paper-with-me

홈 › Papers

Key-Value Means: Transformers with Expandable Block-Recurrent Compressed Memory

2026-05-11 · Daniel Goldstein, Eugene Cheah arxiv

We present Key-Value Means ("KVM"), a novel block-recurrence for attention that can accommodate either fixed-size or growing state. Equipping a strong transformer baseline with fixed-size KVM attention layers yields a strong $O(N)$ chunked RNN, while adding only an insignificant number of new parameters. We train a transformer with a growable KVM cache and show it performs competitively on long-context tests with only subquadratic prefill time and sublinear state growth. KVM is implementable with standard operations and without custom kernels, and supports chunk-wise parallelizable training and prefill. It provides many of the benefits of both traditional transformers (expandable context memory, chunk-wise parallelizable training and prefill) and linear RNNs in a single unified package. It can be used on every layer, saving KV-cache memory, and allowing a continuous range of choices of prefill time complexity between $O(N)$ and $O(N^2)$. It can also be implemented in a hybrid solution in tandem with LRNN layers in place of traditional attention, to supplement the LRNN with improved sublinear memory growth context length usage and long context decoding. We release our code at https://github.com/featherless-ai/KVM-paper and trained models at https://huggingface.co/collections/featherless-ai/kvm-paper under the Apache 2.0 license.

📄 PDF Abstract BibTeX arXiv:2605.09877

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Block-Recurrent Dynamics in Vision Transformers

2025-12-23 · Mozes Jacobs, Thomas Fel, Richard Hakim, Alessandra Brondetta 외 arxiv

As Vision Transformers (ViTs) become standard vision backbones, a mechanistic account of their computational phenomenology is essential. Despite architectural cues that hint at dynamical structure, there is no settled fr…

Hyperloop Transformers

2026-04-23 · Abbas Zeitoun, Lucas Torroba-Hennigen, Yoon Kim arxiv

LLM architecture research generally aims to maximize model quality subject to fixed compute/latency budgets. However, many applications of interest such as edge and on-device deployment are further constrained by the mod…

TIMBA: Time series Imputation with Bi-directional Mamba Blocks and Diffusion models

2024-10-08 · Javier Solís-García, Belén Vega-Márquez, Juan A. Nepomuceno, Isabel A. Nepomuceno-Chamorro

The problem of imputing multivariate time series spans a wide range of fields, from clinical healthcare to multi-sensor systems. Initially, Recurrent Neural Networks (RNNs) were employed for this task; however, their err…

ImputationMambaMissing ValuesState Space Models+1

bViT: Investigating Single-Block Recurrence in Vision Transformers for Image Recognition

2026-05-11 · Michal Byra, Pawel Olszowiec, Grzegorz Stefanski, Grzegorz Gruszczynski 외 arxiv

Vision Transformers (ViTs) are built by stacking independently parameterized blocks, but it remains unclear how much of this depth requires layer specific transformations and how much can be realized through recurrent co…

parameter-efficient fine-tuning

LoopSpec: Pipelined Self-Speculative Decoding for Looped Transformers

2026-09-15 · SangLyul Cho, Langqing Cui, Sehoon Kim, Dongsu Han 외 arxiv

Looped Transformers achieve strong performance with compact parameter sizes by repeatedly applying a shared stack of Transformer blocks across recurrent depths. However, they incur higher decoding latency than standard T…