English
Related papers

Related papers: Circuit Complexity Bounds for RoPE-based Transform…

200 papers

Looped Transformers provide advantages in parameter efficiency, computational capabilities, and generalization for reasoning tasks. However, their expressive power regarding function approximation remains underexplored. In this paper, we…

Machine Learning · Computer Science 2025-06-06 Kevin Xu , Issei Sato

Rotary Position Embedding (RoPE) is widely adopted in large language models (LLMs) due to its efficient encoding of relative positions with strong extrapolation capabilities. However, while its application in higher-dimensional input…

Machine Learning · Computer Science 2025-07-16 Haiping Liu , Lijing Lin , Jingyuan Sun , Zhegong Shangguan , Mauricio A. Alvarez , Hongpeng Zhou

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

In Transformers, Position Embeddings (PEs) significantly influence Length Generalization (LG) performance, yet their fundamental role remains unclear. In this work, we investigate the limitations and capabilities of PEs in achieving LG. We…

Machine Learning · Computer Science 2025-10-07 Yang Chen , Yitao Liang , Zhouchen Lin

Transformer architectures rely on position encodings to model the spatial structure of input data. Rotary Position Encoding (RoPE) is a widely used method in language models that encodes relative positions through fixed, block-diagonal,…

Computer Vision and Pattern Recognition · Computer Science 2025-08-19 Sophie Ostmeier , Brian Axelrod , Maya Varma , Michael E. Moseley , Akshay Chaudhari , Curtis Langlotz

Rotary Position Embedding (RoPE) is the de facto positional encoding in large language models due to its ability to encode relative positions and support length extrapolation. When adapted to vision transformers, the standard axial…

Computer Vision and Pattern Recognition · Computer Science 2026-02-04 Haoyu Liu , Sucheng Ren , Tingyu Zhu , Peng Wang , Cihang Xie , Alan Yuille , Zeyu Zheng , Feng Wang

Without positional information, attention-based Transformer neural networks are permutation-invariant. Absolute or relative positional embeddings are the most popular ways to feed Transformer models with positional information. Absolute…

Machine Learning · Computer Science 2021-11-10 Tatiana Likhomanenko , Qiantong Xu , Gabriel Synnaeve , Ronan Collobert , Alex Rogozhnikov

Length extrapolation has attracted considerable attention recently since it allows transformers to be tested on longer sequences than those used in training. Previous research has shown that this property can be attained by using carefully…

Computation and Language · Computer Science 2023-12-20 Zhen Qin , Yiran Zhong , Hui Deng

Every Transformer architecture dedicates enormous capacity to learning rich representations in semantic embedding space -- yet the rotation manifold acted upon by Rotary Positional Embeddings (RoPE) has been treated as a fixed, hand-crafted…

Artificial Intelligence · Computer Science 2026-04-28 Hailing Cheng , Daqi Sun , Xinyu Lu

Transformers rely on explicit positional encoding to model structure in data. While Rotary Position Embedding (RoPE) excels in 1D domains, its application to image generation reveals significant limitations such as fine-grained spatial…

Computer Vision and Pattern Recognition · Computer Science 2026-03-13 Jiaye Li , Baoyou Chen , Hui Li , Zilong Dong , Jingdong Wang , Siyu Zhu

Transformer architectures rely on explicit position encodings in order to preserve a notion of word order. In this paper, we argue that existing work does not fully utilize position information. For example, the initial proposal of a…

Computation and Language · Computer Science 2020-09-30 Zhiheng Huang , Davis Liang , Peng Xu , Bing Xiang

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

Recent advances in Transformer models allow for unprecedented sequence lengths, due to linear space and time complexity. In the meantime, relative positional encoding (RPE) was proposed as beneficial for classical Transformers and consists…

Machine Learning · Computer Science 2021-06-11 Antoine Liutkus , Ondřej Cífka , Shih-Lun Wu , Umut Şimşekli , Yi-Hsuan Yang , Gaël Richard

Position representation is crucial for building position-aware representations in Transformers. Existing position representations suffer from a lack of generalization to test data with unseen lengths or high computational cost. We…

Computation and Language · Computer Science 2021-09-14 Shun Kiyono , Sosuke Kobayashi , Jun Suzuki , Kentaro Inui

Rotary Positional Embedding (RoPE) is a widely adopted technique for encoding position in language models, which, while effective, causes performance breakdown when input length exceeds training length. Prior analyses assert (rightly) that…

Machine Learning · Computer Science 2026-03-20 Davis Wertheimer , Aozhong Zhang , Derrick Liu , Penghang Yin , Naigang Wang

This paper studies generalization error bounds for Transformer models. Based on the offset Rademacher complexity, we derive sharper generalization bounds for different Transformer architectures, including single-layer single-head,…

Machine Learning · Computer Science 2026-03-24 Yawen Li , Tao Hu , Zhouhui Lian , Wan Tian , Yijie Peng , Huiming Zhang , Zhongyi Li

Length generalization, the ability to generalize from small training context sizes to larger ones, is a critical challenge in the development of Transformer-based language models. Positional encoding (PE) has been identified as a major…

Computation and Language · Computer Science 2023-11-08 Amirhossein Kazemnejad , Inkit Padhi , Karthikeyan Natesan Ramamurthy , Payel Das , Siva Reddy

Positional encoding (PE) is a core architectural component of Transformers, yet its impact on the Transformer's generalization and robustness remains unclear. In this work, we provide the first generalization analysis for a single-layer…

Machine Learning · Statistics 2026-03-25 Weiyi He , Yue Xing

There are several improvements proposed over the baseline Absolute Positional Encoding (APE) method used in original transformer. In this study, we aim to investigate the implications of inadequately representing positional encoding in…

Computation and Language · Computer Science 2024-05-09 Arpit Aggarwal

Transformers have become a standard neural network architecture for many NLP problems, motivating theoretical analysis of their power in terms of formal languages. Recent work has shown that transformers with hard attention are quite…

Computation and Language · Computer Science 2022-04-12 William Merrill , Ashish Sabharwal , Noah A. Smith