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

Merge pull request #1723 from huggingface/fix-1623

Browse Source 
Fix #1623
This commit is contained in:
Thomas Wolf
2019-11-05 08:36:30 +01:00
committed by GitHub
parent 68f7064a3e 89d6272898
commit d2e2577dd3
8 changed files with 87 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
examples/distillation/run_squad_w_distillation.py
View File

@@ -506,9 +506,15 @@ def main():
args.model_type = args.model_type.lower() args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type] config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path) config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path,
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, do_lower_case=args.do_lower_case) cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path, from_tf=bool('.ckpt' in args.model_name_or_path), config=config) tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
if args.teacher_type is not None: if args.teacher_type is not None:
assert args.teacher_name_or_path is not None assert args.teacher_name_or_path is not None
@@ -516,8 +522,11 @@ def main():
assert args.alpha_ce + args.alpha_squad > 0. assert args.alpha_ce + args.alpha_squad > 0.
assert args.teacher_type != 'distilbert', "We constraint teachers not to be of type DistilBERT." assert args.teacher_type != 'distilbert', "We constraint teachers not to be of type DistilBERT."
teacher_config_class, teacher_model_class, _ = MODEL_CLASSES[args.teacher_type] teacher_config_class, teacher_model_class, _ = MODEL_CLASSES[args.teacher_type]
teacher_config = teacher_config_class.from_pretrained(args.teacher_name_or_path) teacher_config = teacher_config_class.from_pretrained(args.teacher_name_or_path,
teacher = teacher_model_class.from_pretrained(args.teacher_name_or_path, config=teacher_config) cache_dir=args.cache_dir if args.cache_dir else None)
teacher = teacher_model_class.from_pretrained(args.teacher_name_or_path,
config=teacher_config,
cache_dir=args.cache_dir if args.cache_dir else None)
teacher.to(args.device) teacher.to(args.device)
else: else:
teacher = None teacher = None
@@ -553,8 +562,10 @@ def main():
torch.save(args, os.path.join(args.output_dir, 'training_args.bin')) torch.save(args, os.path.join(args.output_dir, 'training_args.bin'))
# Load a trained model and vocabulary that you have fine-tuned # Load a trained model and vocabulary that you have fine-tuned
model = model_class.from_pretrained(args.output_dir) model = model_class.from_pretrained(args.output_dir, cache_dir=args.cache_dir if args.cache_dir else None)
tokenizer = tokenizer_class.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case) tokenizer = tokenizer_class.from_pretrained(args.output_dir,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None)
model.to(args.device) model.to(args.device)
@@ -571,7 +582,7 @@ def main():
for checkpoint in checkpoints: for checkpoint in checkpoints:
# Reload the model # Reload the model
global_step = checkpoint.split('-')[-1] if len(checkpoints) > 1 else "" global_step = checkpoint.split('-')[-1] if len(checkpoints) > 1 else ""
model = model_class.from_pretrained(checkpoint) model = model_class.from_pretrained(checkpoint, cache_dir=args.cache_dir if args.cache_dir else None)
model.to(args.device) model.to(args.device)
# Evaluate # Evaluate

14
examples/run_bertology.py
View File

@@ -304,10 +304,16 @@ def main():
break break
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type] config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path, config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path,
num_labels=num_labels, finetuning_task=args.task_name, num_labels=num_labels,
output_attentions=True) finetuning_task=args.task_name,
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path) output_attentions=True,
model = model_class.from_pretrained(args.model_name_or_path, from_tf=bool('.ckpt' in args.model_name_or_path), config=config) cache_dir=args.cache_dir if args.cache_dir else None)
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
if args.local_rank == 0: if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab

