search

pippy360 / redofcrn

https://arxiv.org/pdf/1606.00373.pdf
1 stars 0 forks source link

do i need biases in the convolutions? #1

Open pippy360 opened 6 years ago

pippy360 commented 6 years ago

what about:

    conv1 = tf.multiply(conv1, multScalar)#try removing these and see what happens
    conv1 = tf.add(conv1, addScalar)#try removing these and see what happens