Machine Learning Primitives
Whenever I go to use a new machine learning algorithm / architecture / technique, I normally write it out myself first using plain Rust and NDArray. Doing all the differentiation and grunt work helps me appreciate this stuff more, and I like that nothing is hidden. I figured I would collect these handwritten architectures as I go. This is not at all a library, but moreso a set of primitives / a toolbox I'll be revisiting from time to time.
Sections
envs: Various simulators used to test stuffnn: Core neural net components, stuff like basic FFN, Attention, Layernorm, CTRNN, etc. TBD some experimentals in thenn/experimentaldir.optim: Gradient collection and training, AdamW, SGDrl: A couple RL algos, SAC is still a bit sketch, but PPO & TD3 are solid.util: Graphing and benchmarkingf: Function set