| ▲ | ACCount37 3 hours ago | |
They can be base models for a bunch of things. Turning text-conditioned video generation models into robotics VLAs is a fun exercise. This one is probably too small to be useful for that, and not diverse enough? But I could be wrong. | ||