geek-ai / Texygen

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

A quick question about g_loss #26

Open xixiddd opened 6 years ago

xixiddd commented 6 years ago

Hi, I have a quick question, Does the self.g_loss need to be divided by self.batch_size ? (as the self.pretrain_loss do) Thank you for your reply.