ETTRL: Balancing Exploration and Exploitation in LLM Test-Time Reinforcement Learning Via Entropy Mechanism
机构 * Beihang University(北京航空航天大学) ; Northwestern Polytechnical University(西北工业大学) ; Kuaishou Technology(快手科技)
专题命中 推理评测 :reasoning(abstract);分类 cs.AI、cs.LG