The Issue of training loss
wangtao201919 opened this issue · comments
First of all, thank you very much to the author for providing this amazing model. I am very interested in NeuCoSVC, so I want to use this model to train my own dataset locally. I followed the data processing process exactly as described in the Readme. However, during training, the “generator_loss” suddenly increased after 100k iterations. Therefore, I would like to ask about the changes in each loss during the training process. Thank you very much!
In the default configuration, the discriminator is added to training at 100k step, which will cause the generator_loss to increase.