Skip to content

Commit

Permalink
Fix arguments in CTC decoding script (pytorch#2315)
Browse files Browse the repository at this point in the history
Summary:
Some arguments in `ArgumentParser` are not used in the `lexicon_decoder`. Fix them to use the ones in the parser.

Pull Request resolved: pytorch#2315

Reviewed By: carolineechen

Differential Revision: D35357678

Pulled By: nateanl

fbshipit-source-id: 4e70418cf03708b82bc158cafd9999a80ad08f92
  • Loading branch information
nateanl authored and xiaohui-zhang committed May 4, 2022
1 parent 3cf05e4 commit 81d820b
Showing 1 changed file with 6 additions and 6 deletions.
12 changes: 6 additions & 6 deletions examples/asr/librispeech_ctc_decoder/inference.py
Original file line number Diff line number Diff line change
Expand Up @@ -35,14 +35,14 @@ def run_inference(args):
lexicon=lexicon_file,
tokens=tokens,
lm=kenlm_file,
nbest=1,
beam_size=1500,
beam_size_token=None,
beam_threshold=50,
nbest=args.nbest,
beam_size=args.beam_size,
beam_size_token=args.beam_size_token,
beam_threshold=args.beam_threshold,
lm_weight=args.lm_weight,
word_score=args.word_score,
unk_score=float("-inf"),
sil_score=0,
unk_score=args.unk_score,
sil_score=args.sil_score,
log_add=False,
)

Expand Down

0 comments on commit 81d820b

Please sign in to comment.