Abstract:We introduce LAMP (Local Attribution Mapping Probe), a method that shines light onto a black-box language model's decision surface and studies how reliably a model maps its stated reasons to its reported predictions by approximating a decision surface. LAMP treats the model's own self-reported explanations as a coordinate system and fits a locally linear surrogate that links those weights to the model's output. By doing so, it reveals how much the stated factors steer the model's decisions. We apply LAMP to three tasks: sentiment analysis, controversial-topic detection, and safety-prompt auditing. Across these tasks, LAMP reveals that many language models' locally approximated linear decision landscapes overall agree with human judgments on explanation quality and, on a clinical case-file data set, align with expert assessments. Since LAMP operates without requiring access to model gradients, logits, or internal activations, it serves as a practical and lightweight framework for auditing proprietary language models, and enabling assessment of whether a model appears to behave consistently with the explanations it provides.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2505.11772 [cs.LG] |
| (or arXiv:2505.11772v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2505.11772 arXiv-issued DOI via DataCite |
From: Ryan Chen [view email]
[v1]
Sat, 17 May 2025 00:43:49 UTC (1,684 KB)
[v2]
Wed, 21 May 2025 03:41:43 UTC (1,684 KB)
[v3]
Sun, 26 Apr 2026 00:14:06 UTC (1,757 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。