Pro@programming.dev to Programming@programming.devEnglish · 4 days agoSurprisingly Fast AI-Generated Kernels We Didn’t Mean to Publish (Yet)crfm.stanford.eduexternal-linkmessage-square5linkfedilinkarrow-up121arrow-down16
arrow-up115arrow-down1external-linkSurprisingly Fast AI-Generated Kernels We Didn’t Mean to Publish (Yet)crfm.stanford.eduPro@programming.dev to Programming@programming.devEnglish · 4 days agomessage-square5linkfedilink
minus-squareSpicyToaster420@sopuli.xyzlinkfedilinkarrow-up4·3 days agoAwesome use of LLMs. I wonder they didn’t use FP8 quantization though, especially since their target hardware was an L40s.
Awesome use of LLMs. I wonder they didn’t use FP8 quantization though, especially since their target hardware was an L40s.