From 4f6ae46815a2d4c14e81da12aadc6956505440b9 Mon Sep 17 00:00:00 2001 From: Ronald Tse Date: Sun, 6 Sep 2026 02:41:19 +0200 Subject: [PATCH] =?UTF-8?q?docs:=20GKD=20verdict=20=E2=80=94=20on-policy?= =?UTF-8?q?=20distillation=20negative=20(6.0036)?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- docs/RESULTS.md | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/docs/RESULTS.md b/docs/RESULTS.md index 4bd869f..a648aea 100644 --- a/docs/RESULTS.md +++ b/docs/RESULTS.md @@ -582,3 +582,23 @@ Consequences, stated plainly: Predictions for all five frontier runs publish alongside this entry (release frontier-predictions-v1) so the numbers above are re-derivable by anyone. + +## ara-diac-small-2-gkd — on-policy distillation NEGATIVE: 6.0036 (2026-09-06) + +The on-policy rung (GKD: student-generated mistakes scored by the +teacher, 10,995 steps, labels sha e70ce991): teacher reproduces 2.289, +student **6.0036** full-set, paired bootstrap delta 3.4083 [3.109, +3.743], p=0 — gate FAILED, and 1.43pp WORSE than the off-policy +sequence-KD rung it was meant to improve (G2a 4.5701). + +This completes the residual-attribution program with a clean pattern: +every lever tested fails to close the teacher-student gap — +* classical corpus add (G2b): flat-negative +* register swap (E6): negative +* on-policy distillation (GKD): negative, worse than off-policy +* memory layers (PKM): real but small (0.70pp), below bar +* epochs: small (0.25pp) +The residual is not data, not domain coverage, not an off/on-policy +deficit. It is a property of the compression itself at this rung — +the honest open question for the paper. On-policy stays listed in +Paper B's future work with its measured negative attached.