From 5f8dd1c4d7106665cf9722e7a394893bb87c9b07 Mon Sep 17 00:00:00 2001 From: Yuren Hao Date: Thu, 16 Jul 2026 04:01:03 -0500 Subject: =?UTF-8?q?RESULT=2028=20caveat:=20fp32-EP=20control=20insufficien?= =?UTF-8?q?t=20=E2=80=94=20BP=20mirror=20is=20the=20decision=20metric=20(u?= =?UTF-8?q?ser=20correction)?= 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 | 6 ++++++ 1 file changed, 6 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 7542863..12751d9 100644 --- a/docs/campaign/CASCADE_ABLATION_PLAN.md +++ b/docs/campaign/CASCADE_ABLATION_PLAN.md @@ -586,6 +586,12 @@ Monotone dose-response; clean (zero guards). Readings: qcomp8/6/4 arms running (gate #1b). - Procurement impact: the Y3/pc AD7528 line is unaffected either way (its role is compute); the acceptance ENOB bar keys on the qcomp verdict. +- CAVEAT (user, 2026-07-16): the fp32-EP control conflates "quantization tax" with "EP-specific + quantization tax". BP MIRROR SUITE chained (bp_qctl + qup10/8/6 + qcomp8/6/4 on the BP twin, + same resume protocol). Decision metric = Delta_EP(bits) vs Delta_BP(bits): equal deltas => + generic quantized-training problem => the BP QAT/shadow toolbox transfers to EP unchanged. + Reading-one above ("naked cells need >=10-12b") is PROVISIONAL until the BP control lands — + naked quantized writes likely hurt BP comparably. ### RESULT 27 (2026-07-15): K-SATURATION ACROSS TRAINING (Alexi's challenge answered with data). Challenge (Alexi Gladstone): K=3 nudge sweeps seems very few; as training roughens the landscape, -- cgit v1.2.3