Wendi Li

LLM agents struggle with exploration in multi-agent settings, leading to poor coordination and increased regret, but a new framework can turn this around.

Progress advantage reveals a powerful, annotation-free scoring mechanism that outperforms traditional reward models in LLM agentic settings.

Discovering when a robot's about to fail just got easier: Hide-and-Seek pinpoints failure signals in VLA trajectories using only coarse, trajectory-level labels, ditching the need for expensive step-by-step annotations.

Ditch reward maximization: a new RL objective learns the *distribution* of reasoning advantages, boosting LLM accuracy and diversity without extra training costs.