duplicate use of activation

Kyubyong / tacotron

A TensorFlow Implementation of Tacotron: A Fully End-to-End Text-To-Speech Synthesis Model

Apache License 2.0

1.83k stars 436 forks source link

Closed candlewill closed 7 years ago

candlewill commented 7 years ago

When type=="bn", the activation is applied twice in normalization.

Kyubyong commented 7 years ago

@candlewill Thanks but can you change the acitivatiion argument of bn to None instead? It's better I think.