Hi, the paper claimed "We have additionally collected a testing dataset for evaluation with 100-300 image condition-prompt triplets for each task. The source data is collected from Laion and COCO."
Is these test dataset available to the public?
Looking forward to your reply!