Kevin Frans(@kvfrans):Here's a fun one with @preston_fu et al -- how should we reward RL policies in a way that scales to longer and more difficult tasks? Our answer lies in-between RL and imitation learning, and provides a simple way to assign dense per-token credit to long trajectories.

Kevin Frans

@kvfrans

phd @berkeley_ai prev mit, reflection, openai read my thoughts:

加入 August 2013

518 正在關注 4.3K 粉絲

Kevin Frans@kvfrans

2026.09.08 18:52

Here's a fun one with @preston_fu et al -- how should we reward RL policies in a way that scales to longer and more difficult tasks? Our answer lies in-between RL and imitation learning, and provides a simple way to assign dense per-token credit to long trajectories.