The Learning Interpretability Tool (LIT) provides a visual interface for analyzing ML models to understand their predictions and failure modes. It supports salience maps for local explanations, aggregate metrics with slicing and binning, counterfactual generation, and side-by-side model comparison. The tool is framework-agnostic, working with TensorFlow, PyTorch, and other ML frameworks, and can run as a standalone server or embedded in notebook environments like Jupyter and Google Colab.