
index.add_documents(['''foo''', '''bar''', '''foobar'''] )
# single query
_a : str = '''foo'''
_a : List[str] = {'''hits''': {'''hits''': [{'''_score''': 1, '''_id''': 0}]}}
_a : Optional[Any] = index.search(_a )
self.assertEqual(scores[0] , 1 )
self.assertEqual(indices[0] , 0 )
# single query with timeout
_a : Any = '''foo'''
_a : List[str] = {'''hits''': {'''hits''': [{'''_score''': 1, '''_id''': 0}]}}
_a : Optional[int] = index.search(_a , request_timeout=3_0 )
self.assertEqual(scores[0] , 1 )
self.assertEqual(indices[0] , 0 )
# batched queries
_a : str = ['''foo''', '''bar''', '''foobar''']
_a : Optional[int] = {'''hits''': {'''hits''': [{'''_score''': 1, '''_id''': 1}]}}
_a : Union[str, Any] = index.search_batch(_a )
_a : Union[str, Any] = [scores[0] for scores in total_scores]
_a : Optional[int] = [indices[0] for indices in total_indices]
self.assertGreater(np.min(_a ) , 0 )
self.assertListEqual([1, 1, 1] , _a )
# batched queries with timeout
_a : Tuple = ['''foo''', '''bar''', '''foobar''']
_a : str = {'''hits''': {'''hits''': [{'''_score''': 1, '''_id''': 1}]}}
_a : int = index.search_batch(_a , request_timeout=3_0 )
_a : Tuple = [scores[0] for scores in total_scores]
_a : Optional[int] = [indices[0] for indices in total_indices]
self.assertGreater(np.min(_a ) , 0 )
self.assertListEqual([1, 1, 1] , _a )
| 351
|
from math import ceil
def __UpperCAmelCase ( __a : int = 1_001 ) -> int:
"""simple docstring"""
_a : Dict = 1
for i in range(1 ,int(ceil(n / 2.0 ) ) ):
_a : int = 2 * i + 1
_a : str = 2 * i
_a : Any = total + 4 * odd**2 - 6 * even
return total
if __name__ == "__main__":
import sys
if len(sys.argv) == 1:
print(solution())
else:
try:
a__ = int(sys.argv[1])
print(solution(n))
except ValueError:
print('''Invalid entry - please enter a number''')
| 15
| 0
|
"""simple docstring"""
import os
import shutil
import tempfile
from unittest import TestCase
from unittest.mock import patch
import numpy as np
from datasets import Dataset
from transformers.models.realm.configuration_realm import RealmConfig
from transformers.models.realm.retrieval_realm import _REALM_BLOCK_RECORDS_FILENAME, RealmRetriever
from transformers.models.realm.tokenization_realm import VOCAB_FILES_NAMES, RealmTokenizer
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
def __lowercase ( self ) -> Tuple:
_a : Tuple = tempfile.mkdtemp()
_a : List[str] = 5
# Realm tok
_a : Optional[int] = [
'''[UNK]''',
'''[CLS]''',
'''[SEP]''',
'''[PAD]''',
'''[MASK]''',
'''test''',
'''question''',
'''this''',
'''is''',
'''the''',
'''first''',
'''second''',
'''third''',
'''fourth''',
'''fifth''',
'''record''',
'''want''',
'''##want''',
'''##ed''',
'''wa''',
'''un''',
'''runn''',
'''##ing''',
''',''',
'''low''',
'''lowest''',
]
_a : Optional[int] = os.path.join(self.tmpdirname , '''realm_tokenizer''' )
os.makedirs(_a , exist_ok=_a )
_a : str = os.path.join(_a , 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] ) )
_a : List[str] = os.path.join(self.tmpdirname , '''realm_block_records''' )
os.makedirs(_a , exist_ok=_a )
def __lowercase ( self ) -> RealmTokenizer:
return RealmTokenizer.from_pretrained(os.path.join(self.tmpdirname , '''realm_tokenizer''' ) )
def __lowercase ( self ) -> str:
shutil.rmtree(self.tmpdirname )
def __lowercase ( self ) -> List[str]:
_a : int = RealmConfig(num_block_records=self.num_block_records )
return config
def __lowercase ( self ) -> Optional[int]:
_a : Optional[Any] = Dataset.from_dict(
{
'''id''': ['''0''', '''1'''],
'''question''': ['''foo''', '''bar'''],
'''answers''': [['''Foo''', '''Bar'''], ['''Bar''']],
} )
return dataset
def __lowercase ( self ) -> Dict:
_a : Any = np.array(
[
b'''This is the first record''',
b'''This is the second record''',
b'''This is the third record''',
b'''This is the fourth record''',
b'''This is the fifth record''',
b'''This is a longer longer longer record''',
] , dtype=_a , )
return block_records
