How to set lr_mult for convolutional layer in pytorch?

In caffe, it has the option to set the learning multiple for convolution as follows

layer {
  name: "conv1a"
  type: "Convolution"
  bottom: "data"
  top: "conv1a"
  param {
    lr_mult: 1
  param {
    lr_mult: 2
  convolution_param {
    num_output: 64
    kernel_size: 3
    pad: 1
    stride: 1    
    bias_filler {
      type: "constant"
      value: 0

In pytorch, how to set the lr_mult for convolutional layer only with SGD? Thanks

I think currently this parameters lr_mult is not supported in PyTorch.

Similar issue here: Does it exist ‘lr_mult' and 'decay_mult' in optimizer?

1 Like

So how they can reproduce deeplab network. It is common thing in deep lab architecture