I am looking to solve a multi-class classification problem with long sequences of text with some rows having 1000's of tokens. Some of the state of the art methods such as BERT have a token limit and I was wondering what is currently being done to handle longer text sequences when dealing with classification?
What are some of the available methods for handling multi-label classification for longer sequences of text
Traditional methods don't have such a limit: Naive Bayes, SVM, decision trees...
Also see https://stackoverflow.com/questions/58636587/how-to-use-bert-for-long-text-classification