小模型大胜：知识蒸馏与自训练在减少产品问答智能体幻觉中的对比
Winning Big with Small Models: Knowledge Distillation vs. Self-Training for Reducing Hallucination in Product QA Agents
AI总结：
本文针对客户支持中LLM幻觉与高成本问题，提出检索增强QA流水线，证明合成数据优于众包数据，且自训练在减少幻觉上可与知识蒸馏相媲美，为低成本可扩展QA系统提供了新思路。
摘要：
大型语言模型（LLMs）在客户支持中的部署受到幻觉（生成虚假信息）和专有模型高成本的制约。为应对这些挑战，我们提出了一种检索增强的问答（QA）流水线，并探索如何在人工输入与自动化之间取得平衡。利用一个关于三星智能电视用户手册的问题数据集，我们证明，由LLMs生成的合成数据在减少微调模型幻觉方面优于众包数据。我们还比较了自训练（在模型自身输出上进行微调）和知识蒸馏（在更强模型的输出上进行微调，例如GPT-4o），发现自训练实现了可比的幻觉减少效果。我们推测，这一令人意外的发现可归因于知识蒸馏情况下增加的曝光偏差问题，并通过事后分析支持了这一推测。我们还通过上下文化的"我不知道"回应，提高了对不可回答问题和检索失败的鲁棒性。这些发现表明，可以使用合成数据和基于开源模型的自训练来构建可扩展、成本高效的QA系统，从而减少对专有工具或昂贵人工标注的依赖。
英文摘要：
The deployment of Large Language Models (LLMs) in customer support is constrained by hallucination (generating false information) and the high cost of proprietary models. To address these challenges, we propose a retrieval-augmented question-answering (QA) pipeline and explore how to balance human input and automation. Using a dataset of questions about a Samsung Smart TV user manual, we demonstrate that synthetic data generated by LLMs outperforms crowdsourced data in reducing hallucination in finetuned models. We also compare self-training (fine-tuning models on their own outputs) and knowledge distillation (fine-tuning on stronger models' outputs, e.g., GPT-4o), and find that self-training achieves comparable hallucination reduction. We conjecture that this surprising finding can be attributed to increased exposure bias issues in the knowledge distillation case and support this conjecture with post hoc analysis. We also improve robustness to unanswerable questions and retrieval failures with contextualized "I don't know" responses. These findings show that scalable, cost-efficient QA systems can be built using synthetic data and self-training with open-source models, reducing reliance on proprietary tools or costly human annotations.
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。
