通过控制嵌入模型偏差，显著降低RAG系统整体偏见。
Mitigating Bias in RAG: Controlling the Embedder
- 分析嵌入模型、大模型与语料库间的偏见冲突关系。
- 实验证明反向调节嵌入模型可有效减少系统偏见。
- 不同大模型对嵌入偏差敏感度不同，需针对性调控。
在检索增强生成（RAG）系统中，大语言模型、嵌入模型和语料库各自可能引入偏向特定观点或身份的偏见。本文研究各组件间的偏见冲突及其与系统整体偏见的关系，以性别和政治偏见为案例，发现尽管复杂，但各组件间存在线性关系。通过120组不同偏见水平的嵌入模型微调实验，证明在保持实用性的前提下，可通过反向调节嵌入模型来缓解整体系统偏见。此外，我们发现大模型和任务对嵌入偏见的敏感度各异，是去偏的关键因素。结果表明，通过精细控制嵌入模型偏见，比单纯提升其公平性更能实现更公正的RAG系统。
原文摘要 · Abstract (English)
In retrieval augmented generation (RAG) systems, each individual component -- the LLM, embedder, and corpus -- could introduce biases in the form of skews towards outputting certain perspectives or identities. In this work, we study the conflict between biases of each component and their relationship to the overall bias of the RAG system, which we call bias conflict. Examining both gender and political biases as case studies, we show that bias conflict can be characterized through a linear relationship among components despite its complexity in 6 different LLMs. Through comprehensive fine-tuning experiments creating 120 differently biased embedders, we demonstrate how to control bias while maintaining utility and reveal the importance of reverse-biasing the embedder to mitigate bias in the overall system. Additionally, we find that LLMs and tasks exhibit varying sensitivities to the embedder bias, a crucial factor to consider for debiasing. Our results underscore that a fair RAG system can be better achieved by carefully controlling the bias of the embedder rather than increasing its fairness.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。