Are Large Language Models Good Data Preprocessors?

Elyas Meguellati, Nardiena Pratama, Shazia Sadiq +1

High-quality textual training data is essential for the success of multimodal data processing tasks, yet outputs from image captioning models like BLIP and GIT often contain errors…