Hi,
During the 2nd programming assignment of week 1, we get the following network:
I don’t understand why we have 128 parameters for the batch norm layer. I know that a batch norm layer has only two trainable parameres.
Thank you!
Hi,
During the 2nd programming assignment of week 1, we get the following network:
I don’t understand why we have 128 parameters for the batch norm layer. I know that a batch norm layer has only two trainable parameres.
Thank you!
Check out tensorflow’s doc page which said “for each channel”.
This should be a good hint for you to work out the number of parameters in a batch norm layer. Don’t forget that you can put aside the assignment, experiment the layer with different settings and input data shape.
Raymond
Oh, I see. Thank you!!