arXiv 2310.054702023-12-07cs.CLcs.AI

用于评估对齐的生成式评判器

Generative Judge for Evaluating Alignment

Generative AI Research Lab (GAIR)（生成式人工智能研究实验室（GAIR））；Hong Kong Polytechnic University（香港理工大学）；New York University（纽约大学）；Chinese Academy of Sciences（中国科学院）；Shanghai Jiao Tong University（上海交通大学）；Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）

Junlong Li, Shichao Sun, Weizhe Yuan, Run-Ze Fan, Hai Zhao, Pengfei Liu

备注：: Fix typos in Table 1

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对对齐模型在通用性、灵活性和可解释性评估上的需求，本文提出13B参数生成式评判器Auto-J，并在覆盖58个场景的测试平台上显著超越开源与闭源竞品。

摘要：

大语言模型（LLM）的快速发展大幅扩展了其能够处理的任务范围。在自然语言处理（NLP）领域，研究人员的关注点已从序列标注和句法分析等传统NLP任务，转向围绕契合人类需求的任务，例如头脑风暴和电子邮件撰写。这种任务分布的转变对评估这些对齐模型提出了新要求，涵盖通用性（即在多样场景中评估性能）、灵活性（即在不同协议下检验）和可解释性（即通过解释细致审查模型）。本文提出一个具有13B参数的生成式评判器Auto-J，旨在应对这些挑战。该模型基于海量真实世界场景中的用户查询及LLM生成回复进行训练，可通过结构良好的自然语言评论支持成对回复比较、单回复评估等多种评估协议。为证明方法有效性，作者构建了覆盖58种不同场景的新测试平台。实验中，Auto-J大幅优于包括开源与闭源模型在内的一系列强劲竞争者。本文还提供详细分析和案例研究以进一步揭示该方法的潜力，并在 https://github.com/GAIR-NLP/auto-j 公开多种资源。

英文摘要：

The rapid development of Large Language Models (LLMs) has substantially expanded the range of tasks they can address. In the field of Natural Language Processing (NLP), researchers have shifted their focus from conventional NLP tasks (e.g., sequence tagging and parsing) towards tasks that revolve around aligning with human needs (e.g., brainstorming and email writing). This shift in task distribution imposes new requirements on evaluating these aligned models regarding generality (i.e., assessing performance across diverse scenarios), flexibility (i.e., examining under different protocols), and interpretability (i.e., scrutinizing models with explanations). In this paper, we propose a generative judge with 13B parameters, Auto-J, designed to address these challenges. Our model is trained on user queries and LLM-generated responses under massive real-world scenarios and accommodates diverse evaluation protocols (e.g., pairwise response comparison and single-response evaluation) with well-structured natural language critiques. To demonstrate the efficacy of our approach, we construct a new testbed covering 58 different scenarios. Experimentally, Auto-J outperforms a series of strong competitors, including both open-source and closed-source models, by a large margin. We also provide detailed analysis and case studies to further reveal the potential of our method and make a variety of resources public at https://github.com/GAIR-NLP/auto-j.

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。