ViTamin: Designing Scalable Vision Models in the Vision-Language Era | IEEE Conference Publication | IEEE Xplore