ResNet_cifar icon indicating copy to clipboard operation
ResNet_cifar copied to clipboard

learning rate is not updated

Open holibert opened this issue 7 years ago • 1 comments

I set the learning rate update as below: if train_step < 32000: self._lrn_rate = 0.1 elif train_step < 48000: self._lrn_rate = 0.01 else: self._lrn_rate = 0.001 in resnet_main.py, the _LearningRateSetterHook after_run() function doesn't work.

After train, I see the tensorboard. I found the learning rate always to be 0.1.

holibert avatar Dec 07 '17 01:12 holibert

Please show me the screenshot of your tensorboard. [email protected]

chaipangpang avatar Dec 07 '17 03:12 chaipangpang