Yaya Shi

(1) It seems that you only provided a single human rating for each Turing Test and pairwise comparison, or did you combine the results of the 5 workers? (2) What...

Will you release the predictions in Tables 1 and 2? I want to follow your work and do human evaluation use these predictions

似乎您还没有开放 在 Text Summarization 任务上的测试代码，在论文中您提供的关于seq2seq的描述也不是很详细。所以有一些问题想请教您。 您是采用的encoder-decoder 的结构吗？需要另外添加一个decoder, 并在下游数据集上进行微调？

When I try to use torch.utils.checkpoint.checkpoint as follows, and use apex to train the model, I found that the loss is so small as 0.4, but the normal loss is...

I want to know if there is a regression head for MVM during the MILES pretraining phase.

The code for table 1.

1

Thank you very much for releasing the code, but I found that it is not complete, can you provide the code for table 1. Regarding the correlation coefficients of those...

A question about Flickr 8k dataset

1

In your paper, you excluded some sentence when you used flickr 8k. You said: "Because 158 candidates are actual references of target images, we excluded these for further analysis." This...

About Table 1

2

## ❓ Questions & Help I try to implement Table 1. But when calculating the correlation coefficient, I encounter a situation where the output is nan. How do you deal...

it is weird that the function `set_decay_updates` is not used,

I found a "Motion_Excitation" sub-module in the RegionLearner module which operate like a Residual connection. I wonder if using this module would yield better results.