GRPO-LEAD: A Difficulty-Aware Reinforcement Learning Approach for Concise Mathematical Reasoning in Language Models.

Jixiao Zhang, Chunsheng Zuo

Browse the full EMNLP paper archive.