SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Black preview (#17217)

Browse Source 
* Black preview

* Fixup too!

* Fix check copies

* Use the same version as the CI

* Bump black
This commit is contained in:
Sylvain Gugger
2022-05-12 16:25:55 -04:00
committed by GitHub
parent 9bd67ac7bb
commit afe5d42d8d
578 changed files with 8274 additions and 3296 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
examples/pytorch/audio-classification/run_audio_classification.py
View File

@@ -86,8 +86,9 @@ class DataTrainingArguments:
eval_split_name: str = field(
default="validation",
metadata={
"help": "The name of the training data set split to use (via the datasets library). Defaults to "
"'validation'"
"help": (
"The name of the training data set split to use (via the datasets library). Defaults to 'validation'"
)
},
)
audio_column_name: str = field(
@@ -100,15 +101,19 @@ class DataTrainingArguments:
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_length_seconds: float = field(
@@ -149,8 +154,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
freeze_feature_extractor: Optional[bool] = field(

24
examples/pytorch/contrastive-image-text/run_clip.py
View File

@@ -89,8 +89,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
freeze_vision_model: bool = field(
@@ -132,22 +134,28 @@ class DataTrainingArguments:
max_seq_length: Optional[int] = field(
default=128,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
overwrite_cache: bool = field(

18
examples/pytorch/image-classification/run_image_classification.py
View File

@@ -93,15 +93,19 @@ class DataTrainingArguments:
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
@@ -140,8 +144,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)

17
examples/pytorch/image-classification/run_image_classification_no_trainer.py
View File

@@ -62,7 +62,10 @@ def parse_args():
"--dataset_name",
type=str,
default="cifar10",
help="The name of the Dataset (from the HuggingFace hub) to train on (could be your own, possibly private, dataset).",
help=(
"The name of the Dataset (from the HuggingFace hub) to train on (could be your own, possibly private,"
" dataset)."
),
)
parser.add_argument("--train_dir", type=str, default=None, help="A folder containing the training data.")
parser.add_argument("--validation_dir", type=str, default=None, help="A folder containing the validation data.")
@@ -70,15 +73,19 @@ def parse_args():
"--max_train_samples",
type=int,
default=None,
help="For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set.",
help=(
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
),
)
parser.add_argument(
"--max_eval_samples",
type=int,
default=None,
help="For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set.",
help=(
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
),
)
parser.add_argument(
"--train_val_split",

29
examples/pytorch/image-pretraining/run_mae.py
View File

@@ -74,15 +74,19 @@ class DataTrainingArguments:
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
@@ -104,8 +108,9 @@ class ModelArguments:
model_name_or_path: str = field(
default=None,
metadata={
"help": "The model checkpoint for weights initialization."
"Don't set if you want to train a model from scratch."
"help": (
"The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."
)
},
)
config_name: Optional[str] = field(
@@ -114,8 +119,10 @@ class ModelArguments:
config_overrides: Optional[str] = field(
default=None,
metadata={
"help": "Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
"help": (
"Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
)
},
)
cache_dir: Optional[str] = field(
@@ -129,8 +136,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
mask_ratio: float = field(

40
examples/pytorch/image-pretraining/run_mim.py
View File

@@ -87,15 +87,19 @@ class DataTrainingArguments:
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
@@ -117,9 +121,11 @@ class ModelArguments:
model_name_or_path: str = field(
default=None,
metadata={
"help": "The model checkpoint for weights initialization. Can be a local path to a pytorch_model.bin or a "
"checkpoint identifier on the hub. "
"Don't set if you want to train a model from scratch."
"help": (
"The model checkpoint for weights initialization. Can be a local path to a pytorch_model.bin or a "
"checkpoint identifier on the hub. "
"Don't set if you want to train a model from scratch."
)
},
)
model_type: Optional[str] = field(
@@ -132,8 +138,10 @@ class ModelArguments:
config_overrides: Optional[str] = field(
default=None,
metadata={
"help": "Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
"help": (
"Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
)
},
)
cache_dir: Optional[str] = field(
@@ -148,20 +156,26 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
image_size: Optional[int] = field(
default=None,
metadata={
"help": "The size (resolution) of each image. If not specified, will use `image_size` of the configuration."
"help": (
"The size (resolution) of each image. If not specified, will use `image_size` of the configuration."
)
},
)
patch_size: Optional[int] = field(
default=None,
metadata={
"help": "The size (resolution) of each patch. If not specified, will use `patch_size` of the configuration."
"help": (
"The size (resolution) of each patch. If not specified, will use `patch_size` of the configuration."
)
},
)
encoder_stride: Optional[int] = field(

