Research map / Large language models / Language, safety and interpretability
Deriving Neural Scaling Laws from the Statistics of Natural Language
Francesco Cagnetta, Allan Raventos, Surya Ganguli, Matthieu Wyart
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| rescale n-gram losses by n^-gamma and data by n^(2 beta) | n-gram learning curves of GPT-2 style transformers · TinyStories and WikiText | curve collapse | Improved |
| vary architecture (GPT-2 APE, RoPE, LLaMA, Mamba, n-gram) for estimating gamma (ablation) | conditional entropy decay exponent · TinyStories | gamma | No clear effect |
| increase maximal context length T (ablation) | autoregressive loss · TinyStories | power-law decay with P | No clear effect |
| take beta outside its standard-error interval (ablation) | scaling collapse · TinyStories | collapse quality | Hurt |
| take gamma outside [0.31, 0.34] (ablation) | scaling collapse · TinyStories | collapse quality | Hurt |
| take gamma outside [0.23, 0.30] (ablation) | scaling collapse · WikiText | collapse quality | Hurt |
| use beta in [0.6, 0.7] instead of measured beta (ablation) | LLaMA model · TinyStories | collapse sharpness | Improved |
| increase P by 10x beyond threshold; within-horizon excess loss | n-gram losses n<=12 · TinyStories GPT-2 T=128 | decay exponent delta_n: delta_n > gamma/(2beta) | Improved |
Closest papers by meaning
- On the Invariance and Generality of Neural Scaling Laws (NeurIPS 2026 · similarity 0.92)
- Tokens-per-Parameter Coverage Is Critical for Robust LLM Scaling Law Extrapolation (NeurIPS 2026 · similarity 0.91)
- Universal One-third Time Scaling in Learning Peaked Distributions (ICML 2026 · similarity 0.91)
- Scaling Law with Learning Rate Annealing (NeurIPS 2025 · similarity 0.91)
- A Hitchhiker's Guide to Scaling Law Estimation (ICML 2025 · similarity 0.91)
- Neural Neural Scaling Laws (NeurIPS 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
