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

Merge branch 'master' into squad-refactor

Browse Source
This commit is contained in:
Lysandre Debut
2019-12-09 10:41:15 -05:00
committed by GitHub
parent e9217da5ff 0cb163865a
commit 00c4e39581
127 changed files with 7917 additions and 606 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
examples/run_squad.py
View File

@@ -43,7 +43,8 @@ from transformers import (WEIGHTS_NAME, BertConfig,
XLMTokenizer, XLNetConfig,
XLNetForQuestionAnswering,
XLNetTokenizer,
DistilBertConfig, DistilBertForQuestionAnswering, DistilBertTokenizer)
DistilBertConfig, DistilBertForQuestionAnswering, DistilBertTokenizer,
AlbertConfig, AlbertForQuestionAnswering, AlbertTokenizer)
from transformers import AdamW, get_linear_schedule_with_warmup, squad_convert_examples_to_features
@@ -56,7 +57,8 @@ MODEL_CLASSES = {
'bert': (BertConfig, BertForQuestionAnswering, BertTokenizer),
'xlnet': (XLNetConfig, XLNetForQuestionAnswering, XLNetTokenizer),
'xlm': (XLMConfig, XLMForQuestionAnswering, XLMTokenizer),
'distilbert': (DistilBertConfig, DistilBertForQuestionAnswering, DistilBertTokenizer)
'distilbert': (DistilBertConfig, DistilBertForQuestionAnswering, DistilBertTokenizer),
'albert': (AlbertConfig, AlbertForQuestionAnswering, AlbertTokenizer)
}
def set_seed(args):
@@ -121,7 +123,7 @@ def train(args, train_dataset, model, tokenizer):
logger.info(" Gradient Accumulation steps = %d", args.gradient_accumulation_steps)
logger.info(" Total optimization steps = %d", t_total)
global_step = 0
global_step = 1
tr_loss, logging_loss = 0.0, 0.0
model.zero_grad()
train_iterator = trange(int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0])
@@ -214,7 +216,7 @@ def evaluate(args, model, tokenizer, prefix=""):
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(dataset) if args.local_rank == -1 else DistributedSampler(dataset)
eval_sampler = SequentialSampler(dataset)
eval_dataloader = DataLoader(dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# multi-gpu evaluate
@@ -558,7 +560,7 @@ def main():
torch.save(args, os.path.join(args.output_dir, 'training_args.bin'))
# 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, force_download=True)
tokenizer = tokenizer_class.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
model.to(args.device)
@@ -576,7 +578,7 @@ def main():
for checkpoint in checkpoints:
# Reload the model
global_step = checkpoint.split('-')[-1] if len(checkpoints) > 1 else ""
model = model_class.from_pretrained(checkpoint)
model = model_class.from_pretrained(checkpoint, force_download=True)
model.to(args.device)
# Evaluate