torchxai-tools 1.0.0

TorchXAI is a PyTorch-based toolkit designed for evaluating machine learning models using explainability techniques.

TorchXAI

TorchXAI is a PyTorch-based toolkit designed for evaluating machine learning models using explainability techniques.

Description

TorchXAI provides efficient implementations of several XAI (Explainable AI) metrics that seamlessly integrate with the Captum library. The primary goal of this toolkit is to offer batch computation and task/data-agnostic implementations of various explainability metrics, enabling scalable evaluation while maintaining compatibility with the established Captum framework.

Installation with Pip

The project can be installed directly with pip:

pip install git+https://github.com/saifullah3396/torchxai.git

Installation from Source

Clone the repository:

git clone https://github.com/saifullah3396/torchxai.git

Install the project with pip on source directory

pip install -e .

Usage

Import the required metric, such as completeness

from torchxai.metrics.axiomatic import completeness
from captum.attr import Saliency

# ImageClassifier takes a single input tensor of images Nx3x32x32,
# and returns an Nx10 tensor of class probabilities.
net = ImageClassifier()

saliency = Saliency(net)
input = torch.randn(2, 3, 32, 32, requires_grad=True)
baselines = torch.zeros(2, 3, 32, 32)

# Computes saliency maps for class 3.
attribution = saliency.attribute(input, target=3)

# define a perturbation function for the input
# Computes completeness score for saliency maps
completeness_score = completeness(net, input, attribution, baselines)