Incubator-mxnet: How I can set different 'lr_mult' for weight and bias in one layer?

[lugui2009]

I set up one layer ,for example ,fc1=mx.symbol.Fullyconnected(data=data,name='fc1',num_hidden=64)
There are two parameters of this layer : weight and bias。I want to set different 'lr_mult' for these two
parameters . for example,'lr_mult'='0.1' for weight ,and 'lr_mult'='1' for bias.

What should I write the code for this purpose ? Thank you .

[piiswrong]

fc1=mx.symbol.Fullyconnected(data=data,name='fc1',num_hidden=64, attr={'weight_lr_mult':'0.1', 'bias_lr_mult':'1.0'})

[lugui2009]

@piiswrong I used this code .How,maybe I got the wrong attributions as follows.

Why I got the 'fc1_bias_bias_lr_mult' ,'fc1_weight_weight_lr_mult' and so on.

[piiswrong]

those are harmless

On Aug 7, 2016 7:49 PM, "lugui2009" [email protected] wrote:

@piiswrong https://github.com/piiswrong I used this code .How,maybe I
got the wrong attributions as follows.
[image: mymistake]
https://cloud.githubusercontent.com/assets/20158014/17467932/85417470-5d55-11e6-9d61-36447665cf45.jpg
Why I got the 'fc1_bias_bias_lr_mult' ,'fc1_weight_weight_lr_mult' and so
on.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/dmlc/mxnet/issues/2932#issuecomment-238130761, or mute
the thread
https://github.com/notifications/unsubscribe-auth/AAiudH5mD6JiZkCg02BBfztk2gjTDvIzks5qdplJgaJpZM4JdVLp
.

[lugui2009]

@piiswrong Thank you very much .I see

[phunterlau]

This issue is closed due to lack of activity in the last 90 days. Feel free to reopen if this is still an active issue. Thanks!

[pilhoon]

It looks some changes are. You can use a simple var

w=mx.sym.var(lr_mult=0.1)
b=mx.sym.var(lr_mult=0.1)
fc1=mx.symbol.Fullyconnected(data=data,name='fc1',num_hidden=64, weight=w, bias=b)