AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

KERPLE: Kernelized Relative Positional Embedding for Length Extrapolation

Ta-Chung Chi, Ting-Han Fan, Peter Ramadge, Alexander Rudnicky

NeurIPS 2022 · Architectures and efficiency › Attention and transformers · approach: Long-context KV and memory attention · 9 citations (OpenAlex)

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 changedWhereMeasuredEffect
Use logarithmic relative-position kernel rather than ALiBiGPT-NeoX 162M · OpenWebText2, GitHub, and ArXiv at lengths 4096–16384perplexity: 21.4 $\pm$ 0.6 vs. 22.5 $\pm$ 0.5; 2.75 $\pm$ 0.16 vs. 3.02 $\pm$ 0.13; 5.07 $\pm$ 0.16 vs. 5.78 $\pm$ 0.15Improved
Use the logarithmic kernel rather than the power kernel (ablation)KERPLE · OpenWebText2, GitHub, and ArXiv, all tested extrapolation lengthsperplexity: 21.4 $\pm$ 0.6 vs. 21.6 $\pm$ 0.6; 2.75 $\pm$ 0.16 vs. 2.76 $\pm$ 0.13; 5.07 $\pm$ 0.16 vs. 5.07 $\pm$ 0.19Improved
Use the logarithmic kernel rather than T5 relative biasGPT-NeoX 162M · OpenWebText2, GitHub, and ArXiv at lengths longer than 2048perplexity: OpenWebText2 21.2 vs. 22.5; GitHub 2.74 vs. 2.78; ArXiv 5.20 vs. 5.32Improved
Use the logarithmic kernel rather than T5 relative biasGPT-NeoX 162M · OpenWebText2, GitHub, and ArXiv at length 512perplexity: 23.9 $\pm$ 0.6 vs. 23.7 $\pmb{\pm}$ 0.6; 3.40 $\pm$ 0.20 vs. 3.38 $\pm$ 0.21; 6.07 $\pm$ 0.26 vs. 6.03 $\pmb{\pm}$ 0.26Hurt
Use the logarithmic kernel rather than T5 relative biasGPT-NeoX 162M · GitHub trainingsec/step: 0.307 vs. 0.340Improved
Add a third learnable parameter to the logarithmic kernel (ablation)KERPLE · OpenWebText2, GitHub, and ArXiv at lengths 512–8192perplexity: 21.3 vs. 21.3; 2.68 vs. 2.71; 5.00 vs. 5.01 at length 8192Mixed
Add a distance-dependent weight kernel alongside the power bias (ablation)KERPLE · OpenWebText2, GitHub, and ArXiv, lengths 512–8192perplexity: 24.1 vs. 23.9; 3.44 vs. 3.40; 6.11 vs. 6.07 at length 512Hurt
Use Gaussian-like kernels instead of the logarithmic kernel (ablation)KERPLE · OpenWebText2, GitHub, and ArXiv at extrapolation lengthsperplexity: At length 8192: OpenWebText2 23.2 vs. 21.3; GitHub 3.05 vs. 2.71; ArXiv 5.50 vs. 5.01Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy