From ea43f161b9ef0b87edb16c5d78d8d7a448f16844 Mon Sep 17 00:00:00 2001 From: Ronald Tse Date: Sun, 25 Oct 2026 11:49:28 +0800 Subject: [PATCH] =?UTF-8?q?data(WO88):=20khm-latn=20corpus=20=E2=80=94=20t?= =?UTF-8?q?he=20secryst=20lineage=20recorded,=20retrain=20blockers=20marke?= =?UTF-8?q?d=20UNVERIFIED?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- corpora/khm-latn.yaml | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) create mode 100644 corpora/khm-latn.yaml diff --git a/corpora/khm-latn.yaml b/corpora/khm-latn.yaml new file mode 100644 index 00000000..43379ced --- /dev/null +++ b/corpora/khm-latn.yaml @@ -0,0 +1,16 @@ +id: khm-latn +task: transliteration +sources: + # WO88 migration: the secryst-era khm-latn training lineage, recorded + # with honest provenance markers — the original corpus predates the + # pinning contract. Decode evidence + the khmer-dict-spice source + # (Chuon Nath + SBBIC) are documented; the exact training tsv/tsv + # generation script lives in the secryst repo + # (scripts/export_onnx_byt5.py lineage). UNVERIFIED fields are the + # retrain blockers listed in WO68/88. + - kind: repo + id: secryst (~/src/interscript/secryst lineage) + license: UNVERIFIED + provenance: UNVERIFIED — recover the exact training corpus before any retrain +splits: + train: {path: UNVERIFIED, rows: 0, sha256: UNVERIFIED}