TorchQL Documentation

TorchQL is a framework for programming integrity constraints over machine learning applications in Python. It supports PyTorch to allow for easier integration of TorchQL queries into the machine learning pipeline. Users can write queries that specify integrity constraints through a combination of query operations and arbitrary user-defined functions. These integrity constraints can then be directly evaluated over models and datasets.

Explore TorchQL

You can get started by exploring TorchQL and its features over the iWildCam Dataset through our Google Colab Notebook Demo, which does not require any local installation.

Quick Start

To get started with TorchQL locally, check out the Getting Started guide.