Archai Documentation

Archai is a platform for Neural Network Search (NAS) that allows you to generate efficient deep networks for your applications. It offers the following advantages:

  • 🔬 Easy mix-and-match between different algorithms;

  • 📈 Self-documented hyper-parameters and fair comparison;

  • ⚡ Extensible and modular to allow rapid experimentation;

  • 📂 Powerful configuration system and easy-to-use tools.

Initial Steps

To install the latest release:

pip install archai

Please refer to the installation guide for additional help.


Archai requires Python 3.6+ and PyTorch 1.2+.

Toy Mode Experiment

The fastest way to try out the basic functionalities of Archai is to run a toy mode experiment using every available algorithm:

python scripts/

By switching to toy mode, algorithms will use tiny batches and a single epoch.

Support and Contributions

Archai is maintained on Github by the Neural Architecture Search team of Microsoft Research at Redmond. Please feel free to:

  • Open an issue to request for support;

  • Open a pull request to contribute with changes;

  • Join the Facebook group to stay up-to-date.

Citing Archai

If you use Archai in a scientific publication, please consider citing it:

    author = {Hu, Hanzhang and Langford, John and Caruana, Rich and Mukherjee, Saurajit and Horvitz, Eric J and Dey, Debadeepta},
    booktitle = {Advances in Neural Information Processing Systems},
    editor = {H. Wallach and H. Larochelle and A. Beygelzimer and F. d\textquotesingle Alch\'{e}-Buc and E. Fox and R. Garnett},
    publisher = {Curran Associates, Inc.},
    title = {Efficient Forward Architecture Search},
    url = {},
    volume = {32},
    year = {2019}