paper-with-me

홈 › Papers

Context Compression for Auto-regressive Transformers with Sentinel Tokens

2023-10-12 · Siyu Ren, Qi Jia, Kenny Q. Zhu

The quadratic complexity of the attention module makes it gradually become the bulk of compute in Transformer-based LLMs during generation. Moreover, the excessive key-value cache that arises when dealing with long inputs also brings severe issues on memory footprint and inference latency. In this work, we propose a plug-and-play approach that is able to incrementally compress the intermediate activation of a specified span of tokens into compact ones, thereby reducing both memory and computational cost when processing subsequent context. Experiments on both in-domain language modeling and zero-shot open-ended document generation demonstrate the advantage of our approach over sparse attention baselines in terms of fluency, n-gram matching, and semantic similarity. At last, we comprehensively profile the benefit of context compression on improving the system throughout. Code is available at https://github.com/DRSY/KV_Compression.

📄 PDF Abstract BibTeX arXiv:2310.08152

Code (1)

DRSY/KV_Compression 공식 구현 pytorch

Tasks

Language ModelingLanguage ModellingSemantic SimilaritySemantic Textual Similarity

Similar Papers 제목 키워드 기반

Sentinel: Attention Probing of Proxy Models for LLM Context Compression with an Understanding Perspective

2025-05-29 · Yong Zhang, Yanwen Huang, Ning Cheng, Yang Guo 외

Retrieval-augmented generation (RAG) enhances large language models (LLMs) with external context, but retrieved passages are often lengthy, noisy, or exceed input limits. Existing compression methods typically require su…

DecoderRAGRetrieval-augmented GenerationSentence

Exploring the Limits of KV Cache Compression in Visual Autoregressive Transformers

2025-03-19 · Bo Chen, Xiaoyu Li, Yekun Ke, YIngyu Liang 외

A fundamental challenge in Visual Autoregressive models is the substantial memory overhead required during inference to store previously generated representations. Despite various attempts to mitigate this issue through …

Dimensionality Reduction

General-purpose, long-context autoregressive modeling with Perceiver AR

2022-02-15 · Curtis Hawthorne, Andrew Jaegle, Cătălina Cangea, Sebastian Borgeaud 외

Real-world data is high-dimensional: a book, image, or musical performance can easily contain hundreds of thousands of elements even after compression. However, the most commonly used autoregressive models, Transformers,…

Density EstimationLanguage Modelling

Contextformer: A Transformer with Spatio-Channel Attention for Context Modeling in Learned Image Compression

2022-03-04 · A. Burakhan Koyuncu, Han Gao, Atanas Boev, Georgii Gaikov 외

Entropy modeling is a key component for high-performance image compression algorithms. Recent developments in autoregressive context modeling helped learning-based methods to surpass their classical counterparts. However…

Image CompressionMS-SSIMSSIM

Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference

2024-03-14 · Piotr Nawrot, Adrian Łańcucki, Marcin Chochowski, David Tarjan 외

Transformers have emerged as the backbone of large language models (LLMs). However, generation remains inefficient due to the need to store in memory a cache of key-value representations for past tokens, whose size scale…

GPU