用信息瓶颈改进大模型不确定性估计，让模型更可信。
Calibrating LLMs with Information-Theoretic Evidential Deep Learning
- 引入信息瓶颈约束证据生成，抑制无关信息干扰。
- 在多个任务上显著提升校准度，优于传统EDL和非不确定性方法。
- 适合对置信度要求高的应用场景，如医疗、金融决策。
微调后的大型语言模型（LLM）在小数据集上常表现出过度自信，导致校准性差和不确定性估计不准。证据深度学习（EDL）是一种可实现单次前向传播的不确定性感知方法，具有计算效率优势，但其训练目标易引发过拟合，导致概率分布过于集中。为此，本文提出通过引入信息瓶颈（IB）正则化EDL，构建IB-EDL。该方法抑制模型生成证据中的虚假信息，促进真正预测性信息同时影响输出与不确定性估计。在多种微调过的LLM和任务上的大量实验表明，IB-EDL在校准性能上超越现有EDL及非EDL方法。该方法提升了LLM的可信度，推动其在需高置信度校准的领域（如医疗、金融）的应用。代码已开源：https://github.com/sandylaker/ib-edl。
原文摘要 · Abstract (English)
Fine-tuned large language models (LLMs) often exhibit overconfidence, particularly when trained on small datasets, resulting in poor calibration and inaccurate uncertainty estimates. Evidential Deep Learning (EDL), an uncertainty-aware approach, enables uncertainty estimation in a single forward pass, making it a promising method for calibrating fine-tuned LLMs. However, despite its computational efficiency, EDL is prone to overfitting, as its training objective can result in overly concentrated probability distributions. To mitigate this, we propose regularizing EDL by incorporating an information bottleneck (IB). Our approach IB-EDL suppresses spurious information in the evidence generated by the model and encourages truly predictive information to influence both the predictions and uncertainty estimates. Extensive experiments across various fine-tuned LLMs and tasks demonstrate that IB-EDL outperforms both existing EDL and non-EDL approaches. By improving the trustworthiness of LLMs, IB-EDL facilitates their broader adoption in domains requiring high levels of confidence calibration. Code is available at https://github.com/sandylaker/ib-edl.
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。