Live page · Day archive

LaLa-World: Scaling Latent World Models with Natural Language Actions

Read the original at HF Daily Papers

Summary

LaLa-World is a language-conditioned latent world model that learns from simulation, robot trajectories, and video data. It uses a language model to predict future states and achieves comparable planning performance with six times less data than a video generation baseline.

Carried by: HF Daily Papers. First seen: .