Why do you use math.exp?

[KazutoshiShinoda]

https://github.com/ZiJianZhao/SeqGAN-PyTorch/blob/master/main.py#L87

why?

In the original implementation by LantaoYu, exp is not used for loss values.

[ShengleiH]

@shinochin hello, because 'NLLLoss' is used which is 'NLLLoss(x) = log(softmax(x))'. Therefore, to get the softmax results, you need to do this: 'softmax(x) = exp(NLLLoss(x))'

[deepylt]

But total_loss / total_words seems weird since the discriminator loss is not related to the sequence length.