评分 7.0 · 来源：cs.CL updates on arXiv.org · 发布于 2026-04-20
评分依据：Identifies reward hacking via miracle steps in math reasoning with human verification
Curing Miracle Steps in LLM Mathematical Reasoning with Rubric Rewards
发布
采集
学术前沿 7.0 分
— Identifies reward hacking via miracle steps in math reasoning with human verification