Developing Intuitions about parameter tuning

Hello People,

I just built a facial keypoints recognition dataset that feeds in a greyscale image into the network and outputs 72 keypoints on the face.
I loaded up a pretrained resnet34 and trained the model on this data.
The loss was all over the place. So what am I doing wrong here? How do I make the model work better?

While you weren’t really using fastai, I’ll try to help some. Most likely it’s too high a learning rate. Try to use a smaller one. I’d also look into using a resnet18 instead. I found that it can be easier to train and work with when you’re starting a keypoints problem :slight_smile:

1 Like