Request for Option to Save Best Model During Training

[amine-boukriba]

Thank you for providing this wonderful library! It features cutting-edge models and an easy-to-use interface. However, I'm encountering an issue: is there a way to save the best model during training? I've noticed that it currently saves the model from the last epoch trained, rather than the one with the best evaluation loss. I've searched extensively throughout the library but couldn't find any parameter to modify this behaviour. Thanks for your assistance.

[massquantity]

Hi, unfortunately, this feature is not supported. We will consider adding it.