说“不”的艺术:语言模型中的情境性不遵从
The Art of Saying No: Contextual Noncompliance in Language Models
AI总结:
本文提出情境性不遵从分类体系及1000条评估提示,揭示现有模型在非安全类别中过度遵从的问题,并证明低秩适配器训练能在不损害通用能力的前提下改善不遵从表现。
摘要:
基于聊天的语言模型旨在提供帮助,但它们不应遵从每一个用户请求。虽然现有工作主要关注对“不安全”查询的拒绝,但我们认为不遵从的范围应当被拓宽。我们提出了一个全面的情境性不遵从分类体系,用以描述模型在何时以及如何不应遵从用户请求。该分类体系涵盖广泛的类别,包括不完整请求、无依据请求、不确定请求和拟人化请求(除不安全请求外)。为了测试语言模型的不遵从能力,我们利用该分类体系开发了一个包含1000条不遵从提示的新评估套件。我们发现,大多数现有模型在某些此前未被充分研究的类别中表现出显著较高的遵从率,例如GPT-4等模型错误遵从的请求比例高达30%。为了解决这些差距,我们使用合成生成的请求与预期不遵从响应训练集,探索了不同的训练策略。我们的实验表明,虽然对指令微调模型进行直接微调可能导致过度拒绝和通用能力下降,但使用低秩适配器等参数高效方法有助于在适当的不遵从与其他能力之间取得良好平衡。
英文摘要:
Chat-based language models are designed to be helpful, yet they should not comply with every user request. While most existing work primarily focuses on refusal of "unsafe" queries, we posit that the scope of noncompliance should be broadened. We introduce a comprehensive taxonomy of contextual noncompliance describing when and how models should not comply with user requests. Our taxonomy spans a wide range of categories including incomplete, unsupported, indeterminate, and humanizing requests (in addition to unsafe requests). To test noncompliance capabilities of language models, we use this taxonomy to develop a new evaluation suite of 1000 noncompliance prompts. We find that most existing models show significantly high compliance rates in certain previously understudied categories with models like GPT-4 incorrectly complying with as many as 30% of requests. To address these gaps, we explore different training strategies using a synthetically-generated training set of requests and expected noncompliant responses. Our experiments demonstrate that while direct finetuning of instruction-tuned models can lead to both over-refusal and a decline in general capabilities, using parameter efficient methods like low rank adapters helps to strike a good balance between appropriate noncompliance and other capabilities.
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。