English

LLMs as Noisy Channels: A Shannon Perspective on Model Capacity and Scaling Laws

Machine Learning 2026-05-25 v1 Artificial Intelligence Information Theory math.IT

Abstract

Existing scaling laws for Large Language Models (LLMs), predominantly monotonic power laws, fail to explain emerging non-monotonic phenomena such as catastrophic overtraining and quantization-induced degradation, where performance deteriorates despite increased compute. We propose the Shannon Scaling Law, a unified theoretical framework that models LLM training as information transmission over a noisy channel, grounded in the Shannon-Hartley theorem. By mapping model parameters to channel bandwidth and training tokens to signal power, our formulation explicitly captures the interaction between learning signal and intrinsic noise. This perspective reveals a fundamental Shannon capacity for LLMs: scaling model size or data without preserving a sufficient signal-to-noise ratio (SNR) inevitably amplifies noise, inducing a transition from monotonic improvement to U-shaped performance degradation. We validate our theory through experiments on Pythia and OLMo2 under perturbations, including Gaussian noise, quantization and supervised fine-tuning on math, QA and code tasks. The Shannon Scaling Law consistently outperforms classical scaling laws and recent perturbation-aware laws, achieving strong R2R^2 scores and accurately capturing loss basins missed by prior approaches. It also extrapolates: fitted on \leq6.9B Pythia models with \leq180B tokens, it predicts the unseen 12B model up to 307B tokens at pooled R2=0.847R^2{=}0.847, while monotonic baselines collapse.

Keywords

Cite

@article{arxiv.2605.23901,
  title  = {LLMs as Noisy Channels: A Shannon Perspective on Model Capacity and Scaling Laws},
  author = {Xu Ouyang and Deyi Liu and Yuhang Cai and Jing Liu and Yuan Yang and Chen Zheng and Thomas Hartvigsen and Yiyuan Ma},
  journal= {arXiv preprint arXiv:2605.23901},
  year   = {2026}
}

Comments

Accepted by ICML 2026