1 paper
Jiarui Yang, Jiawei Li, Jiale Zhang +5
Latent action models (LAMs) learn action-related representations from action-free videos via self-supervised future prediction, offering a scalable paradigm for embodied intelligen…