作者位次 #5/8ACL 2026Findings· Findings of ACL
Targeted Exploration via Unified Entropy Control for Reinforcement Learning
Chen Wang · Lai Wei · Yanzhi Zhang · Chenyang Shao · Zedong Dan · Weiran Huang · Ge Lan · Yue Wang
摘要
Recent advances in reinforcement learning (RL) have improved the reasoning capabilities of large language models (LLMs) and vision-language models (VLMs). However, the widely used Group Relative Policy Optimization (GRPO) consistently suffers from entropy collapse, causing the policy to converge...