40
examples/pytorch/language-modeling/run_clm.py
View File

@@ -73,8 +73,9 @@ class ModelArguments:
model_name_or_path: Optional[str] = field(
default=None,
metadata={
"help": "The model checkpoint for weights initialization."
"Don't set if you want to train a model from scratch."
"help": (
"The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."
)
},
)
model_type: Optional[str] = field(
@@ -84,8 +85,10 @@ class ModelArguments:
config_overrides: Optional[str] = field(
default=None,
metadata={
"help": "Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
"help": (
"Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
)
},
)
config_name: Optional[str] = field(
@@ -109,8 +112,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -141,24 +146,30 @@ class DataTrainingArguments:
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
block_size: Optional[int] = field(
default=None,
metadata={
"help": "Optional input sequence length after tokenization. "
"The training dataset will be truncated in block of this size for training. "
"Default to the model max input length for single sentence inputs (take into account special tokens)."
"help": (
"Optional input sequence length after tokenization. "
"The training dataset will be truncated in block of this size for training. "
"Default to the model max input length for single sentence inputs (take into account special tokens)."
)
},
)
overwrite_cache: bool = field(
@@ -390,7 +401,8 @@ def main():
# clm input could be much much longer than block_size
if "Token indices sequence length is longer than the" in cl.out:
tok_logger.warning(
"^^^^^^^^^^^^^^^^ Please ignore the warning above - this long input will be chunked into smaller bits before being passed to the model."
"^^^^^^^^^^^^^^^^ Please ignore the warning above - this long input will be chunked into smaller bits"
" before being passed to the model."
)
return output

6
examples/pytorch/language-modeling/run_clm_no_trainer.py
View File

@@ -168,7 +168,11 @@ def parse_args():
"--block_size",
type=int,
default=None,
help="Optional input sequence length after tokenization. The training dataset will be truncated in block of this size for training. Default to the model max input length for single sentence inputs (take into account special tokens).",
help=(
"Optional input sequence length after tokenization. The training dataset will be truncated in block of"
" this size for training. Default to the model max input length for single sentence inputs (take into"
" account special tokens)."
),
)
parser.add_argument(
"--preprocessing_num_workers",

41
examples/pytorch/language-modeling/run_mlm.py
View File

@@ -70,8 +70,9 @@ class ModelArguments:
model_name_or_path: Optional[str] = field(
default=None,
metadata={
"help": "The model checkpoint for weights initialization."
"Don't set if you want to train a model from scratch."
"help": (
"The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."
)
},
)
model_type: Optional[str] = field(
@@ -81,8 +82,10 @@ class ModelArguments:
config_overrides: Optional[str] = field(
default=None,
metadata={
"help": "Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
"help": (
"Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
)
},
)
config_name: Optional[str] = field(
@@ -106,8 +109,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -147,8 +152,10 @@ class DataTrainingArguments:
max_seq_length: Optional[int] = field(
default=None,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated."
)
},
)
preprocessing_num_workers: Optional[int] = field(
@@ -165,22 +172,28 @@ class DataTrainingArguments:
pad_to_max_length: bool = field(
default=False,
metadata={
"help": "Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
"help": (
"Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)

4
examples/pytorch/language-modeling/run_mlm_no_trainer.py
View File

@@ -171,7 +171,9 @@ def parse_args():
"--max_seq_length",
type=int,
default=None,
help="The maximum total input sequence length after tokenization. Sequences longer than this will be truncated.",
help=(
"The maximum total input sequence length after tokenization. Sequences longer than this will be truncated."
),
)
parser.add_argument(
"--line_by_line",

47
examples/pytorch/language-modeling/run_plm.py
View File

@@ -63,8 +63,9 @@ class ModelArguments:
model_name_or_path: Optional[str] = field(
default=None,
metadata={
"help": "The model checkpoint for weights initialization."
"Don't set if you want to train a model from scratch."
"help": (
"The model checkpoint for weights initialization.Don't set if you want to train a model from scratch."
)
},
)
config_name: Optional[str] = field(
@@ -73,8 +74,10 @@ class ModelArguments:
config_overrides: Optional[str] = field(
default=None,
metadata={
"help": "Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
"help": (
"Override some existing default config settings when a model is trained from scratch. Example: "
"n_embd=10,resid_pdrop=0.2,scale_attn_weights=false,summary_type=cls_index"
)
},
)
tokenizer_name: Optional[str] = field(
@@ -95,8 +98,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -136,8 +141,10 @@ class DataTrainingArguments:
max_seq_length: int = field(
default=512,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated."
)
},
)
preprocessing_num_workers: Optional[int] = field(
@@ -147,8 +154,10 @@ class DataTrainingArguments:
plm_probability: float = field(
default=1 / 6,
metadata={
"help": "Ratio of length of a span of masked tokens to surrounding context length for "
"permutation language modeling."
"help": (
"Ratio of length of a span of masked tokens to surrounding context length for "
"permutation language modeling."
)
},
)
max_span_length: int = field(
@@ -161,22 +170,28 @@ class DataTrainingArguments:
pad_to_max_length: bool = field(
default=False,
metadata={
"help": "Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
"help": (
"Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)

32
examples/pytorch/multiple-choice/run_swag.py
View File

@@ -82,8 +82,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -109,30 +111,38 @@ class DataTrainingArguments:
max_seq_length: Optional[int] = field(
default=None,
metadata={
"help": "The maximum total input sequence length after tokenization. If passed, sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. If passed, sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": "Whether to pad all samples to the maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
"help": (
"Whether to pad all samples to the maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)

