通过基于激活的置信度校准与引导解码增强语言模型事实性
Enhancing Language Model Factuality via Activation-Based Confidence Calibration and Guided Decoding
AI总结：
本文提出基于激活的校准方法ActCab和置信度引导解码策略CoDec，通过在LM最后一层激活上训练线性层捕捉知识表示，并利用置信度信号引导生成高置信度真实答案，在多个QA基准上显著降低校准误差并提升事实性。
摘要：
校准语言模型（LMs）使其生成置信度与答案正确的实际可能性对齐，这能告知用户LMs的可靠性并减轻幻觉内容。然而，先前的校准方法，如基于自洽性和基于logit的方法，要么在推理时效率有限，要么无法提供有信息量的信号。此外，简单地过滤掉低置信度响应会在答案正确时降低LM的帮助性。因此，有效使用校准技术增强LM的事实性仍是一个未解决的挑战。在本文中，我们首先提出一种基于激活的校准方法ActCab，该方法在LM的最后一层激活之上训练一个线性层，能更好地捕捉知识的表示。在ActCab基础上，我们进一步提出CoDec，一种置信度引导的解码策略，以从LMs中引出高置信度的真实答案。通过在五个流行的QA基准上评估，ActCab实现了优于所有竞争基线的校准性能，例如，将平均预期校准误差（ECE）分数降低高达39%。对CoDec的进一步实验表明，在TruthfulQA等具有挑战性的QA数据集上，多个LMs的事实性有一致提升，突显了置信度信号在增强事实性方面的价值。
英文摘要：
Calibrating language models (LMs) aligns their generation confidence with the actual likelihood of answer correctness, which can inform users about LMs' reliability and mitigate hallucinated content. However, prior calibration methods, such as self-consistency-based and logit-based approaches, are either limited in inference-time efficiency or fall short of providing informative signals. Moreover, simply filtering out low-confidence responses reduces the LM's helpfulness when the answers are correct. Therefore, effectively using calibration techniques to enhance an LM's factuality remains an unsolved challenge. In this paper, we first propose an activation-based calibration method, ActCab, which trains a linear layer on top of the LM's last-layer activations that can better capture the representations of knowledge. Built on top of ActCab, we further propose CoDec, a confidence-guided decoding strategy to elicit truthful answers with high confidence from LMs. By evaluating on five popular QA benchmarks, ActCab achieves superior calibration performance than all competitive baselines, e.g., by reducing the average expected calibration error (ECE) score by up to 39%. Further experiments on CoDec show consistent improvements in several LMs' factuality on challenging QA datasets, such as TruthfulQA, highlighting the value of confidence signals in enhancing factuality.
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。
