AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

RePo: Language Models with Context Re-Positioning

Huayang Li, Tianyu Zhao, Deng Cai, Richard Sproat

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Attention head mechanisms · 0 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
replace linear RoPE positions with learned re-positioning (RePo)OLMo-2 1B · RULER NIAHaverage accuracy: 85.9 -> 91.3Improved
replace linear RoPE positions with RePoOLMo-2 7B · RULER NIAHaverage accuracy: 97.3 -> 97.9Improved
replace linear RoPE positions with RePoOLMo-2 1B · HybridQA tableexact match: 24.43 -> 26.70Improved
replace linear RoPE positions with RePoOLMo-2 7B · HybridQA tableexact match: 33.52 -> 37.61Improved
replace linear RoPE positions with RePoOLMo-2 1B · LongBench (<16K, YaRN)average score: 20.93 -> 27.86Improved
replace linear RoPE positions with RePoOLMo-2 7B · LongBench (<16K, YaRN)average score: 25.65 -> 32.03Improved
replace linear RoPE positions with RePoOLMo-2 1B · general short-context benchmarksaverage: 53.70 -> 53.73No clear effect
replace linear RoPE positions with RePoOLMo-2 7B · general short-context benchmarksaverage: 70.92 -> 70.30Hurt
remove positional encoding entirely (NoPE) instead of RoPE (ablation)OLMo-2 1B continual pretraining · RULER NIAHaverage accuracy: 85.9 -> 80.2Hurt
NoPE baseline on long-context (ablation)OLMo-2 1B continual pretraining · LongBench (<16K, YaRN)average score: 20.93 -> 6.98Hurt
interleave 2 RoPE and 1 NoPE layers (R2N1) instead of RePo (ablation)OLMo-2 1B · LongBench (<16K, YaRN)average score: 20.93 -> 22.95Mixed
interleave 1 RoPE and 2 NoPE layers (N2R1) (ablation)OLMo-2 1B · LongBench (<16K, YaRN)average score: 20.93 -> 11.29Hurt

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