High-performance, scalable gradient-boosted decision tree library for regression, classification, ranking and custom objectives. Multi-language bindings (Python, R, Java, Scala, C++), single-node, distributed and GPU training — widely used for tabular data and ML competitions.
