• Flash Attention Notes #papers#llm#gpu

    A first-principles walkthrough of tiled attention, online softmax, and GPU memory movement.