silencelamb

naked_llama. build llama inference compute from scrath, only using torch/numpy base ops

VibeKernel. Benchmarking coding-agent paradigms (plain prompt, /goal, skills, autoresearch...) on one fixed task: hand-writing the fastest fp16 GEMM kernel for the NVIDIA A100.

alpa. Auto parallelization for large-scale neural networks

tensorflow-alpa. C++