Song Park
Song Park
Hi, sorry for the late reply. "seen_chars" is a list of characters to be used for the training, and "unseen_chars" is a list of characters to exclude during the training....
Hello, sorry for the late reply. It is weird... I think it is related to the Pillow version. I will check the code. If available, please provide me the font...
It seems that the training is unstable. "ValueError: matrix contains invalid numeric entrie." means that the input contains NaN, which can caused by various factors. I cannot know what the...
Hi, 1. Yes, we use all the characters in data/chn/train_chars.json for training. As mentioned in our paper, our method uses large number of characters using training phase. 2. We use...
It is the label of font style, which is used for computing the adversarial loss. You can find the detailed explanation of the adversarial loss in Appendix B.4 of our...
Note that, our method is **_not a few-shot training method_**. Our method assumes that the model can access large number of samples during training. How many style fonts (not characters)...
Hi, sorry for the late reply. I've updated the readme and attached the download link for the classifiers. You can download the weights from [here](https://drive.google.com/file/d/1RzB8AwV7RgsoDk9IUpUpkKK6xUvlmAeG/view?usp=share_link). Thanks for letting me know...
Hi, I also failed to reproduce the reported values in the paper. Please see #1, this is a previous issue about that problem.