提出新评估框架，揭示现有不确定性探测方法在分布外场景下普遍不鲁棒。
A Robust Evaluation of Probe Robustness: Lessons for Reliable OOD Uncertainty Quantification
- 构建覆盖多种分布偏移的系统性评估框架ProbeDrift
- 2000+探针实验证明当前方法在远距离分布外时性能显著下降
- 发现隐藏层特征类型等设计细节对鲁棒性有隐性影响，适合关注模型可信度的研究者
近期研究发现大语言模型的隐藏状态包含可用于不确定性估计的信号，推动了高效探针式方法的发展。然而，现有方法的鲁棒性尚不明确，先前工作在不同评估设置下得出矛盾结论。为此，我们提出ProbeDrift——一个涵盖多模型、多任务及多种分布偏移的监督式不确定性探针系统评估框架。利用该框架，我们训练超过2000个探针，分离关键设计选择的影响，发现当前方法在近似分布外（near-OOD）之外的场景中表现极差。我们发现，鲁棒性主要由在分布内表现隐蔽的设计决策驱动，包括特征类型、聚合策略和训练信号的选择。我们认为，可靠的不确定性估计需依赖鲁棒的评估。为此，我们开源ProbeDrift轻量级Python库，包含支撑全面评估的训练与测试划分数据集。此外，我们通过简单混合回退（HBO）策略展示如何直接利用评估洞见提升方法鲁棒性。
原文摘要 · Abstract (English)
Recent work has shown that the hidden states of large language models contain signals useful for uncertainty estimation, motivating a growing interest in efficient probe-based approaches. Yet it remains unclear how robust existing methods are, with prior work reporting conflicting conclusions under substantially different evaluation settings. We address this by introducing ProbeDrift, a systematic evaluation framework for supervised uncertainty probes covering a wide range of OOD settings across models, tasks, and distributional shifts. Using ProbeDrift, we train over 2,000 probes to disentangle the effect of key design choices, showing poor robustness of current methods beyond near-OOD settings. We find that robustness is driven by design decisions that have a largely invisible effect in-distribution, including the choice of feature type, aggregation strategy, and training signal. We argue that robust uncertainty estimation requires robust evaluation. To support this, we release ProbeDrift as a lightweight Python library that contains the train and test splits underpinning our extensive evaluation. We also show how insights from our evaluation can directly lead to more robust methods through a simple Hybrid Back-Off (HBO) strategy.
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。