Google's open-source diffusion language model generates 256 tokens in parallel and self-corrects, hitting 4x speed on one GPU ...
Stuck on Easy, Medium, or Hard? We have everything you need right here.