AI Evaluation

Captum

Captum is a library for model interpretability in PyTorch.

About

Captum provides tools for understanding and interpreting machine learning models built with PyTorch. It supports various modalities, including vision and text, and allows users to implement and benchmark new interpretability algorithms. As an open-source library, it is designed for researchers and developers looking to enhance the interpretability of their models with minimal modifications.

Resources & Links

Related Categories