[{"data":1,"prerenderedAt":33},["ShallowReactive",2],{"maker-tspeterkim":3},[4],{"id":5,"name":6,"line":7,"ai":8,"cat":9,"year":10,"points":11,"comments":12,"rec":13,"s":14,"top":15,"kw":16,"maker":21,"domain":22,"tier":23,"attempts":24,"strip":25,"url":26,"mix":27,"praise":30,"objection":31,"sub":32},39726781,"flash-attention-minimal","Flash Attention in ~100 lines of CUDA",false,"ai",2024,230,39,"debated",0.02,10,[17,18,19,20],"flash attention","cuda kernel","transformer attention","gpu kernels","tspeterkim","github.com","notable",0,"300000100000000000000000300000103503500","https:\u002F\u002Fgithub.com\u002Ftspeterkim\u002Fflash-attention-minimal",[28,29,24,28],2,4,"Neat, clear, minimal implementation","Line count is misleading given reliance on PyTorch codebase","ml kernels",1793991287149]