From 185c0dc51e9fbeda3a36e84bf659af0d105c80e0 Mon Sep 17 00:00:00 2001 From: Yuren Hao Date: Mon, 20 Jul 2026 06:33:50 -0500 Subject: =?UTF-8?q?RESULT=2053:=20=CE=B2=E6=95=99=E6=9D=A1=E9=87=8D?= =?UTF-8?q?=E5=BB=BA=20=E2=80=94=20Codex=20=CF=81=E6=8E=A2=E9=92=88?= =?UTF-8?q?=E6=B5=8B=E5=87=BA=E7=9C=9F=E5=AE=9E=E4=B8=8A=E7=95=8C=CE=B2*?= =?UTF-8?q?=E2=89=880.7@s35000(0.4=E6=94=B6=E6=95=9B/0.7=E5=8F=91=E6=95=A3?= =?UTF-8?q?);=20v1=E7=82=B8@0.09=E5=A4=84=CF=81=3D0.73=E8=BF=9C=E6=9C=AA?= =?UTF-8?q?=E8=B6=8A=E7=95=8C=3D=E6=8E=A7=E5=88=B6=E5=99=A8=E8=87=AA?= =?UTF-8?q?=E7=82=B8=E9=9D=9E=CE=B2=E4=B8=8D=E7=A8=B3;=20=E5=9B=BA?= =?UTF-8?q?=E5=AE=9A=CE=B2=E6=B0=B8=E4=B8=8D=E7=82=B8(=E9=81=87=E4=B8=8B?= =?UTF-8?q?=E6=B2=89=E4=B8=8A=E7=95=8C=E2=86=92skip=E9=98=B2=E5=BE=A1,plai?= =?UTF-8?q?n2=E6=94=B6=E5=AE=9811527skip=E4=BB=8D3.3437=E5=AE=8C=E8=B5=9B)?= =?UTF-8?q?,=E5=8F=AA=E6=9C=89=E6=8E=A7=E5=88=B6=E5=99=A8=E7=82=B8;=20?= =?UTF-8?q?=E7=95=8C=E5=86=85CE=E5=B9=B3=E5=9D=A6+=E5=81=8F=E7=BD=AE?= =?UTF-8?q?=E2=88=9D=CE=B2+=E4=B8=8A=E7=95=8C=E4=B8=8B=E6=B2=89=E2=86=92?= =?UTF-8?q?=E6=94=BF=E7=AD=96=3D=E5=9B=BA=E5=AE=9A=E5=B0=8F=CE=B2,?= =?UTF-8?q?=E6=8E=A7=E5=88=B6=E5=99=A8=E7=BA=BF=E6=8B=86=E9=99=A4(?= =?UTF-8?q?=E9=94=99=E8=AF=AF=E4=B8=94=E5=8D=B1=E9=99=A9=E7=9A=84=E8=BD=B4?= =?UTF-8?q?,=E7=94=A8=E6=88=B7=E5=88=A4=E5=AE=9A=E6=88=90=E7=AB=8B);=201e-?= =?UTF-8?q?3=E5=85=A8=E7=A8=8B=E9=AA=8C=E8=AF=81=E5=B7=B2=E6=8E=92?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Co-Authored-By: Claude Fable 5 Claude-Session: https://claude.ai/code/session_014FAPDWQ49M5Ye3NpTndTpn --- docs/campaign/CASCADE_ABLATION_PLAN.md | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+) (limited to 'docs/campaign/CASCADE_ABLATION_PLAN.md') diff --git a/docs/campaign/CASCADE_ABLATION_PLAN.md b/docs/campaign/CASCADE_ABLATION_PLAN.md index ada9379..f0a82be 100644 --- a/docs/campaign/CASCADE_ABLATION_PLAN.md +++ b/docs/campaign/CASCADE_ABLATION_PLAN.md @@ -568,6 +568,36 @@ direction), not training-under-fault; wave-2 = co-training with faults injected +### RESULT 53 (2026-07-20): THE β DOGMA REBUILT — ceiling is REAL (β*≈0.7 @s35000) but the v1 +collapse did NOT cross it; fixed β never explodes (only the controller does); policy = FIXED SMALL β. +Codex ρ-probe (probe_cx2_rho, s35000, one relax, no training): ρ = res_k/res_{k-1} vs β: +3e-2→0.36, 9e-2→0.73, 0.2→0.93, 0.4→0.95(all converge), 0.7→1.04 (first DIVERGENCE). Stability +ceiling β* bracketed 0.4<β*<=0.7 in the TRAINED state. So: +- CEILING IS REAL (user's dichotomy half 1 ✓) but at 0.7, NOT 0.09. +- v1 blew at β~0.09 where ρ=0.73 — nowhere near β*. hibeta fixed-9e-2 arm ran 10.7k steps ZERO + skips, normal learning (runs/hibeta_9e-2.log) — confirms 0.09 is comfortably stable. v1 death = + controller self-excitation (β×392=1.01^600) + gate bug + 30k boundary-riding damage, NOT a β + instability (user's dichotomy half 2 ✓). BOTH halves true; v1 is half 2. +- FIXED β NEVER EXPLODES: hits the (sinking) ceiling → triggers skips (defense works). plain2 + fixed-3e-3 full 234k: 11527 endgame skips yet best 3.3437 clean completion. ONLY the controller + explodes. => the entire adaptive-controller line (servo/ride/wsync/simple) was the wrong AND + dangerous axis; drop it. +- CEILING SINKS with training: β*=0.7 @s35000 (mid), but endgame ceiling ~3e-3 (plain2's 11527 + skips at fixed 3e-3 = the sunk ceiling biting). So no single fixed β is "safe" in the sense of + never skipping — but skipping ≠ exploding. +- β-CURVE FLAT (RESULT bcurve, s35000+10k, controller-free): 1e-3→3.6305, 3e-3→3.6207, + 1e-2→3.6255, 3e-2→3.6133. Δ0.017 over 30× = seed noise. NO SNR benefit to large β in the digital + sim (consistent with the BBP no-floor finding). Large β only COSTS (single-sided bias ∝β, + Laborieux; less headroom below the sinking ceiling). +POLICY (corrected — my earlier "fixed large β" was backwards, user caught it): three constraints +(bias∝β↓, sinking-ceiling↓, SNR-floor absent in digital per BBP) all point SMALL. Use fixed small +β; controller removed. Digital ceiling never binds via explosion, only via defensive skips that a +small β minimizes. HARDWARE is the only regime with a real SNR floor (additive readout noise → +β*_HW by ENOB) — a separate calc, not settable a priori. +VALIDATION: fw72m_b1e3 (fixed 1e-3, full 234k, controller-free) queued vs plain2 fixed-3e-3 +(best 3.3437). hibeta 0.2/0.4 arms still measuring slow-accumulation (hypothesis B) at ρ≈0.95. +plain2 crown = current best 1.0x-cost result at 3.3437 (vs cent 3.3318 @1.67x). + ### RESULT 52 (2026-07-19): 42M 3x3 MATRIX COMPLETE — cent's "zero-gap" was LUCKY-SEED; estimator choice does NOT move final CE. cent seeds: 1.2334(s1)/1.2716/1.2579 mean 1.2543. Full table: BP 1.2141 | ride 1.2449 | cent 1.2543 | plain 1.2558 — the three EP recipes are -- cgit v1.2.3