Commit Graph

10 Commits

Author SHA1 Message Date
wassname 5fd980244b docs: note SGTM is the latest gradient-routing paper (same authors)
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
2026-06-07 11:56:58 +00:00
wassname 637f9388c8 docs: cite SGTM paper in AGENTS.md (absorption/leakage vocab source)
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
2026-06-07 11:40:40 +00:00
wassname 52619519dc docs: drop dead refs (spec.md link, verify_gate_anchor.py paragraph)
- spec.md never existed at root or docs/; removed the link from AGENTS.md +
  README.md (the live plan is in docs/spec/ dated files).
- RESEARCH_JOURNAL.md link pointed at docs/; it lives at repo root. Fixed.
- Trimmed the no-cheat-leak paragraph citing scripts/verify_gate_anchor.py
  (that file doesn't exist); kept the general 'gate every load-bearing
  invariant in the same commit' rule.

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
2026-06-07 11:01:31 +00:00
wassname 83cae4ef72 docs: reframe no-cheat in VECTOR terms; move it README->AGENTS.md
The 'weak detector for hack A, generalize to B' framing was wrong for this repo.
That is the weak-LABEL setup (labelA -> labelNotA), which is NOT ours. Ours is
vec -> routing: vec extracted from hand-built synthetic pairs, route the live
GRPO gradient by cosine alignment to vec; no detector ever runs over student
rollouts at train time. Generalization = does vec (from pairs covering some
modes) suppress held-out modes -- vector generalization, not detector-label.

- AGENTS.md: rewrote the no-cheat bullet to the 3-way distinction (oracle grader
  = cheat; weak-label setup = not ours; vec->routing = ours). For coding agents.
- README: removed the 'We cannot cheat' section (belongs in agent instructions,
  not the new-reader overview).
- spec: dropped the stray 'validation uses known-A detector' line; pointed the
  no-cheat reference at AGENTS.md.

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
2026-06-06 02:39:48 +00:00
wassname f0cbbacaf0 save per-eval deploy-adapter ckpts (rescore w/o retrain) + CLAUDE.md test lesson
save_eval_ckpts (default on): write the deploy adapter (δS only, ~2.3MB) at each
deploy-eval step, step-tagged, so a run can be re-scored later (more prompts /
different eval) without retraining. The A5 run saved only final+first_hack, which
is why the leak needed a full retrain rather than a rescore.

AGENTS.md: every load-bearing invariant gets a verify_*.py gate. The no-cheat leak
shipped because the green gates never covered the property -- 'tests passed' is
meaningless if the property was never tested.

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
2026-06-05 03:58:26 +00:00
wassname efdf86a0cb wip 2026-05-30 04:33:33 +00:00
wassname c1f8ca4e7b tidy 2026-05-29 06:29:43 +00:00
wassname f27c658ca9 docs 2026-05-29 05:42:28 +00:00
wassname 646edfc7af purge dead modules and stale recipes
Deletes 7 source files that were superseded but never removed:
  run.py, grad_proj.py, extract_vhack.py (older twin-NLL extractor),
  grpo_smoke.py, grpo_proj_smoke.py (smoke harnesses replaced by
  train.py "smoke" subcommand), phase2_analyze.py (pilot is past),
  probe_uat.py (UAT pipeline is past).

Drops matching justfile recipes (vhack-check, phase2-analyze,
probe-uat) and the BASE constant that pointed at run.py. Updates
AGENTS/README references to the stale fast-dev-run recipe (now
just smoke / smoke-vanilla).

Verified by running just smoke-vanilla --steps=2 end-to-end.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-28 08:42:15 +00:00
wassname 120400c5f5 setup 2026-05-23 10:40:02 +08:00