Fetching the paper…

DiG: Scalable and Efficient Diffusion Models with Gated Linear Attention · Around