AI can learn to show its workings through trial and error
成果类型：
News Item
署名作者：
Ippolito, Daphne; Zhang, Yiming
署名单位：
Carnegie Mellon University
刊物名称：
Nature
ISSN/ISSBN：
0028-1116
DOI：
10.1038/d41586-025-02703-7
发表日期：
2025-09-18
页码：
594-595
关键词：
摘要：
Large language models (LLMs) are more accurate when they output intermediate steps. A strategy called reinforcement can teach them to do this without being told.
来源URL：