# HeadWiseKV: Budgeted Per-Head Cache Residency for Hybrid Long-Context Language Models

arXiv: https://arxiv.org/abs/2609.02029 · HTML: https://mlsanity.com/paper/2609.02029/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 24 of its 31 references that are in the map.
- First posted on arXiv 2 Sep 2026.

## Builds on (most cited first, 24 of 24)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Pointer Sentinel Mixture Models](https://mlsanity.com/paper/1609.07843/) (2016)
- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains](https://mlsanity.com/paper/2406.12045/) (2024)
- [RULER: What's the Real Context Size of Your Long-Context Language Models?](https://mlsanity.com/paper/2404.06654/) (2024)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [Evaluating Very Long-Term Conversational Memory of LLM Agents](https://mlsanity.com/paper/2402.17753/) (2024)
- [LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory](https://mlsanity.com/paper/2410.10813/) (2024)
- [KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache](https://mlsanity.com/paper/2402.02750/) (2024)
- [Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference](https://mlsanity.com/paper/2406.10774/) (2024)
- [Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs](https://mlsanity.com/paper/2310.01801/) (2023)
- [Scissorhands: Exploiting the Persistence of Importance Hypothesis for LLM KV Cache Compression at Test Time](https://mlsanity.com/paper/2305.17118/) (2023)
- [DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads](https://mlsanity.com/paper/2410.10819/) (2024)
- [Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inference](https://mlsanity.com/paper/2407.11550/) (2024)
- [InfLLM: Training-Free Long-Context Extrapolation for LLMs with an Efficient Context Memory](https://mlsanity.com/paper/2402.04617/) (2024)
- [PyramidInfer: Pyramid KV Cache Compression for High-throughput LLM Inference](https://mlsanity.com/paper/2405.12532/) (2024)
- [RazorAttention: Efficient KV Cache Compression Through Retrieval Heads](https://mlsanity.com/paper/2407.15891/) (2024)
- [vAttention: Dynamic Memory Management for Serving LLMs without PagedAttention](https://mlsanity.com/paper/2405.04437/) (2024)
- [Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning](https://mlsanity.com/paper/2410.19258/) (2024)
- [CAKE: Cascading and Adaptive KV Cache Eviction with Layer Preferences](https://mlsanity.com/paper/2503.12491/) (2025)
- [TokenSelect: Efficient Long-Context Inference and Length Extrapolation for LLMs via Dynamic Token-Level KV Cache Selection](https://mlsanity.com/paper/2411.02886/) (2024)
- [DiffKV: Differentiated Memory Management for Large Language Models with Parallel KV Compaction](https://mlsanity.com/paper/2412.03131/) (2024)
- [SqueezeAttention: 2D Management of KV-Cache in LLM Inference via Layer-wise Optimal Budget](https://mlsanity.com/paper/2404.04793/) (2024)
- [KV-Compress: Paged KV-Cache Compression with Variable Compression Rates per Attention Head](https://mlsanity.com/paper/2410.00161/) (2024)

## Abstract (arXiv)

Long-context inference retains a growing key--value (KV) cache during decoding, which consumes substantial GPU memory and can reduce generation throughput. This bottleneck remains in hybrid language models because their residual global-attention layers can dominate context-dependent cache demand. We study how to allocate this state under an aggregate KV-residency budget. We introduce HeadWiseKV, a training-free framework that compresses the residual global KV caches of hybrid language models while preserving their native local, recurrent, and linear paths. It assigns each physical KV head a static, multilevel history window, making cache demand predictable before serving. We formulate this allocation as a restricted operational rate--distortion problem and propose SeqCalib as the core policy-generation algorithm in HeadWiseKV. SeqCalib processes layers in execution order and conditions each decision on the lower-layer policy used at deployment, thereby accounting for interactions across depth. A grouped-cache runtime materializes the selected policy as actual per-head KV residency rather than a mask over a full cache. We evaluate downstream quality across four hybrid long-context models and study physical residency and serving behavior on Qwen3.6-27B. HeadWiseKV retains near-Full-KV RULER and LoCoMo quality across the evaluated models. In the fixed-model systems study, it reduces sampled peak device memory by 8.59% at a 112K context length and extends the largest verified successful context from 114K to 161K.