This notebook contains the code for the Medium post BYOL: Bring You Own Loss Function about potential benefits from customization of a loss function and its parctical application in Gradient Boosting model
Additionally, consider installing this library, that implements the functionality of both of the custom loss functions https://github.com/pashna/gbm_custom_loss
This notebook is also uploaded to Colab to play around.