Research map / Architectures and efficiency / Attention and transformers
Quantitative Bounds for Length Generalization in Transformers
Zachary Izzo, Eshaan Nichani, Jason D. Lee
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 sequence length | one-layer transformer · SimpleTask | plateau test loss | Improved |
| Increase frequency omega of target function (larger weight norm L) | one-layer transformer · SimpleTask | plateau test loss | Hurt |
| Increase training sequence length | one-layer transformer · ModPTask | plateau test loss | Improved |
| Increase period Delta | one-layer transformer · ModPTask | plateau test loss | Hurt |
| Increase training sequence length | two-layer transformer · in-context k-gram | plateau test loss | Improved |
| Increase alphabet size S or k | two-layer transformer · in-context k-gram | plateau test loss | Hurt |
Closest papers by meaning
- A Formal Framework for Understanding Length Generalization in Transformers (ICLR 2025 · similarity 0.95)
- Arithmetic Transformers Can Length-Generalize in Both Operand Length and Count (ICLR 2025 · similarity 0.92)
- Length Generalization for Transformers via Compression (NeurIPS 2026 · similarity 0.91)
- Two Heads are Better than One: Simulating Large Transformers with Small Ones (NeurIPS 2025 · similarity 0.90)
- What Algorithms can Transformers Learn? A Study in Length Generalization (ICLR 2024 · similarity 0.90)
- Extrapolation by Association: Length Generalization Transfer In Transformers (NeurIPS 2025 · 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.
