How to handle error in Lightgbm with custom loss?

Question

I\'m getting an error when trying to use my own custom loss function in lightgbm.cv whereas the same function works fine within the sklearn API of lightgbm.

My loss funct