Practical overview
When it fits
Use LightGBM when a machine-learning workflow calls for decision-tree gradient boosting, particularly for ranking or classification, or when you want to investigate its parallel, distributed or GPU learning options.
Who it is for
- Machine-learning practitioners training tree-based models for classification or ranking.
- Teams exploring parallel, distributed or GPU-based gradient-boosting workflows.
Documented uses
- Train gradient-boosted decision-tree models for classification.
- Build ranking models using gradient boosting.
- Explore parallel, distributed or GPU learning for model training.
Requirements & limits
The project is open source under the MIT license. Its documentation links to installation instructions and package listings.
- The framework is based on decision-tree gradient boosting.
- The project documents parallel, distributed and GPU learning options.
- The repository identifies C++ as its primary language; the README also links to Python and R packages.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Build LightGBM from source on Linux with GCC
Compile LightGBM on Linux using CMake and GCC.
Before you start
- Linux
- CMake
- GCC
Steps
Clone the LightGBM repository recursively: the command below.
git clone --recursive https://github.com/lightgbm-org/LightGBMEnter the repository directory:
cd LightGBM.Configure the build:
cmake -B build -S ..Compile:
cmake --build build -j4.
Expected outcome
The Linux build produces an executable and shared library files in the LightGBM directory.
Guide limitations
- The instructions target a 64-bit build.
- The outcome is a compiled executable and library; model training is a separate stage.
Run the LightGBM CLI with a configuration file
Start LightGBM CLI using a configuration file, with optional command-line parameters.
Before you start
- LightGBM CLI installed
- A LightGBM configuration file
Steps
Prepare training data in CSV, TSV, or zero-based LibSVM format; the data may include a header.
Specify parameters in a config file using
key=valueentries.Run
lightgbm config=your_config_file other_args ....To override a parameter from the config file, pass it on the command line, for example
lightgbm config=train.conf num_trees=10.
Expected outcome
LightGBM CLI starts with the config parameters; command-line values take precedence over matching config values.
Guide limitations
- This guide does not specify a complete training configuration or confirm model output.
- Ranking requires query data; weighted training requires weight data. These may be supplied as additional data or as columns in the training data.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 12 October 2026
- Original website title
- lightgbm-org/LightGBM · Official homepage title · retained in its source language