tweet training dataset

As mentioned in DeepMoji GitHub repo [https://github.com/bfelbo/DeepMoji](https://github.com/bfelbo/DeepMoji), the large Twitter dataset of tweets with emojis is not released.

I wonder if there is still a chance to get the original training dataset, even a permission is required. If I understand correctly, torchMoji is also trained on the same dataset, right? Could you share how you get the training dataset? In the original paper, I saw the authors wrote

> The authors would like to thank Janys Analytics for generously allowing us to use their dataset ofhuman-rated tweets

Should I contact [**Janys Analytics**](http://www.janysanalytics.com/) in order to get the training dataset?