Skip to main content
Search

Yijun Li

1 related articles

Subscribe to updates
  1. arXiv

    Rubrics that everyone passes or fails no longer go to waste: RuACT turns them into teachers and finer critics

    In rubric-based reinforcement learning, a rubric that every sampled response fails or passes cancels out of the group-relative advantage; in Qwen3-4B medical training such inactive rubrics account for 60.88% of query–rubric observations. RuACT assigns each inactive rubric the role that supplies what it lacks: unreachable rubrics become privileged information for reward-gated on-policy self-distillation from a frozen self-teacher, and saturated rubrics are decomposed into child rubrics that separate current responses while sharing the parent's weight. On four medical and scientific benchmarks, RuACT with Qwen3-4B beats GRPO, rubric reweighting and rubric-guided exploration on every benchmark, raising ResearchQA from 64.82% to 69.44%.

You’re all caught up