Recently we have been doing some experiments to cluster semantically similar messages, by leveraging pre-trained models so we can get something off the ground using no labelled data. Task here is…