跳到正文
arXiv cs.AI· Nathan Sheffield, Alek Westover, Lukas Finnveden, Alexa Pan, Julian Stastny, Ryan Greenblatt·· 4 小时前AI 评分55

论文提出 NLS 深度:量化模型不透明串行思考深度的可操作指标

An Operationalization of Opaque Serial Depth

AI 导读

Redwood Research 研究者在 arXiv 论文(arXiv:2611.17495)中提出不透明串行深度的一种可操作化度量 NLS depth,即模型计算中最长的不经过可解释瓶颈的路径长度,并定义了 NL-rooted 节点作为瓶颈标准。

正文 · 原文

View PDF HTML (experimental)

Abstract:Chain-of-thought (CoT) is currently a valuable tool for overseeing AI models, but some architectural shifts could significantly reduce CoT monitorability. We operationalize a measure that serves as a proxy for the amount of unverbalized serial cognition a model can perform, as a specific instantiation of the notion of opaque serial depth: roughly, the length of the longest path through a model's computation that does not pass through an interpretable bottleneck. Our main contribution is a particular standard for what counts as such a bottleneck. We define natural-language-rooted (NL-rooted) nodes as nodes that were initialized from a pre-training prior to produce human-understandable outputs, whose output space has not been expanded beyond that prior, and whose outputs the model has not been modified to interpret as a different type of data. By construction, the resulting measure, NLS depth, can be adjudicated before training from an architecture, training recipe, and deployment configuration alone, and is insensitive to ordinary post-training such as prompting, distillation, and RL. We find that, across open-weight models, NLS depth correlates strongly with layer count, with some exceptions, and has grown slowly over recent years. It would likely grow slowly under continued scaling of standard transformers, while latent-reasoning architectures such as full-bandwidth transformers could increase it by a large amount. We work through how NLS depth applies to a range of existing and proposed architectures, discuss alternative operationalizations, and give a worked example bounding the depth of Gemma3-1B. We argue that AI developers should track and report the NLS depth of their models as a complement to direct measurements of CoT monitorability.
Comments: 32 pages, 6 figures. An earlier version appeared on the Redwood Research blog
Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Cite as: arXiv:2611.17495 [cs.AI]
  (or arXiv:2611.17495v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.17495

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Alexa Pan [view email]
[v1] Thu, 12 Nov 2026 20:34:11 UTC (3,329 KB)

来源:arXiv cs.AI · arxiv.org