Please read the description for the models i put out. There is some good information that may increase the quality of your output when using my models.
Description
-Big dataset 120k images
-Lots of images from rule 34
-Images that has a high score from r34 are tagged with "highly rated"
-Images have been aesthetically predicted and images with a higher score than 6.2 are tagged with "high aesthetic". Images with a score lower than 6.2 are rated with "low aesthetic
-Trained at 640x640
-Clip skip 1