English
Related papers

Related papers: OBCache: Optimal Brain KV Cache Pruning for Effici…

200 papers

The integration of visual information into Large Language Models (LLMs) has enabled Multimodal LLMs (MLLMs), but the quadratic memory and computational costs of Transformer architectures remain a bottleneck. Existing KV cache eviction…

Machine Learning · Computer Science 2026-02-03 Xindian Ma , Yidi Lu , Peng Zhang , Jing Zhang

Reasoning LLMs produce thousands of chain-of-thought tokens whose KV cache must reside in scarce GPU HBM. The dominant response -- permanently evicting low-importance tokens -- is catastrophic for reasoning: accuracy collapses to 0-2.5%…

Computation and Language · Computer Science 2026-05-12 Aojie Yuan , Tianqi Shen , Dajun Zhang

Transformer-based large language models (LLMs) rely on key-value (KV) caching to avoid redundant computation during autoregressive inference. While this mechanism greatly improves efficiency, the cache size grows linearly with the input…

Machine Learning · Computer Science 2026-03-12 Jinwoo Ahn , Ingyu Seong , Akhil Kedia , Junhan Kim , Hyemi Jang , Kangwook Lee , Yongkweon Jeon

Vision-Language Models (VLMs) have demonstrated impressive performance across a versatile set of tasks. A key challenge in accelerating VLMs is storing and accessing the large Key-Value (KV) cache that encodes long visual contexts, such as…

Computer Vision and Pattern Recognition · Computer Science 2024-11-01 Dezhan Tu , Danylo Vashchilenko , Yuzhe Lu , Panpan Xu

Vision-Language Models (VLMs) face a critical memory bottleneck when processing long-form video content due to the linear growth of the Key-Value (KV) cache with sequence length. Existing solutions predominantly employ reactive eviction…

Computer Vision and Pattern Recognition · Computer Science 2026-02-17 Vishnu Sai , Dheeraj Sai , Srinath B , Girish Varma , Priyesh Shukla

Vision-language models (VLMs) typically encode substantially more visual tokens than text tokens, resulting in significant token redundancy. Pruning uninformative visual tokens is therefore crucial for improving computational efficiency,…

Computer Vision and Pattern Recognition · Computer Science 2026-01-19 Kai Zhao , Wubang Yuan , Yuchen Lin , Liting Ruan , Xiaofeng Lu , Deng-Ping Fan , Ming-Ming Cheng , Dan Zeng

Transformer-based large language models (LLMs) use the key-value (KV) cache to significantly accelerate inference by storing the key and value embeddings of past tokens. However, this cache consumes significant GPU memory. In this work, we…

Training and serving long-context large language models (LLMs) incurs substantial overhead. To address this, two critical steps are often required: a pretrained LLM typically undergoes a separate stage for context length extension by…

Computation and Language · Computer Science 2024-12-06 Suyu Ge , Xihui Lin , Yunan Zhang , Jiawei Han , Hao Peng

Large Language Models (LLMs) encounter significant performance bottlenecks in long-sequence tasks due to the computational complexity and memory overhead inherent in the self-attention mechanism. To address these challenges, we introduce…

Artificial Intelligence · Computer Science 2026-02-17 Ziming Wang , Xiang Wang , Kailong Peng , Lang Qin , Juan Gabriel Kostelec , Christos Sourmpis , Axel Laborieux , Qinghai Guo

Modern Large Language Models (LLMs) are increasingly trained to support very large context windows. We present Compactor, a training-free, query-agnostic KV compression strategy that uses approximate leverage scores to determine token…

Computation and Language · Computer Science 2025-12-10 Vivek Chari , Benjamin Van Durme

Excessive memory requirements of key and value features (KV-cache) present significant challenges in the autoregressive inference of large language models (LLMs), restricting both the speed and length of text generation. Approaches such as…

Computation and Language · Computer Science 2024-06-18 Vinay Joshi , Prashant Laddha , Shambhavi Sinha , Om Ji Omer , Sreenivas Subramoney

Recent advances in long-text understanding have pushed the context length of large language models (LLMs) up to one million tokens. It boosts LLMs's accuracy and reasoning capacity but causes exorbitant computational costs and…

Computation and Language · Computer Science 2025-05-19 Huan Yang , Renji Zhang , Mingzhe Huang , Weijun Wang , Yin Tang , Yuanchun Li , Yunxin Liu , Deyu Zhang

Large language models (LLMs) have achieved impressive performance in numerous domains but often struggle to process lengthy inputs effectively and efficiently due to limited length generalization and attention's quadratic computational…

Computation and Language · Computer Science 2024-03-26 Yi Lu , Xin Zhou , Wei He , Jun Zhao , Tao Ji , Tao Gui , Qi Zhang , Xuanjing Huang

Long-context inference in large language models (LLMs) is bottlenecked by the linear growth of the self-attention key-value (KV) cache. Top-k sparse attention alleviates this by loading only a small fraction of the KV cache, but accurately…

Computation and Language · Computer Science 2026-05-28 Keqi Deng , Shaoshi Ling , Ruchao Fan , Jinyu Li

The linear growth of key-value (KV) cache memory and quadratic computational in attention mechanisms complexity pose significant bottlenecks for large language models (LLMs) in long-context processing. While existing KV cache optimization…

Computation and Language · Computer Science 2025-10-07 Xin Liu , Xudong Wang , Pei Liu , Guoming Tang

Autoregressive Transformers adopted in Large Language Models (LLMs) are hard to scale to long sequences. Despite several works trying to reduce their computational cost, most of LLMs still adopt attention layers between all pairs of tokens…

Computation and Language · Computer Science 2024-06-03 Sotiris Anagnostidis , Dario Pavllo , Luca Biggio , Lorenzo Noci , Aurelien Lucchi , Thomas Hofmann

Despite the significant success of large language models (LLMs), their extensive memory requirements pose challenges for deploying them in long-context token generation. The substantial memory footprint of LLM decoders arises from the…

Machine Learning · Computer Science 2024-02-12 Amir Zandieh , Insu Han , Vahab Mirrokni , Amin Karbasi

Large Language Model (LLM) serving is increasingly constrained by the growing size of the key-value (KV) cache, which scales with both context length and generation length. Prior work shows that attention is dominated by a small subset of…

Machine Learning · Computer Science 2026-04-21 Nazmul Takbir , Hamidreza Alikhani , Nikil Dutt , Sangeetha Abdu Jyothi

Standard sequence mixing layers used in language models struggle to balance efficiency and performance. Self-attention performs well on long context tasks but has expensive quadratic compute and linear memory costs, while linear attention…

Machine Learning · Computer Science 2026-05-18 Nick Alonso , Tomas Figliolia , Beren Millidge

Despite the recent success associated with Large Language Models (LLMs), they are notably cost-prohibitive to deploy in resource-constrained environments due to their excessive memory and computational demands. In addition to model…

Computation and Language · Computer Science 2024-02-20 Siyu Ren , Kenny Q. Zhu