LightGBM
An open-source gradient boosting framework for classification and ranking, with MIT licensing, GPU training and support for multiple machines.
LightGBM is a decision-tree gradient boosting framework for classification, regression and ranking. It trains models on your own machine or across multiple machines, with GPU support for compatible configurations. The code is open source under the MIT license.
Python users can install it with pip install lightgbm. The framework also has an R interface and a C++ implementation. Its Python API accepts common data structures, including NumPy arrays, sparse matrices and dataframes, and can work directly with categorical features.
LightGBM focuses on efficient training and memory use for large datasets. Parallel and distributed learning support workloads that exceed a single machine, while the trained model can be saved for later prediction. Training performance depends on the data, parameters and hardware.
The official repository is maintained by lightgbm-org. Independently maintained ecosystem projects provide additional compilation, conversion and deployment options; those integrations have their own requirements.
