# microsoft/hummingbird

Hummingbird compiles trained ML models into tensor computation for faster inference.

Repository: https://github.com/microsoft/hummingbird
Canonical: https://ross.abutalabs.com/products/hummingbird
Language: Python
License: MIT
License Family: permissive
Topics: machine-learning, neural-networks, scikit-learn, pytorch, tensor-computation
Last push: 2025-07-17T17:52:29+00:00

## Health v2 (maintenance only)
Score: 37/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 32, release rhythm 8, longevity 100
- inputs: {"age_days": 2365, "days_push": 412, "days_rel": 678, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3544, forks 290 (observed 2026-08-28T04:08:09.362889+00:00)

## What it is
Hummingbird is a Microsoft library that compiles trained traditional ML models (decision trees, random forests, XGBoost, LightGBM) into tensor computations for PyTorch, TorchScript, ONNX, and TVM. It lets users accelerate inference with GPU and neural network framework optimizations without re-engineering their models.

## Use cases
- speed up scikit-learn model inference on GPU
- convert XGBoost or LightGBM models to PyTorch
- export decision trees to ONNX for deployment
- serve sklearn models with TorchServe
- unify traditional and neural network model inference
- accelerate random forest predictions with tensor ops

## When to choose
- you need faster inference for tree-based models on GPU
- you want a single serving platform for both sklearn and neural network models
- you want to export traditional ML models to TorchScript or ONNX

## When to avoid
- your models are already neural networks
- you need CPU-only inference with no performance bottleneck
- you use ML operators not in Hummingbird's supported list

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, compiler, llm-inference
- domain: machine-learning, deep-learning, developer-tools
- platform: python, cross-platform
- tags: model-compilation, scikit-learn, pytorch, onnx, tvm, inference-optimization, tree-models, xgboost, lightgbm, gpu

## Member repositories
- microsoft/hummingbird (main) score 37

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:09.362889+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:34:26.208524+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/hummingbird (fetched 2026-08-28T04:08:09.362889+00:00, sha ff5874046dc1)
- Data as of 2026-08-30T08:39:29.467469+00:00.