CT Multi-Task Learning with a Large Image-Text (LIT) Model