English
Related papers

Related papers: How Smooth Is Attention?

200 papers

Attention is a powerful and ubiquitous mechanism for allowing neural models to focus on particular salient pieces of information by taking their weighted average when making predictions. In particular, multi-headed attention is a driving…

Computation and Language · Computer Science 2019-11-05 Paul Michel , Omer Levy , Graham Neubig

The attention mechanism within the transformer architecture enables the model to weigh and combine tokens based on their relevance to the query. While self-attention has enjoyed major success, it notably treats all queries $q$ in the same…

Machine Learning · Computer Science 2024-11-21 Xuechen Zhang , Xiangyu Chang , Mingchen Li , Amit Roy-Chowdhury , Jiasi Chen , Samet Oymak

Transformers provide a class of expressive architectures that are extremely effective for sequence modeling. However, the key limitation of transformers is their quadratic memory and time complexity $\mathcal{O}(L^2)$ with respect to the…

Machine Learning · Computer Science 2021-10-29 Hongyu Ren , Hanjun Dai , Zihang Dai , Mengjiao Yang , Jure Leskovec , Dale Schuurmans , Bo Dai

Recently, transformer-based methods have dominated 3D instance segmentation, where mask attention is commonly involved. Specifically, object queries are guided by the initial instance masks in the first cross-attention, and then iteratively…

Computer Vision and Pattern Recognition · Computer Science 2023-09-06 Xin Lai , Yuhui Yuan , Ruihang Chu , Yukang Chen , Han Hu , Jiaya Jia

Transformers owe much of their empirical success in natural language processing to the self-attention blocks. Recent perspectives interpret attention blocks as interacting particle systems, whose mean-field limits correspond to gradient…

Machine Learning · Computer Science 2026-03-18 Viktor Stein , Wuchen Li , Gabriele Steidl

This paper investigates the limitations of transformers for entity-tracking tasks in large language models. We identify a theoretical constraint, showing that transformers require at least $\log_2 (n+1)$ layers to handle entity tracking…

Machine Learning · Computer Science 2024-12-12 Erwan Fagnou , Paul Caillon , Blaise Delattre , Alexandre Allauzen

Attention, specifically scaled dot-product attention, has proven effective for natural language, but it does not have a mechanism for handling hierarchical patterns of arbitrary nesting depth, which limits its ability to recognize certain…

Computation and Language · Computer Science 2024-01-25 Brian DuSell , David Chiang

Transformers have proven highly effective across modalities, but standard softmax attention scales quadratically with sequence length, limiting long context modeling. Linear attention mitigates this by approximating attention with kernel…

Machine Learning · Computer Science 2026-02-10 Ashkan Shahbazi , Chayne Thrash , Yikun Bai , Keaton Hamm , Navid NaderiAlizadeh , Soheil Kolouri

We characterize the pre-softmax attention matrix $\mathbf{QK^\top}$ in transformers as an associative memory matrix encoding pairwise associations between input features. By decomposing this matrix into its symmetric and skew-symmetric…

Machine Learning · Computer Science 2026-05-28 Hyunmin Cho , Woo Kyoung Han , Kyong Hwan Jin

We prove that with linear transformations, both (i) two-layer self-attention and (ii) one-layer self-attention followed by a softmax function are universal approximators for continuous sequence-to-sequence functions on compact domains. Our…

Machine Learning · Computer Science 2025-12-17 Jerry Yao-Chieh Hu , Hude Liu , Hong-Yu Chen , Weimin Wu , Han Liu

End-to-end models reach state-of-the-art performance for speech recognition, but global soft attention is not monotonic, which might lead to convergence problems, to instability, to bad generalisation, cannot be used for online streaming,…

Computation and Language · Computer Science 2021-04-01 Albert Zeyer , Ralf Schlüter , Hermann Ney

Impressive performance of Transformer has been attributed to self-attention, where dependencies between entire input in a sequence are considered at every position. In this work, we reform the neural $n$-gram model, which focuses on only…

Computation and Language · Computer Science 2022-07-28 Mengsay Loem , Sho Takase , Masahiro Kaneko , Naoaki Okazaki

Transformers have demonstrated remarkable performance across diverse domains. The key component of Transformers is self-attention, which learns the relationship between any two tokens in the input sequence. Recent studies have revealed that…

Machine Learning · Computer Science 2025-05-14 Hyowon Wi , Jeongwhan Choi , Noseong Park

Transformers with self-attention modules as their core components have become an integral architecture in modern large language and foundation models. In this paper, we study the evolution of tokens in deep encoder-only transformers at…

Analysis of PDEs · Mathematics 2026-05-12 Albert Alcalde , Leon Bungert , Konstantin Riedl , Tim Roith

Motivated by the factorization inherent in the original fast multipole method and the improved fast Gauss transform we introduce a factorable form of attention that operates efficiently in high dimensions. This approach reduces the…

Machine Learning · Computer Science 2024-02-13 Armin Gerami , Monte Hoover , Pranav S. Dulepet , Ramani Duraiswami

An attention matrix of a transformer self-attention sublayer can provably be decomposed into two components and only one of them (effective attention) contributes to the model output. This leads us to ask whether visualizing effective…

Computation and Language · Computer Science 2021-05-20 Kaiser Sun , Ana Marasović

Transformers allow attention between all pairs of tokens, but there is reason to believe that most of these connections - and their quadratic time and memory - may not be necessary. But which ones? We evaluate the impact of sparsification…

Computation and Language · Computer Science 2022-10-11 Siddhartha Brahma , Polina Zablotskaia , David Mimno

The Transformer is a sequence model that forgoes traditional recurrent architectures in favor of a fully attention-based approach. Besides improving performance, an advantage of using attention is that it can also help to interpret a model…

Human-Computer Interaction · Computer Science 2019-06-14 Jesse Vig

Existing analyses of the expressive capacity of Transformer models have required excessively deep layers for data memorization, leading to a discrepancy with the Transformers actually used in practice. This is primarily due to the…

Machine Learning · Computer Science 2024-01-30 Tokio Kajitsuka , Issei Sato

Since their introduction the Trasformer architectures emerged as the dominating architectures for both natural language processing and, more recently, computer vision applications. An intrinsic limitation of this family of "fully-attentive"…

Machine Learning · Computer Science 2023-03-16 Carmelo Scribano , Giorgia Franchini , Marco Prato , Marko Bertogna