Nrgeup

> 要是shuffle的话，那很可能您实验的分数会更高呀。不管怎么样，您的idea肯定都是work的。 谢谢，学到了很多。👍

请教不敢当，我的邮箱是[email protected] 您可以联系我或者在邮件里附上微信号。

Thank you! "fgim_attack" aims to generate the outputs under a fixed modification weight W, to show that different modification weights W can control the obviousness of the target attribute in...

For privacy reasons, I cannot publish the Beer Advocate dataset here, you can find it in the paper "Zachary C Lipton, Sharad Vikram, and Julian McAuley. Generative concatenative nets jointly...

Thank you for your reminder!

抱歉回复的有点晚，多属性风格迁移所用的代码是一样的。只需要修改鉴别器的输出维度这个参数就可以。

Sorry to reply to you so late! I provided the outputs in https://github.com/Nrgeup/controllable-text-attribute-transfer/tree/master/outputs, or you can use the checkpoints I provided in https://github.com/Nrgeup/controllable-text-attribute-transfer/tree/master/method/mymodel-***/save/***/***.pkl to generate outputs.

这个是为了对encode完了的states算一个权重，就是简单的相乘。最后乘上encode states送到GRU里面。

Sorry to reply to you so late, the outputs presented in https://github.com/Nrgeup/controllable-text-attribute-transfer/blob/master/outputs/ were generated under dynamic modification weights. We detailed it in Section#3.3 (what we call “Dynamic-weight-initialization method” in our...

抱歉抱歉，我暂时只是搞着玩的，还不完善（不能用），你发我邮箱你的微信吧( [email protected] )，我加你