57
examples/pytorch/question-answering/run_qa.py
View File

@@ -81,8 +81,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -118,37 +120,46 @@ class DataTrainingArguments:
max_seq_length: int = field(
default=384,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
pad_to_max_length: bool = field(
default=True,
metadata={
"help": "Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch (which can "
"be faster on GPU but will be slower on TPU)."
"help": (
"Whether to pad all samples to `max_seq_length`. If False, will pad the samples dynamically when"
" batching to the maximum length in the batch (which can be faster on GPU but will be slower on TPU)."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
version_2_with_negative: bool = field(
@@ -157,9 +168,11 @@ class DataTrainingArguments:
null_score_diff_threshold: float = field(
default=0.0,
metadata={
"help": "The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
"help": (
"The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
)
},
)
doc_stride: int = field(
@@ -173,8 +186,10 @@ class DataTrainingArguments:
max_answer_length: int = field(
default=30,
metadata={
"help": "The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
"help": (
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
)
},
)
@@ -319,9 +334,9 @@ def main():
# Tokenizer check: this script requires a fast tokenizer.
if not isinstance(tokenizer, PreTrainedTokenizerFast):
raise ValueError(
"This example script only works for models that have a fast tokenizer. Checkout the big table of models "
"at https://huggingface.co/transformers/index.html#supported-frameworks to find the model types that meet this "
"requirement"
"This example script only works for models that have a fast tokenizer. Checkout the big table of models at"
" https://huggingface.co/transformers/index.html#supported-frameworks to find the model types that meet"
" this requirement"
)
# Preprocessing the datasets.

51
examples/pytorch/question-answering/run_qa_beam_search.py
View File

@@ -80,8 +80,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -117,37 +119,46 @@ class DataTrainingArguments:
max_seq_length: int = field(
default=384,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
pad_to_max_length: bool = field(
default=True,
metadata={
"help": "Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch (which can "
"be faster on GPU but will be slower on TPU)."
"help": (
"Whether to pad all samples to `max_seq_length`. If False, will pad the samples dynamically when"
" batching to the maximum length in the batch (which can be faster on GPU but will be slower on TPU)."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
version_2_with_negative: bool = field(
@@ -156,9 +167,11 @@ class DataTrainingArguments:
null_score_diff_threshold: float = field(
default=0.0,
metadata={
"help": "The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
"help": (
"The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
)
},
)
doc_stride: int = field(
@@ -172,8 +185,10 @@ class DataTrainingArguments:
max_answer_length: int = field(
default=30,
metadata={
"help": "The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
"help": (
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
)
},
)

32
examples/pytorch/question-answering/run_qa_beam_search_no_trainer.py
View File

@@ -116,8 +116,10 @@ def parse_args():
"--max_seq_length",
type=int,
default=384,
help="The maximum total input sequence length after tokenization. Sequences longer than this will be truncated,"
" sequences shorter will be padded if `--pad_to_max_lengh` is passed.",
help=(
"The maximum total input sequence length after tokenization. Sequences longer than this will be truncated,"
" sequences shorter will be padded if `--pad_to_max_lengh` is passed."
),
)
parser.add_argument(
"--pad_to_max_length",
@@ -190,9 +192,11 @@ def parse_args():
"--null_score_diff_threshold",
type=float,
default=0.0,
help="The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`.",
help=(
"The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
),
)
parser.add_argument(
"--version_2_with_negative",
@@ -203,22 +207,28 @@ def parse_args():
"--max_answer_length",
type=int,
default=30,
help="The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another.",
help=(
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
),
)
parser.add_argument(
"--max_train_samples",
type=int,
default=None,
help="For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set.",
help=(
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
),
)
parser.add_argument(
"--max_eval_samples",
type=int,
default=None,
help="For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set.",
help=(
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
),
)
parser.add_argument(
"--overwrite_cache", type=bool, default=False, help="Overwrite the cached training and evaluation sets"

32
examples/pytorch/question-answering/run_qa_no_trainer.py
View File

@@ -121,8 +121,10 @@ def parse_args():
"--max_seq_length",
type=int,
default=384,
help="The maximum total input sequence length after tokenization. Sequences longer than this will be truncated,"
" sequences shorter will be padded if `--pad_to_max_lengh` is passed.",
help=(
"The maximum total input sequence length after tokenization. Sequences longer than this will be truncated,"
" sequences shorter will be padded if `--pad_to_max_lengh` is passed."
),
)
parser.add_argument(
"--pad_to_max_length",
@@ -212,9 +214,11 @@ def parse_args():
"--null_score_diff_threshold",
type=float,
default=0.0,
help="The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`.",
help=(
"The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
),
)
parser.add_argument(
"--version_2_with_negative",
@@ -225,22 +229,28 @@ def parse_args():
"--max_answer_length",
type=int,
default=30,
help="The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another.",
help=(
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
),
)
parser.add_argument(
"--max_train_samples",
type=int,
default=None,
help="For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set.",
help=(
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
),
)
parser.add_argument(
"--max_eval_samples",
type=int,
default=None,
help="For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set.",
help=(
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
),
)
parser.add_argument(
"--overwrite_cache", type=bool, default=False, help="Overwrite the cached training and evaluation sets"

