Can large language models identify and correct their mistakes?

Posted by Gladys Tyen, Intern, Google Research LLMs are increasingly popular for reasoning tasks, such as multi-turn QA, task completion, code gene...