Task Residual for Tuning Vision-Language Models | IEEE Conference Publication | IEEE Xplore