中文
相关论文

相关论文: Limitations of Normalization in Attention Mechanis…

200 篇论文

Attention architectures are widely used; they recently gained renewed popularity with Transformers yielding a streak of state of the art results. Yet, the geometrical implications of softmax-attention remain largely unexplored. In this work…

机器学习 · 计算机科学 2020-05-20 Oliver Richter , Roger Wattenhofer

Attention is a core component of transformer architecture, whether encoder-only, decoder-only, or encoder-decoder model. However, the standard softmax attention often produces noisy probability distribution, which can impair effective…

计算与语言 · 计算机科学 2025-11-11 Dhananjay Ram , Wei Xia , Stefano Soatto

The softmax content-based attention mechanism has proven to be very beneficial in many applications of recurrent neural networks. Nevertheless it suffers from two major computational limitations. First, its computations for an attention…

机器学习 · 计算机科学 2016-09-20 Alexandre de Brébisson , Pascal Vincent

Transformer attention is typically implemented using softmax normalization, which enforces attention weights with unit sum normalization. While effective in many settings, this constraint can limit flexibility in controlling attention…

计算与语言 · 计算机科学 2026-02-27 Jeongin Bae , Baeseong Park , Gunho Park , Minsub Kim , Joonhyung Lee , Junhee Yoo , Sunghyeon Woo , Jiwon Ryu , Se Jung Kwon , Dongsoo Lee

Large language models rely on attention mechanisms with a softmax activation. Yet the dominance of softmax over alternatives (e.g., component-wise or linear) remains poorly understood, and many theoretical works have focused on the…

机器学习 · 计算机科学 2026-02-27 O. Duranthon , P. Marion , C. Boyer , B. Loureiro , L. Zdeborová

Large transformer models have achieved state-of-the-art results in numerous natural language processing tasks. Among the pivotal components of the transformer architecture, the attention mechanism plays a crucial role in capturing token…

计算与语言 · 计算机科学 2026-03-16 Yichuan Deng , Zhao Song , Kaijun Yuan , Tianyi Zhou

The attention mechanism is the computational core of modern Transformer architectures, but its quadratic complexity in the input sequence length is the bottleneck for large-scale inference. This has motivated a rapidly growing body of work…

The self-attention mechanism traditionally relies on the softmax operator, necessitating positional embeddings like RoPE, or position biases to account for token order. But current methods using still face length generalisation challenges.…

机器学习 · 计算机科学 2025-05-21 Shawn Tan , Songlin Yang , Aaron Courville , Rameswar Panda , Yikang Shen

Attention mechanism is a central component of the transformer architecture which led to the phenomenal success of large language models. However, the theoretical principles underlying the attention mechanism are poorly understood,…

机器学习 · 计算机科学 2023-12-11 Davoud Ataee Tarzanagh , Yingcong Li , Xuechen Zhang , Samet Oymak

Attention mechanisms have revolutionized several domains of artificial intelligence, such as natural language processing and computer vision, by enabling models to selectively focus on relevant parts of the input data. While recent work has…

机器学习 · 计算机科学 2026-02-03 Addison Kristanto Julistiono , Davoud Ataee Tarzanagh , Navid Azizan

Attention mechanism is contributing to the majority of recent advances in machine learning for natural language processing. Additionally, it results in an attention map that shows the proportional influence of each input in its decision.…

计算与语言 · 计算机科学 2025-01-23 Duc Hau Nguyen , Cyrielle Mallart , Guillaume Gravier , Pascale Sébillot

We study transformer language models, analyzing attention heads whose attention patterns are spread out, and whose attention scores depend weakly on content. We argue that the softmax denominators of these heads are stable when the…

计算与语言 · 计算机科学 2025-10-07 Alex Gibson

Attention mechanism is a fundamental component of the transformer model and plays a significant role in its success. However, the theoretical understanding of how attention learns to select tokens is still an emerging area of research. In…

机器学习 · 计算机科学 2025-05-20 Keitaro Sakamoto , Issei Sato

This paper questions whether the strong performance of softmax attention in transformers stems from producing a probability distribution over inputs. Instead, we argue that softmax's effectiveness lies in its implicit regularization of the…

机器学习 · 计算机科学 2026-03-16 Hemanth Saratchandran , Jianqiao Zheng , Yiping Ji , Wenbo Zhang , Simon Lucey

Softmax attention is a central component of transformer architectures, yet its nonlinear structure poses significant challenges for theoretical analysis. We develop a unified, measure-based framework for studying single-layer softmax…

机器学习 · 计算机科学 2025-12-15 Etienne Boursier , Claire Boyer

Transformer-based architectures traditionally employ softmax to compute attention weights, which produces dense distributions over all tokens in a sequence. While effective in many settings, this density has been shown to be detrimental for…

计算与语言 · 计算机科学 2026-03-03 Pavlo Vasylenko , Hugo Pitorro , André F. T. Martins , Marcos Treviso

At the core of the popular Transformer architecture is the self-attention mechanism, which dynamically assigns softmax weights to each input token so that the model can focus on the most salient information. However, the softmax structure…

机器学习 · 计算机科学 2025-05-27 Fanqi Yan , Huy Nguyen , Pedram Akbarian , Nhat Ho , Alessandro Rinaldo

Transformer models have achieved remarkable results in a wide range of applications. However, their scalability is hampered by the quadratic time and memory complexity of the self-attention mechanism concerning the sequence length. This…

机器学习 · 计算机科学 2024-02-27 Yury Nahshan , Joseph Kampeas , Emir Haleva

Transformers have become the go-to architecture for language and vision tasks, yet their theoretical properties, especially memorization capacity, remain elusive. This paper investigates the memorization abilities of multi-head attention…

机器学习 · 计算机科学 2024-03-05 Sadegh Mahdavi , Renjie Liao , Christos Thrampoulidis

Multi-head attention enables transformer models to represent multiple attention patterns simultaneously. Empirically, head specialization emerges in distinct stages during training, while many heads remain redundant and learn similar…

机器学习 · 计算机科学 2026-03-05 M. Sagitova , O. Duranthon , L. Zdeborová
‹ 上一页 1 2 3 10 下一页 ›