From 48cbf267c988b56c71a2380f748a3e6092ccaed3 Mon Sep 17 00:00:00 2001 From: VictorSanh Date: Tue, 3 Dec 2019 11:01:37 -0500 Subject: [PATCH] Use full dataset for eval (SequentialSampler in Distributed setting) --- examples/run_glue.py | 2 +- examples/run_lm_finetuning.py | 2 +- examples/run_multiple_choice.py | 2 +- examples/run_xnli.py | 2 +- 4 files changed, 4 insertions(+), 4 deletions(-) diff --git a/examples/run_glue.py b/examples/run_glue.py index 601e9a34c2..369a7110ab 100644 --- a/examples/run_glue.py +++ b/examples/run_glue.py @@ -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) # 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) # multi-gpu eval diff --git a/examples/run_lm_finetuning.py b/examples/run_lm_finetuning.py index 4acea00c55..0bb7460353 100644 --- a/examples/run_lm_finetuning.py +++ b/examples/run_lm_finetuning.py @@ -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) # 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) # multi-gpu evaluate diff --git a/examples/run_multiple_choice.py b/examples/run_multiple_choice.py index 30c3332929..9d1ca7f300 100644 --- a/examples/run_multiple_choice.py +++ b/examples/run_multiple_choice.py @@ -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) # 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) # multi-gpu evaluate diff --git a/examples/run_xnli.py b/examples/run_xnli.py index a3bc0d4604..42d134a43a 100644 --- a/examples/run_xnli.py +++ b/examples/run_xnli.py @@ -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) # 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) # multi-gpu eval