kacky24 / stylenet Goto Github PK
View Code? Open in Web Editor NEWA pytorch implemention of "StyleNet: Generating Attractive Visual Captions with Styles"
A pytorch implemention of "StyleNet: Generating Attractive Visual Captions with Styles"
In test.py, captions with humorous style are generated, but the caption generation process used only training text rather than image. Why don't use image feature to generate humorous captions?
I would like to categorically state that this Paper "StyleNet: Generating Attractive Visual Captions with Styles" from Microsoft is non-reproducible. This is not just from the code based on this repo, but our own extensive experiments have lead us to believe that this paper is just a work of fiction put together. We have also contacted the lead authors Chuang Gan & Zhe Gan. However, we did not get any reasonable explanation about why this architecture does not work. It is unfortunate to see that this paper also have significant citations. At this point, how this was accepted at CVPR remains a big question.
Also the new dataset as mentioned in the paper, is not available as a whole. Only a part of this dataset is available, which makes this task even more questionable.
Overall, I would request readers stumbling across this not to waste their time reproducing this paper!!
I think as a developer it is important to mention the current state of the repo.
You need to update the readme that it is buggy and no further fix is done so that people who are just looking to run the code do not really waste time in setting it up if they don't plan on finding/fixing the bug(wrong/random caption for images).
Also, it is a good practice to keep the issues open until it is fixed. You have closed so many issues which are BUG reports.
Please update the readme file about the BUG in the beginning.
I am sorry to bother you, but i am a recruit in caption. I am very interested in this paper and would be very appreciate if you show more about your data fold. Thanks.
Hi,
sorry to bother you again.
I encountered the same problem as you when generating sentences . And I reimplemented each step by step but still couldn't find where the problem is. If you already find the problem or possible reason, would you please let me know? Thanks a lot !
Sincerely,
Tingyao
Hi, It's a great work!
I did not run the code. But I suspect there may be a problem.
In the paper, the author said that in the multi-task training(ie.training with romantic sentences or humorous sentences) he only update the matrix set {S}. But in your code, the whole LSTM was updated. I didn't find that you frozen other matrixs in somewhere. am I right?
Thank you!
No problem, it's my fault!
Hi, thanks for the hard work!
I had some questions about the state of the code right now:
Best,
Sam
Hi,
I'm a graduate student majoring NLP.
I came across your work recent, and I found it very interesting.
I thought I would implement your model(Factored LSTM), and to do so I would need your dataset.
It is written in the paper that you would release the data to community, but apparently it is not available at the moment.
Is there any chance you will release the dataset to the public?
Best
Hwijeen
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.