67
examples/pytorch/question-answering/run_seq2seq_qa.py
View File

@@ -81,8 +81,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -130,53 +132,66 @@ class DataTrainingArguments:
max_seq_length: int = field(
default=384,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
max_answer_length: int = field(
default=30,
metadata={
"help": "The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
"help": (
"The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
)
},
)
val_max_answer_length: Optional[int] = field(
default=None,
metadata={
"help": "The maximum total sequence length for validation target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded. Will default to `max_answer_length`."
"This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
"during ``evaluate`` and ``predict``."
"help": (
"The maximum total sequence length for validation target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded. Will default to `max_answer_length`."
"This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
"during ``evaluate`` and ``predict``."
)
},
)
pad_to_max_length: bool = field(
default=True,
metadata={
"help": "Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch (which can "
"be faster on GPU but will be slower on TPU)."
"help": (
"Whether to pad all samples to `max_seq_length`. If False, will pad the samples dynamically when"
" batching to the maximum length in the batch (which can be faster on GPU but will be slower on TPU)."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
version_2_with_negative: bool = field(
@@ -185,9 +200,11 @@ class DataTrainingArguments:
null_score_diff_threshold: float = field(
default=0.0,
metadata={
"help": "The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
"help": (
"The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
)
},
)
doc_stride: int = field(
@@ -201,8 +218,10 @@ class DataTrainingArguments:
num_beams: Optional[int] = field(
default=None,
metadata={
"help": "Number of beams to use for evaluation. This argument will be passed to ``model.generate``, "
"which is used during ``evaluate`` and ``predict``."
"help": (
"Number of beams to use for evaluation. This argument will be passed to ``model.generate``, "
"which is used during ``evaluate`` and ``predict``."
)
},
)
ignore_pad_token_for_loss: bool = field(

18
examples/pytorch/semantic-segmentation/run_semantic_segmentation.py
View File

@@ -194,15 +194,19 @@ class DataTrainingArguments:
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
reduce_labels: Optional[bool] = field(
@@ -241,8 +245,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)

12
examples/pytorch/speech-pretraining/run_wav2vec2_pretraining_no_trainer.py
View File

@@ -219,7 +219,10 @@ def parse_args():
"--pad_to_multiple_of",
type=int,
default=None,
help="If set will pad the sequence to a multiple of the provided value. This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).",
help=(
"If set will pad the sequence to a multiple of the provided value. This is especially useful to enable the"
" use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta)."
),
)
parser.add_argument(
"--adam_beta1",
@@ -440,7 +443,7 @@ def main():
# only normalized-inputs-training is supported
if not feature_extractor.do_normalize:
raise ValueError(
"Training is only supported for normalized inputs. " "Make sure ``feature_extractor.do_normalize == True``"
"Training is only supported for normalized inputs. Make sure ``feature_extractor.do_normalize == True``"
)
# set max & min audio length in number of samples
@@ -496,7 +499,8 @@ def main():
# apply_spec_augment has to be True, mask_feature_prob has to be 0.0
if not config.do_stable_layer_norm or config.feat_extract_norm != "layer":
raise ValueError(
"PreTraining is only supported for ``config.do_stable_layer_norm=True`` and ``config.feat_extract_norm='layer'"
"PreTraining is only supported for ``config.do_stable_layer_norm=True`` and"
" ``config.feat_extract_norm='layer'"
)
# initialize random model
@@ -615,7 +619,7 @@ def main():
lr_scheduler.step()
elif accelerator.is_local_main_process:
progress_bar.write(
"Gradients have overflown - skipping update step... " f"Updating gradient scale to {scale}..."
f"Gradients have overflown - skipping update step... Updating gradient scale to {scale}..."
)
# update gumbel temperature

