arXiv cs.CL· Maor Farid·· 9 小时前AI 评分34
大语言模型作为非线性动力系统:深度、训练与生成的三重时钟
Large Language Models as Nonlinear Dynamical Systems: The Three Clocks of Depth, Training, and Generation
AI 导读
研究者将约200篇论文凝练成“三重时钟词典”,把15个非线性动力学概念映射到Transformer的层迭代、训练步和生成token三个维度,并标注每项对应关系为已证明、已测量或推测。
正文 · 原文
Abstract:Language models are dynamical systems. A transformer is iterated over its layers, over the steps of its training and over the tokens it generates, and on each of these three clocks it shows behavior that nonlinear dynamics has studied for a century: fixed points, bifurcations, synchronization, chaos and noise-driven escape from potential wells. The two communities rarely meet, so this behavior is often rediscovered under new names. We build a bridge between them. To describe, we condense about 200 papers into the Three-Clock Dictionary, which maps fifteen concepts of nonlinear dynamics onto each clock and marks every correspondence as proved, measured or conjectured, and we show each concept in a mechanical system and in a language model side by side. To explain, we derive reduced-order models: the softmax is a Boltzmann distribution, gradient descent an Euler step on a nonlinear spring, a loss spike a ball thrown out of a stiff valley, attention on the circle a Kuramoto model, and a repetition loop a well that deepens while it is occupied. We then test these explanations on open language models, on a laptop CPU. Escape from repetition loops in GPT-2 and Pythia follows the Arrhenius law with the logit gap as the barrier, and the barrier deepens and then saturates, so the temperature needed to leave a loop falls like $1/\ln N$ with the generation length $N$ and rises with model size. Coherent text lives in a narrow window of temperatures, which a small reasoning model shares at its recommended temperature. Training makes each Pythia layer more expansive, and a small transformer trains at a multi-mode edge of stability. The third step, solving problems of language models, is still ahead: controllers derived from the escape analysis remove loops but do not yet beat the best fixed temperature. We state twelve open problems where nonlinear dynamics could get there.
| Comments: | 52 pages, 19 figures. Code and data: this https URL |
| Subjects: | Machine Learning (cs.LG); Computation and Language (cs.CL); Dynamical Systems (math.DS); Chaotic Dynamics (nlin.CD) |
| Cite as: | arXiv:2611.21983 [cs.LG] |
| (or arXiv:2611.21983v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2611.21983 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Maor Farid [view email]
[v1]
Mon, 16 Nov 2026 23:11:52 UTC (847 KB)
来源:arXiv cs.CL · arxiv.org