Skip to content

Home

Welcome to the AutoML-Toolkit framework docs.

Tip

See the navigation links in the header or side-bars. Click the button (top left) on mobile.

For a quick-start, check out examples for copy-pastable snippets to start from. For a more guided tour through what AutoML-Toolkit can offer, please check out our guides. If you've used AutoML-Toolkit before but need some refreshers, you can look through our reference pages or the API docs.

What is AutoML-Toolkit?#

AutoML-Toolkit is a highly-flexible set of modules and components, allowing you to define, search and build machine learning systems.

  • Python

    Use the programming language that defines modern machine learning research. We use mypy internally and for external API so you can identify and fix errors before a single line of code runs.

  • Minimal Dependencies

    AutoML-Toolkit was designed to not introduce dependencies on your code. We support some tool integrations but only if they are optionally installed!.

  • Plug-and-play

    We can't support all frameworks, and thankfully we don't have to. AutoML-Toolkit was designed to be plug-and-play. Integrate in your own optimizers, search spaces, execution backends, builders and more.

    We've worked hard to make sure that how we integrate tools can be done for your own tools we don't cover.

  • Event Driven

    AutoML-Toolkit is event driven, meaning you write code that reacts to events as they happen. You can ignore, extend and create new events that have meaning to the systems you build.

    This enables tools built from AutoML-Toolkit to support greater forms of interaction, automation and deployment.

  • Task Agnostic

    AutoML-Toolkit is task agnostic, meaning you can use it for any machine learning task. We provide a base Task which you can extend with events and functionality specific to the tasks you care about.