75
examples/pytorch/speech-recognition/run_speech_recognition_ctc.py
View File

@@ -101,9 +101,11 @@ class ModelArguments:
mask_time_prob: float = field(
default=0.05,
metadata={
"help": "Probability of each feature vector along the time axis to be chosen as the start of the vector"
"span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"
"vectors will be masked along the time axis."
"help": (
"Probability of each feature vector along the time axis to be chosen as the start of the vector"
"span to be masked. Approximately ``mask_time_prob * sequence_length // mask_time_length`` feature"
"vectors will be masked along the time axis."
)
},
)
mask_time_length: int = field(
@@ -113,8 +115,11 @@ class ModelArguments:
mask_feature_prob: float = field(
default=0.0,
metadata={
"help": "Probability of each feature vector along the feature axis to be chosen as the start of the vector"
"span to be masked. Approximately ``mask_feature_prob * sequence_length // mask_feature_length`` feature bins will be masked along the time axis."
"help": (
"Probability of each feature vector along the feature axis to be chosen as the start of the vectorspan"
" to be masked. Approximately ``mask_feature_prob * sequence_length // mask_feature_length`` feature"
" bins will be masked along the time axis."
)
},
)
mask_feature_length: int = field(
@@ -146,8 +151,10 @@ class DataTrainingArguments:
train_split_name: str = field(
default="train+validation",
metadata={
"help": "The name of the training data set split to use (via the datasets library). Defaults to "
"'train+validation'"
"help": (
"The name of the training data set split to use (via the datasets library). Defaults to "
"'train+validation'"
)
},
)
eval_split_name: str = field(
@@ -174,15 +181,19 @@ class DataTrainingArguments:
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of validation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of validation examples to this "
"value if set."
)
},
)
chars_to_ignore: Optional[List[str]] = list_field(
@@ -196,7 +207,10 @@ class DataTrainingArguments:
max_duration_in_seconds: float = field(
default=20.0,
metadata={
"help": "Filter audio files that are longer than `max_duration_in_seconds` seconds to 'max_duration_in_seconds`"
"help": (
"Filter audio files that are longer than `max_duration_in_seconds` seconds to"
" 'max_duration_in_seconds`"
)
},
)
min_duration_in_seconds: float = field(
@@ -205,17 +219,21 @@ class DataTrainingArguments:
preprocessing_only: bool = field(
default=False,
metadata={
"help": "Whether to only do data preprocessing and skip training. "
"This is especially useful when data preprocessing errors out in distributed training due to timeout. "
"In this case, one should run the preprocessing in a non-distributed setup with `preprocessing_only=True` "
"so that the cached datasets can consequently be loaded in distributed training"
"help": (
"Whether to only do data preprocessing and skip training. This is especially useful when data"
" preprocessing errors out in distributed training due to timeout. In this case, one should run the"
" preprocessing in a non-distributed setup with `preprocessing_only=True` so that the cached datasets"
" can consequently be loaded in distributed training"
)
},
)
use_auth_token: bool = field(
default=False,
metadata={
"help": "If :obj:`True`, will use the token generated when running"
":obj:`transformers-cli login` as HTTP bearer authorization for remote files."
"help": (
"If :obj:`True`, will use the token generated when running"
":obj:`transformers-cli login` as HTTP bearer authorization for remote files."
)
},
)
unk_token: str = field(
@@ -233,10 +251,12 @@ class DataTrainingArguments:
phoneme_language: Optional[str] = field(
default=None,
metadata={
"help": "The target language that should be used be"
" passed to the tokenizer for tokenization. Note that"
" this is only relevant if the model classifies the"
" input audio to a sequence of phoneme sequences."
"help": (
"The target language that should be used be"
" passed to the tokenizer for tokenization. Note that"
" this is only relevant if the model classifies the"
" input audio to a sequence of phoneme sequences."
)
},
)
@@ -405,9 +425,9 @@ def main():
if data_args.audio_column_name not in raw_datasets["train"].column_names:
raise ValueError(
f"--audio_column_name '{data_args.audio_column_name}' not found in dataset '{data_args.dataset_name}'. "
"Make sure to set `--audio_column_name` to the correct audio column - one of "
f"{', '.join(raw_datasets['train'].column_names)}."
f"--audio_column_name '{data_args.audio_column_name}' not found in dataset '{data_args.dataset_name}'."
" Make sure to set `--audio_column_name` to the correct audio column - one of"
f" {', '.join(raw_datasets['train'].column_names)}."
)
if data_args.text_column_name not in raw_datasets["train"].column_names:
@@ -720,7 +740,10 @@ def main():
"finetuned_from": model_args.model_name_or_path,
"tasks": "speech-recognition",
"tags": ["automatic-speech-recognition", data_args.dataset_name],
"dataset_args": f"Config: {config_name}, Training split: {data_args.train_split_name}, Eval split: {data_args.eval_split_name}",
"dataset_args": (
f"Config: {config_name}, Training split: {data_args.train_split_name}, Eval split:"
f" {data_args.eval_split_name}"
),
"dataset": f"{data_args.dataset_name.upper()} - {config_name.upper()}",
}
if "common_voice" in data_args.dataset_name:

33
examples/pytorch/speech-recognition/run_speech_recognition_seq2seq.py
View File

@@ -87,8 +87,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
freeze_feature_encoder: bool = field(
@@ -122,15 +124,19 @@ class DataTrainingArguments:
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
audio_column_name: str = field(
@@ -144,7 +150,10 @@ class DataTrainingArguments:
max_duration_in_seconds: float = field(
default=20.0,
metadata={
"help": "Truncate audio files that are longer than `max_duration_in_seconds` seconds to 'max_duration_in_seconds`"
"help": (
"Truncate audio files that are longer than `max_duration_in_seconds` seconds to"
" 'max_duration_in_seconds`"
)
},
)
min_duration_in_seconds: float = field(
@@ -153,10 +162,12 @@ class DataTrainingArguments:
preprocessing_only: bool = field(
default=False,
metadata={
"help": "Whether to only do data preprocessing and skip training. "
"This is especially useful when data preprocessing errors out in distributed training due to timeout. "
"In this case, one should run the preprocessing in a non-distributed setup with `preprocessing_only=True` "
"so that the cached datasets can consequently be loaded in distributed training"
"help": (
"Whether to only do data preprocessing and skip training. This is especially useful when data"
" preprocessing errors out in distributed training due to timeout. In this case, one should run the"
" preprocessing in a non-distributed setup with `preprocessing_only=True` so that the cached datasets"
" can consequently be loaded in distributed training"
)
},
)
train_split_name: str = field(

