Finalize env_clip rollout report verification
Browse files
workspace/reports/2026-07-03.md
CHANGED
|
@@ -637,6 +637,10 @@ Action-bound audit status: new rollout artifacts now log action-bound safety lab
|
|
| 637 |
- `.venv/bin/python -m py_compile scripts/eval_ctt_generated_rollout.py`
|
| 638 |
- `.venv/bin/python -m pytest tests/test_ctt.py -q`: 12 passed.
|
| 639 |
- direct Apptainer `env_clip` rollout smoke completed and wrote measured metrics/artifacts.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 640 |
|
| 641 |
## Hugging Face Sync Status
|
| 642 |
|
|
@@ -678,6 +682,12 @@ Action-bound audit status: new rollout artifacts now log action-bound safety lab
|
|
| 678 |
- uploaded per-dim aggregates: validation commit `b198d7df479af1b21dd1b61ed544e1e044fe3431`, test commit `9d92c6ec4e93d10fc9b52c36527cab21d52db7cc`.
|
| 679 |
- uploaded per-dim seed dirs: validation seeds commits `46e23ea051ac6e950a234eaff0ca740b404c8ac6`, `f82d3994336421abf7cba7b30c2ce7c9cc119bef`, `ee520662d1dfd67e03429aa9c42c524069725573`; test seeds commits `ce7933c0e2e1561ee8d591330047db2512772adb`, `c09fae3968c98f9d1a0741c287708e6300efdb65`, `8a0e8b285272280cd6d268c1869dd0c2ec77b76b`; train seed artifacts commits `8132e1f730f3ffaa6cb7d3b6ea7b4ab283e451c9`, `49a5fc7f79deb22016613d289a2e19b8d2a2d2d4`, `132ddc494614bf899aeb1177022aff2e911e4649`.
|
| 680 |
- verified SHA-256 prefixes: `latex/main.tex` `0d1266d52de3`, `latex/main.pdf` `2018c89d5b0d`, report `fcbb15db1807`, rollout wrapper `613edd1f4fc7`, validation aggregate metrics `c709b1a2c536`, test aggregate metrics `8d13b92ef786`, validation seed0 candidates `52f991963730`, test seed0 candidates `0b4628a2ebf9`, train seed1 candidates `04bc78b3e1ba`.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 681 |
|
| 682 |
## Next Actions
|
| 683 |
|
|
|
|
| 637 |
- `.venv/bin/python -m py_compile scripts/eval_ctt_generated_rollout.py`
|
| 638 |
- `.venv/bin/python -m pytest tests/test_ctt.py -q`: 12 passed.
|
| 639 |
- direct Apptainer `env_clip` rollout smoke completed and wrote measured metrics/artifacts.
|
| 640 |
+
- Verification after adding full `env_clip` rollout evidence to the paper/report:
|
| 641 |
+
- `latexmk -pdf -interaction=nonstopmode main.tex`: PDF rebuilt successfully at 30 pages; only existing overfull/underfull layout and font warnings remain.
|
| 642 |
+
- Overclaim scan for selector-gap closure, submission-tier, leaderboard-style, and internal acceptance phrasing in `latex/main.tex`, `reports/2026-07-03.md`, `reports/draft_audit_ctt.md`, and `docs/cil_format.md` returned no matches.
|
| 643 |
+
- HF forced-download SHA verification matched local files for `workspace/latex/main.{tex,pdf}`, `workspace/scripts/eval_ctt_generated_rollout.py`, `workspace/tests/test_ctt.py`, all three `env_clip` aggregate metrics, the best train-clean `env_clip` learned-dominance metrics, and representative validation/test/train-calibration seed measured-candidate files.
|
| 644 |
|
| 645 |
## Hugging Face Sync Status
|
| 646 |
|
|
|
|
| 682 |
- uploaded per-dim aggregates: validation commit `b198d7df479af1b21dd1b61ed544e1e044fe3431`, test commit `9d92c6ec4e93d10fc9b52c36527cab21d52db7cc`.
|
| 683 |
- uploaded per-dim seed dirs: validation seeds commits `46e23ea051ac6e950a234eaff0ca740b404c8ac6`, `f82d3994336421abf7cba7b30c2ce7c9cc119bef`, `ee520662d1dfd67e03429aa9c42c524069725573`; test seeds commits `ce7933c0e2e1561ee8d591330047db2512772adb`, `c09fae3968c98f9d1a0741c287708e6300efdb65`, `8a0e8b285272280cd6d268c1869dd0c2ec77b76b`; train seed artifacts commits `8132e1f730f3ffaa6cb7d3b6ea7b4ab283e451c9`, `49a5fc7f79deb22016613d289a2e19b8d2a2d2d4`, `132ddc494614bf899aeb1177022aff2e911e4649`.
|
| 684 |
- verified SHA-256 prefixes: `latex/main.tex` `0d1266d52de3`, `latex/main.pdf` `2018c89d5b0d`, report `fcbb15db1807`, rollout wrapper `613edd1f4fc7`, validation aggregate metrics `c709b1a2c536`, test aggregate metrics `8d13b92ef786`, validation seed0 candidates `52f991963730`, test seed0 candidates `0b4628a2ebf9`, train seed1 candidates `04bc78b3e1ba`.
|
| 685 |
+
- `env_clip` targeted HF uploads completed and were verified by forced download SHA-256 comparison.
|
| 686 |
+
- code/test uploads were skipped by HF because remote files already matched the local evaluator and tests.
|
| 687 |
+
- uploaded/checked LaTeX artifacts under commit `7ea4e4ab75b963ebf889bca2f292754a3da7206f`.
|
| 688 |
+
- uploaded/checked aggregate and train-to-test dominance artifacts under commit `d78d4fe43424c713aa4dffb9bdee31f84ff178b7`.
|
| 689 |
+
- uploaded/checked validation/test/train-calibration seed rollout artifacts under commit `ee45c3c3c8c26cd1d19726ba00dd2c4a7ff4b53e`.
|
| 690 |
+
- verified SHA-256 prefixes: `latex/main.tex` `d7c8d1007d12`, `latex/main.pdf` `121e6bd969e8`, rollout evaluator `443e11c5e6ad`, `tests/test_ctt.py` `b36c71068eff`, validation aggregate metrics `92075c45030c`, held-out test aggregate metrics `a0447025c9b6`, train-calibration aggregate metrics `3ea089764ef9`, best learned-dominance metrics `88d8c26c0f31`, validation seed0 candidates `2cf3b8c30987`, test seed0 candidates `9f1bfc68f58d`, and train-calibration seed0 candidates `0ef3bf6e29b7`.
|
| 691 |
|
| 692 |
## Next Actions
|
| 693 |
|