CONTROLLING VISION-LANGUAGE MODELS FOR MULTI-TASK IMAGE RESTORATION
CONTROLLINGVISION-LANGUAGEMODELSFORMULTI-TASKIMAGERESTORATION(Paperreading)ZiweiLuo,UppsalaUniversity,ICLRunderreview(6663),Cited:None,Stars:350+,Code,Paper.1.前言像CLIP这样的视觉语言模型已经显示出对零样本或无标签预测的各种下游任务的巨大