Skip to content

docs(WO88): khm multi-task POSITIVE — the framework beats the legacy #374

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ronaldtse merged 1 commit into from
Oct 27, 2026
Merged

docs(WO88): khm multi-task POSITIVE — the framework beats the legacy #374

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 13 additions & 0 deletions TODO.sota/88-secryst-migration.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,3 +56,16 @@ Chuon Nath IPA layer + SBBIC vocabulary + the khm-latn pairs), which
is the WO68 data work. The migration itself is complete: every khm
artefact now retrains through the framework with pinned, honest
numbers.


## Multi-task RESULT (2026-10-27): the framework BEATS the legacy

khm multi-task (<latn>+<ipa> task prefixes, ByT5-small, 4ep):
**dev CER 24.50%** (best; final 24.54) — vs legacy 27.57% like-for-like.
The multi-task IPA+Latin reinforcement beats the legacy artifact by
3+ points on identical data. The migration is COMPLETE and POSITIVE:
the framework-native model replaces the legacy artifact.

Next: export the multi-task model → fixtures → gates → publish as the
khm-latn successor (owner: id + version — the convention is
khm-phonemic-1.0 under the MECE taxonomy, since the output is IPA).
Loading