search

geek-ai / Texygen

A text generation benchmarking platform
MIT License
863 stars 203 forks source link

Why the leakgan implementation is different from the paper? #40

Open EternalFeather opened 5 years ago

EternalFeather commented 5 years ago

I found that your leakgan is hard to train if I change the generator model to BERT or Transformer_Autoregression