Prevent network weights to be updated when performing backward pass

Asked Mar 27 '17 at 17:43

Active Mar 29 '17 at 15:59

Viewed 43 times

I am trying to perform a backward pass through my network and I don't want to update my network weights of my network when, I do a backward pass.

output = net:forward(input)
err = criterion:forward(output, label)
df_do = criterion:backward(output, label)
net:backward(input, df_do)

I'm assuming this can be done using by any of the two methods

accGradParameters(input, gradOutput, scale)
accUpdateGradParameters(input, gradOutput, learningRate)

Can,I do this using the optim package in torch?

edited Mar 29 '17 at 15:59

Tom

asked Mar 27 '17 at 17:43

rahul_raghavan

Possible duplicate of [Finetune a Torch model](http://stackoverflow.com/questions/37459812/finetune-a-torch-model) – Manuel Lagunas Mar 28 '17 at 09:20
This is not a duplicate, I know about finetuning a network.However, I would like to perform backpropogation without updating the weights for a network – rahul_raghavan Mar 29 '17 at 16:56

0 Answers0