Recall Before You Rank: Similarity-Guided Top- K Reuse for Efficient Long-Context Attention

arXiv:2607.27692 · cs.CL, cs.LG · Submitted 2026-07-30 · Read on arXiv

Wenshuai Yao, Wenyong Zhou, Hanyong Shao, Yizhe Chen, Zhiyuan Ning, Yuannuo Feng, Ru Huang, Kechao Tang

cs.CL, cs.LG

Submitted: 2026-07-30

Comments: 9 pages, 9 figures, and 5 tables

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• KV cache — 17× in this paper · explained in 4 episodes
• Sparse Attention — 11× in this paper · explained in 5 episodes
• Perplexity — 6× in this paper · explained in 3 episodes
• Ablation — 4× in this paper · explained in 1 episode
• Fidelity — 4× in this paper · explained in 2 episodes
• Cosine Similarity — 3× in this paper · explained in 3 episodes
• Attention Sinks — 2× in this paper · explained in 2 episodes
• Benchmark — 2× in this paper · explained in 2 episodes
• KV Cache Compression — 2× in this paper · explained in 2 episodes
• Papers — 2× in this paper · explained in 3 episodes
• Pareto Frontier — 2× in this paper · explained in 2 episodes
• Sparsity — 2× in this paper · explained in 5 episodes
• Composition — 1× in this paper · explained in 2 episodes
• Decoder — 1× in this paper · explained in 2 episodes
• Generalization — 1× in this paper · explained in 14 episodes
• Language Models (LLMs) — 1× in this paper · explained in 2 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 9 episodes
• Pruning — 1× in this paper · explained in 4 episodes
• Trace — 1× in this paper · explained in 2 episodes
• Transformer — 1× in this paper · explained in 2 episodes
• Transformers — 1× in this paper · explained in 2 episodes

Sources

• IndexCache: Accelerating Sparse Attention via Cross-Layer Index Reuse
• Kascade: A Practical Sparse Attention Method for Long-Context LLM Inference
• The Llama 3 Herd of Models
• Qwen2.5 Technical Report
• Predict, Reuse, and Repair: Accelerating Dynamic Sparse Attention for Long-Context LLM Decoding
• Qwen2.5-1M Technical Report

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering