About captions generation

[behome]

I got a problem when generating captions. The model trained by myself always generate the word 'a' which is the most frequent word. I have done some changes on your code. So I wanna know what is the most likely problem.

[xinli2008]

Have you soloved this problem?

[behome]

I have solved this problem. The model which is implemented by myself had something wrong. And the hyper-parameters need to adjust. The best solution is to find the original open source code of the paper.

[xinli2008]

congratulation，can you give me your email ?Please do me a favor .I want to send you some pictures, and can you put the picture to your model ,and send me the result? Because i want to know the result about same picture when putting into image caption model and visual storytelling model. Thank you ！

[behome]

[email protected] That's my email. I have been very busy with my graduation thesis recently. So I may not be able to help you finish it in a short time.