一作ACL 2026Poster· Natural Language Processing
Free Energy-Driven Reinforcement Learning with Adaptive Advantage Shaping for Unsupervised Reasoning in LLM s
Yiming Huang · Zhenbo Shi · Xin-Cheng Wen · Jichuan Zeng · Cuiyun Gao · Peiyi Han · Chuanyi Liu
摘要
Unsupervised reinforcement learning (RL) has emerged as a promising paradigm for enabling self-improvement in large language models (LLMs). However, existing unsupervised RL-based methods often lack the capacity to adapt to the model’s evolving reasoning capabilities during training. Therefore, t...