Somebody help me to get the gradient values for each iteration , while training using SGD?

autograd
1

i need to get the gradient value of a loss function with respect to the weight(s) in a neural network while training it.

query_pytorchforum
query_pytorchforum1031×510 67.7 KB

2

Hi,
After loss.backward(), you could loop over the net instance’s parameters and extract their grad attribute, like so:

for param in net.parameters():
    print(param.grad)
1 Like