jiant Karma 37 Created 1 month ago Recent Submissions 1. ▲ Show HN: We compressed a 31B video+audio diffusion model to 14B, 6 steps, INT8 (github.com) 84 points · 6 days ago · 14 comments 2. ▲ HyperFlow – one LoRA for all MiniMax-H3 tasks (ref2va/t2va/fl2va) (github.com) 2 points · 1 month ago · 0 comments All submissions on HN · View profile on HN