manlp-suda/mmesgn

Transformer-based Label Set Generation for Multi-modal Multi-label Emotion Detection

mmesgn's Issues

mosei

你好！请问你使用的数据就是(http://immortal.multicomp.cs.cmu.edu/raw_datasets/processed_data/cmu-mosei/seq_length_50/mosei_senti_data.pkl)数据吗？

关于标签问题

您好，实验中使用的test data的数据就是这种文本标签序列的形式吗？还是您经过转换以后将one-hot向量转化为文本标签的？（就是测试数据中的 {sadness happiness} ）还有一些空行，是因为测试数据里没有情感标记吗？

Question about your code structure

Appreciation for your great work on MMESGN! For the few days, I have tried to fix out the network structure of your work. However, no affense, I failed since you encapsulated the code too well that I can not understand the structure in a short time, can you release a simple example to illustrate how to train your model, lots of thanks!

Can you share the data set

Can you share the data set of "mosei_emotion_aligned_60.pkl"?

关于论文中的Single-model Emotion Set Generation Network base on transformer

您好，很感激您上次为我解答预测标签的问题，这次我想询问您一下有无关于使用transforemer进行序列多标签预测的模型代码，或者有论文和链接参考。谢谢。