Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning
Paper • 2506.01939 • Published • 190
Xiwei Pan
isPANN
AI & ML interests
None yet
Recent Activity
authored a paper about 14 hours ago
Problem Reductions at Scale: Agentic Integration of Computationally Hard ProblemsOrganizations
None yet
upvoted 4
papers 6 months ago
DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree Search
Paper • 2509.25454 • Published • 148
The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain
Paper • 2509.26507 • Published • 550
Less is More: Recursive Reasoning with Tiny Networks
Paper • 2510.04871 • Published • 513