Unlocking Agentic RL Training for GPT-OSS: A Practical Retrospective
•
58
AI & ML interests
None defined yet.
Recent Activity
View all activity
Papers
Overconfident Errors Need Stronger Correction: Asymmetric Confidence Penalties for Reinforcement Learning
SemCoT: Accelerating Chain-of-Thought Reasoning through Semantically-Aligned Implicit Tokens
Articles
models 0
None public yet
datasets 0
None public yet