Spaces:

mazpie

/

genrl

like 3

Sleeping

I would like to request a GPU (with at least 16 GB VRAM) for my project.

The project is an academic project about multimodal foundation world models. These new class of models allows to connect the representation of multimodal foundation models with the representation of world models for reinforcement learning.

The idea is to release a demo of our pre-trained model where we show how the knowledge of the foundation model is reflect into the world, by showing a list of prompt and how the world model interprets them through reconstructions (decoded videos).

See attachment for example.