Documentation
Welcome to Raven's documentation. Raven is an ecosystem of OCaml libraries for numerical computing, machine learning, and data science.
Start Here
- Quickstart — zero to gradient in 5 minutes
- Coming from Python — map NumPy, PyTorch, pandas concepts to Raven
- Ecosystem Overview — how the libraries relate and which to use
Libraries
| Library | Like | What it does | |
|---|---|---|---|
| nx | NumPy | N-dimensional arrays with pluggable backends | |
| ᚱ | rune | JAX | Automatic differentiation and functional transformations |
| ᚲ | kaun | PyTorch / Flax | Neural networks and training |
| ᚹ | vega | Optax | Gradient-based optimizers and schedules |
| ᚨ | brot | HF Tokenizers | Fast tokenization for language models |
| ᛃ | talon | Pandas / Polars | DataFrames with type-safe columns |
| ᛞ | hugin | Matplotlib | Data visualization and plotting |
| ᛈ | quill | Jupyter + IPython | Interactive REPL and markdown notebooks |
| ᚠ | fehu | Gymnasium | Reinforcement learning environments |
| ᛋ | sowilo | OpenCV | Differentiable computer vision |
Project
- Installation — system dependencies, opam setup, building from source
- Roadmap — what works today and what's coming
- Updates — progress reports on Raven's development
- Introduction — vision and philosophy
- Support Raven — sponsorship and contributing