From 52e204d85884fa0e53b117024c03b68771a4c2b7 Mon Sep 17 00:00:00 2001 From: Ronald Tse Date: Mon, 31 Aug 2026 10:49:01 +0200 Subject: [PATCH] =?UTF-8?q?docs:=20G2a=20full-set=204.5701=20=E2=80=94=20a?= =?UTF-8?q?ttribution=20stands,=204th=20overstatement=20instance?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- docs/RESULTS.md | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/docs/RESULTS.md b/docs/RESULTS.md index 6500740..798d7da 100644 --- a/docs/RESULTS.md +++ b/docs/RESULTS.md @@ -435,3 +435,16 @@ residual attribution**: the 2.25pp residual was mostly optimization (undertraining), not domain coverage. Full-set gate running - given three prior subset-overstatement instances, the honest number waits there. Provenance: labels sha256 e70ce991 (137.2MB) recorded. + +## ara-diac-small-2-6ep — full-set verdict: 4.5701; the discipline vindicated (2026-08-31) + +Full 1,200-paragraph gate (teacher in-run 2.289): student **4.5701**, +gate_delta 2.28. Doubling epochs bought 0.25pp over the 3-epoch 2.0 +rung (4.8218 -> 4.5701) - the residual is NOT mostly undertraining; +the E2/E3 domain attribution substantially stands (~0.25pp epochs, +~2.0pp domain/other). The subset had said 2.0062 / delta 0.72 - a +near-overturn of the decomposition that the full set corrects: the +FOURTH subset-overstatement instance and the most dramatic (subset +delta 0.72 -> full delta 2.28, a 3.2x inflation). The full-set-only +publication rule earns its keep here: this entry is the paper's +strongest measurement-discipline exhibit.