Captum · Model Interpretability for PyTorch Introduction
What is Captum · Model Interpretability for PyTorch
Captum is a comprehensive tool for model interpretability, designed to facilitate understanding of complex PyTorch models. It offers a wide range of algorithms and visualization tools to help researchers and developers identify the key features driving model predictions. Captum supports most types of PyTorch models and can be used with minimal modification to the original neural network.