Show HN: We compressed a 31B video+audio diffusion model to 14B, 6 steps, INT8

 (github.com)

2 points | by jiant 1 hour ago

0 comments