Image de couverture de livre

Density-aware Soft Context Compression with Semi-Dynamic Compression Ratio

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Yu, Yijiong, Yuan, Shuai, Zheng, Jie, Wang, Huazheng, Pei, Ji
Format:	Preprint
Publié:	2026
Sujets:	Computation and Language
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Speculative Pipeline Decoding: Higher-Accruacy and Zero-Bubble Speculation via Pipeline Parallelism
par: Yu, Yijiong, et autres
Publié: (2026)

SWAA: Sliding Window Attention Adaptation for Efficient and Quality Preserving Long Context Processing
par: Yu, Yijiong, et autres
Publié: (2025)

DAST: Context-Aware Compression in LLMs via Dynamic Allocation of Soft Tokens
par: Chen, Shaoshen, et autres
Publié: (2025)

Patience Is The Key to Large Language Model Reasoning
par: Yu, Yijiong
Publié: (2024)

An Effective Framework to Help Large Language Models Handle Numeric-involved Long-context Tasks
par: Yu, Yijiong
Publié: (2024)

Accelerate Parallelizable Reasoning via Parallel Decoding within One Sequence
par: Yu, Yijiong
Publié: (2025)

Submodular Context Partitioning and Compression for In-Context Learning
par: Zheng, Shaoyi, et autres
Publié: (2025)

Do LLMs Really Think Step-by-step In Implicit Reasoning?
par: Yu, Yijiong
Publié: (2024)

OpenCSG Chinese Corpus: A Series of High-quality Chinese Datasets for LLM Training
par: Yu, Yijiong, et autres
Publié: (2025)

COMI: Coarse-to-fine Context Compression via Marginal Information Gain
par: Tang, Jiwei, et autres
Publié: (2026)

Compressing Lengthy Context With UltraGist
par: Zhang, Peitian, et autres
Publié: (2024)

AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration
par: Lin, Ji, et autres
Publié: (2023)

Adapting LLMs for Efficient Context Processing through Soft Prompt Compression
par: Wang, Cangqing, et autres
Publié: (2024)

SVD-LLM: Truncation-aware Singular Value Decomposition for Large Language Model Compression
par: Wang, Xin, et autres
Publié: (2024)

LongAttnComp: Cross-Family Context Compression for Long-Context Reasoning
par: Ji, Mengmeng, et autres
Publié: (2026)

Long Context Compression with Activation Beacon
par: Zhang, Peitian, et autres
Publié: (2024)

Read As Human: Compressing Context via Parallelizable Close Reading and Skimming
par: Tang, Jiwei, et autres
Publié: (2026)

Task-agnostic Prompt Compression with Context-aware Sentence Embedding and Reward-guided Task Descriptor
par: Liskavets, Barys, et autres
Publié: (2025)

Long-context Language Models Fail in Basic Retrieval Tasks Without Sufficient Reasoning Steps
par: Yu, Yijiong, et autres
Publié: (2024)

DAC: A Dynamic Attention-aware Approach for Task-Agnostic Prompt Compression
par: Zhao, Yi, et autres
Publié: (2025)

CCF: A Context Compression Framework for Efficient Long-Sequence Language Modeling
par: Li, Wenhao, et autres
Publié: (2025)

ZSMerge: Zero-Shot KV Cache Compression for Memory-Efficient Long-Context LLMs
par: Liu, Xin, et autres
Publié: (2025)

AdmTree: Compressing Lengthy Context with Adaptive Semantic Trees
par: Li, Yangning, et autres
Publié: (2025)

Beyond Hard and Soft: Hybrid Context Compression for Balancing Local and Global Information Retention
par: Liao, Huanxuan, et autres
Publié: (2025)

Context Cascade Compression: Exploring the Upper Limits of Text Compression
par: Liu, Fanfan, et autres
Publié: (2025)

OpenBA-V2: Reaching 77.3% High Compression Ratio with Fast Multi-Stage Pruning
par: Qiao, Dan, et autres
Publié: (2024)

Compress the Context, Keep the Commitments: A Formal Framework for Verifiable LLM Context Compression
par: Trukhina, Natalia, et autres
Publié: (2026)

ASVD: Activation-aware Singular Value Decomposition for Compressing Large Language Models
par: Yuan, Zhihang, et autres
Publié: (2023)

Compressing KV Cache for Long-Context LLM Inference with Inter-Layer Attention Similarity
par: Ma, Da, et autres
Publié: (2024)

In-Context Former: Lightning-fast Compressing Context for Large Language Model
par: Wang, Xiangfeng, et autres
Publié: (2024)

DynamicKV: Task-Aware Adaptive KV Cache Compression for Long Context LLMs
par: Zhou, Xiabin, et autres
Publié: (2024)

VQKV: High-Fidelity and High-Ratio Cache Compression via Vector-Quantization
par: Wang, Yixuan, et autres
Publié: (2026)

Effective In-Context Example Selection through Data Compression
par: Sun, Zhongxiang, et autres
Publié: (2024)

Data Distribution Matters: A Data-Centric Perspective on Context Compression for Large Language Model
par: Lv, Kangtao, et autres
Publié: (2026)

Retaining Key Information under High Compression Ratios: Query-Guided Compressor for LLMs
par: Cao, Zhiwei, et autres
Publié: (2024)

Recurrent Context Compression: Efficiently Expanding the Context Window of LLM
par: Huang, Chensen, et autres
Publié: (2024)

OSCAR: Online Soft Compression And Reranking
par: Louis, Maxime, et autres
Publié: (2025)

Beyond Position Bias: Shifting Context Compression from Position-Driven to Semantic-Driven
par: Tang, Jiwei, et autres
Publié: (2026)

Evaluating Zero-Shot Long-Context LLM Compression
par: Wang, Chenyu, et autres
Publié: (2024)

Attn-GS: Attention-Guided Context Compression for Efficient Personalized LLMs
par: Zeng, Shenglai, et autres
Publié: (2026)