English
Related papers

Related papers: Thin Keys, Full Values: Reducing KV Cache via Low-…

200 papers

The dot product attention mechanism, originally designed for natural language processing tasks, is a cornerstone of modern Transformers. It adeptly captures semantic relationships between word pairs in sentences by computing a similarity…

Disordered Systems and Neural Networks · Physics 2025-01-14 Riccardo Rende , Luciano Loris Viteritti

With the development of large language models (LLMs), efficient inference through Key-Value (KV) cache compression has attracted considerable attention, especially for long-context generation. To compress the KV cache, recent methods…

Computation and Language · Computer Science 2025-10-28 Qingyue Yang , Jie Wang , Xing Li , Zhihai Wang , Chen Chen , Lei Chen , Xianzhi Yu , Wulong Liu , Jianye Hao , Mingxuan Yuan , Bin Li

A central problem related to transformers can be stated as follows: given two $n \times d$ matrices $Q$ and $K$, and a non-negative function $f$, define the matrix $A$ as follows: (1) apply the function $f$ to each entry of the $n \times n$…

Machine Learning · Computer Science 2024-10-10 Ravindran Kannan , Chiranjib Bhattacharyya , Praneeth Kacham , David P. Woodruff

In large-scale recommender systems, ultra-long user behavior sequences encode rich signals of evolving interests. Extending sequence length generally improves accuracy, but directly modeling such sequences in production is infeasible due to…

Information Retrieval · Computer Science 2025-08-26 Kaiyuan Li , Yongxiang Tang , Yanhua Cheng , Yong Bai , Yanxiang Zeng , Chao Wang , Xialong Liu , Peng Jiang

We introduce the Block Transformer which adopts hierarchical global-to-local modeling to autoregressive transformers to mitigate the inference bottlenecks associated with self-attention. Self-attention requires the key-value (KV) cache of…

Computation and Language · Computer Science 2024-11-04 Namgyu Ho , Sangmin Bae , Taehyeon Kim , Hyunjik Jo , Yireun Kim , Tal Schuster , Adam Fisch , James Thorne , Se-Young Yun

Efficient real-world deployments of large language models (LLMs) rely on Key-Value (KV) caching for processing and generating long outputs, reducing the need for repetitive computation. For large contexts, Key-Value caches can take up tens…

Recently, transformer-based generative recommendation has garnered significant attention for user behavior modeling. However, it often requires discretizing items into multi-code representations (e.g., typically four code tokens or more),…

Information Retrieval · Computer Science 2025-12-18 Longtao Xiao , Haolin Zhang , Guohao Cai , Jieming Zhu , Yifan Wang , Heng Chang , Zhenhua Dong , Xiu Li , Ruixuan Li

The increasing size of the Key-Value (KV) cache during the Large Language Models long-context inference is the main obstacle for its balance between the deployment cost and task accuracy. To reduce the KV cache size in such scenarios, most…

Machine Learning · Computer Science 2025-07-25 Manlai Liang , JiaMing Zhang , Xiong Li , Jinlong Li

Since transformer was firstly published in 2017, several works have been proposed to optimize it. However, the major structure of transformer remains unchanged, ignoring one of its main intrinsic limitations, which is the same static value…

Machine Learning · Computer Science 2025-12-30 Xiaowei Wang

LLMs are seeing growing use for applications which require large context windows, and with these large context windows KV cache activations surface as the dominant contributor to memory consumption during inference. Quantization is a…

Machine Learning · Computer Science 2025-05-30 Coleman Hooper , Sehoon Kim , Hiva Mohammadzadeh , Michael W. Mahoney , Yakun Sophia Shao , Kurt Keutzer , Amir Gholami

Serving transformer language models with high throughput requires caching Key-Values (KVs) to avoid redundant computation during autoregressive generation. The memory footprint of KV caching is significant and heavily impacts serving costs.…

Machine Learning · Computer Science 2026-04-28 Anastasiia Filippova , David Grangier , Marco Cuturi , João Monteiro

Vision-Language Models suffer severe KV cache pressure at inference, as a single image often encodes into thousands of tokens. Most existing methods exploit token sparsity through token pruning, but permanently discarding visual content…

Computer Vision and Pattern Recognition · Computer Science 2026-05-20 Beomseok Kang , Dongwon Jo , Jiwon Song , Donghwee Son , Jae-Joon Kim

The advent of pre-trained large language models (LLMs) has revolutionized various natural language processing tasks. These models predominantly employ an auto-regressive decoding mechanism that utilizes Key-Value (KV) caches to eliminate…

Computation and Language · Computer Science 2024-06-12 Hao Yu , Zelan Yang , Shen Li , Yong Li , Jianxin Wu

In modern machine learning, inner product attention computation is a fundamental task for training large language models such as Transformer, GPT-1, BERT, GPT-2, GPT-3 and ChatGPT. Formally, in this problem, one is given as input three…

Machine Learning · Computer Science 2023-05-11 Josh Alman , Zhao Song

The quadratic complexity of self-attention in Transformer models remains a significant bottleneck for processing long sequences and deploying large language models efficiently. For this approach, there has been significant research into…

Computation and Language · Computer Science 2026-05-26 Spandan Pratyush

Transformer model with multi-head attention requires caching intermediate results for efficient inference in generation tasks. However, cache brings new memory-related costs and prevents leveraging larger batch size for faster speed. We…

Computation and Language · Computer Science 2021-06-15 Yu Yan , Jiusheng Chen , Weizhen Qi , Nikhil Bhendawade , Yeyun Gong , Nan Duan , Ruofei Zhang

Linear attention offers a computationally efficient yet expressive alternative to softmax attention. However, recent empirical results indicate that the hidden state of trained linear attention models often exhibits a low-rank structure,…

Machine Learning · Computer Science 2026-02-13 Philipp Nazari , T. Konstantin Rusch

We compare two strategies for compressing the KV cache in transformer inference: rank reduction (discard dimensions) and quantization (keep all dimensions, reduce precision). At matched storage budgets across five models (124M-14B, MHA and…

Machine Learning · Computer Science 2026-04-14 Samuel Salfati

The Key-Value (KV) cache is a crucial component in serving transformer-based autoregressive large language models (LLMs), enabling faster inference by storing previously computed KV vectors. However, its memory consumption scales linearly…

Machine Learning · Computer Science 2024-10-07 Rongzhi Zhang , Kuang Wang , Liyuan Liu , Shuohang Wang , Hao Cheng , Chao Zhang , Yelong Shen

The efficiency of large language models (LLMs) remains a critical challenge, particularly in contexts where computational resources are limited. Traditional attention mechanisms in these models, while powerful, require significant…

Computation and Language · Computer Science 2024-07-19 Bingli Liao , Danilo Vasconcellos Vargas