def __lowercase ( self ) -> Optional[int]:
_a : List[str] = RealmRetriever(
block_records=self.get_dummy_block_records() , tokenizer=self.get_tokenizer() , )
return retriever
def __lowercase ( self ) -> Optional[int]:
_a : Dict = self.get_config()
_a : Tuple = self.get_dummy_retriever()
_a : Dict = retriever.tokenizer
_a : Union[str, Any] = np.array([0, 3] , dtype='''long''' )
_a : Any = tokenizer(['''Test question'''] ).input_ids
_a : int = tokenizer(
['''the fourth'''] , add_special_tokens=_a , return_token_type_ids=_a , return_attention_mask=_a , ).input_ids
_a : str = config.reader_seq_len
_a : Any = retriever(
_a , _a , answer_ids=_a , max_length=_a , return_tensors='''np''' )
self.assertEqual(len(_a ) , 2 )
self.assertEqual(len(_a ) , 2 )
self.assertEqual(len(_a ) , 2 )
self.assertEqual(concat_inputs.input_ids.shape , (2, 1_0) )
self.assertEqual(concat_inputs.attention_mask.shape , (2, 1_0) )
self.assertEqual(concat_inputs.token_type_ids.shape , (2, 1_0) )
self.assertEqual(concat_inputs.special_tokens_mask.shape , (2, 1_0) )
self.assertEqual(
tokenizer.convert_ids_to_tokens(concat_inputs.input_ids[0] ) , ['''[CLS]''', '''test''', '''question''', '''[SEP]''', '''this''', '''is''', '''the''', '''first''', '''record''', '''[SEP]'''] , )
self.assertEqual(
tokenizer.convert_ids_to_tokens(concat_inputs.input_ids[1] ) , ['''[CLS]''', '''test''', '''question''', '''[SEP]''', '''this''', '''is''', '''the''', '''fourth''', '''record''', '''[SEP]'''] , )
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[int] = self.get_config()
_a : Union[str, Any] = self.get_dummy_retriever()
_a : Tuple = retriever.tokenizer
_a : List[str] = np.array([0, 3, 5] , dtype='''long''' )
_a : int = tokenizer(['''Test question'''] ).input_ids
_a : List[str] = tokenizer(
['''the fourth''', '''longer longer'''] , add_special_tokens=_a , return_token_type_ids=_a , return_attention_mask=_a , ).input_ids
_a : Union[str, Any] = config.reader_seq_len
_a : List[Any] = retriever(
_a , _a , answer_ids=_a , max_length=_a , return_tensors='''np''' )
self.assertEqual([False, True, True] , _a )
self.assertEqual([[-1, -1, -1], [6, -1, -1], [6, 7, 8]] , _a )
self.assertEqual([[-1, -1, -1], [7, -1, -1], [7, 8, 9]] , _a )
def __lowercase ( self ) -> int:
_a : Any = self.get_dummy_retriever()
retriever.save_pretrained(os.path.join(self.tmpdirname , '''realm_block_records''' ) )
# Test local path
_a : str = retriever.from_pretrained(os.path.join(self.tmpdirname , '''realm_block_records''' ) )
self.assertEqual(retriever.block_records[0] , b'''This is the first record''' )
# Test mocked remote path
with patch('''transformers.models.realm.retrieval_realm.hf_hub_download''' ) as mock_hf_hub_download:
_a : Tuple = os.path.join(
os.path.join(self.tmpdirname , '''realm_block_records''' ) , _REALM_BLOCK_RECORDS_FILENAME )
_a : str = RealmRetriever.from_pretrained('''google/realm-cc-news-pretrained-openqa''' )
self.assertEqual(retriever.block_records[0] , b'''This is the first record''' )
| 352
|
from typing import Dict, Iterable, Optional, Union
import numpy as np
from ...image_processing_utils import BaseImageProcessor, BatchFeature, get_size_dict
from ...image_transforms import normalize, rescale, resize, to_channel_dimension_format, to_pil_image
from ...image_utils import (
IMAGENET_STANDARD_MEAN,
IMAGENET_STANDARD_STD,
ChannelDimension,
ImageInput,
PILImageResampling,
make_list_of_images,
to_numpy_array,
valid_images,
)
from ...utils import TensorType, is_pytesseract_available, is_vision_available, logging, requires_backends
if is_vision_available():
import PIL
# soft dependency
if is_pytesseract_available():
import pytesseract
a__ = logging.get_logger(__name__)
def __UpperCAmelCase ( __a : Union[str, Any] ,__a : str ,__a : Union[str, Any] ) -> List[str]:
"""simple docstring"""
return [
int(1_000 * (box[0] / width) ),
int(1_000 * (box[1] / height) ),
int(1_000 * (box[2] / width) ),
