NVIDIA / OpenSeq2Seq

Toolkit for efficient experimentation with Speech Recognition, Text2Speech and NLP
https://nvidia.github.io/OpenSeq2Seq
Apache License 2.0
1.54k stars 371 forks source link

Change the shape of FC layer (vocab size) #442

Open mtbadakhshan opened 5 years ago

mtbadakhshan commented 5 years ago

Hi everyone! I want to change the size of fully connected layers of a pre-trained model. In other words, I want to replace fully connected layers with other fully connected layers with different shapes.

borisgin commented 5 years ago

Take a look at https://nvidia.github.io/OpenSeq2Seq/html/sentiment-analysis.html example