[Feedback] C3W2A crop function fails in test_expanding_block but passes test_expanding_block_crop

Course Q&A Generative Adversarial Networks (GANS) Apply Generative Adversarial Networks
1

Hello,
thank you for providing such a nice class on GANs.

I found this issue while doing assignment C3W2A. I made a crop function which passed test_expanding_block_crop() but was then failing in the next part with test_expanding_block() .

I narrowed down the issue to how my code was handling cases with inputs having odd and even width and height (test_size parameter in test_expanding_block_crop() ).
I solved this anyways, but maybe there could be an extra check in test_expanding_block_crop() that checks the crop function correctly works when the input has odd/even width and height.

Not sure if I can post code here but If needed I can provide some more concrete example of my code.

Thanks

2

Hi Maevil!
Hope you are doing well. I guess you are talking about the issues when you are using PyTorch’s CenterCrop function. There is a version mismatch which is leading to such issues. Do check this conversation and let me know if this is the one that you are talking about.
Regards,
Nithin