machine learning - Which metric should we examine while making the model?

Question

I have built the model using transfer learning. Used MobileNet, I want to know that should I see validation loss or validation accuracy to identify the overfitting. My validation accuracy is too good, but validation loss is high. Is my model overfitted or not?

I have used Early Stopping callback also!

The image is:

question from:https://stackoverflow.com/questions/65921421/which-metric-should-we-examine-while-making-the-model

Answer 1

Waitting for answers