controllable-text-attribute-transfer

controllable-text-attribute-transfer copied to clipboard

Dinxin

'./data/imagecaption/sentiment.test.0' and './data/imagecaption/sentiment.test.1' are identical.

zylMozart

For why "No response to the reproduction issues".

2

I apologize for not checking the issues on the Github repo in time, because I did not regularly check the GitHub account after publishing the code. I will respond to...

Nrgeup

关于outputs中的my-model-yelp1.txt和my-model-yelp2.txt

2

请问这两个文件就是风格转换之后的输出吗？分别使用了w=1和w=2？论文中的acc,bleu这些值就是通过这两个文件的结果计算的吗？

Zhenwen-NLP

Multi-Aspect Sentiment Transfer experiment in detail

1

Could you provide detailed information of for Multi-Aspect Sentiment Transfer experiment ? (model code, input data and output data)

zhou1xiang2

the return of "fgim_attack"

1

Great work! but there is an issue about "fgim_attack" since it return nothing, but "eval_iters" wants to save the results of it to the file. Thanks again for your job!

Icyzrw

How to obtain the examples listed in the paper?

4

Thanks for sharing the code and great work! The paper picks several example sentences and they look amazing. But after running the code, I could hardly see such good examples....

RenqinCai

Usage of torch.nn.GRU

1

torch.nn.GRU 's input shape is [[sequence_size, batch_size, hidden_size]](https://pytorch.org/docs/stable/nn.html#torch.nn.GRU). But your code such as https://github.com/Nrgeup/controllable-text-attribute-transfer/blob/master/method/mymodel-yelp/model2.py#L315 is [batch_size, sequence_size, hidden_size]. I think your code is something wrong...

MokkeMeguru

How to get BLEU score?

1

Hi I'm trying to reproduce your work about YELP dataset. However, the BLEU score that i've gotten is quite different. I trained the model 200 times and got the score...

kinggodhj

论文中出现了多属性风格迁移，但代码未提供多属性，能否提供相关的代码供学习一下

1

archersama