SOTAI

The new standard in AI interpretability.

SOTAI is:

• Transparent: SOTAI combines robust interpretable modeling techniques with supporting analysis tooling to help organizations make faster, well-informed decisions effortlessly.
• Simple: Keep it straightforward and avoid unnecessary complexity, just like our approach to AI interpretability.
• Confidence-Boosting: With our comprehensive interpretability tools, SOTAI empowers you to trust and act on AI model predictions, enhancing your decision-making confidence.

SOTAI is a Library For Interpretable Machine Learning. This library is a PyTorch implementation of modeling techniques found in Monotonic Calibrated Interpolated Look-Up Tables.

You can get started in minutes after downloading the package, see our Quickstart guide or follow along below.

Installing the package:

pip install sotai

Importing the package:

import sotai

SDK Documentation

You can find documentation for this SDK at https://docs.sotai.ai/v/sdk-ref or in the repo docs folder.

Web Client User Documentation

You can find documentation for how to use the hosted web client at https://docs.sotai.ai/

SDK Code Generator

Inference Results Side-By-Side Analysis

Contribution Guidelines

See the guide on contributing for full details on how to contribute to the library. For any feature and/or bug requests, visit our Issues.

Examples

For detailed examples on how to use the library, see examples.

Questions and Help

If you have questions about the SOTAI SDK or using the web client, we encourage you to reach out to the community and SOTAI dev team for help.

We actively monitor our Discord and welcome new community members.

License

MIT