Glad to see others saying this. I haven't looked at it in some months, but I previously realized it's mostly a very complicated way to optimize few-shot learnin

Glad to see others saying this. I haven't looked at it in some months, but I previously realized it's mostly a very complicated way to optimize few-shot learning prompts. It's hardly whatever magical blackbox optimizer they try to market it as.