Diffusion text models get 3x faster
A small helper model learns the right order to fill in text, cutting diffusion language model inference costs by up to 6.8x.
A small helper model learns the right order to fill in text, cutting diffusion language model inference costs by up to 6.8x.