this is a really cool paper and well written, worth a read, but it really didn't do anything to disabuse my feeling that diffusion is the wrong inductive bias for language modeling.like look at this…

Posters: voooooogel. voooooogel

5 likes recorded by the source corpus.