English
Related papers

Related papers: Trellis: Learning to Compress Key-Value Memory in …

200 papers

The quadratic cost of attention in transformers motivated the development of efficient approaches: namely sparse and sliding window attention, convolutions and linear attention. Although these approaches result in impressive reductions in…

Machine Learning · Computer Science 2025-11-10 Jatin Prakash , Aahlad Puli , Rajesh Ranganath

Autoregressive Transformers rely on Key-Value (KV) caching to accelerate inference. However, the linear growth of the KV cache with context length leads to excessive memory consumption and bandwidth constraints. This bottleneck is…

Computation and Language · Computer Science 2025-06-10 Ravi Ghadia , Avinash Kumar , Gaurav Jain , Prashant Nair , Poulami Das

Large Language Models (LLMs) face limitations due to the high demand on GPU memory and computational resources when handling long contexts. While sparsify the Key-Value (KV) cache of transformer model is a typical strategy to alleviate…

Machine Learning · Computer Science 2024-02-06 Yumeng Wang , Zhenyang Xiao

The key to a Transformer model is the self-attention mechanism, which allows the model to analyze an entire sequence in a computationally efficient manner. Recent work has suggested the possibility that general attention mechanisms used by…

Machine Learning · Computer Science 2020-01-01 Thomas Dowdell , Hongyu Zhang

Memory consumption of the Key-Value (KV) cache represents a major bottleneck for efficient large language model inference. While attention-score-based KV cache pruning shows promise, it faces critical practical limitations: attention scores…

Artificial Intelligence · Computer Science 2025-10-02 Alessio Devoto , Maximilian Jeblick , Simon Jégou

Context lengths of Large Language Models (LLMs) have exploded in recent years, with 128k-token context becoming a standard and million-token context becoming a reality. Efficiently supporting long-context inference remains challenging as…

Computation and Language · Computer Science 2024-10-08 Isaac Rehg

Vision Transformers have achieved state-of-the-art performance in many visual tasks. Due to the quadratic computational and memory complexities of self-attention, recent works either apply attention only to low-resolution inputs or restrict…

Computer Vision and Pattern Recognition · Computer Science 2022-07-05 Cheng Li , Yangxin Liu

Transformers have been established as the most popular backbones in sequence modeling, mainly due to their effectiveness in in-context retrieval tasks and the ability to learn at scale. Their quadratic memory and time complexity, however,…

Computation and Language · Computer Science 2025-05-30 Ali Behrouz , Zeman Li , Praneeth Kacham , Majid Daliri , Yuan Deng , Peilin Zhong , Meisam Razaviyayn , Vahab Mirrokni

An important development in deep learning from the earliest MLPs has been a move towards architectures with structural inductive biases which enable the model to keep distinct sources of information and routes of processing well-separated.…

Machine Learning · Computer Science 2021-03-02 Alex Lamb , Di He , Anirudh Goyal , Guolin Ke , Chien-Feng Liao , Mirco Ravanelli , Yoshua Bengio

Autoregressive Transformer KV caches grow linearly with context length; sliding-window caching bounds memory but discards evicted tokens entirely, so relevant evidence outside the window becomes inaccessible. We introduce \emph{Tensor…

Machine Learning · Computer Science 2026-05-25 Kabir Swain , Sijie Han , Daniel Karl I. Weidele , Mauro Martino , Antonio Torralba

As the context length of current large language models (LLMs) rapidly increases, the memory demand for the Key-Value (KV) cache is becoming a bottleneck for LLM deployment and batch processing. Traditional KV cache compression methods…

Computation and Language · Computer Science 2025-12-23 Aomufei Yuan , Zhiming Wang , Ruijie Miao , Dayu Wang , Yuxuan Tian , Zihan Wang , Yebo Peng , Yuhan Wu , Bairen Yi , Xin Liu , Tong Yang

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

Systems that require high-throughput and fault tolerance, such as key-value stores and databases, are looking to persistent memory to combine the performance of in-memory systems with the data-consistent fault-tolerance of nonvolatile…

Databases · Computer Science 2020-02-07 Brian Choi , Parv Saxena , Ryan Huang , Randal Burns

Transformers have achieved significant success across various domains, relying on self-attention to capture dependencies. However, the standard first-order attention mechanism is often limited by a low-rank bottleneck, struggling to capture…

Computation and Language · Computer Science 2025-12-05 Hanting Chen , Chong Zhu , Kai Han , Yuchuan Tian , Yuchen Liang , Tianyu Guo , Xinghao Chen , Dacheng Tao , Yunhe Wang

Despite scale driving substantial recent advancements in machine learning, reinforcement learning (RL) methods still primarily use small value functions. Naively scaling value functions -- including with a transformer architecture, which is…

Machine Learning · Computer Science 2026-02-03 Perry Dong , Kuo-Han Hung , Alexander Swerdlow , Dorsa Sadigh , Chelsea Finn

Standard Transformer attention uses identical dimensionality for queries, keys, and values, yet these components serve different roles: queries and keys produce scalar attention weights (selection), while values carry rich representations…

Machine Learning · Computer Science 2026-03-31 Hengshuai Yao , Xing Chen , Ahmed Murtadha , Guan Wang

As Large Language Models (LLMs) scale to support context windows exceeding one million tokens, the linear growth of Key-Value (KV) cache imposes severe memory capacity and bandwidth bottlenecks, constraining the efficiency of long-context…

Computation and Language · Computer Science 2026-04-09 Zhirui Chen , Peiyang Liu , Ling Shao

The deployment of large language models (LLMs) is often hindered by the extensive memory requirements of the Key-Value (KV) cache, especially as context lengths increase. Existing approaches to reduce the KV cache size involve either…

Computation and Language · Computer Science 2024-11-05 Alessio Devoto , Yu Zhao , Simone Scardapane , Pasquale Minervini

In pursuit of faster computation, Efficient Transformers demonstrate an impressive variety of approaches -- models attaining sub-quadratic attention complexity can utilize a notion of sparsity or a low-rank approximation of inputs to reduce…

Machine Learning · Computer Science 2022-11-09 Uladzislau Yorsh , Alexander Kovalenko

Test-time training (TTT) with KV binding as sequence modeling layer is commonly interpreted as a form of online meta-learning that memorizes a key-value mapping at test time. However, our analysis reveals multiple phenomena that contradict…

Machine Learning · Computer Science 2026-05-14 Junchen Liu , Sven Elflein , Or Litany , Zan Gojcic , Ruilong Li