[**LightGBM**](https://github.com/microsoft/LightGBM) is a high-performance, gradient boosting framework by Microsoft, optimized for speed and accuracy in classification, regression, and ranking tasks. It’s designed for large-scale data and supports GPU training, categorical features natively, and distributed learning across machines. Often outperforms XGBoost in both speed and memory usage while maintaining comparable accuracy.