During stitching, extra rows and columns are generated due to the non-square patches created in loader.py. The stitching code has been editted to account for this. This PR addresses the AssertionErrors generated that prediction size does not equal the size of the original input (X_t).
This is a draft fix to resolve this problem. I do believe this may generate errors in different scenarios, and ideally changes to loader.py should be made to prevent non-square patches from being produced.
During stitching, extra rows and columns are generated due to the non-square patches created in
loader.py
. The stitching code has been editted to account for this. This PR addresses the AssertionErrors generated that prediction size does not equal the size of the original input (X_t).This is a draft fix to resolve this problem. I do believe this may generate errors in different scenarios, and ideally changes to loader.py should be made to prevent non-square patches from being produced.