important thing to declare is the target_spcae_id in different T2T…

important thing to declare is the target_spcae_id in different T2T implementation is different, default 4 in 1.0.14, but 0 in 1.6.5

scripts/convert_t2t_new_to_fairseq.py

@@ -698,7 +698,7 @@ if __name__ == '__main__':
   shard = args.shard_num
   # used for position embedding
   # this is `EN_BPE_TOK`, which is used target-space-id for Problem `wmt_zhen_tokens_32k`
-  # target_space_id = 4
+  # target_space_id = 0 in t2t-1.6.5
   target_space_id = args.tgt_space_id
 
   src_max_length = 1024