ГЛОБАЛЬНЫЙ РЕФАКТОРИНГ АРХИТЕКТУРЫ - Stage 4.7: внедрить stage4_contract_compliance_rate в P0 quality-gap и baseline gate

This commit is contained in:
2026-04-12 09:32:27 +03:00
parent ce1ebae8ec
commit a49212608f
8 changed files with 75 additions and 7 deletions
+19 -1
View File
@@ -2728,7 +2728,25 @@ Implemented in current pass (Stage 4.6 eval metric for answer-contract complianc
- `assistantEvalHarness.test.ts`: `6 passed` (run with extended timeout budget);
- `npm --prefix llm_normalizer/backend run build` passed.
Status: In progress (Stage 4.1-4.6 completed; continue with focused wave/manual-comment quality backlog)
Implemented in current pass (Stage 4.7 Stage4 compliance rollout into P0 quality-gap loop, 2026-04-12):
1. Extended P0 quality-gap metric contract:
- added `stage4_contract_compliance_rate` to `P0QualityGapMetricVector`;
- added metric definition in `P0_QUALITY_GAP_METRIC_DEFINITIONS`.
2. Extended P0 quality-gap thresholds and baseline gate checks:
- added `stage4_contract_compliance_rate_min` to `P0QualityGapThresholds`;
- default threshold set to `0.9`;
- `buildQualityGapChecks(...)` now includes Stage4 compliance check.
3. Integrated Stage4 compliance metric computation in P0 eval runner:
- source: `buildStage4AnswerContractAuditV1(assistant_reply).is_stage4_shape`;
- included in `quality_gap_metrics.raw`;
- added denominator `stage4_contract_audited_cases_total`.
4. Regression updates:
- `assistantP0EvalHarness.test.ts` updated for new quality-gap metric key.
5. Validation snapshot:
- `assistantP0EvalHarness.test.ts`: `5 passed` (extended timeout budget);
- `npm --prefix llm_normalizer/backend run build` passed.
Status: In progress (Stage 4.1-4.7 completed; continue with focused wave/manual-comment quality backlog)
## Stage 5 (P3): Quality Loop Driven By GUI Markup