The abstract from the paper is the following:
We present BERTweet, the first public large-scale pre-trained language model for English Tweets.