Thank you very much for exposing the MCAN implementation.
I downloaded the Weibo dataset from the link you provided, but when I tried to use it, I found that the posts and pictures in it did not match exactly. For different posts, the data set seemed to provide different numbers of pictures (for example, some provided 0 picture while some provided many pictures), so I was not sure what to do after I got the data set to make proper use of it.
I did not find a solution to the relevant problem in the data_process_weibo.py . Could you please provide more detailed preprocessing method for weibo dataset?
It will be beneficial if you are willing to reply to this issue.
Hi Yang Wu,
Thank you very much for exposing the MCAN implementation.
I downloaded the Weibo dataset from the link you provided, but when I tried to use it, I found that the posts and pictures in it did not match exactly. For different posts, the data set seemed to provide different numbers of pictures (for example, some provided 0 picture while some provided many pictures), so I was not sure what to do after I got the data set to make proper use of it.
I did not find a solution to the relevant problem in the
data_process_weibo.py
. Could you please provide more detailed preprocessing method for weibo dataset?It will be beneficial if you are willing to reply to this issue.
Thank you and sorry for bothering you.