MPF：通过多视角融合在部署后对齐和去偏语言模型
MPF: Aligning and Debiasing Language Models post Deployment via Multi Perspective Fusion
AI总结：
本文提出多视角融合框架 MPF，通过将基线分布分解为可解释视角成分并加权采样响应，在部署后以无需微调的方式对齐和缓解 LLM 偏见，实验验证了其在 KL 散度、校准误差和泛化性上的优势。
摘要：
多视角融合（MPF）是一种新颖的大型语言模型（LLMs）训练后对齐框架，旨在应对日益增长的便捷偏见缓解需求。MPF 构建于 SAGED 流水线之上，该流水线是一个用于构建偏见基准并提取可解释基线分布的自动化系统。MPF 利用多视角生成来暴露 LLM 输出中的偏见，并将其与细致且类人的基线对齐。通过将基线（例如来自人力资源专业人员的情绪分布）分解为可解释的视角成分，MPF 在生成过程中通过采样和平衡响应来引导输出，并以分解所得的概率作为权重。在实验上，我们证明了 MPF 能够将 LLM 的情绪分布同时与反事实基线（绝对平等）和 HR 基线（偏向顶尖大学）对齐，从而产生较小的 KL 散度、降低校准误差，并能够泛化到未见问题。这表明 MPF 提供了一种可扩展且可解释的对齐与偏见缓解方法，兼容已部署的 LLM，且无需大量提示工程或微调。
英文摘要：
Multiperspective Fusion (MPF) is a novel posttraining alignment framework for large language models (LLMs) developed in response to the growing need for easy bias mitigation. Built on top of the SAGED pipeline, an automated system for constructing bias benchmarks and extracting interpretable baseline distributions, MPF leverages multiperspective generations to expose and align biases in LLM outputs with nuanced, humanlike baselines. By decomposing baseline, such as sentiment distributions from HR professionals, into interpretable perspective components, MPF guides generation through sampling and balancing of responses, weighted by the probabilities obtained in the decomposition. Empirically, we demonstrate its ability to align LLM sentiment distributions with both counterfactual baselines (absolute equality) and the HR baseline (biased for Top Univeristy), resulting in small KL divergence, reduction of calibration error and generalization to unseen questions. This shows that MPF offers a scalable and interpretable method for alignment and bias mitigation, compatible with deployed LLMs and requiring no extensive prompt engineering or finetuning.
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。
