Fix dimension off by not adding things to dictionary in intermediate eval (#573)

Summary:
Pull Request resolved: #573

Repeat title. The long term solution could be to add an option to distinguish intermediate eval. But confirmed with Juan setting it to False shouldn't affect existing stuff.

Reviewed By: jmp84

Differential Revision: D15793470

fbshipit-source-id: 670de05741607e6765cb8f839cab1301999ba1f0

Author: cndn

pytorch_translate/generate.py

@@ -205,7 +205,9 @@ def _generate_score(models, args, task, dataset):
             maxlen_b=args.max_len_b,
             cuda=use_cuda,
             timer=gen_timer,
-            prefix_size=1 if pytorch_translate_data.is_multilingual_many_to_one(args) else 0,
+            prefix_size=1
+            if pytorch_translate_data.is_multilingual_many_to_one(args)
+            else 0,
         )
 
         for trans_info in _iter_translations(
@@ -432,7 +434,7 @@ def _iter_translations(args, task, dataset, translations, align_dict, rescorer):
                     # Convert back to tokens for evaluation with unk replacement
                     # and/or without BPE
                     target_tokens = task.target_dictionary.encode_line(
-                        target_str, add_if_not_exist=True
+                        target_str, add_if_not_exist=False
                     )
                 # The probs score for the hypo_str; whether it's normalized by
                 # sequence length or not depends on normalize_scores, which is