Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning
Paper
•
2506.01939
•
Published
•
188
LLM RL
-
-
ToolOrchestra: Elevating Intelligence via Efficient Model and Tool Orchestration
Paper • 2511.21689 • Published • 124 -
PretrainZero: Reinforcement Active Pretraining
Paper • 2512.03442 • Published • 48 -
DSGym: A Holistic Framework for Evaluating and Training Data Science Agents
Paper • 2601.16344 • Published • 11