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

output node name to freeze the model?

Open dreamibor opened this issue 7 years ago • 0 comments

Are there anyone who freeze the model checkpoint to a protobuf file successfully? What's the output node name that we should use? I read the train.py and generate.py which makes me more confused since the author is creating input nodes in the source code, rather than calling and creating input / output node extracted from the model.

Any ideas?

dreamibor avatar Mar 08 '18 14:03 dreamibor