You know, one could probably build a training loop to finetine LLaMA/Alpaca for SD prompting. - Generate a randomized prompt with the LLM - Run it with SD -

You know, one could probably build a training loop to finetine LLaMA/Alpaca for SD prompting. - Generate a randomized prompt with the LLM - Run it with SD - Rate it with an aesthetic score - Train on that with a high score as a positive weight - Repeat