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

(github.com) by jiant | Nov 7, 2026 | 0 comments on HN