eli5

Why we included this project

ELI5 is a debugging library for people who need to see why a trained model made a particular call. It gives data scientists and ML engineers one Python API for inspecting classifiers and explaining individual predictions across scikit-learn, XGBoost, LightGBM, CatBoost, and Keras. For linear and tree models it reports feature weights directly; for black-box estimators it provides LIME-style local explanations and permutation importance to show which inputs actually drive outcomes. It also covers text pipelines and named-entity recognition through sklearn-crfsuite, and newer versions add token-probability visualizations for LLM outputs. When a model misbehaves and you need to know what it is actually looking at, this is a good place to start.