fix multi-gpu eval

examples/run_glue.py

@@ -224,6 +224,10 @@ def evaluate(args, model, tokenizer, prefix=""):
         eval_sampler = SequentialSampler(eval_dataset) if args.local_rank == -1 else DistributedSampler(eval_dataset)
         eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
 
+        # multi-gpu eval
+        if args.n_gpu > 1:
+            model = torch.nn.DataParallel(model)
+
         # Eval!
         logger.info("***** Running evaluation {} *****".format(prefix))
         logger.info("  Num examples = %d", len(eval_dataset))

examples/run_lm_finetuning.py

@@ -300,6 +300,10 @@ def evaluate(args, model, tokenizer, prefix=""):
     eval_sampler = SequentialSampler(eval_dataset) if args.local_rank == -1 else DistributedSampler(eval_dataset)
     eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
 
+    # multi-gpu evaluate
+    if args.n_gpu > 1:
+        model = torch.nn.DataParallel(model)
+
     # Eval!
     logger.info("***** Running evaluation {} *****".format(prefix))
     logger.info("  Num examples = %d", len(eval_dataset))

examples/run_multiple_choice.py

@@ -229,6 +229,10 @@ def evaluate(args, model, tokenizer, prefix="", test=False):
         eval_sampler = SequentialSampler(eval_dataset) if args.local_rank == -1 else DistributedSampler(eval_dataset)
         eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
 
+        # multi-gpu evaluate
+        if args.n_gpu > 1:
+            model = torch.nn.DataParallel(model)
+
         # Eval!
         logger.info("***** Running evaluation {} *****".format(prefix))
         logger.info("  Num examples = %d", len(eval_dataset))

examples/run_ner.py

@@ -191,6 +191,10 @@ def evaluate(args, model, tokenizer, labels, pad_token_label_id, mode, prefix=""
     eval_sampler = SequentialSampler(eval_dataset) if args.local_rank == -1 else DistributedSampler(eval_dataset)
     eval_dataloader = DataLoader(eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
 
+    # multi-gpu evaluate
+    if args.n_gpu > 1:
+        model = torch.nn.DataParallel(model)
+
     # Eval!
     logger.info("***** Running evaluation %s *****", prefix)
     logger.info("  Num examples = %d", len(eval_dataset))

examples/run_squad.py

@@ -217,6 +217,10 @@ def evaluate(args, model, tokenizer, prefix=""):
     eval_sampler = SequentialSampler(dataset) if args.local_rank == -1 else DistributedSampler(dataset)
     eval_dataloader = DataLoader(dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
 
+    # multi-gpu evaluate
+    if args.n_gpu > 1:
+        model = torch.nn.DataParallel(model)
+
     # Eval!
     logger.info("***** Running evaluation {} *****".format(prefix))
     logger.info("  Num examples = %d", len(dataset))

examples/run_summarization_finetuning.py

@@ -275,6 +275,10 @@ def evaluate(args, model, tokenizer, prefix=""):
         eval_dataset, sampler=eval_sampler, batch_size=args.eval_batch_size
     )
 
+    # multi-gpu evaluate
+    if args.n_gpu > 1:
+        model = torch.nn.DataParallel(model)
+        
     logger.info("***** Running evaluation {} *****".format(prefix))
     logger.info("  Num examples = %d", len(eval_dataset))
     logger.info("  Batch size = %d", args.eval_batch_size)