Logo image
As Language Models Scale, Low-order Linear Depth Dynamics Emerge
Preprint   Open access

As Language Models Scale, Low-order Linear Depth Dynamics Emerge

Buddhika Nettasinghe and Geethu Joseph
ArXiv.org
Cornell University
03/13/2026
DOI: 10.48550/arxiv.2603.12541
url
https://doi.org/10.48550/arxiv.2603.12541View
Preprint (Author's original)This preprint has not been evaluated by subject experts through peer review. Preprints may undergo extensive changes and/or become peer-reviewed journal articles. Open Access

Abstract

Large language models are often viewed as high-dimensional nonlinear systems and treated as black boxes. Here, we show that transformer depth dynamics admit accurate low-order linear surrogates within context. Across tasks including toxicity, irony, hate speech and sentiment, a 32-dimensional linear surrogate reproduces the layerwise sensitivity profile of GPT-2-large with near-perfect agreement, capturing how the final output shifts under additive injections at each layer. We then uncover a surprising scaling principle: for a fixed-order linear surrogate, agreement with the full model improves monotonically with model size across the GPT-2 family. This linear surrogate also enables principled multi-layer interventions that require less energy than standard heuristic schedules when applied to the full model. Together, our results reveal that as language models scale, low-order linear depth dynamics emerge within contexts, offering a systems-theoretic foundation for analyzing and controlling them.
Computer Science - Learning Computer Science - Systems and Control

Details

Metrics

1 Record Views
Logo image