Reduced image count, and improved tag dataset. Does anyone else think this is somehow worse?
This was inspired by a similar model for SD 1.5.
Does anyone else think V2 is somehow worse? Let me know.
