A rumour stance classification model, intended to classify the stance of a "reply" post with respect to the post to which it is replying. This is a target oblivious model, i.e. it considers only the reply itself, not the original target post. It is the model underlying the GATE Cloud Target Oblivious Stance Classifier.

For background, see this paper.

This model is fine-tuned from a multilingual base model (cardiffnlp/twitter-xlm-roberta-base), however the fine-tuning data is English-only.