KandinskyVideo — multilingual end-to-end text2video latent diffusion model - View it on GitHub