CORTEXA
← Browse
arxivcs.LGcs.CL2026-06-30

RaBitQCache: Rotated Binary Quantization for KVCache in Long Context LLM Inference

Wenhao Li, Jinhao Dong, Hailin Zhang, Wenhang Shi, Wei Lu, Xiaoyong Du

Long-context Large Language Model inference is severely bottlenecked by the massive Key-Value (KV) cache, yet existing sparse attention methods often suffer from static fixed-budget (Top-k) retrieval or rely on proxy scores that are computationally expensive and biased. To address these limitations, we propose RaBitQCache, a novel sparse attention framework that utilizes randomized rotated binary quantization and high-throughput binary-INT4 arithmetic to efficiently estimate attention weights. Our proxy score serves as an unbiased estimator with a proven error bound, enabling adaptive Top-p retrieval that dynamically adjusts the token budget based on actual attention sparsity. We further implement a hardware-aware system with asynchronous pipelining and lazy updates to mask overhead. Evaluations demonstrate that RaBitQCache significantly accelerates inference and reduces memory I/O while preserving generation quality compared to state-of-the-art baselines. Code is available at https://github.com/Sakuraaa0/RaBitQCache.git.

View free PDFSource page

Related papers

arxivcs.CLcs.AIcs.LG2026-06-29

MATCH: Modulating Attention via In-Context Retrieval for Long-Context Transformers

Linrui Ma, Chun Hei Lo, Xinyu Wang, Peng Lu, Xihao Yuan, Hanting Chen, et al.

The quadratic computational cost of traditional attention mechanisms poses a major bottleneck to the scalability and practical deployment of large language models (LLMs), particularly in long-context scenarios. To improve efficiency, existing approaches often enforce rigid struct…

View free PDFSource page
arxivcs.CLcs.LG2026-07-16

VarRate: Training-Free Variable-Rate KV Cache Compression for Long-Context LLMs

Shahrzad Esmat, Dhawal Shah, Ali Jannesari

The key-value (KV) cache is the main memory bottleneck in long-context large language model (LLM) inference. Two leading training-free families are both structurally limited: token-selection methods (SnapKV, Ada-KV) score importance from an observation window and evict low-scorin…

View free PDFSource page
arxivcs.LGcs.AIcs.CL2026-06-30

QVal: Cheaply Evaluating Dense Supervision Signals for Long-Horizon LLM Agents

Sergio Hernández-Gutiérrez, Matteo Merler, Ilze Amanda Auzina, Joschka Strüber, Ameya Prabhu, Matthias Bethge

LLM agents increasingly act over long horizons, where a single trajectory can contain hundreds or thousands of actions. In these settings, outcome-only rewards provide too sparse guidance, failing to inform the model about the goodness of intermediate actions. Dense supervision m…

View free PDFSource page
arxivcs.LGcs.AIcs.ARcs.CLcs.PF2026-07-05

HiFA4: Training-Free 4-bit FlashAttention on Ascend HIF4 NPUs for LLM Inference

Hui Dong, Yanzhao Li, Jie Gao, Chunlu Li, Zhiyuan Zhang, Yupeng Sun, et al.

We present HiFA4, a post-training operator-level design that executes both QK^T and PV in FlashAttention as 4-bit HIF4 Cube GEMMs for LLM inference on Ascend NPUs, while maintaining the online softmax state in FP16. To our knowledge, HiFA4 is the first Ascend-HIF4-targeted design…

View free PDFSource page
arxivcs.LGcs.AIcs.CLcs.IR2026-07-11

Context by Distinct Information: An Auditable Dirichlet-Process Working Memory for Long, Redundant Context Streams

Siddharth Pal, Viktoria Rojkova

Context engineering decides what information a model carries forward, and current designs meter it in tokens: compressing the past into a bounded recurrent state, keeping a key-value entry for every token, or imposing a fixed budget through a window or eviction rule. All three ma…

View free PDFSource page