Cong Wei
@CongWei1230
๊ฐ์
July 2023
476 ํ๋ก์ ์ค 1.2K ํฌ
Let's remove VAEs and ViTs from video models!
๐ Introducing ๐ฃ๐ถ๐
๐ฒ๐น๐จ๐ ๐ : an encoder-free unified multimodal model for image and video understanding and generation, directly in pixel space.
Code and model available today!
๐
๐
๋ ๋ณด๊ธฐ
