Hi, I saw you assignment on the private message; a couple of things you need to fix:
- At the downsampling path you need to start with 32 filters up to 256.
- Maybe you should try another optimizer at model compile as well.
- You should also increase the number of epochs to be trained as well.
Assuming the others are alright, you should re submit and hopefully pass the grader.
Is your problem fixed now? I’m having the same issue with model.h5 submission.
Yhh, my problem is fixed now.