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

Use full dataset for eval (SequentialSampler in Distributed setting)

Browse Source
This commit is contained in:
VictorSanh
2019-12-03 11:01:37 -05:00
parent f434bfc623
commit 48cbf267c9
4 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/run_glue.py
View File

@@ -231,7 +231,7 @@ def evaluate(args, model, tokenizer, prefix=""):
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu) args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly # Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(eval_dataset) if args.local_rank == -1 else DistributedSampler(eval_dataset) eval_sampler = SequentialSampler(eval_dataset)
eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size) eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# multi-gpu eval # multi-gpu eval

2
examples/run_lm_finetuning.py
View File

@@ -300,7 +300,7 @@ def evaluate(args, model, tokenizer, prefix=""):
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu) args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly # Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(eval_dataset) if args.local_rank == -1 else DistributedSampler(eval_dataset) eval_sampler = SequentialSampler(eval_dataset)
eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size) eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# multi-gpu evaluate # multi-gpu evaluate

2
examples/run_multiple_choice.py
View File

@@ -226,7 +226,7 @@ def evaluate(args, model, tokenizer, prefix="", test=False):
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu) args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly # Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(eval_dataset) if args.local_rank == -1 else DistributedSampler(eval_dataset) eval_sampler = SequentialSampler(eval_dataset)
eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size) eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# multi-gpu evaluate # multi-gpu evaluate

2
examples/run_xnli.py
View File

@@ -206,7 +206,7 @@ def evaluate(args, model, tokenizer, prefix=""):
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu) args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly # Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(eval_dataset) if args.local_rank == -1 else DistributedSampler(eval_dataset) eval_sampler = SequentialSampler(eval_dataset)
eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size) eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# multi-gpu eval # multi-gpu eval