Full threadalcomatt·Why is it surprising? All the models were trained on more or less the same content.View on HN