Jamie

Question about Eq(8)

2

My question is, in Eq(8), **v** represents the spatial image features, and I think it should be time-invariant, so why it has a subscript "t"? Thanks.

When I train the model with RL using run_att_d.sh, the CIDEr score got a significant drop, I saw that you have provided the curve of training VSE model#11, I wonder...

The previous code will swap the current box with the last box if the confidence score of previous box falls below threshold. However, it does not check the confidence score...

The split in this repo is different from the split on https://cs.stanford.edu/people/karpathy/deepimagesent/caption_datasets.zip. (AKA Karpathy's split in many papers. see also https://github.com/kelvinxu/arctic-captions/tree/master/splits) In this repo, images of training and val set...

when I use the keyword: '热水器+浴室' to search on the website of Baidu, I got the correct result:  however, when I use the same keyword in icrawler, I got...

Didn't use Karpathy's split

1

The result of the paper was reported using Karpathy's split(http://cs.stanford.edu/people/karpathy/ deepimagesent/), while this repo didn't do so, just note that for your information.

Is there any way to visualize the attention weights when generating captions?

2

like this 

set -g @resurrect-save 'S' # 修改保存指令为S set -g @resurrect-restore 'R' 修改恢复指令为R -> set -g @resurrect-save 'S' # 修改保存指令为S set -g @resurrect-restore 'R' # 修改恢复指令为R 少了一个#