arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2408.02666cs.CLcs.AI
收藏

自学评估器

Self-Taught Evaluators

Meta FAIR

Tianlu Wang, Ilia Kulikov, Olga Golovneva, Ping Yu, Weizhe Yuan, Jane Dwivedi-Yu, Richard Yuanzhe Pang, Maryam Fazel-Zarandi, Jason Weston, Xian Li

AI总结:

本文提出自学评估器方法,仅用合成数据迭代训练LLM-as-a-Judge,无需人工标注即可将Llama3-70B-Instruct在RewardBench上从75.4提升至88.3,超越GPT-4并媲美顶级标注训练的奖励模型。

摘要:

基于模型的评估是成功开发模型的核心——既可作为训练的奖励模型,也可替代人工评估。训练此类评估器的标准方法是收集大量关于模型响应的人工偏好判断,这既昂贵,且随着模型改进数据会变得过时。在这项工作中,我们提出一种旨在无需人工标注、仅使用合成训练数据来改进评估器的方法。从未标注的指令开始,我们的迭代自改进方案生成对比性模型输出,并训练一个LLM-as-a-Judge以产生推理轨迹和最终判断,在每次新迭代中使用改进后的预测重复此训练。在没有任何标注偏好数据的情况下,我们的自学评估器能将一个强大的LLM(Llama3-70B-Instruct)在RewardBench上从75.4提升到88.3(多数投票下为88.7)。这优于常用的LLM评判器(如GPT-4),并可与使用标注样本训练的顶级奖励模型的表现相媲美。

英文摘要:

Model-based evaluation is at the heart of successful model development -- as a reward model for training, and as a replacement for human evaluation. To train such evaluators, the standard approach is to collect a large amount of human preference judgments over model responses, which is costly and the data becomes stale as models improve. In this work, we present an approach that aims to im-prove evaluators without human annotations, using synthetic training data only. Starting from unlabeled instructions, our iterative self-improvement scheme generates contrasting model outputs and trains an LLM-as-a-Judge to produce reasoning traces and final judgments, repeating this training at each new iteration using the improved predictions. Without any labeled preference data, our Self-Taught Evaluator can improve a strong LLM (Llama3-70B-Instruct) from 75.4 to 88.3 (88.7 with majority vote) on RewardBench. This outperforms commonly used LLM judges such as GPT-4 and matches the performance of the top-performing reward models trained with labeled examples.

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。

↑