标签: tensorflow recurrent-neural-network seq2seq
我想知道如何以tie_rnn_seq2seq模式在编码器和解码器之间共享变量,因为编码器和解码器的单元数不同?
https://github.com/tensorflow/tensorflow/blob/r1.14/tensorflow/contrib/legacy_seq2seq/python/ops/seq2seq.py