Why Transpose?

Open MachineJeff opened this issue 6 years ago • 1 comments

Hi there In tf_qrnn.py, line 122

_w = tf.transpose(_weighted, [1, 0, 2])

I am confused about the transpose operation, could you explain it ?

Jul 19 '19 05:07 MachineJeff

I think it can be replaced by follows: # _w = tf.transpose(_weighted, [1, 0, 2])

    _ws = tf.split(_w, num_or_size_splits=3, axis=1) # make 3(f, z, o) conved_size x  batch_size x size

do u think so?

Jul 19 '19 06:07 MachineJeff