int(1_000 * (box[3] / height) ),
]
def __UpperCAmelCase ( __a : np.ndarray ,__a : Optional[str] ,__a : Optional[str] ) -> List[Any]:
"""simple docstring"""
_a : str = to_pil_image(__a )
_a , _a : Optional[Any] = pil_image.size
_a : Tuple = pytesseract.image_to_data(__a ,lang=__a ,output_type='''dict''' ,config=__a )
_a , _a , _a , _a , _a : List[str] = data['''text'''], data['''left'''], data['''top'''], data['''width'''], data['''height''']
# filter empty words and corresponding coordinates
_a : Dict = [idx for idx, word in enumerate(__a ) if not word.strip()]
_a : str = [word for idx, word in enumerate(__a ) if idx not in irrelevant_indices]
_a : List[str] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : str = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
_a : Union[str, Any] = [coord for idx, coord in enumerate(__a ) if idx not in irrelevant_indices]
# turn coordinates into (left, top, left+width, top+height) format
_a : int = []
for x, y, w, h in zip(__a ,__a ,__a ,__a ):
_a : List[str] = [x, y, x + w, y + h]
actual_boxes.append(__a )
# finally, normalize the bounding boxes
_a : Dict = []
for box in actual_boxes:
normalized_boxes.append(normalize_box(__a ,__a ,__a ) )
assert len(__a ) == len(__a ), "Not as many words as there are bounding boxes"
return words, normalized_boxes
class UpperCAmelCase_ ( __lowercase ):
"""simple docstring"""
UpperCAmelCase__ : Optional[int] = ["pixel_values"]
def __init__( self , _a = True , _a = None , _a = PILImageResampling.BILINEAR , _a = True , _a = 1 / 2_5_5 , _a = True , _a = None , _a = None , _a = True , _a = None , _a = "" , **_a , ) -> None:
super().__init__(**_a )
_a : List[str] = size if size is not None else {'''height''': 2_2_4, '''width''': 2_2_4}
_a : Union[str, Any] = get_size_dict(_a )
_a : int = do_resize
_a : Optional[int] = size
_a : str = resample
_a : str = do_rescale
_a : Any = rescale_value
_a : Optional[Any] = do_normalize
_a : int = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
_a : List[str] = image_std if image_std is not None else IMAGENET_STANDARD_STD
_a : List[Any] = apply_ocr
_a : Optional[int] = ocr_lang
_a : Tuple = tesseract_config
def __lowercase ( self , _a , _a , _a = PILImageResampling.BILINEAR , _a = None , **_a , ) -> np.ndarray:
_a : Any = get_size_dict(_a )
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()}""" )
_a : Optional[int] = (size['''height'''], size['''width'''])
return resize(_a , size=_a , resample=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a = None , **_a , ) -> np.ndarray:
return rescale(_a , scale=_a , data_format=_a , **_a )
def __lowercase ( self , _a , _a , _a , _a = None , **_a , ) -> np.ndarray:
return normalize(_a , mean=_a , std=_a , data_format=_a , **_a )
def __lowercase ( 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:
_a : Optional[int] = do_resize if do_resize is not None else self.do_resize
_a : Union[str, Any] = size if size is not None else self.size
_a : Any = get_size_dict(_a )
_a : List[str] = resample if resample is not None else self.resample
_a : int = do_rescale if do_rescale is not None else self.do_rescale
_a : Union[str, Any] = rescale_factor if rescale_factor is not None else self.rescale_factor
_a : int = do_normalize if do_normalize is not None else self.do_normalize
_a : str = image_mean if image_mean is not None else self.image_mean
_a : Tuple = image_std if image_std is not None else self.image_std
_a : Any = apply_ocr if apply_ocr is not None else self.apply_ocr
_a : int = ocr_lang if ocr_lang is not None else self.ocr_lang
_a : Optional[int] = tesseract_config if tesseract_config is not None else self.tesseract_config
_a : List[Any] = make_list_of_images(_a )
if not valid_images(_a ):
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_normalize and (image_mean is None or image_std is None):
raise ValueError('''If do_normalize is True, image_mean and image_std must be specified.''' )
