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

13
examples/research_projects/movement-pruning/bertarize.py
View File

@@ -103,15 +103,20 @@ if __name__ == "__main__":
choices=["l0", "magnitude", "topK", "sigmoied_threshold"],
type=str,
required=True,
help="Pruning Method (l0 = L0 regularization, magnitude = Magnitude pruning, topK = Movement pruning, sigmoied_threshold = Soft movement pruning)",
help=(
"Pruning Method (l0 = L0 regularization, magnitude = Magnitude pruning, topK = Movement pruning,"
" sigmoied_threshold = Soft movement pruning)"
),
)
parser.add_argument(
"--threshold",
type=float,
required=False,
help="For `magnitude` and `topK`, it is the level of remaining weights (in %) in the fine-pruned model."
"For `sigmoied_threshold`, it is the threshold \tau against which the (sigmoied) scores are compared."
"Not needed for `l0`",
help=(
"For `magnitude` and `topK`, it is the level of remaining weights (in %) in the fine-pruned model."
"For `sigmoied_threshold`, it is the threshold \tau against which the (sigmoied) scores are compared."
"Not needed for `l0`"
),
)
parser.add_argument(
"--model_name_or_path",

13
examples/research_projects/movement-pruning/counts_parameters.py
View File

@@ -70,15 +70,20 @@ if __name__ == "__main__":
choices=["l0", "topK", "sigmoied_threshold"],
type=str,
required=True,
help="Pruning Method (l0 = L0 regularization, topK = Movement pruning, sigmoied_threshold = Soft movement pruning)",
help=(
"Pruning Method (l0 = L0 regularization, topK = Movement pruning, sigmoied_threshold = Soft movement"
" pruning)"
),
)
parser.add_argument(
"--threshold",
type=float,
required=False,
help="For `topK`, it is the level of remaining weights (in %) in the fine-pruned model."
"For `sigmoied_threshold`, it is the threshold \tau against which the (sigmoied) scores are compared."
"Not needed for `l0`",
help=(
"For `topK`, it is the level of remaining weights (in %) in the fine-pruned model."
"For `sigmoied_threshold`, it is the threshold \tau against which the (sigmoied) scores are compared."
"Not needed for `l0`"
),
)
parser.add_argument(
"--serialization_dir",

4
examples/research_projects/movement-pruning/emmental/modeling_bert_masked.py
View File

@@ -80,8 +80,8 @@ class BertSelfAttention(nn.Module):
super().__init__()
if config.hidden_size % config.num_attention_heads != 0 and not hasattr(config, "embedding_size"):
raise ValueError(
"The hidden size (%d) is not a multiple of the number of attention "
"heads (%d)" % (config.hidden_size, config.num_attention_heads)
"The hidden size (%d) is not a multiple of the number of attention heads (%d)"
% (config.hidden_size, config.num_attention_heads)
)
self.output_attentions = config.output_attentions

37
examples/research_projects/movement-pruning/masked_run_glue.py
View File

@@ -622,8 +622,10 @@ def main():
"--max_seq_length",
default=128,
type=int,
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("--do_train", action="store_true", help="Whether to run training.")
parser.add_argument("--do_eval", action="store_true", help="Whether to run eval on the dev set.")
@@ -669,22 +671,29 @@ def main():
"--initial_warmup",
default=1,
type=int,
help="Run `initial_warmup` * `warmup_steps` steps of threshold warmup during which threshold stays"
"at its `initial_threshold` value (sparsity schedule).",
help=(
"Run `initial_warmup` * `warmup_steps` steps of threshold warmup during which threshold stays"
"at its `initial_threshold` value (sparsity schedule)."
),
)
parser.add_argument(
"--final_warmup",
default=2,
type=int,
help="Run `final_warmup` * `warmup_steps` steps of threshold cool-down during which threshold stays"
"at its final_threshold value (sparsity schedule).",
help=(
"Run `final_warmup` * `warmup_steps` steps of threshold cool-down during which threshold stays"
"at its final_threshold value (sparsity schedule)."
),
)
parser.add_argument(
"--pruning_method",
default="topK",
type=str,
help="Pruning Method (l0 = L0 regularization, magnitude = Magnitude pruning, topK = Movement pruning, sigmoied_threshold = Soft movement pruning).",
help=(
"Pruning Method (l0 = L0 regularization, magnitude = Magnitude pruning, topK = Movement pruning,"
" sigmoied_threshold = Soft movement pruning)."
),
)
parser.add_argument(
"--mask_init",
@@ -717,7 +726,10 @@ def main():
"--teacher_type",
default=None,
type=str,
help="Teacher type. Teacher tokenizer and student (model) tokenizer must output the same tokenization. Only for distillation.",
help=(
"Teacher type. Teacher tokenizer and student (model) tokenizer must output the same tokenization. Only for"
" distillation."
),
)
parser.add_argument(
"--teacher_name_or_path",
@@ -787,8 +799,10 @@ def main():
"--fp16_opt_level",
type=str,
default="O1",
help="For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
"See details at https://nvidia.github.io/apex/amp.html",
help=(
"For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
"See details at https://nvidia.github.io/apex/amp.html"
),
)
parser.add_argument("--local_rank", type=int, default=-1, help="For distributed training: local_rank")
@@ -805,7 +819,8 @@ def main():
and not args.overwrite_output_dir
):
raise ValueError(
f"Output directory ({args.output_dir}) already exists and is not empty. Use --overwrite_output_dir to overcome."
f"Output directory ({args.output_dir}) already exists and is not empty. Use --overwrite_output_dir to"
" overcome."
)
# Setup CUDA, GPU & distributed training

