paper-with-me

홈 › Papers

Adaptive Mass-Segmented KV Compression for Long-Context Reasoning

2026-05-22 · Junzhe Yang, Xiaoyu Shen arxiv

The linear growth of the Key-Value (KV) cache is a critical bottleneck in long-form LLM inference. Existing KV compression methods mitigate this by evicting tokens based on importance scores. However, we show that their reliance on global Top-k selection triggers Region Wipe-out: the severe eviction of contiguous reasoning blocks that derails logical coherence. To address this, we propose Adaptive Mass-Segmented (AMS) KV Compression, a framework that shifts the paradigm from token-level competition to region-aware quota allocation. AMS adaptively partitions the KV cache based on the spatial distribution of attention mass, ensuring structurally vital reasoning segments receive guaranteed memory quotas. To ensure stability during iterative decoding, an EMA-based smoothing mechanism is incorporated to prevent jitter in segment boundaries. Crucially, AMS is a universal plug-and-play layer that is orthogonal to existing scorers. It can be seamlessly integrated into representative methods such as TOVA, Expected Attention, KeyDiff, R-KV and TriAttention. AMS is also system-compatible with modern paged-KV serving frameworks such as vLLM, supporting efficient gather-and-compact KV execution without introducing additional steady-state attention overhead. Extensive experiments across a diverse suite of tasks, including mathematical reasoning (MATH500, AIME, GSM8K), code completion, open-domain QA, and sparse retrieval, demonstrate that AMS consistently mitigates structural fragmentation and boosts model performance.

📄 PDF Abstract BibTeX arXiv:2605.23200

Code (0)

등록된 구현이 없습니다.

Tasks

Mathematical ReasoningCode Completion

Similar Papers 제목 키워드 기반

ATACompressor: Adaptive Task-Aware Compression for Efficient Long-Context Processing in LLMs

2026-02-03 · Xuancheng Li, Haitao Li, Yujia Zhou, Qingyao Ai 외 arxiv

Long-context inputs in large language models (LLMs) often suffer from the "lost in the middle" problem, where critical information becomes diluted or ignored due to excessive length. Context compression methods aim to ad…

Developing Adaptive Context Compression Techniques for Large Language Models (LLMs) in Long-Running Interactions

2026-03-31 · Payal Fofadiya, Sunil Tiwari arxiv

Large Language Models (LLMs) often experience performance degradation during long-running interactions due to increasing context length, memory saturation, and computational overhead. This paper presents an adaptive cont…

Computational Efficiency

ArcAligner: Adaptive Recursive Aligner for Compressed Context Embeddings in RAG

2026-01-08 · Jianbo Li, Yi Jiang, Sendong Zhao, Bairui Hu 외 arxiv

Retrieval-Augmented Generation (RAG) helps LLMs stay accurate, but feeding long documents into a prompt makes the model slow and expensive. This has motivated context compression, ranging from token pruning and summariza…

DynTok: Dynamic Compression of Visual Tokens for Efficient and Effective Video Understanding

2025-06-04 · Hongzhi Zhang, Jingyuan Zhang, Xingguang Ji, Qi Wang 외

Typical video modeling methods, such as LLava, represent videos as sequences of visual tokens, which are then processed by the LLM backbone for effective video understanding. However, this approach leads to a massive num…

MMEVideo MMEVideo Understanding

LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding

2024-10-22 · Xiaoqian Shen, Yunyang Xiong, Changsheng Zhao, Lemeng Wu 외

Multimodal Large Language Models (MLLMs) have shown promising progress in understanding and analyzing video content. However, processing long videos remains a significant challenge constrained by LLM's context size. To a…

Token ReductionVideo Question AnsweringVideo UnderstandingZero-Shot Video Question Answer