Hi, all,
I want to assign the parameters of a conv layer by hand and fix it. Is it possible to set some parameters such as requires_grad = False for the conv layer?
The easy way is
conv_layer.train(False)
Which tells pytorch not to update the weights on that particular layer.