How can I add a new dropout layer in the model and train it?

[Calvin-Pang]

I want to add a new dropout layer before the last fully-connected classification layer in the model. So is there any method that I can make it?
Also, if I can make it, how can I still use the train.py? I mean, the train.py seems like it can only create models which are predefined in the file named "timm.models".