Dioptra Overview#
Dioptra is a software test platform for assessing the trustworthy characteristics of Artificial Intelligence (AI) models.
Dioptra is a modular, microservice-based environment for creating reproducible, trackable, and reusable AI workflows. It is designed to measure, analyze, and track AI risks, making it particularly useful for addressing the “combinatorial explosion” of testing possibilities across attacks.
Dioptra Documentation#
Below are some useful guides to help get you started. To see the entire list of documentation topics, view the table of contents on the left sidebar.
Understand Dioptra#
Motivation for Dioptra - What is Dioptra for and how does it compare to other tools?
Workflow Architecture - Learn how all the high level Dioptra components orchestrate together to execute jobs
Use Dioptra#
Follow the installation guide to build the Docker containers and start the services
Run the Hello World tutorial to make sure things are set up correctly
Dive Into the Details#
Read through the Dioptra component explainers to dive deeper on each part of the workflow architecture.
Learn how to customize your Dioptra deployment
Reference the API endpoints and learn how to use the Python client
Progress through the intermediate tutorial to learn about complex workflows, artifacts, and more
About#
Dioptra is open-source software developed by the National Institute of Standards and Technology (NIST). Contributions and feedback are welcome from the community. You can find the source code, license information, and more on the NIST GitHub repository.