8월 7일까지 워킹트리에만 남아 있던 미커밋 작업을 커밋한다. 여러 사본 폴더(worktree·clone)에 흩어져 있던 중간 스냅샷을 정리하기 전에 원본을 git 이력으로 고정하는 것이 목적이다. - contracts/routes/services: measurement, outcome_trajectory, rupture_repair, deliberate_practice, calibration_transfer, supervision_research, multimodal_alliance, continuous_improvement 계열 신규 모듈과 테스트 - infra/db/init: 07~16 마이그레이션(측정 기반~calibration transfer 실행) - apps/web: 세션 리뷰 카드·관리 화면·E2E 스펙 추가 - docs/ops: G0~G8 라이브 통합·배포·롤백 증거 문서와 evidence JSON/PNG - scripts: smoke·ledger·릴리스 에이전트·NAS 프리뷰 운영 스크립트 engine.public 로그 .bak과 apps/web/test-results 산출물은 커밋에서 제외했다.
32 lines
3.1 KiB
JSON
32 lines
3.1 KiB
JSON
{
|
|
"schema_version": "vignette.multimodal-alliance-benchmark.v1",
|
|
"version": "1.0.0",
|
|
"data_classification": "synthetic_educational",
|
|
"clinical_claim_allowed": false,
|
|
"cases": [
|
|
{
|
|
"case_id":"oas-g7-bench-001","title_ko":"검증된 음성 추가 이득이 있는 bond 축",
|
|
"text_measurement":{"measurement_id":"oas-g7-measurement-text-help","axis":"bond","modality":"text","status":"ready","value":0.62,"confidence":0.82,"uncertainty":0.18,"evidence_refs":["turn-text-1"],"model_run_id":"11111111-1111-1111-1111-111111111111","error_code":null},
|
|
"voice_measurement":{"measurement_id":"oas-g7-measurement-voice-help","axis":"bond","modality":"voice","status":"ready","value":0.48,"confidence":0.75,"uncertainty":0.25,"evidence_refs":["oas-g7-event-silence-1"],"model_run_id":"22222222-2222-2222-2222-222222222222","error_code":null},
|
|
"calibration":{"calibration_id":"oas-g7-fusion-bond-help","axis":"bond","text_weight":0.7,"voice_weight":0.3,"text_only_accuracy":0.70,"fused_accuracy":0.76,"benchmark_version":"1.0.0","minimum_incremental_gain":0.01},
|
|
"target_value":0.48,
|
|
"expected_fusion_applied":true
|
|
},
|
|
{
|
|
"case_id":"oas-g7-bench-002","title_ko":"음성 추가 이득이 없어 text-only를 유지하는 task 축",
|
|
"text_measurement":{"measurement_id":"oas-g7-measurement-text-no-gain","axis":"task","modality":"text","status":"ready","value":0.72,"confidence":0.88,"uncertainty":0.12,"evidence_refs":["turn-text-2"],"model_run_id":"33333333-3333-3333-3333-333333333333","error_code":null},
|
|
"voice_measurement":{"measurement_id":"oas-g7-measurement-voice-no-gain","axis":"task","modality":"voice","status":"ready","value":0.40,"confidence":0.60,"uncertainty":0.40,"evidence_refs":["oas-g7-event-pace-2"],"model_run_id":"44444444-4444-4444-4444-444444444444","error_code":null},
|
|
"calibration":{"calibration_id":"oas-g7-fusion-task-no-gain","axis":"task","text_weight":0.8,"voice_weight":0.2,"text_only_accuracy":0.82,"fused_accuracy":0.80,"benchmark_version":"1.0.0","minimum_incremental_gain":0.01},
|
|
"target_value":0.72,
|
|
"expected_fusion_applied":false
|
|
},
|
|
{
|
|
"case_id":"oas-g7-bench-003","title_ko":"음성 평가 오류를 정상값으로 보간하지 않는 goal 축",
|
|
"text_measurement":{"measurement_id":"oas-g7-measurement-text-error-fallback","axis":"goal","modality":"text","status":"ready","value":0.67,"confidence":0.84,"uncertainty":0.16,"evidence_refs":["turn-text-3"],"model_run_id":"55555555-5555-5555-5555-555555555555","error_code":null},
|
|
"voice_measurement":{"measurement_id":"oas-g7-measurement-voice-error","axis":"goal","modality":"voice","status":"error","value":null,"confidence":null,"uncertainty":1.0,"evidence_refs":[],"model_run_id":null,"error_code":"audio_quality_unusable"},
|
|
"calibration":{"calibration_id":"oas-g7-fusion-goal-error","axis":"goal","text_weight":0.75,"voice_weight":0.25,"text_only_accuracy":0.74,"fused_accuracy":0.79,"benchmark_version":"1.0.0","minimum_incremental_gain":0.01},
|
|
"target_value":0.67,
|
|
"expected_fusion_applied":false
|
|
}
|
|
]
|
|
}
|