Stop manually testing your ML models and don’t let your users be beta testers.

Efemarai is a platform dedicated to stress testing your ML models. By providing a trained model, a dataset, and using our visual domain builder you can let Efemarai find rare cases that make your model misbehave.

You can integrate Efemarai as part of your standard CI check through our CLI and SDK, as well as start tests manually through the browser. We take care of computing infrastructure and let you get the performance results of your model in just a few lines of code/commands.

  • The Local Tutorials guide will get you started using Efemarai in under 3 minutes, locally, on your computer.

  • The Hub Tutorials guide explains the basics behind running Efemarai on your problem in the Hub and is the recommended way to begin your exploration into automatically testing your models and longer vulnerability checks.

The SDK Reference contains the definition of how to create and manipulate each of the Efemarai objects.

One of the main building blocks that need to exist in each Efemarai-based project is the configuration file - efemarai.yaml. Have a look at the specifications here for more details on how to construct it.

To quickly interact with the platform, you can locally setup the Efemarai CLI. Checkout the CLI Quickstart or run efemarai --help in your terminal.

If you have any issues, questions or comments, don’t hesitate to contact us at feedback@efemarai.com, we love hearing what you like and what we can do better!