# All transformations expect numpy arrays.
_a : Any = [to_numpy_array(_a ) for image in images]
# Tesseract OCR to get words + normalized bounding boxes
if apply_ocr:
requires_backends(self , '''pytesseract''' )
_a : str = []
_a : str = []
for image in images:
_a , _a : Union[str, Any] = apply_tesseract(_a , _a , _a )
words_batch.append(_a )
boxes_batch.append(_a )
if do_resize:
_a : List[str] = [self.resize(image=_a , size=_a , resample=_a ) for image in images]
if do_rescale:
_a : Optional[Any] = [self.rescale(image=_a , scale=_a ) for image in images]
if do_normalize:
_a : List[Any] = [self.normalize(image=_a , mean=_a , std=_a ) for image in images]
_a : List[str] = [to_channel_dimension_format(_a , _a ) for image in images]
_a : List[str] = BatchFeature(data={'''pixel_values''': images} , tensor_type=_a )
if apply_ocr:
_a : Optional[int] = words_batch
_a : List[Any] = boxes_batch
return data
| 15
| 0
|
"""simple docstring"""
import json
import os
import shutil
import tempfile
import unittest
import numpy as np
from transformers import BertTokenizerFast
from transformers.models.bert.tokenization_bert import VOCAB_FILES_NAMES, BertTokenizer
from transformers.testing_utils import require_tokenizers, require_vision
from transformers.utils import IMAGE_PROCESSOR_NAME, is_vision_available
if is_vision_available():
from PIL import Image
from transformers import VisionTextDualEncoderProcessor, ViTImageProcessor
@require_tokenizers
@require_vision
class UpperCAmelCase_ ( unittest.TestCase ):
"""simple docstring"""
def __lowercase ( self ) -> Union[str, Any]:
_a : Optional[Any] = tempfile.mkdtemp()
# fmt: off
_a : Optional[int] = ['''[UNK]''', '''[CLS]''', '''[SEP]''', '''[PAD]''', '''[MASK]''', '''want''', '''##want''', '''##ed''', '''wa''', '''un''', '''runn''', '''##ing''', ''',''', '''low''', '''lowest''']
# fmt: on
_a : Tuple = 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] ) )
_a : Any = {
'''do_resize''': True,
'''size''': {'''height''': 1_8, '''width''': 1_8},
'''do_normalize''': True,
'''image_mean''': [0.5, 0.5, 0.5],
'''image_std''': [0.5, 0.5, 0.5],
}
_a : str = os.path.join(self.tmpdirname , _a )
with open(self.image_processor_file , '''w''' , encoding='''utf-8''' ) as fp:
json.dump(_a , _a )
def __lowercase ( self , **_a ) -> Any:
return BertTokenizer.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self , **_a ) -> str:
return ViTImageProcessor.from_pretrained(self.tmpdirname , **_a )
def __lowercase ( self ) -> List[Any]:
shutil.rmtree(self.tmpdirname )
def __lowercase ( self ) -> Any:
_a : Union[str, Any] = [np.random.randint(2_5_5 , size=(3, 3_0, 4_0_0) , dtype=np.uinta )]
_a : Tuple = [Image.fromarray(np.moveaxis(_a , 0 , -1 ) ) for x in image_inputs]
return image_inputs
def __lowercase ( self ) -> str:
_a : List[str] = self.get_tokenizer()
_a : Tuple = self.get_image_processor()
_a : Union[str, Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
processor.save_pretrained(self.tmpdirname )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(self.tmpdirname )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Dict:
_a : List[str] = VisionTextDualEncoderProcessor(
tokenizer=self.get_tokenizer() , image_processor=self.get_image_processor() )
processor.save_pretrained(self.tmpdirname )
_a : Any = self.get_tokenizer(bos_token='''(BOS)''' , eos_token='''(EOS)''' )
_a : List[Any] = self.get_image_processor(do_normalize=_a , padding_value=1.0 )
_a : Dict = VisionTextDualEncoderProcessor.from_pretrained(
