Vision-Language Models are Zero-Shot Reward Models for Reinforcement Learning.

Juan Rocamonde, Victoriano Montesinos, Elvis Nava, Ethan Perez, David Lindner

Browse the full ICLR paper archive.