Image Segmentation_Unet _v2 Conv_block code

[Askhan]

I am getting error in the conv_block, I can not figure out where the problem is. I am including a copy of the script and the error messages, please help me to find out the problem. Thanks

[TMosh]

Don’t use hard-coded values unless you’re told to specifically.
So, don’t use 32, use “n_filters”.
Check for those types of problems.

Also, please edit your message and remove the code. Posting your code breaks the course Honor Code.

[TMosh]

Also, instead of posting your code, just post any error messages you get from the unit tests.
Those are usually enough to point to the problem in your code.

[Askhan]

Thanks. I will try to edit the message.

[TMosh]

You should have a “pencil” icon in the toolbar below your message. That’s the ‘edit’ tool.
If you don’t see it, let me know and I can edit your message.

[Askhan]

Thanks Mosh for the suggestions. I removed the code as you suggested.

The error is as following:
Test failed
Expected value

[‘Conv2D’, (None, 96, 128, 1024), 28672, ‘same’, ‘relu’, ‘HeNormal’]

does not match the input value:

[‘Conv2D’, (None, 96, 128, 32), 896, ‘same’, ‘relu’, ‘HeNormal’]

[TMosh]

Most likely, in the first Conv2D layer, either your “filters=…” argument is wrong, or your “kernel_size=…” is wrong.

Be sure you use the identifiers for the arguments, because if you leave them out, you’re just relying on getting them in exactly the correct order.

[TMosh]

Another problem might be with embedding comments inside the argument definition. Try taking those out.

[paulinpaloalto]

This was also asked and answered on this other parallel thread. The bug is that you must be hard-coding the value of the number of filters to 32 instead of using the parameter that was passed in.

Oh, sorry, I see that @TMosh already also gave this answer earlier on this thread. So it’s been covered!

[TMosh]

Note also that in the “conv10 = …” layer, you don’t use n_filters there.