LWM documentation

LWM

Large World Model (LWM) is a general-purpose large-context multimodal autoregressive model. It is trained on a large dataset of diverse long videos and books using RingAttention, and can perform language, image, and video understanding and generation.

👍🏼 What I like

• I am impressed by the details it can capture from one hour long videos.
• I think Readme file is very clarifying.

👎🏼 What I dislike

• It does not have wiki page