English
Related papers

Related papers: Rethinking RoPE: A Mathematical Blueprint for N-di…

200 papers

Rotary Positional Encoding (RoPE) is widely used in modern large language models. However, when sequences are extended beyond the range seen during training, rotary phases can enter out-of-distribution regimes, leading to spurious…

Machine Learning · Computer Science 2026-05-12 Riccardo Ali , Alessio Borgi , Christopher Irwin , Mario Severino , Pietro Liò

Length extrapolation algorithms based on Rotary position embedding (RoPE) have shown promising results in extending the context length of language models. However, understanding how position embedding can capture longer-range contextual…

Computation and Language · Computer Science 2024-10-22 Xiangyu Hong , Che Jiang , Biqing Qi , Fandong Meng , Mo Yu , Bowen Zhou , Jie Zhou

Large Language Models (LLMs) often struggle to process and generate coherent context when the number of input tokens exceeds the pre-trained length. Recent advancements in long-context extension have significantly expanded the context…

Computation and Language · Computer Science 2025-04-29 Yi Lu , Wanxu Zhao , Xin Zhou , Chenxin An , Chenglong Wang , Shuo Li , Yuming Yang , Jun Zhao , Tao Ji , Tao Gui , Qi Zhang , Xuanjing Huang

We study the extent to which rotary position encodings (RoPE), a recent transformer position encoding algorithm broadly adopted in large language models (LLMs) and vision transformers (ViTs), can be applied to graph-structured data. We find…

We identify intrinsic limitations of Rotary Positional Embeddings (RoPE) in Transformer-based long-context language models. Our theoretical analysis abstracts away from the specific content of the context and depends only on its length. We…

Computation and Language · Computer Science 2026-05-18 Yufeng Du , Phillip Harris , Minyang Tian , Eliu A Huerta , Srikanth Ronanki , Subendhu Rongali , Aram Galstyan , Hao Peng

The attention mechanism in a Transformer architecture matches key to query based on both content -- the what -- and position in a sequence -- the where. We present an analysis indicating that what and where are entangled in the popular RoPE…

Machine Learning · Computer Science 2025-12-24 Anand Gopalakrishnan , Robert Csordás , Jürgen Schmidhuber , Michael C. Mozer

Positional embeddings (PE) play a crucial role in Vision Transformers (ViTs) by providing spatial information otherwise lost due to the permutation invariant nature of self attention. While absolute positional embeddings (APE) have shown…

Computer Vision and Pattern Recognition · Computer Science 2026-04-14 Md Abtahi Majeed Chowdhury , Md Rifat Ur Rahman , Akil Ahmad Taki

Relative position encoding (RPE) is important for transformer to capture sequence ordering of input tokens. General efficacy has been proven in natural language processing. However, in computer vision, its efficacy is not well studied and…

Computer Vision and Pattern Recognition · Computer Science 2021-07-30 Kan Wu , Houwen Peng , Minghao Chen , Jianlong Fu , Hongyang Chao

Although large language models (LLMs) have achieved significant progress in handling long-context inputs, they still suffer from the ``lost-in-the-middle'' problem, where crucial information in the middle of the context is often…

Computation and Language · Computer Science 2025-03-07 Zhenghua Wang , Yiran Ding , Changze Lv , Zhibo Xu , Tianlong Li , Tianyuan Shi , Xiaoqing Zheng , Xuanjing Huang

Rotary Position Embedding (RoPE) is an efficient position encoding approach and is widely utilized in numerous large language models (LLMs). Recently, a lot of methods have been put forward to further expand the context window based on…

Computation and Language · Computer Science 2025-05-20 Wenqiao Zhu , Chao Xu , Lulu Wang , Jun Wu

Characterizing the express power of the Transformer architecture is critical to understanding its capacity limits and scaling law. Recent works provide the circuit complexity bounds to Transformer-like architecture. On the other hand,…

Machine Learning · Computer Science 2024-12-03 Bo Chen , Xiaoyu Li , Yingyu Liang , Jiangxuan Long , Zhenmei Shi , Zhao Song

Enabling LLMs to handle lengthy context is currently a research hotspot. Most LLMs are built upon rotary position embedding (RoPE), a popular position encoding method. Therefore, a prominent path is to extrapolate the RoPE trained on…

Computation and Language · Computer Science 2024-12-13 Meizhi Zhong , Chen Zhang , Yikun Lei , Xikai Liu , Yan Gao , Yao Hu , Kehai Chen , Min Zhang

Since self-attention layers in Transformers are permutation invariant by design, positional encodings must be explicitly incorporated to enable spatial understanding. However, fixed-size lookup tables used in traditional learnable position…

Machine Learning · Computer Science 2025-06-18 Huayang Li , Yahui Liu , Hongyu Sun , Deng Cai , Leyang Cui , Wei Bi , Peilin Zhao , Taro Watanabe

Positional encodings are essential to transformer-based generative models, yet their behavior in multimodal and attention-sharing settings is not fully understood. In this work, we present a principled analysis of Rotary Positional…

Graphics · Computer Science 2026-02-06 Aryan Mikaeili , Or Patashnik , Andrea Tagliasacchi , Daniel Cohen-Or , Ali Mahdavi-Amiri

The ability to process ultra-long contexts is crucial for large language models (LLMs) to perform long-horizon tasks. While recent efforts have extended context windows to 1M and beyond, model performance degrades when sequence length…

Computation and Language · Computer Science 2026-05-28 Simin Huo

This paper addresses the challenge of train-short-test-long (TSTL) scenarios in Large Language Models (LLMs) equipped with Rotary Position Embedding (RoPE), where models pre-trained on shorter sequences face difficulty with…

Computation and Language · Computer Science 2024-09-05 Suyuchen Wang , Ivan Kobyzev , Peng Lu , Mehdi Rezagholizadeh , Bang Liu

Rotary Position Embedding (RoPE)-extension refers to modifying or generalizing the Rotary Position Embedding scheme to handle longer sequences than those encountered during pre-training. However, current extension strategies are highly…

Computation and Language · Computer Science 2026-02-02 Qingyuan Tian , Wenhong Zhu , Xiaoran Liu , Xiaofeng Wang , Rui Wang

We propose Parabolic Position Encoding (PaPE), a parabola-based position encoding for vision modalities in attention-based architectures. Given a set of vision tokens-such as from videos, event camera streams, images, or point clouds-our…

Embedding models play a pivot role in modern NLP applications such as IR and RAG. While the context limit of LLMs has been pushed beyond 1 million tokens, embedding models are still confined to a narrow context window not exceeding 8k…

Computation and Language · Computer Science 2024-11-08 Dawei Zhu , Liang Wang , Nan Yang , Yifan Song , Wenhao Wu , Furu Wei , Sujian Li

We prove under practical assumptions that Rotary Positional Embedding (RoPE) introduces an intrinsic distance-dependent bias in attention scores that limits RoPE's ability to model long-context. RoPE extension methods may alleviate this…

Computation and Language · Computer Science 2026-05-12 Yu Wang , Sheng Shen , Rémi Munos , Hongyuan Zhan , Yuandong Tian