92
examples/pytorch/summarization/run_summarization.py
View File

@@ -101,15 +101,19 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
resize_position_embeddings: Optional[bool] = field(
default=None,
metadata={
"help": "Whether to automatically resize the position embeddings if `max_source_length` exceeds "
"the model's position embeddings."
"help": (
"Whether to automatically resize the position embeddings if `max_source_length` exceeds "
"the model's position embeddings."
)
},
)
@@ -142,14 +146,15 @@ class DataTrainingArguments:
validation_file: Optional[str] = field(
default=None,
metadata={
"help": "An optional input evaluation data file to evaluate the metrics (rouge) on "
"(a jsonlines or csv file)."
"help": (
"An optional input evaluation data file to evaluate the metrics (rouge) on (a jsonlines or csv file)."
)
},
)
test_file: Optional[str] = field(
default=None,
metadata={
"help": "An optional input test data file to evaluate the metrics (rouge) on " "(a jsonlines or csv file)."
"help": "An optional input test data file to evaluate the metrics (rouge) on (a jsonlines or csv file)."
},
)
overwrite_cache: bool = field(
@@ -162,60 +167,76 @@ class DataTrainingArguments:
max_source_length: Optional[int] = field(
default=1024,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
max_target_length: Optional[int] = field(
default=128,
metadata={
"help": "The maximum total sequence length for target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total sequence length for target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
val_max_target_length: Optional[int] = field(
default=None,
metadata={
"help": "The maximum total sequence length for validation target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded. Will default to `max_target_length`."
"This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
"during ``evaluate`` and ``predict``."
"help": (
"The maximum total sequence length for validation target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded. Will default to `max_target_length`."
"This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
"during ``evaluate`` and ``predict``."
)
},
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": "Whether to pad all samples to model maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
"help": (
"Whether to pad all samples to model maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
num_beams: Optional[int] = field(
default=None,
metadata={
"help": "Number of beams to use for evaluation. This argument will be passed to ``model.generate``, "
"which is used during ``evaluate`` and ``predict``."
"help": (
"Number of beams to use for evaluation. This argument will be passed to ``model.generate``, "
"which is used during ``evaluate`` and ``predict``."
)
},
)
ignore_pad_token_for_loss: bool = field(
@@ -231,9 +252,11 @@ class DataTrainingArguments:
forced_bos_token: Optional[str] = field(
default=None,
metadata={
"help": "The token to force as the first generated token after the decoder_start_token_id."
"Useful for multilingual models like mBART where the first generated token"
"needs to be the target language token (Usually it is the target language token)"
"help": (
"The token to force as the first generated token after the decoder_start_token_id."
"Useful for multilingual models like mBART where the first generated token"
"needs to be the target language token (Usually it is the target language token)"
)
},
)
@@ -410,17 +433,18 @@ def main():
):
if model_args.resize_position_embeddings is None:
logger.warning(
f"Increasing the model's number of position embedding vectors from {model.config.max_position_embeddings} "
f"to {data_args.max_source_length}."
"Increasing the model's number of position embedding vectors from"
f" {model.config.max_position_embeddings} to {data_args.max_source_length}."
)
model.resize_position_embeddings(data_args.max_source_length)
elif model_args.resize_position_embeddings:
model.resize_position_embeddings(data_args.max_source_length)
else:
raise ValueError(
f"`--max_source_length` is set to {data_args.max_source_length}, but the model only has {model.config.max_position_embeddings}"
f" position encodings. Consider either reducing `--max_source_length` to {model.config.max_position_embeddings} or to automatically "
"resize the model's position encodings by passing `--resize_position_embeddings`."
f"`--max_source_length` is set to {data_args.max_source_length}, but the model only has"
f" {model.config.max_position_embeddings} position encodings. Consider either reducing"
f" `--max_source_length` to {model.config.max_position_embeddings} or to automatically resize the"
" model's position encodings by passing `--resize_position_embeddings`."
)
prefix = data_args.source_prefix if data_args.source_prefix is not None else ""

34
examples/pytorch/summarization/run_summarization_no_trainer.py
View File

@@ -111,20 +111,22 @@ def parse_args():
"--ignore_pad_token_for_loss",
type=bool,
default=True,
help="Whether to ignore the tokens corresponding to " "padded labels in the loss computation or not.",
help="Whether to ignore the tokens corresponding to padded labels in the loss computation or not.",
)
parser.add_argument(
"--max_source_length",
type=int,
default=1024,
help="The maximum total input sequence length after "
"tokenization.Sequences longer than this will be truncated, sequences shorter will be padded.",
help=(
"The maximum total input sequence length after "
"tokenization.Sequences longer than this will be truncated, sequences shorter will be padded."
),
)
parser.add_argument(
"--source_prefix",
type=str,
default=None,
help="A prefix to add before every source text " "(useful for T5 models).",
help="A prefix to add before every source text (useful for T5 models).",
)
parser.add_argument(
"--preprocessing_num_workers",
@@ -139,18 +141,22 @@ def parse_args():
"--max_target_length",
type=int,
default=128,
help="The maximum total sequence length for target text after "
"tokenization. Sequences longer than this will be truncated, sequences shorter will be padded."
"during ``evaluate`` and ``predict``.",
help=(
"The maximum total sequence length for target text after "
"tokenization. Sequences longer than this will be truncated, sequences shorter will be padded."
"during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--val_max_target_length",
type=int,
default=None,
help="The maximum total sequence length for validation "
"target text after tokenization.Sequences longer than this will be truncated, sequences shorter will be "
"padded. Will default to `max_target_length`.This argument is also used to override the ``max_length`` "
"param of ``model.generate``, which is used during ``evaluate`` and ``predict``.",
help=(
"The maximum total sequence length for validation "
"target text after tokenization.Sequences longer than this will be truncated, sequences shorter will be "
"padded. Will default to `max_target_length`.This argument is also used to override the ``max_length`` "
"param of ``model.generate``, which is used during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--max_length",
@@ -165,8 +171,10 @@ def parse_args():
"--num_beams",
type=int,
default=None,
help="Number of beams to use for evaluation. This argument will be "
"passed to ``model.generate``, which is used during ``evaluate`` and ``predict``.",
help=(
"Number of beams to use for evaluation. This argument will be "
"passed to ``model.generate``, which is used during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--pad_to_max_length",

