Order of maxpool / relu after CNN 5th layer

[kelvin-21]

Hi! I wonder after conv5, should we do maxpool2d first or relu first?

From the instruction in jupyter notebook, it seems we do maxpool2d before relu. I thought it should be the other way round.

The architecture should be:

(* skipped ... *)

conv2d, 5x5, 64->128, padding=2

maxpool2d

Relu()

flatten the feature map

...

[TianshuoY]

Good question. You may conduct an experimental analysis and include your findings in your report. Please also email me your results. I will address this issue during the assignment review session.

[kelvin-21]

Sure, will do, thanks.