Learning from Reward-Free Offline Data: A Case for Planning with Latent Dynamics Models

no code implementations • 20 Feb 2025 • Vlad Sobal, Wancong Zhang, Kynghyun Cho, Randall Balestriero, Tim G. J. Rudner, Yann Lecun

On the control side, we train a latent dynamics model using the Joint Embedding Predictive Architecture (JEPA) and use it for planning.

Reinforcement Learning (RL) Zero-shot Generalization