self.tmpdirname , bos_token='''(BOS)''' , eos_token='''(EOS)''' , do_normalize=_a , padding_value=1.0 )
self.assertEqual(processor.tokenizer.get_vocab() , tokenizer_add_kwargs.get_vocab() )
self.assertIsInstance(processor.tokenizer , (BertTokenizer, BertTokenizerFast) )
self.assertEqual(processor.image_processor.to_json_string() , image_processor_add_kwargs.to_json_string() )
self.assertIsInstance(processor.image_processor , _a )
def __lowercase ( self ) -> Any:
_a : Dict = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : int = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[str] = self.prepare_image_inputs()
_a : List[Any] = image_processor(_a , return_tensors='''np''' )
_a : Dict = processor(images=_a , return_tensors='''np''' )
for key in input_feat_extract.keys():
self.assertAlmostEqual(input_feat_extract[key].sum() , input_processor[key].sum() , delta=1e-2 )
def __lowercase ( self ) -> List[str]:
_a : Union[str, Any] = self.get_image_processor()
_a : Dict = self.get_tokenizer()
_a : Optional[Any] = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Tuple = '''lower newer'''
_a : int = processor(text=_a )
_a : str = tokenizer(_a )
for key in encoded_tok.keys():
self.assertListEqual(encoded_tok[key] , encoded_processor[key] )
def __lowercase ( self ) -> List[Any]:
_a : Any = self.get_image_processor()
_a : str = self.get_tokenizer()
_a : Tuple = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : List[Any] = '''lower newer'''
_a : Union[str, Any] = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , ['''input_ids''', '''token_type_ids''', '''attention_mask''', '''pixel_values'''] )
# test if it raises when no input is passed
with self.assertRaises(_a ):
processor()
def __lowercase ( self ) -> Optional[int]:
_a : Union[str, Any] = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : Any = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Any = [[1, 4, 5, 8, 1, 0, 8], [3, 4, 3, 1, 1, 8, 9]]
_a : int = processor.batch_decode(_a )
_a : int = tokenizer.batch_decode(_a )
self.assertListEqual(_a , _a )
def __lowercase ( self ) -> List[Any]:
_a : Tuple = self.get_image_processor()
_a : List[str] = self.get_tokenizer()
_a : str = VisionTextDualEncoderProcessor(tokenizer=_a , image_processor=_a )
_a : Optional[int] = '''lower newer'''
_a : Dict = self.prepare_image_inputs()
_a : Any = processor(text=_a , images=_a )
self.assertListEqual(list(inputs.keys() ) , processor.model_input_names )
| 353
|
# Copyright 2021 The HuggingFace Team. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from argparse import ArgumentParser
from accelerate.commands.config import get_config_parser
from accelerate.commands.env import env_command_parser
from accelerate.commands.launch import launch_command_parser
from accelerate.commands.test import test_command_parser
from accelerate.commands.tpu import tpu_command_parser
def __UpperCAmelCase ( ) -> Optional[Any]:
"""simple docstring"""
_a : int = ArgumentParser('''Accelerate CLI tool''' ,usage='''accelerate <command> [<args>]''' ,allow_abbrev=__a )
_a : Optional[int] = parser.add_subparsers(help='''accelerate command helpers''' )
# Register commands
get_config_parser(subparsers=__a )
env_command_parser(subparsers=__a )
launch_command_parser(subparsers=__a )
tpu_command_parser(subparsers=__a )
test_command_parser(subparsers=__a )
# Let's go
_a : Dict = parser.parse_args()
if not hasattr(__a ,'''func''' ):
parser.print_help()
exit(1 )
# Run
args.func(__a )
if __name__ == "__main__":
main()
| 15
| 0
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.