AutoMegaKernel: Compiling a LLM into a single CUDA kernel

(arxiv.org) by OsamaJaber | Jun 9, 2026 | 0 comments on HN