Tribuo

What it does

Tribuo is a machine learning library written in Java for building and deploying models. It takes labelled datasets, such as CSV or LibSVM data, and produces trained models that make predictions and produce evaluation reports, including accuracy, recall and F1 scores. Its models consume Examples and return typed Predictions rather than raw float arrays.

Developers use Tribuo to load data, split it into training and test sets, train models with different trainers, and evaluate them in a few lines of Java. Tribuo's distinct features include provenance records that capture the parameters, transformations and files used to build each model, and strong typing that stops mismatched models from being used. It also wraps third-party libraries such as XGBoost and liblinear and supports ONNX model exchange, so models trained in Python can be deployed from Java.

Tribuo suits Java developers and data scientists who want ML inside Java applications. It is distributed under the Apache 2.0 licence and added to a project through Maven or Gradle. It runs on Java 8+, and native-code interfaces are only supported on Windows 10, macOS and Linux on x86_64 architectures. The model card and reproducibility packages require Java 17 and are not included in the tribuo-all Maven artefact.

What it can do

• Classification with linear models, SVMs, trees, ensembles and deep learning
• Regression with linear, penalised linear, factorization machine and tree models
• Clustering with HDBSCAN and K-Means
• Anomaly detection with SVMs and multi-label classification
• Model and evaluation provenance for reproducible rebuilds
• ONNX model import and export for cross-language deployment

Works with

• XGBoost
• liblinear
• libsvm
• TensorFlow
• ONNX Runtime
• scikit-learn
• PyTorch

Example workflows

Train and evaluate a classifier

Illustrative example. Input: a labelled CSV dataset such as the Iris data. Steps: load the data with CSVLoader, split 70/30 with TrainTestSplitter, train a LogisticRegressionTrainer, then run LabelEvaluator on the test set. Output: a formatted evaluation with per-class recall, precision, F1 and overall accuracy.

Reproduce a model from provenance

Illustrative example. Input: a trained model with its serialised provenance object. Steps: extract the provenance to JSON to inspect the trainer hyperparameters, data source and transformations. Output: a record of the exact configuration needed to rebuild the model from scratch.

Before you choose

• Native-code interfaces to XGBoost, TensorFlow and ONNX Runtime are supported only on x86_64 Windows 10, macOS and Linux; other platforms need the native library developers.
• The model card and reproducibility packages require Java 17 and are excluded from the tribuo-all Maven Central deployment.
• Tribuo's tutorial notebooks use the IJava kernel and Java 10+; converting them to Java 8 requires replacing var with explicit types.

Listed under

• Coding
• Classification
• Clustering
• Developer library
• Java
• Machine learning
• Open source

Get started

Add tribuo-all 4.3.2 via Maven or Gradle, or clone the repository.

1. Add the tribuo-all 4.3.2 pom dependency to a Maven project, or the equivalent Gradle implementation line with transitive set to true.
2. Load a dataset with a DataSource such as CSVLoader or LibSVMDataSource and convert it to a MutableDataset.
3. Train a model with a trainer such as LogisticRegressionTrainer, then evaluate it with LabelEvaluator on held-out data.

Tribuo’s Apache-2.0-licensed code is free to use. You supply the Java environment, training data and compute; any external service costs are separate.

Tribuo is free to use. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Machine Learning in Java - Tribuo: Machine Learning in Java ↗tribuo.org
• Docs - Tribuo: Machine Learning in Java ↗tribuo.org

Company website

tribuo.org: product, pricing and policy pages

Web archive

When the site first appeared

YC directory mirror

Batch and company name

Web search

Confirms the official site

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.