An experimental type-safe machine learning library. Implementing concepts from scratch is cool, so that's what I'm doing here.
Currently, I'm working on building a somewhat fast1 Rust library from the ground up (yes, no external math libraries) with API design and "parse, don't validate" as a large focus.
- Multi-layer perceptrons
- Convolutional neural networks
- Automatic differentiation
And possibly a lot more in the future
Footnotes
-
yes. and also we don't talk about Rust compile times ↩