From Trial-and-Error to Improvement：A Systematic Analysis of LLM Exploration Mechanisms in RLVR 08-28
Beyond the 80 20 Rule：High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning 08-21
ETTRL：BALANCING EXPLORATION AND EXPLOITATION IN LLM TEST-TIME REINFORCEMENT LEARNING VIA ENTROPY MECHANISM 08-19
vllbc02