This AI Paper from UCSD and ByteDance Proposes a Novel Machine Learning Framework for Filtering Image-Text Data by Leveraging Fine-Tuned Multimodal Language Models (MLMs)