Grader less for convolution model

I tried twice. first it showed with callback should have fired earlier.
So I change the accuracy to 0.995

still I am getting the same grade. Can anyone guide me how to improve the convolution model.

I am sharing the screenshot fo the grader result and convolution model

Thank You

Hi there,

Im not quite familiar right now with this assignment. But in general you should keep in mind that the assignments are similar to the labs.

In this context you should be trying different architectures, possibly optimizers, and learning rates…etc.

The grader feedback says that your model training should’ve fired the callback in less than 10 epochs.

Looking at your training log, this cutoff was reached at end of 5th epoch but the callback didn’t stop model training. Please refer this link to fix the issue.

this part I understood with the feedback, what I meant to ask where do I need to fix the issue, I mean I felt the code in convolution model need to be addressed. But I was not sure, so wanted to ask. I mean do I need to address the code in my callback or the convolution model, or both?

The link you sent, seems I need to improve my code in the myCallback??

Please check the following regarding the callback class:

  1. You’ve specified an object of the callback class in the list of callbacks to method.
  2. on_epoch_end checks for the proper threshold and stops further training if required.
1 Like

Hello sir,

thank you so much. you gave a detailed solution for my problem although you could have just talk work on the callback. So that was generous of you. Just got 100 grade. So wanted to express gratitude for your work and time. Keep learning and teaching !

thank you gent for replying, I solved the problem. my epoch_end code was not right which I corrected. thank you for your time and work sir.

1 Like