From the course: Transfer Learning for Images Using PyTorch: Essential Training
Unlock the full course today
Join today to access over 22,600 courses taught by industry experts or purchase this course individually.
Differential learning rates
From the course: Transfer Learning for Images Using PyTorch: Essential Training
Differential learning rates
- [Man] So this is almost identical to the fine tuning notebook so I'm going to just focus on the sections which are relevant to differential learning rate. So I'm just going to run through the first couple of cells. So we get our VGG-16 pretrained model and download it. With its parameters and then we freeze the VGG model. We unfreeze the classifier. We define a new classifier. Define the criterion. And this is new. And we're specifying the same learning rate across the group. In the next stage when we unfreeze a group of convolution neuronetworks, that group will have a different learning rate to the one used by the classifier. Let's train our model. Now although this is our differential rate notebook we're using the same learning rate as our starting point for the classifier as we used in the learning rate notebook. So let's check the accuracy of the model. Then you can see that the accuracy that we've got here…
Practice while you learn with exercise files
Download the files the instructor uses to teach the course. Follow along and learn by watching, listening and practicing.