Update generate.py (#831)

Summary:
Repeated use of 'i' in evaluate may cause some problems.
Pull Request resolved: https://github.com/pytorch/fairseq/pull/831

Differential Revision: D15980227

Pulled By: myleott

fbshipit-source-id: 7b6b54a6b54938ad63ed1720d930505b56e5c84b

generate.py

@@ -135,7 +135,7 @@ def main(args):
                         print('T-{}\t{}'.format(sample_id, target_str))
 
                 # Process top predictions
-                for i, hypo in enumerate(hypos[i][:min(len(hypos), args.nbest)]):
+                for j, hypo in enumerate(hypos[i][:min(len(hypos), args.nbest)]):
                     hypo_tokens, hypo_str, alignment = utils.post_process_prediction(
                         hypo_tokens=hypo['tokens'].int().cpu(),
                         src_str=src_str,
@@ -162,7 +162,7 @@ def main(args):
                             ))
 
                     # Score only the top hypothesis
-                    if has_target and i == 0:
+                    if has_target and j == 0:
                         if align_dict is not None or args.remove_bpe is not None:
                             # Convert back to tokens for evaluation with unk replacement and/or without BPE
                             target_tokens = tgt_dict.encode_line(target_str, add_if_not_exist=True)