Lucid

DeepDream, but sane. Home of cats, dreams, and interpretable neural networks.

Lucid is a collection of infrastructure and tools for research in neural network interpretability.

In particular, it provides state of the art implementations of feature visualization techniques, and flexible abstractions that make it very easy to explore new research directions.

Dive In with Colab Notebooks

Start visualizing neural networks with no setup. The following notebooks run in your browser.

Beginner notebooks:

• lucid tutorial (TODO) -- Introduction to the core ideas of lucid.
• DeepDream (TODO) -- Make some dog slugs and crazy art.

More advanced:

• Aligned interpolation (TODO(colah))
• xy2rgb??? (???)
• 3d??? (???)

Project Structure

How lucid is structured:

• modelzoo: Easily import models for visualization
• optvis: Framework for optimization-based feature visualization
• scratch: Incubating code that needs to be shared between notebooks.
• misc: More mature code that doesn't fit into a large cluster.
• recipes: Less general code that makes a particular visualization.

Note that we do a lot of our research in colab notebooks and transition code here as it matures.

License and Disclaimer

You may use this software under the Apache 2.0 License. See LICENSE.

This project is research code. It is not an official Google product.

Running tests

Use tox to run the test suite on all supported environments.

To run tests only for a specific module, pass a folder to tox: tox tests/misc/io

To run tests only in a specific environment, pass the environment's identifier via the -e flag: tox -e py27.

After adding dependencies to setup.py, run tox with the --recreate flag to update the environments' dependencies.