Revert `dim` in `F.softmax` for backwards compatibility

fairseq/models/fconv.py

@@ -128,7 +128,7 @@ class AttentionLayer(nn.Module):
 
         # softmax over last dim
         sz = x.size()
-        x = F.softmax(x.view(sz[0] * sz[1], sz[2]), dim=1)
+        x = F.softmax(x.view(sz[0] * sz[1], sz[2]))
         x = x.view(sz)
         attn_scores = x
 

fairseq/sequence_generator.py

@@ -330,7 +330,7 @@ class SequenceGenerator(object):
                 decoder_out, attn = model.decoder.incremental_forward(tokens, encoder_out)
             else:
                 decoder_out, attn = model.decoder.forward(tokens, encoder_out)
-            probs = F.softmax(decoder_out[:, -1, :], dim=1).data
+            probs = F.softmax(decoder_out[:, -1, :]).data
             attn = attn[:, -1, :].data
             if avg_probs is None or avg_attn is None:
                 avg_probs = probs