A machine-learning framework on the kurumi engine: NumPy-ergonomic
tensors, a real nn, training, and a single-file .hodu deploy artifact. CPU and Metal.

use hodu::prelude::*;

let ctx = Ctx::cpu();                  // or Ctx::metal() on Apple Silicon
let lin = Linear::new(&ctx, 1, 1, 0);  // build the graph once, feed batches, step an optimizer, then:
save("model.hodu", &lin).unwrap();     // a self-describing, re-runnable artifact

• hodu - the ML framework: nn, optim, loss, data, .hodu artifacts
• kurumi - the tensor + autograd engine (CPU & GPU) beneath hodu
• coming soon: hodu-py (hodu for Python) and hodu-rok (hodu for rok)

_{Built by DaminStudio}