tensorflow-wavenet icon indicating copy to clipboard operation
tensorflow-wavenet copied to clipboard

Published 20 hours ago • verified publisher icon ibab

tf.layers.conv1d(dilations)

Open JamesX2015 opened this issue 6 years ago • 1 comments

I'm wondering why we use time_to_batch + causal_conv + batch_to_time in ops.py, rather than tf.layers.conv1d with dilation_rate>1. Is it because it computes faster or old versions of tf don't support?

JamesX2015 avatar Aug 07 '18 05:08 JamesX2015

I simplified the code using tf.layers.conv1d.

See the Issue

hccho2 avatar Nov 25 '18 08:11 hccho2