SmartCLIP: Modular Vision-language Alignment with Identification Guarantees

Shaoan Xie, Lingjing Kong, Yujia Zheng +5

Contrastive Language-Image Pre-training (CLIP)~\citep{radford2021learning} has emerged as a pivotal model in computer vision and multimodal learning, achieving state-of-the-art per…