Training Foundation Models on a Full-Stack AMD Platformarxiv.org26 points·ngaut··1 commentOpen articleSaveView on HN