Research map / Large language models / Language, safety and interpretability
Intrinsic Entropy of Context Length Scaling in LLMs
Jingzhe Shi, Qinwei Ma, Hongyi Liu, Hang Zhao, Jenq-Neng Hwang, Lei Li
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 |
|---|---|---|---|
| increase training context length beyond an optimum with fixed training data (ablation) | GPT-2 style model (6 layers) · OpenWebText subset | validation loss | Hurt |
| increase training dataset size (ablation) | GPT-2 style model · OpenWebText subset (200M to 750M tokens) | optimal context length | Improved |
| increase visible context beyond optimum at inference (ablation) | Llama-3.1-8B · Ruler-QA1 (uniform, 16k max) | accuracy: peak near 6k then drops | Hurt |
| use a task needing more long context (smaller gamma) (ablation) | Llama-3.1-8B · Position-Weighted Ruler-QA1 | optimal context length | Improved |
| use larger model size (ablation) | Qwen3 series · RULER cwe | optimal context length | Improved |
| increase visible context (ablation) | Qwen3 series · RULER vt (variable tracking) | accuracy | Improved |
Closest papers by meaning
- Understanding LLM Behaviors via Compression: Data Generation, Knowledge Acquisition and Scaling Laws (NeurIPS 2025 · similarity 0.91)
- Scaling Laws with Vocabulary: Larger Models Deserve Larger Vocabularies (NeurIPS 2024 · similarity 0.90)
- L$^2$M: Mutual Information Scaling Law for Long-Context Language Modeling (NeurIPS 2025 · similarity 0.90)
- LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws (ICML 2025 · similarity 0.90)
- What is Wrong with Perplexity for Long-context Language Modeling? (ICLR 2025 · similarity 0.89)
- Position: Enough of Scaling LLMs! Lets Focus on Downscaling (ICML 2025 · similarity 0.89)
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.
