From 215035e383b39aa248acb673c96af8acd1fd3025 Mon Sep 17 00:00:00 2001 From: Lucas Winther Date: Fri, 30 Oct 2026 00:43:51 +0100 Subject: [PATCH] Assert whole-file WER per clip in the real-model test (plan 001-3 Tasks 8-9 done) Chunked file path and long-form (explicit and auto language) held to the per-clip bars; full real-model test green under the heavy-job lock. Co-Authored-By: Claude Opus 5.5 --- docs/plans/001-3-longform-pipeline.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/plans/001-3-longform-pipeline.md b/docs/plans/001-3-longform-pipeline.md index 505b13d..09a7a77 100644 --- a/docs/plans/001-3-longform-pipeline.md +++ b/docs/plans/001-3-longform-pipeline.md @@ -232,10 +232,10 @@ empty finished transcript. ### Task 8: File-mode accuracy gate **Files:** Modify `tests/live-real-model.test.ts` (one new `describe` block at the end). -- [ ] Per clip: whole-file WER of the default file path (long-form with real deps if default, else +- [x] Per clip: whole-file WER of the default file path (long-form with real deps if default, else old options) ≤ `MAX_WER_BY_CLIP[clip] ?? MAX_WER`; long-form additionally ≤ old + 0.01 if it is the default. -- [ ] Run under the lock → PASS → commit "Assert whole-file WER per clip in the real-model test". +- [x] Run under the lock → PASS → commit "Assert whole-file WER per clip in the real-model test". ### Task 9: Docs and finish @@ -243,4 +243,4 @@ empty finished transcript. so, rollout decision) and `D-WS3b` (VAD model, loading, offline, live kept on the energy gate). - [x] `docs/Architecture.md` §3 layout + §4.1 flow (long-form branch) + §4.2 offline note; `docs/WorkBreakdown.md` ticks + "Noticed, not fixed"; `tests/offline.md` manual VAD step. -- [ ] typecheck, fast suite, build; real-model test under the lock; `git rm` this plan; commit. +- [x] typecheck, fast suite, build; real-model test under the lock; `git rm` this plan; commit.