if dropout_p>0:
c1 = tf.nn.dropout(c1, keep_prob=1.-dropout_p)
c2 = conv(c1, num_filters*2, nonlinearity=None, init_scale=0.1)
c3 = tf.tanh(c2[:,:,:,:num_filters]) * tf.nn.sigmoid(c2[:,:,:,num_filters:])
return x+c3
""" utilities for shifting the image around, efficient alternative to masking convolutions """
After Change
if dropout_p>0:
c1 = tf.nn.dropout(c1, keep_prob=1.-dropout_p)
c2 = conv(c1, num_filters*2, nonlinearity=None, init_scale=0.1)
a, b = tf.split(3, 2, c2)
c3 = a * tf.nn.sigmoid(b)
return x+c3
""" utilities for shifting the image around, efficient alternative to masking convolutions """
