Skip to content

scaling-group/icon-core

Repository files navigation

ICON-CORE

python pytorch lightning
hydra ruff
license PRs contributors

ICON-CORE is a project to facilitate research in scientific machine learning,
specializing in In-Context Operator Networks (ICON).
Click on fork or use this template to initialize your own project.

Description

ICON-CORE is an open-source project designed to accelerate research in Scientific Machine Learning (SciML), specializing in In-Context Operator Networks (ICON). Originally developed as the internal infrastructure for the Scientific Computing and Intelligence Group (Scaling Group), it is now an open platform for the community.

Core Features:

  • Models and algorithms on in-context operator networks.
  • Standardized datasets and dataloaders supporting seamless multi-dataset integration.
  • Streamlined workflows for training and evaluation.
  • Comprehensive utilities, including visualization, logging, and experiment management.
  • AI-Native Development: Designed with AI-friendly conventions and skills to maximize productivity when using AI coding assistants (e.g., Claude Code).

For more details on the repository architecture, please refer to structure.md.

🚀 Join the Community

To build a truly accessible and versatile platform, ICON-CORE thrives on community contributions. Whether you are implementing a new model, sharing a specialized dataset, or optimizing core infrastructure, your input is invaluable.

We invite you to join us in shaping the project by opening a pull request!

How to use this repository

There are two ways to use this repository:

  • Use this template to create your own repository. This is essentially copying the files you need, and the generated repository is independent of this one. Of course, you can also create your own repository from scratch and manually copy the files you need.

  • Fork this repository and use it as an upstream for your own project. This makes it easier to sync up with the latest features. Note that your forked repository will be public, and when syncing the updates, the changes may break your code, even silently changing your training results.

Contributors

Thanks to all the people who contribute:

Also to contributors before open-sourcing (may not shown in the contributors list above):

Acknowledgement

Please include the NOTICE file (already included in this repository) in your code base that uses this repository.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors