Rumour Evaluation with Very Large Language Models