RoBERTa

We propose pre-trained language models for political Twitter data. We evaluate all models and report results. We release both data and pre-trained models.