🧑🔬 Shenzhi Wang — papers of 2025 — 1 paper, the most cited first1 paper · all years of Shenzhi WangBeyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM ReasoningConference on Neural Information Processing Systems (NeurIPS)Shenzhi Wang, Le Yu, Chang Gao, et al.2025 · 152 citationsThis page in Markdown → https://md.research.elamur.ai/free/en/authors/shenzhi-wang/year/2025/