Mamba Explained: The State Space Model Taking On Transformers(kolaayonrinde.com)232 points by koayon 2 years ago | 80 comments
Gram Newton-Schulz: A Fast, Hardware-Aware Newton-Schulz Algorithm for Muon(tridao.me)17 points by jxmorris12 4 months ago | 3 comments
The MilkV Jupiter 2/SpacemiT K3 (RISC-V vector compute)(taoofmac.com)32 points by rcarmo 4 months ago | 6 comments
'Make invalid states unrepresentable' considered harmful(seangoedecke.com)49 points by zdw 1 year ago | 49 comments
DeepScaleR: Surpassing O1-Preview with a 1.5B Model by Scaling RL(pretty-radio-b75.notion.site)279 points by sijuntan 1 year ago | 111 comments