Commit 398d12c3 398d12c3e5108e3f690fbd5ab7aaa84922b63bb5 by cnb.bofCdSsphPA

Preserve the first cap64 score before the second strategy finishes

Constraint: The cap64 run has only produced the high_energy leg so far, so any larger conclusion must wait for hybrid and the final report
Rejected: Wait for report.json before checkpointing | Would lose the verified cap64 high_energy score and the proof that execution has already switched into the hybrid branch
Confidence: high
Scope-risk: narrow
Directive: Do not compare cap64 strategy winners until both legs and the final report land; treat the current 0.625 high_energy score as an intermediate checkpoint only
Tested: Verified high_energy eval.json reports num_queries=32, top1=0.625, topk=1.0; verified progress.json records the same result; verified the active process has switched to the hybrid smoke-local branch and report.json is still absent
Not-tested: Final cap64 hybrid metrics, final report.json, and any cap64-based strategy conclusion
1 parent 3243aebb
## 2026-06-02 cap64 high_energy 首个结果 checkpoint
完成项:
- 已拿到 cap64 中 `high_energy` 的首个评测结果。
- 主流程已从 `high_energy` 切换到 `hybrid` 分支,说明 cap64 仍在继续。
验证证据:
- `high_energy/fma_reports_smoke/eval.json`
- `num_queries=32`
- `top1=0.625`
- `topk=1.0`
- `progress.json` 已同步记录同一结果。
- 当前进程显示:
- `external_adapters.py smoke-local ... /tmp/ab_smoke_seg_cap64_top2/hybrid`
- `manifest_tools.py audio-dir-to-splits ... --query-strategy hybrid`
说明:
- 截至本 checkpoint,`hybrid` 结果尚未生成,总 `report.json` 也尚未生成。
## 2026-06-02 cap64 索引完成并进入评测 checkpoint
完成项:
......
......@@ -68,3 +68,5 @@ cd /workspace/acr-engine
- 已记录 cap64 benchmark 已启动,并确认进入 `high_energy` 训练阶段。
- 已补充 cap64 新鲜证据:`high_energy` 索引完成(`64 refs / 657 windows / 192-d`)并进入 `evaluate.py`
- 已补充 cap64 首个结果:`high_energy = top1 0.625 / topk 1.0 / num_queries 32`,并记录主流程切换到 `hybrid`
......
......@@ -60,5 +60,6 @@ test -f /tmp/ab_smoke_seg_cap48_top2_seed999/report.json && cat /tmp/ab_smoke_se
## 下一轮已启动
- 新 benchmark:`/tmp/ab_smoke_seg_cap64_top2`
- 当前阶段:`high_energy` 索引已完成,现处于 evaluate 中
- 下一 session 应优先检查 `report.json` 是否生成
- 当前阶段:`high_energy` 已完成评测,结果为 `top1=0.625 / topk=1.0 / num_queries=32`
- 当前已切换到 `hybrid` 分支
- 下一 session 应优先检查 `hybrid` 结果与 `report.json` 是否生成
......
......@@ -240,7 +240,7 @@
- `hybrid``mean_top1=0.8750, min=0.7917, max=0.9583, stdev=0.0680`
### 最优先待办
1. 跟进正在运行的 cap64 benchmark:`/tmp/ab_smoke_seg_cap64_top2/report.json`
1. 跟进 cap64 的 `hybrid` 结果与最终 `/tmp/ab_smoke_seg_cap64_top2/report.json`
2. 在 cap64 完成后更新 `open-dataset-workflow.md / session-handoff.md / CHANGELOG.md`
3. 接着增加 bucket/style-aware benchmark。
4. 继续优化 `hybrid`,重点降低波动并提升 hard case 稳定性。
......@@ -675,6 +675,7 @@ seed123 最终结论:
- 已启动:`/tmp/ab_smoke_seg_cap64_top2`
- 配置:`subset_size=64`, `max_test_queries=32`, `seed=42`
- 当前最新证据:
- `high_energy` reference index 已完成:`64 refs / 657 windows / 192-d`
- 当前已进入 `evaluate.py`,尚未产出最终 `report.json`
- `high_energy` 已完成评测:`num_queries=32, top1=0.625, topk=1.0`
- cap64 主流程已切换到 `hybrid` 分支
-`report.json` 尚未生成
......