sciencekit

🌐 Official site: https://neurono-ml.github.io/sciencekit/ — full API reference, guides and examples.

A machine learning library in Rust reimplementing all of scikit-learn — extreme performance, zero-copy APIs and native out-of-core processing.

Highlights

• Full scikit-learn surface with idiomatic Rust builder-pattern APIs
• Zero-copy public APIs built on ndarray views and sparse sprs views
• Pluggable execution via SKExecutionMode: CPU (rayon), SIMD/BLAS, GPU (OpenCL/CUDA/ROCm), streaming/out-of-core

Architecture

flowchart LR
    subgraph Data["Data In"]
        A[("ndarray dense")]:::data
        B[("sprs sparse")]:::data
    end
    subgraph Core["sciencekit Core"]
        C["Estimators · Transformers · Metrics"]:::algo
    end
    subgraph Exec["SKExecutionMode"]
        D["CPU rayon"]:::exec
        E["SIMD / BLAS"]:::exec
        F["GPU OpenCL / CUDA / ROCm"]:::exec
        G["Streaming out-of-core"]:::exec
    end
    H["Python Bindings"]:::bind
    I["mdBook Docs"]:::doc
    A --> C
    B --> C
    C --> D
    C --> E
    C --> F
    C --> G
    D --> H
    E --> H
    F --> H
    G --> H
    C --> I
    classDef data fill:#ffd166,stroke:#b07d00,color:#3b2f00
    classDef algo fill:#06d6a0,stroke:#007a58,color:#00301f
    classDef exec fill:#118ab2,stroke:#0b5d78,color:#ffffff
    classDef bind fill:#ef476f,stroke:#9c2e49,color:#ffffff
    classDef doc fill:#8338ec,stroke:#5b1fa8,color:#ffffff

Loading

Documentation

The full documentation — API reference, usage examples and a description of every function — lives in the mdBook on the docs/documentations branch, published via GitHub Pages at https://neurono-ml.github.io/sciencekit/.

Release notes: CHANGELOG.md. Contribution workflow for agents and humans: AGENTS.md.

Status

Early development — the Cargo workspace lands with Phase 0 of the roadmap (docs/PRD.md).