From 94b9428d21fa8478ba21c92af54053c7c8ff3fac Mon Sep 17 00:00:00 2001 From: Ronald Tse Date: Sun, 6 Sep 2026 13:02:03 +0200 Subject: [PATCH] =?UTF-8?q?docs:=20Paper=20A=20artifacts=20=E2=80=94=20tea?= =?UTF-8?q?chers=20release,=20eval=20CLI,=20preds=20release?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- docs/paper-a.adoc | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/docs/paper-a.adoc b/docs/paper-a.adoc index 55af4fe..db7f2a0 100644 --- a/docs/paper-a.adoc +++ b/docs/paper-a.adoc @@ -254,12 +254,16 @@ Sections 4-6 depends on small-model feasibility. == 9. Artifacts -r5/r6/r7/r8 checkpoints (rababa lineage); the windowed DER-CE -evaluation harness with label-provenance sha256 recording; per-run -`final_preds.jsonl` for bootstrap replication; the full measurement -ledger (RESULTS.md) with every verdict, interval, and provenance hash. -Paper B carries the client-tier artifacts (distilled IMF v1 zips, -parity and margin reports, cross-runtime byte parity). +r5/r6/r7/r8 checkpoints — released as GitHub Release +`teachers-arabic-v1` (split archives with sha256 sidecars) together +with the two provenance-pinned teacher-label files; the windowed +DER-CE scoring protocol as the `interscript-sadeed-eval` CLI (PyPI: +`interscript-ml-tools[sadeed]`); per-run `final_preds.jsonl` (Release +`frontier-predictions-v1`) so every frontier number re-derives from +raw predictions and the public benchmark; the full measurement ledger +(RESULTS.md) with every verdict, interval, and provenance hash. Paper +B carries the client-tier artifacts (distilled IMF v1 zips resolvable +from index-v4, parity and margin reports, cross-runtime byte parity). == References