Upload the ML Model on Hugging Face Hub

[tztsai]

For enhancing accessibility and collaboration, we can upload the CAM model and its weights to the Hugging Face Hub. This requires converting the model weights into PyTorch's .pt format to ensure compatibility with the Hub's infrastructure, thus necessitating the adoption of ftorch. The conversion process will enable the broader climate science and machine learning communities to easily access, utilize, and contribute to the model's development. This task involves:

• Converting Weights: Develop a script or utilize existing tools to convert the Fortran model weights into .pt format.
• Documentation: Provide detailed documentation on the model architecture, conversion process, and instructions for use, ensuring that users can effectively leverage the model.
• Testing and Validation: Before uploading, test the converted weights in PyTorch to verify their integrity and performance.
• Hugging Face Upload: Create a repository on the Hugging Face Hub and upload the .pt formatted weights, along with the necessary documentation.

[tztsai]

I've uploaded the ML model (as an initial commit) on the Hugging Face Hub.

[jatkinson1000]

This should be added to the documentation to close this issue.
Note also that the weights have changed (see #66) so would need re-uploading.

@MarionBWeinzierl perhaps this is related to #65 but can be considered of lower priority?

[MarionBWeinzierl]

This should be added to the documentation to close this issue. Note also that the weights have changed (see #66) so would need re-uploading.

@MarionBWeinzierl perhaps this is related to #65 but can be considered of lower priority?

I would leave this as a separate issue (maybe mentioned in #65), but agree that it should be added to the documentation.

[tztsai]

I've updated the NN weights (from qobs...nc to NN_weights_YOG_convection.nc) and updated the documentation on https://huggingface.co/ICCS/cam-ml-yog-v0.