DetoxAI documentation

Welcome to the documentation for DetoxAI!

DetoxAI is a Python package for debiasing neural networks. It provides a simple and efficient way to remove bias from your models while maintaining their performance. The package is designed to be easy to use and integrate into existing projects.

We hosted a website with a demo and an overview of the package, which can be found at https://detoxai.github.io. DetoxAI is also available on GitHub at https://github.com/DetoxAI/detoxai.

Getting Started

DetoxAI is available on PyPI, and can be installed by running the following command:

pip install detoxai

Then, import the library and debias your model using the api:

import detoxai

model = ... # your torch model
dataloader = ... # your torch dataloader returning (image, label, prot. attr) tuples
results = detoxai.debias(model, dataloader)

To get started with DetoxAI, please refer to the following examples:

Notebooks

• A short example
• A full example

We have also put together a bunch small tutorials to help you get started with DetoxAI in your own projects:

Tutorials

• Adding a new method
  • Adding a New Method to DetoxAI
• Datasets in DetoxAI
  • Using internal tools for datasets

API Reference

For detailed information on the library’s functions and classes, see the API reference:

API

• DetoxAI
  • Modules
    • detoxai.cavs
      • Submodules
      • detoxai.cavs.cav submodule
      • detoxai.cavs.extract_activations submodule
      • detoxai.cavs.mass_mean_probe submodule
      • Module contents
    • detoxai.core
      • Submodules
      • detoxai.core.evaluation submodule
      • detoxai.core.interface submodule
      • detoxai.core.interface_helpers submodule
      • detoxai.core.mcda_helpers submodule
      • detoxai.core.model_wrappers submodule
      • detoxai.core.results_class submodule
      • detoxai.core.xai submodule
      • Module contents
    • detoxai.datasets
      • Subpackages
      • Module contents
    • detoxai.methods
      • Subpackages
      • Submodules
      • detoxai.methods.model_correction submodule
      • detoxai.methods.utils submodule
      • Module contents
    • detoxai.metrics
      • Submodules
      • detoxai.metrics.bias_metrics submodule
      • detoxai.metrics.distance_metrics submodule
      • detoxai.metrics.fairness_metrics submodule
      • detoxai.metrics.metrics submodule
      • Module contents
    • detoxai.utils
      • Submodules
      • detoxai.utils.dataloader submodule
      • detoxai.utils.datasets submodule
      • detoxai.utils.decorators submodule
      • detoxai.utils.experiment_logger submodule
      • detoxai.utils.transformations submodule
      • Module contents
    • detoxai.visualization
      • Submodules
      • detoxai.visualization.DataVisualizer submodule
      • detoxai.visualization.HeatmapVisualizer submodule
      • detoxai.visualization.ImageVisualizer submodule
      • detoxai.visualization.LRPHandler submodule
      • detoxai.visualization.MetricsVisualizer submodule
      • detoxai.visualization.SSVisualizer submodule
      • detoxai.visualization.ScatterVisualizer submodule
      • detoxai.visualization.TableVisualizer submodule
      • detoxai.visualization.Visualizer submodule
      • detoxai.visualization.enums submodule
      • Module contents
  • Module contents

Contributing

Interested in contributing to DetoxAI? Check out our contribution guidelines on GitHub.

License

DetoxAI is licensed under the MIT License.

Index

• Module Index