RightNow-AI/AutoMegaKernel: An agent harness that compiles a model into one provably-correct, self-retargeting CUDA megakernel and self-tunes it past cuBLAS at batch-1 LLM decode. (opens in new tab) 聽馃惂Operating Systems 聽Content type: Code
An agent harness that compiles a model into one provably-correct, self-retargeting CUDA megakernel and self-tunes it past cuBLAS at batch-1 LLM decode. - RightNow-AI/AutoMegaKernel
Read the original article