Skip to content

fix: seq2seq decode returns primary choice (strip alternates) - #108

Merged
ronaldtse merged 1 commit into
mainfrom
fix/first-alternates-decode
Oct 11, 2026
Merged

fix: seq2seq decode returns primary choice (strip alternates)#108
ronaldtse merged 1 commit into
mainfrom
fix/first-alternates-decode

Conversation

@ronaldtse

@ronaldtse ronaldtse commented Oct 10, 2026

Copy link
Copy Markdown
Contributor

Summary

  • Models trained on multi-reference corpora (QCRI silver) emit slash-separated alternate sets per token; the runtime returned them verbatim
  • firstAlternates in tokens.ts, applied in translateDirect (windowed path delegates through it); parity with the Python and Ruby runtimes
  • AR news specialist corrects from 10.03 to 4.10 WER on WikiNews-2024 multiref under the same benchmark, first-alternate decode

Test plan

  • vitest test/ml/first-alternates.test.ts (3 tests, TDD red-green)
  • full suite: 296 passed, 3 skipped (two consecutive runs)

@ronaldtse
ronaldtse merged commit 6678a79 into main Oct 11, 2026
6 of 8 checks passed
@ronaldtse
ronaldtse deleted the fix/first-alternates-decode branch October 11, 2026 08:29
@ronaldtse ronaldtse mentioned this pull request Oct 14, 2026
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@ronaldtse