Updating word_language_model examples to evaluate with no_grad instead of volatile

examples/word_language_model/main.py

@@ -134,9 +134,9 @@ def repackage_hidden(h):
 # by the batchify function. The chunks are along dimension 0, corresponding
 # to the seq_len dimension in the LSTM.
 
-def get_batch(source, i, evaluation=False):
+def get_batch(source, i):
     seq_len = min(args.bptt, len(source) - 1 - i)
-    data = Variable(source[i:i+seq_len], volatile=evaluation)
+    data = Variable(source[i:i+seq_len])
     target = Variable(source[i+1:i+1+seq_len].view(-1))
     return data, target
 
@@ -147,13 +147,14 @@ def evaluate(data_source):
     total_loss = 0
     ntokens = len(corpus.dictionary)
     hidden = model.init_hidden(eval_batch_size)
-    for i in range(0, data_source.size(0) - 1, args.bptt):
-        data, targets = get_batch(data_source, i, evaluation=True)
-        output, hidden = model(data, hidden)
-        output_flat = output.view(-1, ntokens)
-        #total loss can overflow if accumulated in fp16.
-        total_loss += len(data) * criterion(output_flat, targets).data.float()
-        hidden = repackage_hidden(hidden)
+    with torch.no_grad():
+        for i in range(0, data_source.size(0) - 1, args.bptt):
+            data, targets = get_batch(data_source, i)
+            output, hidden = model(data, hidden)
+            output_flat = output.view(-1, ntokens)
+            #total loss can overflow if accumulated in fp16.
+            total_loss += len(data) * criterion(output_flat, targets).data.float()
+            hidden = repackage_hidden(hidden)
     return to_python_float(total_loss) / len(data_source)
 
 

examples/word_language_model/main_fp16_optimizer.py

@@ -149,9 +149,9 @@ def repackage_hidden(h):
 # by the batchify function. The chunks are along dimension 0, corresponding
 # to the seq_len dimension in the LSTM.
 
-def get_batch(source, i, evaluation=False):
+def get_batch(source, i):
     seq_len = min(args.bptt, len(source) - 1 - i)
-    data = Variable(source[i:i+seq_len], volatile=evaluation)
+    data = Variable(source[i:i+seq_len])
     target = Variable(source[i+1:i+1+seq_len].view(-1))
     return data, target
 
@@ -162,13 +162,14 @@ def evaluate(data_source):
     total_loss = 0
     ntokens = len(corpus.dictionary)
     hidden = model.init_hidden(eval_batch_size)
-    for i in range(0, data_source.size(0) - 1, args.bptt):
-        data, targets = get_batch(data_source, i, evaluation=True)
-        output, hidden = model(data, hidden)
-        output_flat = output.view(-1, ntokens)
-        #total loss can overflow if accumulated in fp16.
-        total_loss += len(data) * criterion(output_flat, targets).data.float()
-        hidden = repackage_hidden(hidden)
+    with torch.no_grad():
+        for i in range(0, data_source.size(0) - 1, args.bptt):
+            data, targets = get_batch(data_source, i)
+            output, hidden = model(data, hidden)
+            output_flat = output.view(-1, ntokens)
+            #total loss can overflow if accumulated in fp16.
+            total_loss += len(data) * criterion(output_flat, targets).data.float()
+            hidden = repackage_hidden(hidden)
     return to_python_float(total_loss) / len(data_source)