BayesWitnesses/m2cgen

Health v2 · maintenance only

Adoption not part of the score

2998 stars · 264 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

m2cgen is a lightweight Python library that transpiles trained statistical machine learning models into native code in languages such as Python, C, Java, Go, JavaScript, Rust, and more. It enables zero-dependency model inference by generating standalone scoring code from scikit-learn, XGBoost, LightGBM, StatsModels, and lightning models.

Use cases

• deploy ML models in languages without Python runtimes
• generate dependency-free scoring code from a trained sklearn model
• transpile an XGBoost model to Java or C for embedded inference
• export a LightGBM classifier to Go or Rust source code
• run model inference in environments where installing ML libraries is impossible
• convert a trained regression model to JavaScript for browser-side prediction

When to choose

• you need to serve a trained model in a language without ML library support
• you want zero-dependency, low-latency inference without shipping heavy ML runtimes
• you target embedded or constrained environments where only native code can run

When to avoid

• you need to serve deep learning or neural network models
• you want a full model serving platform with APIs and monitoring
• your model uses estimators not in the supported list

Facets

library · maturity maintenance

machine-learning compiler developer-tools machine-learning developer-tools programming-languages python cross-platform transpilation code-generation scikit-learn xgboost lightgbm zero-dependency-inference

Member repositories

For agents

markdown · JSON · MCP: product_card(name="BayesWitnesses/m2cgen")