Refresh handoff with latest live FMA smoke checkpoint
Persist a newer runtime checkpoint so restart docs continue to prove that the real FMA smoke is still progressing inside Epoch 1 without yet saving a model or entering downstream stages. Constraint: Verification is still limited to live runtime evidence because Epoch 1 has not completed Rejected: Reuse the prior 22:10 checkpoint | would leave handoff docs behind the latest verified state Confidence: high Scope-risk: narrow Directive: Keep monitoring until the first saved model file or stage transition appears Tested: ps on PID 311629; validate-splits on /tmp/fma_real_smoke_stopcheck/fma/manifests; find on /tmp/fma_real_smoke_stopcheck/fma_models_smoke Not-tested: End-of-epoch artifacts, build-index, evaluate, final metrics
Showing
3 changed files
with
46 additions
and
0 deletions
| 1 | ## 2026-06-02 真实 FMA smoke fresh evidence 22:58 checkpoint | ||
| 2 | |||
| 3 | 完成项: | ||
| 4 | - 再次检查真实 FMA smoke 运行态,确认 `train.py` elapsed 已推进到 22:58。 | ||
| 5 | - 更新 `docs/session-handoff.md` 与 `docs/changelist-2026-06-02.md`,同步更晚的 live evidence。 | ||
| 6 | |||
| 7 | 验证结果: | ||
| 8 | - `ps -p 311629 -o pid,etime,%cpu,%mem,cmd` => `ELAPSED=22:58` | ||
| 9 | - 仍未出现 `build-index/evaluate` 相关新进程 | ||
| 10 | - `validate-splits /tmp/fma_real_smoke_stopcheck/fma/manifests` => `ok=true` | ||
| 11 | - `fma_models_smoke/` 仍仅有目录本身 | ||
| 12 | |||
| 13 | 结论: | ||
| 14 | - 真实 FMA 全量 smoke 仍在 epoch 内推进,没有中断迹象。 | ||
| 15 | - 到该时点仍未产生首个模型文件或下游阶段切换证据。 | ||
| 16 | |||
| 1 | ## 2026-06-02 真实 FMA smoke fresh evidence 22:10 checkpoint | 17 | ## 2026-06-02 真实 FMA smoke fresh evidence 22:10 checkpoint |
| 2 | 18 | ||
| 3 | 完成项: | 19 | 完成项: | ... | ... |
| ... | @@ -192,3 +192,11 @@ cd /workspace/acr-engine | ... | @@ -192,3 +192,11 @@ cd /workspace/acr-engine |
| 192 | - 当前 CPU / 内存观测:`%CPU≈615`, `%MEM≈10.7`。 | 192 | - 当前 CPU / 内存观测:`%CPU≈615`, `%MEM≈10.7`。 |
| 193 | - 仍未出现模型文件,也未切换到 `build-index/evaluate`。 | 193 | - 仍未出现模型文件,也未切换到 `build-index/evaluate`。 |
| 194 | - manifest 复核继续通过,统计保持不变。 | 194 | - manifest 复核继续通过,统计保持不变。 |
| 195 | |||
| 196 | |||
| 197 | ## 12:20 UTC 时间推进补充 | ||
| 198 | |||
| 199 | - 最新 live 证据已推进到:`train.py ELAPSED=22:58`。 | ||
| 200 | - 当前 CPU / 内存观测:`%CPU≈615`, `%MEM≈10.8`。 | ||
| 201 | - 仍未出现模型文件,也未切换到 `build-index/evaluate`。 | ||
| 202 | - manifest 复核继续通过,统计保持不变。 | ... | ... |
| ... | @@ -233,6 +233,28 @@ | ... | @@ -233,6 +233,28 @@ |
| 233 | - 当前依旧处于第 1 个 epoch 内部的持续训练阶段。 | 233 | - 当前依旧处于第 1 个 epoch 内部的持续训练阶段。 |
| 234 | - 到 12:19 UTC 为止,仍没有首个模型文件或下游检索/评测阶段证据。 | 234 | - 到 12:19 UTC 为止,仍没有首个模型文件或下游检索/评测阶段证据。 |
| 235 | 235 | ||
| 236 | ### 持续推进的 fresh evidence(2026-06-02 12:20 UTC) | ||
| 237 | |||
| 238 | - 真实 FMA smoke 继续推进到: | ||
| 239 | - `train.py ELAPSED=22:58` | ||
| 240 | - `%CPU≈615` | ||
| 241 | - `%MEM≈10.8` | ||
| 242 | - 当前进程结构仍未发生阶段切换: | ||
| 243 | - `PID=311494`:`external_adapters.py smoke-local fma ...` | ||
| 244 | - `PID=311629`:`train.py --data /tmp/fma_real_smoke_stopcheck/fma/manifests ...` | ||
| 245 | - 仍未出现 `build-index` / `evaluate` 相关新进程。 | ||
| 246 | - `fma_models_smoke/` 仍只有目录本身,没有模型文件。 | ||
| 247 | - manifest 再次复核仍通过: | ||
| 248 | - `ok=true` | ||
| 249 | - `catalog_references=8000` | ||
| 250 | - `train_queries=6401` | ||
| 251 | - `test_queries=1593` | ||
| 252 | - `val_queries=0` | ||
| 253 | |||
| 254 | 这说明: | ||
| 255 | - 当前依旧处于第 1 个 epoch 内部的持续训练阶段。 | ||
| 256 | - 到 12:20 UTC 为止,仍没有首个模型文件或下游检索/评测阶段证据。 | ||
| 257 | |||
| 236 | ### 重启后第一优先级动作 | 258 | ### 重启后第一优先级动作 |
| 237 | 259 | ||
| 238 | 1. 先检查真实 FMA smoke 是否完成: | 260 | 1. 先检查真实 FMA smoke 是否完成: | ... | ... |
-
Please register or sign in to post a comment