exlib 0.1.0

Toolkit for explainability by BrachioLab

exlib

exlib is a comprehensive package showcasing our lab's work on explanation methods, featuring user-friendly modules for easy application of various techniques.

Installation

To install:

pip install exlib

If you have exlib already installed, please check that you have the latest version:

python -c "import exlib; print(exlib.__version__)"
# This should print "0.0.2". If it does not, update the package by running:
pip install -U exlib

In order to use pytorch-gradcam, install the lab version to handle more diverse shapes at

grad-cam@git+https://github.com/brachiolab/pytorch-grad-cam

Projects

We list below some relevant projects that use exlib heavily.

The FIX Benchmark: Extracting Features Interpretable to eXperts

• You can find the FIX benchmark README here.
• [Paper] [Website]