Inference Optimization : GEMM

If you look under the hood of any modern LLM, past the attention mechanisms and layer norms, it basically boils down to one thing: Matrix… Continue reading on Medium »