张宗平 cd6209119a eval(m5): full-scale retrain report — VUS-PR 0.14→0.22, model beats pure_llm ...

Full-scale retrain (stage1 31250 steps on 500k align, stage2 9375 steps on
100k sft) vs the prior small-scale run (5万/5k steps):

  metric              small-scale    full-scale
  stage2 EMA          0.81           0.75
  model VUS-PR        0.14           0.22  (+57%)
  model Aff-F1        0.21           0.25  (vs pure_llm 0.003 — TS modality helps)
  model QA mean       2.45           2.41
  QA vs pure_llm      2.8×           2.7×  (stable lead, 6 categories)

Confirms the hypothesis that weak anomaly localization was mostly a
data-volume problem, not an algorithmic one: full-scale training lifted
VUS-PR 57% and the model now clearly beats the pure-LLM baseline on
affiliation-F1 (0.25 vs 0.003). VUS-PR still trails trivial baselines
(0.22 vs random 0.33) — precise localization remains the open frontier,
but the TS modality is demonstrably contributing (was unclear before).

2026-07-01 13:44:31 +00:00

..

eval-2026-06-30.md

fix(m5): eval pipeline — anomaly_regions schema + nested JSON + pure-LLM baseline

2026-06-30 10:02:59 +00:00

eval-2026-07-01.md

eval(m5): full-scale retrain report — VUS-PR 0.14→0.22, model beats pure_llm

2026-07-01 13:44:31 +00:00

instruct_check.json

fix(m5): eval pipeline — anomaly_regions schema + nested JSON + pure-LLM baseline

2026-06-30 10:02:59 +00:00