Can I customize the loss function used during fine-tuning in the Lag-Llama model?

[XinyuChen-hey]

Hello,

I am currently working with the Lag-Llama model and would like to know if it is possible to customize the loss function during the fine-tuning process. Is there a way to modify or replace the default loss function with a custom one, and if so, how can I implement this change?

Any guidance or examples would be greatly appreciated!

Thank you in advance!