From 170950ff92955133241d4307106e4ad6b175f176 Mon Sep 17 00:00:00 2001 From: Ronald Tse Date: Sat, 29 Aug 2026 06:34:25 +0800 Subject: [PATCH] fix(distill): ara-diac-tiny -> run-005 on clean teacher_labels_v2 run-004 consumed the Aug-23 snapshot (pre decode_joined fix): its best/ decodes as UTF-8-as-Latin1 mojibake and its final_eval.json reproduces the retracted 83.08 exactly - the poisoned-data number, not a capacity result. run-005 is the clean-label re-run. --- src/gpu/modal_distill.py | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/src/gpu/modal_distill.py b/src/gpu/modal_distill.py index 19f6749..a7bd618 100644 --- a/src/gpu/modal_distill.py +++ b/src/gpu/modal_distill.py @@ -216,10 +216,15 @@ def _ensure_src_path() -> None: "unit_limits": [8000, 4000], "max_len": 1450, "label_beams": "1", - "out": "rababa_arabic_distill_tiny/run-004", - "labels_file": "/root/interscript-ml/data/ara-tiny-labels-snapshot.jsonl.gz.b64", + "out": "rababa_arabic_distill_tiny/run-005", + # run-004 used the Aug-23 snapshot = PRE decode_joined fix: every + # label mojibake (verified: its best/ decodes as UTF-8-as-Latin1, + # and its final_eval.json reproduces the retracted 83.08 exactly). + # run-005 is the actual clean-label re-run the retraction calls for. + "labels_file": "teacher_labels_v2.jsonl", + "labels_complete": "true", "mode": "sequence", - "note": "client tier (~30MB int8); r6 teacher (2.5793 DER); gate <= 3.07", + "note": "clean-label tiny re-run; r6 teacher (2.5793 DER); gate <= 3.07", }, "tha-g2p-tiny": { "teacher": "B-K/umt5-thai-g2p-v2-0.5k",