Skip to content

Fix OOV report tokenization and logged paths - #955

Open
p3zo wants to merge 2 commits into
MontrealCorpusTools:mainfrom
p3zo:fix/utterance-oovs-tokenization
Open

Fix OOV report tokenization and logged paths#955
p3zo wants to merge 2 commits into
MontrealCorpusTools:mainfrom
p3zo:fix/utterance-oovs-tokenization

Conversation

@p3zo

@p3zo p3zo commented May 2, 2026

Copy link
Copy Markdown

Two fixes related to OOV reporting:

  1. Utterance.oovs is a space-delimited string but treated as an iterable, causing utterance_oovs.txt and OOV token counts to be character-based. Resolved by splitting the strings into word tokens before writing utterance_oovs.txt. This fixes [BUG] OOV count reflects number of characters not tokens #926.

  2. The logs print oovs_found.txt, while the files that are actually written are dictionary-specific files like oovs_found_english_mfa.txt. Resolved by logging the dictionary-specific OOV report paths.

@p3zo p3zo closed this May 2, 2026
@p3zo p3zo reopened this May 2, 2026
@p3zo
p3zo force-pushed the fix/utterance-oovs-tokenization branch 2 times, most recently from 5462a0c to 85db3b9 Compare May 2, 2026 04:00
@p3zo
p3zo force-pushed the fix/utterance-oovs-tokenization branch from 85db3b9 to 70db128 Compare May 2, 2026 04:01
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[BUG] OOV count reflects number of characters not tokens

1 participant