docs: GLM-5.3-Flash attribution correction — wrong haraqat, not convention

docs/PUBLICATION-NOTES.md

@@ -177,12 +177,19 @@ raw, 8.7978/6.6368 zero-skip** at reasoning_effort=low — ~3.4x worse
 DER than GLM-5.2 (2.5060/2.6911), behind Sadeed-1.5B (7.2915). Two
 protocol facts discovered live: the API rejects disabled thinking
 outright (400 code 1210; valid efforts exactly low/high/max), so the
-plain-completion protocol is inexpressible for this model; and the gap
-is orthography-shaped (Quranic dagger-alif conventions vs the GT's
-MSA; 9.8% not-fully-diacritized). Paper: leaderboard row + protocol
-note (PR #100). This strengthens the dedicated-model thesis: the
-frontier's newest generalist regressed on this classical-knowledge
-task while the dedicated 580M teacher improved.
+plain-completion protocol is inexpressible for this model. ATTRIBUTION
+(2026-09-01, corrects the first orthography-shaped reading): the gap
+is wrong haraqat — 10.05% of positions vs GLM-5.2's 2.64% (matching
+our r7 teacher's 2.62% to 0.02pp); missing 1.01%, extra 0.20%, and
+the whole dagger-alif U+0670 convention effect is 0.125pp (rules
+derived from aligned positions; controls <=0.009pp). Paper: leaderboard
+row + protocol note (PR #100, attribution corrected in PR after #69
+in rababa). This strengthens the dedicated-model thesis: the
+frontier's newest generalist lost classical-Arabic mark knowledge its
+predecessor had, while the dedicated 580M teacher improved. The
+per-position decomposition method (missing/wrong/extra +
+convention-normalization with controls) is reusable — apply to any
+future LLM row before interpreting its DER.
 
 Sibling measurement finding (same day): resumable LLM-eval
 checkpoints silently resume error sentinels — the first pass read

docs/paper.adoc

@@ -214,7 +214,7 @@ External claims are made only where a public benchmark exists and the full proto
 |our client student 1.0 (ByT5-small) |300M |8.259
 |===
 
-A protocol note attaches to the newest generalist: GLM-5.3-Flash's API rejects disabled thinking outright (HTTP 400, code 1210 — valid reasoning_effort values are exactly low/high/max), so its row carries `reasoning_effort=low`, the nearest expressible analog of the plain-completion protocol the other LLM rows used. It scores 8.7978 zero-skip / 8.5721 raw — ~3.4x the DER of its predecessor GLM-5.2 and behind Sadeed-1.5B — driven by Quranic-convention orthography (dagger-alif forms absent from the ground truth; 9.8% not-fully-diacritized words) and the reasoning floor. The frontier's newest generalist regressed on this classical-knowledge task while our dedicated 580M teacher improved; we record it rather than average it away.
+A protocol note attaches to the newest generalist: GLM-5.3-Flash's API rejects disabled thinking outright (HTTP 400, code 1210 — valid reasoning_effort values are exactly low/high/max), so its row carries `reasoning_effort=low`, the nearest expressible analog of the plain-completion protocol the other LLM rows used. It scores 8.7978 zero-skip / 8.5721 raw — ~3.4x the DER of its predecessor GLM-5.2 and behind Sadeed-1.5B. Attribution over the same 1,200 paragraphs (convention-normalized, with our teacher and GLM-5.2 as <=0.01pp controls) locates the regression in wrong haraqat: 10.05% of positions vs GLM-5.2's 2.64% — which matches our dedicated teacher's 2.62% to within 0.02pp — while missing marks run 1.01%, extra 0.20%, and the entire dagger-alif (U+0670) convention effect is 0.125pp. The frontier's newest generalist lost core classical-Arabic mark knowledge its predecessor had, while our 580M teacher improved; we record it rather than average it away.
 
 The teacher is the best dedicated (task-trained, runnable-locally) model measured under this protocol — second only to a frontier proprietary LLM's published number, ahead of a clean GLM-5.2 reproduction, Gemini Flash, and GPT-4 — and it does so at 580M parameters against Sadeed's 1.5B. The current rung (r7) adds a teacher-labeled news-domain mix to the morphological-auxiliary teacher (r6) and improves both surfaces at once: in-domain 2.5793 → 2.2864, and out-of-domain WikiNews-2024 multi-reference 19.82/12.46 → 17.38/11.83 (WER/DER) — the paragraph-specialization trade-off of the earlier lineage is gone.