English
Related papers

Related papers: vAttention: Dynamic Memory Management for Serving …

200 papers

Large Language Models (LLMs) with extended context lengths face significant computational challenges during the pre-filling phase, primarily due to the quadratic complexity of self-attention. Existing methods typically employ dynamic…

Machine Learning · Computer Science 2025-05-30 Yu Zhang , Dong Guo , Fang Wu , Guoliang Zhu , Dian Ding , Yiming Zhang

Compressing the KV cache is a required step to deploy large language models on edge devices. Current quantization methods compress storage but fail to reduce bandwidth as attention calculation requires dequantizing keys from INT4/INT8 to…

Machine Learning · Computer Science 2026-01-16 Aryan Karmore

Converting pretrained attention modules such as grouped-query attention (GQA) into multi-head latent attention (MLA) can improve expressivity without increasing KV-cache cost, making it attractive for efficient inference. However, many…

Existing memory management techniques severely hinder efficient Large Language Model serving on accelerators constrained by poor random-access bandwidth.While static pre-allocation preserves memory contiguity,it incurs significant overhead…

Hardware Architecture · Computer Science 2026-04-22 Guoqiang Zou , Wanyu Wang , Hao Zheng , Longxiang Yin , Yinhe Han

KV cache is a widely used acceleration technique for large language models (LLMs) inference. However, its memory requirement grows rapidly with input length. Previous studies have reduced the size of KV cache by either removing the same…

Computation and Language · Computer Science 2025-01-28 Xingyang He , Jie Liu , Shaowei Chen

Multi-Head Latent Attention (MLA), introduced in DeepSeek-V2, improves the efficiency of large language models by projecting query, key, and value tensors into a compact latent space. This architectural change reduces the KV-cache size and…

Hardware Architecture · Computer Science 2026-04-10 Robin Geens , Marian Verhelst

Linear attention has emerged as a promising direction for scaling Vision Transformers beyond the quadratic cost of dense self-attention. A prevalent strategy is to compress spatial tokens into a compact set of intermediate proxies that…

Computer Vision and Pattern Recognition · Computer Science 2026-05-15 Yuntong Li , Hainuo Wang , Hengxing Liu , Mingjia Li , Xiaojie Guo

Serving long-context LLMs is costly because attention computation grows linearly with context length. Dynamic sparse attention algorithms (DSAs) mitigate this by attending only to the key-value (KV) cache of critical tokens. However, with…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-09-30 Qihui Zhou , Peiqi Yin , Pengfei Zuo , James Cheng

In large language model (LLM) serving systems, executing each request consists of two phases: the compute-intensive prefill phase and the memory-intensive decoding phase. To prevent performance interference between the two phases, current…

Distributed, Parallel, and Cluster Computing · Computer Science 2025-03-27 Yunkai Liang , Zhangyu Chen , Pengfei Zuo , Zhi Zhou , Xu Chen , Zhou Yu

Modern online large language model (LLM) services, such as Retrieval-Augmented Generation (RAG) and agent systems, increasingly expose two prominent characteristics: prompt segmentation (e.g., system instructions, retrieved passages, tool…

Machine Learning · Computer Science 2026-05-12 Xingyu Qu , Tianhao Lin , Yiqi Li , Zhiyu Chen , Sheng Wang

Static-graph LLM decoders provide predictable launches, fixed tensor shapes, and low submission overhead, but online decoding exposes highly irregular KV-cache behavior: request lengths differ, EOS events arrive asynchronously, and logical…

Hardware Architecture · Computer Science 2026-05-12 Zhiqing Zhong , Zhijing Ye , Jian Zhang , Weijian Zheng , Bolun Sun , Xiaodong Yu

KV cache pruning has emerged as a promising technique for reducing memory and computation costs in long-context auto-regressive generation. Existing methods for vision-language models (VLMs) typically rely on self-attention scores from…

Computer Vision and Pattern Recognition · Computer Science 2024-12-09 Xiaohuan Pei , Tao Huang , Chang Xu

Transformers are slow and memory-hungry on long sequences, since the time and memory complexity of self-attention are quadratic in sequence length. Approximate attention methods have attempted to address this problem by trading off model…

Machine Learning · Computer Science 2022-06-24 Tri Dao , Daniel Y. Fu , Stefano Ermon , Atri Rudra , Christopher Ré

The computational and memory challenges of large language models (LLMs) have sparked several optimization approaches towards their efficient implementation. While prior LLM-targeted quantization, and prior works on sparse acceleration have…

Hardware Architecture · Computer Science 2025-03-18 Abhishek Moitra , Arkapravo Ghosh , Shrey Agarwal , Aporva Amarnath , Karthik Swaminathan , Priyadarshini Panda

Effective attention modules have played a crucial role in the success of Transformer-based large language models (LLMs), but the quadratic time and memory complexities of these attention modules also pose a challenge when processing long…

Distributed, Parallel, and Cluster Computing · Computer Science 2024-06-07 Ao Sun , Weilin Zhao , Xu Han , Cheng Yang , Zhiyuan Liu , Chuan Shi , Maosong Sun

Multimodal Large Language Models (MLLMs) have advanced unified reasoning over text, images, and videos, but their inference is hindered by the rapid growth of key-value (KV) caches. Each visual input expands into thousands of tokens,…

Artificial Intelligence · Computer Science 2026-04-08 Bowen Zeng , Feiyang Ren , Jun Zhang , Xiaoling Gu , Ke Chen , Lidan Shou , Huan Li

The original softmax-based attention mechanism (regular attention) in the extremely successful Transformer architecture computes attention between $N$ tokens, each embedded in a $D$-dimensional head, with a time complexity of $O(N^2D)$.…

Machine Learning · Computer Science 2025-10-28 Armin Gerami , Ramani Duraiswami

The attention operator remains a critical performance bottleneck in large language models (LLMs), particularly for long-context scenarios. While FlashAttention is the most widely used and effective GPU-aware acceleration algorithm, it must…

Current high-resolution vision-language models encode images as high-resolution image tokens and exhaustively take all these tokens to compute attention, which significantly increases the computational cost. To address this problem, we…

Computer Vision and Pattern Recognition · Computer Science 2024-07-30 Junyan Li , Delin Chen , Tianle Cai , Peihao Chen , Yining Hong , Zhenfang Chen , Yikang Shen , Chuang Gan

With the rapid development of large language models (LLMs), handling long context has become one of the vital abilities in LLMs. Such long-context ability is accompanied by difficulties in deployment, especially due to the increased…

Computation and Language · Computer Science 2025-08-19 Zhuorui Liu , Chen Zhang , Dawei Song