Post by @paul2webster • Hey

Chameleon: Plug-and-Play Compositional Reasoning with Large Language Models Chameleon with GPT-4 achieves an 86.54% accuracy on ScienceQA, significantly i