36
examples/pytorch/text-classification/run_glue.py
View File

@@ -89,8 +89,10 @@ class DataTrainingArguments:
max_seq_length: int = field(
default=128,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
overwrite_cache: bool = field(
@@ -99,29 +101,37 @@ class DataTrainingArguments:
pad_to_max_length: bool = field(
default=True,
metadata={
"help": "Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
"help": (
"Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
train_file: Optional[str] = field(
@@ -180,8 +190,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)

36
examples/pytorch/text-classification/run_xnli.py
View File

@@ -67,8 +67,10 @@ class DataTrainingArguments:
max_seq_length: Optional[int] = field(
default=128,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
overwrite_cache: bool = field(
@@ -77,29 +79,37 @@ class DataTrainingArguments:
pad_to_max_length: bool = field(
default=True,
metadata={
"help": "Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
"help": (
"Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
server_ip: Optional[str] = field(default=None, metadata={"help": "For distant debugging."})
@@ -146,8 +156,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)

54
examples/pytorch/token-classification/run_ner.py
View File

@@ -81,8 +81,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -127,44 +129,56 @@ class DataTrainingArguments:
max_seq_length: int = field(
default=None,
metadata={
"help": "The maximum total input sequence length after tokenization. If set, sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. If set, sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": "Whether to pad all samples to model maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
"help": (
"Whether to pad all samples to model maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
label_all_tokens: bool = field(
default=False,
metadata={
"help": "Whether to put the label for one word on all tokens of generated by that word or just on the "
"one (in which case the other tokens will have a padding index)."
"help": (
"Whether to put the label for one word on all tokens of generated by that word or just on the "
"one (in which case the other tokens will have a padding index)."
)
},
)
return_entity_level_metrics: bool = field(
@@ -355,9 +369,9 @@ def main():
# Tokenizer check: this script requires a fast tokenizer.
if not isinstance(tokenizer, PreTrainedTokenizerFast):
raise ValueError(
"This example script only works for models that have a fast tokenizer. Checkout the big table of models "
"at https://huggingface.co/transformers/index.html#supported-frameworks to find the model types that meet this "
"requirement"
"This example script only works for models that have a fast tokenizer. Checkout the big table of models at"
" https://huggingface.co/transformers/index.html#supported-frameworks to find the model types that meet"
" this requirement"
)
# Model has labels -> use them.
@@ -373,8 +387,8 @@ def main():
else:
logger.warning(
"Your model seems to have been trained with labels, but they don't match the dataset: ",
f"model labels: {list(sorted(model.config.label2id.keys()))}, dataset labels: {list(sorted(label_list))}."
"\nIgnoring the model labels as a result.",
f"model labels: {list(sorted(model.config.label2id.keys()))}, dataset labels:"
f" {list(sorted(label_list))}.\nIgnoring the model labels as a result.",
)
# Set the correspondences label/ID inside the model config

4
examples/pytorch/token-classification/run_ner_no_trainer.py
View File

@@ -403,8 +403,8 @@ def main():
else:
logger.warning(
"Your model seems to have been trained with labels, but they don't match the dataset: ",
f"model labels: {list(sorted(model.config.label2id.keys()))}, dataset labels: {list(sorted(label_list))}."
"\nIgnoring the model labels as a result.",
f"model labels: {list(sorted(model.config.label2id.keys()))}, dataset labels:"
f" {list(sorted(label_list))}.\nIgnoring the model labels as a result.",
)
# Set the correspondences label/ID inside the model config

