MRACL: Multi-Reward Space Guided Adaptive Curriculum Reinforcement Learning for LLMs.

Wenxuan Liu Liangyu Huo Yi Jing Xiyuan Zhang Jian Xie MRACL: Multi-Reward Space Guided Adaptive Curriculum Reinforcement Learning for LLMs. 37663-37672 2026 AAAI https://doi.org/10.1609/aaai.v40i44.41101 conf/aaai/2026 db/conf/aaai/aaai2026.html#LiuHJZX26