57
examples/research_projects/movement-pruning/masked_run_squad.py
View File

@@ -737,8 +737,10 @@ def main():
"--max_seq_length",
default=384,
type=int,
help="The maximum total input sequence length after WordPiece tokenization. Sequences "
"longer than this will be truncated, and sequences shorter than this will be padded.",
help=(
"The maximum total input sequence length after WordPiece tokenization. Sequences "
"longer than this will be truncated, and sequences shorter than this will be padded."
),
)
parser.add_argument(
"--doc_stride",
@@ -750,8 +752,10 @@ def main():
"--max_query_length",
default=64,
type=int,
help="The maximum number of tokens for the question. Questions longer than this will "
"be truncated to this length.",
help=(
"The maximum number of tokens for the question. Questions longer than this will "
"be truncated to this length."
),
)
parser.add_argument("--do_train", action="store_true", help="Whether to run training.")
parser.add_argument("--do_eval", action="store_true", help="Whether to run eval on the dev set.")
@@ -785,22 +789,29 @@ def main():
"--initial_warmup",
default=1,
type=int,
help="Run `initial_warmup` * `warmup_steps` steps of threshold warmup during which threshold stays"
"at its `initial_threshold` value (sparsity schedule).",
help=(
"Run `initial_warmup` * `warmup_steps` steps of threshold warmup during which threshold stays"
"at its `initial_threshold` value (sparsity schedule)."
),
)
parser.add_argument(
"--final_warmup",
default=2,
type=int,
help="Run `final_warmup` * `warmup_steps` steps of threshold cool-down during which threshold stays"
"at its final_threshold value (sparsity schedule).",
help=(
"Run `final_warmup` * `warmup_steps` steps of threshold cool-down during which threshold stays"
"at its final_threshold value (sparsity schedule)."
),
)
parser.add_argument(
"--pruning_method",
default="topK",
type=str,
help="Pruning Method (l0 = L0 regularization, magnitude = Magnitude pruning, topK = Movement pruning, sigmoied_threshold = Soft movement pruning).",
help=(
"Pruning Method (l0 = L0 regularization, magnitude = Magnitude pruning, topK = Movement pruning,"
" sigmoied_threshold = Soft movement pruning)."
),
)
parser.add_argument(
"--mask_init",
@@ -833,7 +844,10 @@ def main():
"--teacher_type",
default=None,
type=str,
help="Teacher type. Teacher tokenizer and student (model) tokenizer must output the same tokenization. Only for distillation.",
help=(
"Teacher type. Teacher tokenizer and student (model) tokenizer must output the same tokenization. Only for"
" distillation."
),
)
parser.add_argument(
"--teacher_name_or_path",
@@ -883,20 +897,27 @@ def main():
"--max_answer_length",
default=30,
type=int,
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(
"--verbose_logging",
action="store_true",
help="If true, all of the warnings related to data processing will be printed. "
"A number of warnings are expected for a normal SQuAD evaluation.",
help=(
"If true, all of the warnings related to data processing will be printed. "
"A number of warnings are expected for a normal SQuAD evaluation."
),
)
parser.add_argument(
"--lang_id",
default=0,
type=int,
help="language id of input for language-specific xlm models (see tokenization_xlm.PRETRAINED_INIT_CONFIGURATION)",
help=(
"language id of input for language-specific xlm models (see"
" tokenization_xlm.PRETRAINED_INIT_CONFIGURATION)"
),
)
parser.add_argument("--logging_steps", type=int, default=500, help="Log every X updates steps.")
@@ -925,8 +946,10 @@ def main():
"--fp16_opt_level",
type=str,
default="O1",
help="For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
"See details at https://nvidia.github.io/apex/amp.html",
help=(
"For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
"See details at https://nvidia.github.io/apex/amp.html"
),
)
parser.add_argument("--server_ip", type=str, default="", help="Can be used for distant debugging.")
parser.add_argument("--server_port", type=str, default="", help="Can be used for distant debugging.")