Add new scripts for AntiPaSTO and GRPO validation, including v_hack extraction, held-out validation, and smoke tests

docs/RESEARCH_JOURNAL.md

@@ -2,7 +2,57 @@
 
 Append-only. New entries at the top, date-stamped. Never edit old entries.
 
-# 2026-05-23
+# 2026-05-30
+
+## Mechanism end-to-end verified on Qwen3.5-0.8B; H4 falsified at this scale
+
+Closed the smoke loop: AntiPaSTO identity (bf16, max_abs_diff=0) -> v_hack
+extraction from 15 contrastive pairs -> held-out validation (frac>0=0.952,
+median cos=+0.363, n=186 modules) -> 10-step GRPO with subprocess-executed
+LeetCode rewards on vanilla and projected arms. Full writeup in
+[out/proof.md](../out/proof.md).
+
+**Observation (mechanism)**: projected arm shows `cos_out < cos_in` every step,
+`frac_fired ≈ 0.51` averaged over 10 steps. Vanilla arm: `cos_out == cos_in`.
+The one-sided projection removes the v_hack-aligned component of the SVD-basis
+gradient when and only when alignment is positive. This is the core mechanical
+claim of the method and it is verified end-to-end.
+
+**Observation (H4 sanity)**: both arms produce zero hack_rate and zero pass_rate
+on 30 LeetCode medium/hard problems, G=2, 10 steps. Inspection of generations
+shows Qwen3.5-0.8B emits format-only output that saturates the 0.25 format
+bonus but never attempts code or hack patterns. Per [spec.md](../spec.md) §H4,
+this falls below the 30% hack-rate threshold and triggers the model-scaling
+fallback.
+
+**Inference**: 0.8B is too small to exhibit the failure mode the method
+targets. The mechanism is sound; the test substrate is not. Wu & Tang's
+Rebound paper used Qwen2.5-Coder-7B and observed ~50% baseline hack rate;
+Ariahw's benchmark assumes ≥4B class models. Mechanism + scale are
+separable concerns and the smaller scope of this session was mechanism.
+
+**Caveats / what's untested**:
+
+- β=0 (no ref-model KL) to fit 24 GB. Rebound used β=0.04. KL-free GRPO can
+  diverge faster; not a fair comparison to Rebound at this scale.
+- Only 10 steps. Reward-hacking emerges around step 50–200 in Rebound figs.
+- 186 target modules, m=8 SVD rank. Larger models scale this to ~400+ modules.
+- `frac_fired ≈ 0.5` is consistent with random gradient direction wrt v_hack
+  at init; we expect it to rise as training induces hack-aligned grads. Need
+  longer runs to see this.
+
+**Next (queued in [justfile](../justfile), pending ≥80 GB GPU)**:
+
+1. `queue-vanilla`: Qwen2.5-Coder-7B baseline GRPO on full LeetCode set, 200
+   steps, 3 seeds, β=0.04, G=4. Expected hack_rate at convergence: 40–60%
+   (Rebound table 2).
+2. `queue-projected-m16`: same config + per-module v_hack projection at m=16.
+3. `queue-rebound`: H3 baseline arm — Wu-Tang advantage modification.
+
+Confidence in method post-mechanism-verification: ~65% (was ~60%). The bump is
+small because mechanism-works was already high-prior; the real evidence is the
+7B run.
+
 
 ## Project init