Coding guide to NVIDIA's cuTile, Triton, and Flash Attention

Covers building a Colab workflow with Triton kernels and Flash Attention, falling back from cuTile when GPUs lack support. Includes practical code examples and environment probing.