Fetching the paper…

CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing · Around