16
examples/run_glue.py
View File

@@ -477,9 +477,17 @@ def main():
args.model_type = args.model_type.lower() args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type] config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path, num_labels=num_labels, finetuning_task=args.task_name) config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path,
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, do_lower_case=args.do_lower_case) num_labels=num_labels,
model = model_class.from_pretrained(args.model_name_or_path, from_tf=bool('.ckpt' in args.model_name_or_path), config=config) finetuning_task=args.task_name,
cache_dir=args.cache_dir if args.cache_dir else None)
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
if args.local_rank == 0: if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab
@@ -514,7 +522,7 @@ def main():
# Load a trained model and vocabulary that you have fine-tuned # Load a trained model and vocabulary that you have fine-tuned
model = model_class.from_pretrained(args.output_dir) model = model_class.from_pretrained(args.output_dir)
tokenizer = tokenizer_class.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case) tokenizer = tokenizer_class.from_pretrained(args.output_dir)
model.to(args.device) model.to(args.device)

12
examples/run_lm_finetuning.py
View File

@@ -471,12 +471,18 @@ def main():
torch.distributed.barrier() # Barrier to make sure only the first process in distributed training download model & vocab torch.distributed.barrier() # Barrier to make sure only the first process in distributed training download model & vocab
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type] config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path) config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path,
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, do_lower_case=args.do_lower_case) cache_dir=args.cache_dir if args.cache_dir else None)
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None)
if args.block_size <= 0: if args.block_size <= 0:
args.block_size = tokenizer.max_len_single_sentence # Our input block size will be the max possible for the model args.block_size = tokenizer.max_len_single_sentence # Our input block size will be the max possible for the model
args.block_size = min(args.block_size, tokenizer.max_len_single_sentence) args.block_size = min(args.block_size, tokenizer.max_len_single_sentence)
model = model_class.from_pretrained(args.model_name_or_path, from_tf=bool('.ckpt' in args.model_name_or_path), config=config) model = model_class.from_pretrained(args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
model.to(args.device) model.to(args.device)
if args.local_rank == 0: if args.local_rank == 0:

14
examples/run_multiple_choice.py
View File

@@ -464,9 +464,17 @@ def main():
args.model_type = args.model_type.lower() args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type] config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path, num_labels=num_labels, finetuning_task=args.task_name) config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path,
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, do_lower_case=args.do_lower_case) num_labels=num_labels,
model = model_class.from_pretrained(args.model_name_or_path, from_tf=bool('.ckpt' in args.model_name_or_path), config=config) finetuning_task=args.task_name,
cache_dir=args.cache_dir if args.cache_dir else None)
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
if args.local_rank == 0: if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab

12
examples/run_ner.py
View File

@@ -428,11 +428,15 @@ def main():
args.model_type = args.model_type.lower() args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type] config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path, config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path,
num_labels=num_labels) num_labels=num_labels,
cache_dir=args.cache_dir if args.cache_dir else None)
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case) do_lower_case=args.do_lower_case,
model = model_class.from_pretrained(args.model_name_or_path, from_tf=bool(".ckpt" in args.model_name_or_path), cache_dir=args.cache_dir if args.cache_dir else None)
config=config) model = model_class.from_pretrained(args.model_name_or_path,
from_tf=bool(".ckpt" in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
if args.local_rank == 0: if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab

12
examples/run_squad.py
View File

@@ -477,9 +477,15 @@ def main():
args.model_type = args.model_type.lower() args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type] config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path) config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path,
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, do_lower_case=args.do_lower_case) cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path, from_tf=bool('.ckpt' in args.model_name_or_path), config=config) tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
if args.local_rank == 0: if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab

12
templates/adding_a_new_example_script/run_xxx.py
View File

@@ -472,9 +472,15 @@ def main():
args.model_type = args.model_type.lower() args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type] config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path) config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path,
tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, do_lower_case=args.do_lower_case) cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path, from_tf=bool('.ckpt' in args.model_name_or_path), config=config) tokenizer = tokenizer_class.from_pretrained(args.tokenizer_name if args.tokenizer_name else args.model_name_or_path,
do_lower_case=args.do_lower_case,
cache_dir=args.cache_dir if args.cache_dir else None)
model = model_class.from_pretrained(args.model_name_or_path,
from_tf=bool('.ckpt' in args.model_name_or_path),
config=config,
cache_dir=args.cache_dir if args.cache_dir else None)
if args.local_rank == 0: if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab