Toolkit

Introducing the Toolkit

  • Tutorial: Toolkit in Action

Using the Toolkit

  • Jobs
  • Data
  • Permissions
  • Tokens

Use Cases

  • Hyper-parameters Search
    • Ray Tune
    • Shuriken
  • Label Data
  • Large Model Training
  • Error Analysis with Azimuth
  • Running Experiments With mlflow
  • Running Jupyter Lab / Notebooks

eai cli

  • CLI Reference Manual
  • CLI Release Notes

Support

  • Toolkit Docker Tips
  • Walkthrough - Docker
  • FAQ
  • How To
  • Troubleshooting
Toolkit
  • Hyper-parameters Search
  • Edit on GitHub

Hyper-parameters Search

Ray Tune

Ray is a distributed framework allowing the computation of your model to be distributed across several workers. In particular, Ray Tune allows the user to do hyper-parameters search in a distributed fashion.

To support that use case, the Toolkit team maintains the eai-distributed-toolkit python library. The documentation should be able to help some experienced users.

Shuriken

Shuriken is an ElementAI made library specifically built to do hyper-parameters search. Although unmaintained and unsupported by the Toolkit team, it is still reported as working.

  • Github repository

  • Documentation (requires EAI VPN access)

Warning

There are two versions of Shuriken. The one listed above is compatible with Toolkit. The other one has been deprecated earlier and the repository archived.

Previous Next

© Copyright 2024, Element AI (by ServiceNow).

Built with Sphinx using a theme provided by Read the Docs.

Last updated on 2025-05-06 12:29 UTC
Commit: b6c2fc5