OpenNMT / OpenNMT/CTranslate2

Feature request: Word level scores

Open
#1,077 11 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement
Dominant language
C++
Stars
4.7k
Forks
536
Avg merge
12h 12m
Merged PRs (30d)
4

Description

Hi, currently the decoder produces sentence level scores, instead of just outputting the average another option would be produce the score of each word/token.

Beam search might be a harder task to calculate a meaningful individual score, but for greedy I am assuming its just a matter of not averaging the scores together, please correct me if I'm wrong.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by tracing the decoder's current sentence-level score calculation and the output path for greedy decoding and beam search. Determine how per-word or per-token scores should be represented for each decoding mode, then define completion criteria and tests that verify the returned scores without averaging them into a single sentence score.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.