English
Related papers

Related papers: Softpick: No Attention Sink, No Massive Activation…

200 papers

Large language models (LLMs) have demonstrated remarkable performance across a wide range of tasks. However, the quadratic complexity of softmax attention remains a central bottleneck that limits their scalability. Alman and Song (NeurIPS…

Machine Learning · Computer Science 2026-03-20 Maryam Aliakbarpour , Vladimir Braverman , Junze Yin , Haochen Zhang

Multi-head attention is a driving force behind state-of-the-art transformers, which achieve remarkable performance across a variety of natural language processing (NLP) and computer vision tasks. It has been observed that for many…

Machine Learning · Computer Science 2022-06-14 Tam Nguyen , Tan M. Nguyen , Dung D. Le , Duy Khuong Nguyen , Viet-Anh Tran , Richard G. Baraniuk , Nhat Ho , Stanley J. Osher

How much information do NLP tasks really need from a transformer's attention mechanism at application-time (inference)? From recent work, we know that there is sparsity in transformers and that the floating-points within its computation can…

Computation and Language · Computer Science 2021-06-03 Tianchu Ji , Shraddhan Jain , Michael Ferdman , Peter Milder , H. Andrew Schwartz , Niranjan Balasubramanian

This paper introduces and evaluates two novel Hierarchical Attention Network models [Yang et al., 2016] - i) Hierarchical Pruned Attention Networks, which remove the irrelevant words and sentences from the classification process in order to…

Computation and Language · Computer Science 2020-04-10 João G. Ribeiro , Frederico S. Felisberto , Isabel C. Neto

The self-attention mechanism is central to the success of Transformer architectures. However, standard row-stochastic attention has been shown to suffer from significant signal degradation across layers. In particular, it can induce rank…

Machine Learning · Computer Science 2026-04-10 Michela Lapenna , Rita Fioresi , Bahman Gharesifard

Transformers have excelled in many tasks including vision. However, efficient deployment of transformer models in low-latency or high-throughput applications is hindered by the computation in the attention mechanism which involves expensive…

Computer Vision and Pattern Recognition · Computer Science 2024-06-12 John Yang , Le An , Su Inn Park

We introduceDropDim, a structured dropout method designed for regularizing the self-attention mechanism, which is a key component of the transformer. In contrast to the general dropout method, which randomly drops neurons, DropDim drops…

Computation and Language · Computer Science 2023-04-21 Hao Zhang , Dan Qu , Keji Shao , Xukui Yang

Softmax-based dot-product attention is a cornerstone of Transformer architectures, enabling remarkable capabilities such as in-context learning. However, as context lengths increase, a fundamental limitation of the softmax function emerges:…

Machine Learning · Computer Science 2026-02-12 Sai Surya Duvvuri , Nirmal Patel , Nilesh Gupta , Inderjit S. Dhillon

The goal of this paper is to strengthen the reasoning of Omnimodal Large Language Models (Omni-LLMs) at inference time, without additional training. These models jointly process video, audio, and text, and given the large number of tokens…

Computer Vision and Pattern Recognition · Computer Science 2026-05-08 Suho Yoo , Youngjoon Jang , Joon Son Chung

Recently, it has been argued that encoder-decoder models can be made more interpretable by replacing the softmax function in the attention with its sparse variants. In this work, we introduce a novel, simple method for achieving sparsity in…

Computation and Language · Computer Science 2021-10-07 Biao Zhang , Ivan Titov , Rico Sennrich

The quadratic computational cost of the self-attention mechanism is a primary challenge in scaling Transformer models. While attention sparsity is widely studied as a technique to improve computational efficiency, it is almost universally…

Computation and Language · Computer Science 2025-08-11 Sagar Gandhi , Vishal Gandhi

Vision transformers have emerged as a powerful tool across a wide range of applications, yet their inner workings remain only partially understood. In this work, we examine the phenomenon of massive tokens - tokens with exceptionally high…

Computer Vision and Pattern Recognition · Computer Science 2025-07-23 Andrew Lu , Wentinn Liao , Liuhui Wang , Huzheng Yang , Jianbo Shi

Recently, post-training quantization (PTQ) has driven much attention to produce efficient neural networks without long-time retraining. Despite its low cost, current PTQ works tend to fail under the extremely low-bit setting. In this study,…

Computer Vision and Pattern Recognition · Computer Science 2023-02-22 Xiuying Wei , Ruihao Gong , Yuhang Li , Xianglong Liu , Fengwei Yu

The self-attention mechanism distinguishes transformer-based large language models (LLMs) apart from convolutional and recurrent neural networks. Despite the performance improvement, achieving real-time LLM inference on silicon remains…

Hardware Architecture · Computer Science 2024-11-18 Shiwei Liu , Guanchen Tao , Yifei Zou , Derek Chow , Zichen Fan , Kauna Lei , Bangfei Pan , Dennis Sylvester , Gregory Kielian , Mehdi Saligane

Compute-in-memory (CIM) techniques are widely employed in energy-efficient artificial intelligent (AI) processors. They alleviate power and latency bottlenecks caused by extensive data movements between compute and storage units. To extend…

Hardware Architecture · Computer Science 2025-12-15 Jianyi Yu , Tengxiao Wang , Yuxuan Wang , Xiang Fu , Fei Qiao , Ying Wang , Rui Yuan , Liyuan Liu , Cong Shi

Attention mechanisms have been extensively employed in various applications, including time series modeling, owing to their capacity to capture intricate dependencies; however, their utility is often constrained by quadratic computational…

Machine Learning · Computer Science 2025-11-06 Mingtao Zhang , Guoli Yang , Zhanxing Zhu , Mengzhu Wang , Xiaoying Bai

Attention-based large language models (LLMs) have transformed modern AI applications, but the quadratic cost of self-attention imposes significant compute and memory overhead. Dynamic sparsity (DS) attention mitigates this, yet its hardware…

Machine Learning · Computer Science 2025-12-09 Huizheng Wang , Hongbin Wang , Shaojun Wei , Yang Hu , Shouyi Yin

Deep learning models have been widely applied in various aspects of daily life. Many variant models based on deep learning structures have achieved even better performances. Attention-based architectures have become almost ubiquitous in…

Machine Learning · Computer Science 2022-02-25 Zhiying Fang , Yidong Ouyang , Ding-Xuan Zhou , Guang Cheng

The attention mechanism of a transformer has a quadratic complexity, leading to high inference costs and latency for long sequences. However, attention matrices are mostly sparse, which implies that many entries may be omitted from…

Machine Learning · Computer Science 2025-11-25 Jeffrey Willette , Heejun Lee , Sung Ju Hwang

Despite the recent success in many applications, the high computational requirements of vision transformers limit their use in resource-constrained settings. While many existing methods improve the quadratic complexity of attention, in most…

Computer Vision and Pattern Recognition · Computer Science 2023-02-28 Dmitrii Marin , Jen-Hao Rick Chang , Anurag Ranjan , Anish Prabhu , Mohammad Rastegari , Oncel Tuzel