Large Language Models can improve zero-shot information extraction by following annotation guidelines.
GoLLIE improves zero-shot information extraction by following annotation guidelines, outperforming previous attempts.