zjunlp / AutoSciRub

Learning to Evaluate Before Improving: Automatic Rubric Induction for Automatic Research Agents