ai-models-for-all

This release addresses small bugs and issues that have cropped up due to churn in the ecmwf-labs/ai-models package, as well as implements the option for users to initialize forecasts from GFS/GDAS in addition to ERA-5.

This initial feature-complete release of ai-models-for-all contains a full application which runs on Modal and can use any of the published ai-models plugin packages (PanguWeather, FourCastNetV2, or GraphCast). All models have been tested and the core functionality is working. User feedback and suggestions are welcome!

Release of FourCastNet(v2-small) integration.

Incorporates immediate, small tweaks to the model to make it a bit more general for users to avoid using hard-coded values.

Snapshotting the application for quick distribution / feedback from potential users.