← back to paper

Review history

arxiv: 2605.22148 · 2 revisions

Ratchet: How Reliable Must an LLM Judge Be to Retire a Skill?

1. 2026-08-02 CONDITIONAL MODERATE v1.3.0-alltime-deepseek novelty 6.0
   153482 ms 14076 in 15204 out 2026-08-02T13:24:55.873642+00:00
2. 2026-05-22 CONDITIONAL LOW v0.9.0 novelty 6.0
   38823 ms 5856 in 1663 out 2026-05-22T06:22:50.867976+00:00