Hi, I meet a source code that using the Vit Model with the code:
from transformers import ViTModel
backbone = ViTModel.from_pretrained('google/vit-base-patch16-384')
it works good, but if I want to change the Vit Model itself – add some new model, change the number of input(i.e. change the forward process). I should locate the VIT model and change it instead of an VITModel(compact one).
My question is how to find the actual model of ViTModel???