75
examples/pytorch/translation/run_translation.py
View File

@@ -91,8 +91,10 @@ class ModelArguments:
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
"help": (
"Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
)
},
)
@@ -116,15 +118,12 @@ class DataTrainingArguments:
validation_file: Optional[str] = field(
default=None,
metadata={
"help": "An optional input evaluation data file to evaluate the metrics (sacreblue) on "
"a jsonlines file."
"help": "An optional input evaluation data file to evaluate the metrics (sacreblue) on a jsonlines file."
},
)
test_file: Optional[str] = field(
default=None,
metadata={
"help": "An optional input test data file to evaluate the metrics (sacreblue) on " "a jsonlines file."
},
metadata={"help": "An optional input test data file to evaluate the metrics (sacreblue) on a jsonlines file."},
)
overwrite_cache: bool = field(
default=False, metadata={"help": "Overwrite the cached training and evaluation sets"}
@@ -136,60 +135,76 @@ class DataTrainingArguments:
max_source_length: Optional[int] = field(
default=1024,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
max_target_length: Optional[int] = field(
default=128,
metadata={
"help": "The maximum total sequence length for target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
"help": (
"The maximum total sequence length for target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
)
},
)
val_max_target_length: Optional[int] = field(
default=None,
metadata={
"help": "The maximum total sequence length for validation target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded. Will default to `max_target_length`."
"This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
"during ``evaluate`` and ``predict``."
"help": (
"The maximum total sequence length for validation target text after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded. Will default to `max_target_length`."
"This argument is also used to override the ``max_length`` param of ``model.generate``, which is used "
"during ``evaluate`` and ``predict``."
)
},
)
pad_to_max_length: bool = field(
default=False,
metadata={
"help": "Whether to pad all samples to model maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
"help": (
"Whether to pad all samples to model maximum sentence length. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch. More "
"efficient on GPU but very bad for TPU."
)
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
)
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
)
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
"help": (
"For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
)
},
)
num_beams: Optional[int] = field(
default=None,
metadata={
"help": "Number of beams to use for evaluation. This argument will be passed to ``model.generate``, "
"which is used during ``evaluate`` and ``predict``."
"help": (
"Number of beams to use for evaluation. This argument will be passed to ``model.generate``, "
"which is used during ``evaluate`` and ``predict``."
)
},
)
ignore_pad_token_for_loss: bool = field(
@@ -204,9 +219,11 @@ class DataTrainingArguments:
forced_bos_token: Optional[str] = field(
default=None,
metadata={
"help": "The token to force as the first generated token after the :obj:`decoder_start_token_id`."
"Useful for multilingual models like :doc:`mBART <../model_doc/mbart>` where the first generated token "
"needs to be the target language token.(Usually it is the target language token)"
"help": (
"The token to force as the first generated token after the :obj:`decoder_start_token_id`.Useful for"
" multilingual models like :doc:`mBART <../model_doc/mbart>` where the first generated token needs to"
" be the target language token.(Usually it is the target language token)"
)
},
)

42
examples/pytorch/translation/run_translation_no_trainer.py
View File

@@ -95,41 +95,51 @@ def parse_args():
"--num_beams",
type=int,
default=None,
help="Number of beams to use for evaluation. This argument will be "
"passed to ``model.generate``, which is used during ``evaluate`` and ``predict``.",
help=(
"Number of beams to use for evaluation. This argument will be "
"passed to ``model.generate``, which is used during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--max_source_length",
type=int,
default=1024,
help="The maximum total input sequence length after "
"tokenization.Sequences longer than this will be truncated, sequences shorter will be padded.",
help=(
"The maximum total input sequence length after "
"tokenization.Sequences longer than this will be truncated, sequences shorter will be padded."
),
)
parser.add_argument(
"--max_target_length",
type=int,
default=128,
help="The maximum total sequence length for target text after "
"tokenization. Sequences longer than this will be truncated, sequences shorter will be padded."
"during ``evaluate`` and ``predict``.",
help=(
"The maximum total sequence length for target text after "
"tokenization. Sequences longer than this will be truncated, sequences shorter will be padded."
"during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--val_max_target_length",
type=int,
default=None,
help="The maximum total sequence length for validation "
"target text after tokenization.Sequences longer than this will be truncated, sequences shorter will be "
"padded. Will default to `max_target_length`.This argument is also used to override the ``max_length`` "
"param of ``model.generate``, which is used during ``evaluate`` and ``predict``.",
help=(
"The maximum total sequence length for validation "
"target text after tokenization.Sequences longer than this will be truncated, sequences shorter will be "
"padded. Will default to `max_target_length`.This argument is also used to override the ``max_length`` "
"param of ``model.generate``, which is used during ``evaluate`` and ``predict``."
),
)
parser.add_argument(
"--pad_to_max_length",
type=bool,
default=False,
help="Whether to pad all samples to model maximum sentence "
"length. If False, will pad the samples dynamically when batching to the maximum length in the batch. More"
"efficient on GPU but very bad for TPU.",
help=(
"Whether to pad all samples to model maximum sentence "
"length. If False, will pad the samples dynamically when batching to the maximum length in the batch. More"
"efficient on GPU but very bad for TPU."
),
)
parser.add_argument(
"--validation_file", type=str, default=None, help="A csv or a json file containing the validation data."
@@ -138,7 +148,7 @@ def parse_args():
"--ignore_pad_token_for_loss",
type=bool,
default=True,
help="Whether to ignore the tokens corresponding to " "padded labels in the loss computation or not.",
help="Whether to ignore the tokens corresponding to padded labels in the loss computation or not.",
)
parser.add_argument("--source_lang", type=str, default=None, help="Source language id for translation.")
parser.add_argument("--target_lang", type=str, default=None, help="Target language id for translation.")
@@ -146,7 +156,7 @@ def parse_args():
"--source_prefix",
type=str,
default=None,
help="A prefix to add before every source text " "(useful for T5 models).",
help="A prefix to add before every source text (useful for T5 models).",
)
parser.add_argument(
"--preprocessing_num_workers",

4
examples/pytorch/xla_spawn.py
View File

@@ -39,9 +39,7 @@ def parse_args():
"""
parser = ArgumentParser(
description=(
"PyTorch TPU distributed training launch "
"helper utility that will spawn up "
"multiple distributed processes"
"PyTorch TPU distributed training launch helper utility that will spawn up multiple distributed processes"
)
)