Is there anything other than just changing block 6 in section 7.2 from 224 to --> 96 for exercise 3 of the section?
It didnt seem to change the training results at all
I mean yeah the data structure is different, but results seem almost identical?
The results are slightly different. Like, only a miniscule difference. But, since the input size is smaller, it is able to process a lot more examples comparatively.
kernel size 7
model with size 96