Model Freezing Callback

[icedoom888]

Is your feature request related to a problem? Please describe.

No response

Describe the solution you'd like

Implement a callback to freeze parts of the model during training.
This will allow for interesting strategies: train for x epochs, freeze the encoder and decoder, train the processor only for y epochs.

Describe alternatives you've considered

Will call the same method introduced by ecmwf/anemoi-training#189 within a callback.
An issue might arise in DDP due to Synchronization.

Additional context

Take a look here for similar implementation: https://lightning.ai/docs/pytorch/stable/api/lightning.pytorch.callbacks.BaseFinetuning.html

Organisation

MeteoSwiss