Closed JaGeo closed 6 months ago
Check out this pull request on
See visual diffs & provide feedback on Jupyter Notebooks.
Powered by ReviewNB
Now, it is fixed. There was a multiplication with the stride itself missing as the loops give the indices in the corresponding padded matrices. In the very simple, stride=1 case, it worked correctly.
Likely also wrong in the 1D case. Have to check that more carefully as well.
Now, it should be fixed. Sorry for taking a bit longer with this in the pull request.
Thanks so much for the kind update. This is awesome! I added a note & acknowledgement to the errata so I don't forget updating this in the book, too!
The stride in the 2D convolution example was not implemented correctly. The for loops did not consider the stride which lead to issues when changing to another example including a stride different from (1,1).