The instruction under Baseline: Without Self-Supervised Learning said clearly "all weights in the model are trained", but I found something confusing in the corresponding code as follow,
model = Classifier(num_class=len(train_data.classes)).to(device)
for param in model.f.parameters():
param.requires_grad = False
The instruction under Baseline: Without Self-Supervised Learning said clearly "all weights in the model are trained", but I found something confusing in the corresponding code as follow,
and
It's obvious that only the weights of the final fc layer are trained, isn't it?