Research
DreamX uses a pretrained video diffusion model as a universal latent action learner
DreamX compresses each latent video frame 300 times, from 1,200 to 4 tokens, and reports state-of-the-art results among latent action models on LARY-Bench, with zero-shot action transfer across embodiments.
Every outlet that ran it 1 article
Hugging Face PapersResearchLead
DreamX: Video Diffusion Model is a Universal Latent Action Learner