TruthFlow：通过表示流校正实现真实的大语言模型生成
TruthFlow: Truthful LLM Generation via Representation Flow Correction
AI总结：
TruthFlow利用Flow Matching技术学习查询特定的表示校正向量，将LLM表示从幻觉状态引导至真实状态，在TruthfulQA上显著提升多种先进LLM的真实性，并展现出良好的跨基准迁移能力。
摘要：
大语言模型（LLMs）在持续生成真实响应方面存在困难。尽管已经提出了各种表示干预技术，但这些方法通常对所有输入查询应用一个通用的表示校正向量，限制了它们在实践中对多样化查询的有效性。在本研究中，我们提出了TruthFlow，一种利用Flow Matching技术进行查询特定真实表示校正的新方法。具体而言，TruthFlow首先使用一个流模型来学习查询特定的校正向量，这些向量将表示从幻觉状态过渡到真实状态。然后，在推理过程中，训练好的流模型生成这些校正向量，以增强LLM输出的真实性。实验结果表明，TruthFlow在TruthfulQA上评估的各种先进LLMs的开放式生成任务中显著提升了性能。此外，训练好的TruthFlow模型表现出强大的迁移能力，在其他未见过的幻觉基准测试上也能有效运行。
英文摘要：
Large language models (LLMs) are known to struggle with consistently generating truthful responses. While various representation intervention techniques have been proposed, these methods typically apply a universal representation correction vector to all input queries, limiting their effectiveness against diverse queries in practice. In this study, we introduce TruthFlow, a novel method that leverages the Flow Matching technique for query-specific truthful representation correction. Specifically, TruthFlow first uses a flow model to learn query-specific correction vectors that transition representations from hallucinated to truthful states. Then, during inference, the trained flow model generates these correction vectors to enhance the truthfulness of LLM outputs. Experimental results demonstrate that TruthFlow significantly improves performance on open-ended generation tasks across various advanced LLMs evaluated on TruthfulQA. Moreover, the trained TruthFlow model exhibits strong transferability, performing effectively on other unseen hallucination benchmarks.
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。
