STATE OF MARK PROMPTING

Summary from the TLDR AI newsletter

With language models becoming multimodal, we now have more ways than ever to prompt them. State of Mark prompting runs images through Segment Anything first and numbers the detected classes. It then uses that to prompt visual models.