arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2405.01535cs.CL
收藏

Prometheus 2:一款专门用于评估其他语言模型的开源语言模型

Prometheus 2: An Open Source Language Model Specialized in Evaluating Other Language Models

KAIST(韩国科学技术院);LG AI Research(LG人工智能研究院);Carnegie Mellon University(卡内基梅隆大学);MIT(麻省理工学院);Allen Institute for AI(艾伦人工智能研究所);University of Illinois Chicago(芝加哥伊利诺伊大学)

Seungone Kim, Juyoung Suk, Shayne Longpre, Bill Yuchen Lin, Jamin Shin, Sean Welleck, Graham Neubig, Moontae Lee, Kyungjae Lee, Minjoon Seo

AI总结:

针对现有开源评估器LM与人类评分差异大、评估形式单一且无法自定义标准的问题,提出Prometheus 2,其能紧密匹配人类和GPT-4判断,支持直接评估、成对排序及自定义标准,在多基准测试中相关性和一致性居开源评估器首位。

摘要:

GPT-4等专有语言模型(LM)常被用于评估各类LM的响应质量,但透明度、可控性和经济性方面的担忧强烈推动了专门用于评估的开源LM的发展。另一方面,现有开源评估器LM存在关键缺陷:1)其给出的分数与人类评分存在显著差异;2)缺乏执行直接评估和成对排序这两种最常见评估形式的灵活性。此外,它们无法基于自定义评估标准进行评估,而是仅关注有用性和无害性等通用属性。为解决这些问题,我们引入Prometheus 2,这是一款比其前身更强大的评估器LM,能更紧密地反映人类和GPT-4的判断。此外,它能够处理结合用户定义评估标准的直接评估和成对排序格式。在四个直接评估基准和四个成对排序基准上,Prometheus 2在所有测试的开源评估器LM中与人类和专有LM评估者的相关性和一致性最高。我们的模型、代码和数据均可在https://github.com/prometheus-eval/prometheus-eval公开获取。

英文摘要:

Proprietary LMs such as GPT-4 are often employed to assess the quality of responses from various LMs. However, concerns including transparency, controllability, and affordability strongly motivate the development of open-source LMs specialized in evaluations. On the other hand, existing open evaluator LMs exhibit critical shortcomings: 1) they issue scores that significantly diverge from those assigned by humans, and 2) they lack the flexibility to perform both direct assessment and pairwise ranking, the two most prevalent forms of assessment. Additionally, they do not possess the ability to evaluate based on custom evaluation criteria, focusing instead on general attributes like helpfulness and harmlessness. To address these issues, we introduce Prometheus 2, a more powerful evaluator LM than its predecessor that closely mirrors human and GPT-4 judgements. Moreover, it is capable of processing both direct assessment and pair-wise ranking formats grouped with a user-defined evaluation criteria. On four direct assessment benchmarks and four pairwise ranking benchmarks, Prometheus 2 scores the highest correlation and agreement with humans and proprietary LM judges among all tested open evaluator LMs. Our models, code, and data are all publicly available at https://github.com/prometheus-eval/prometheus-eval.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