# DepthWeave-KV: Token-Adaptive Cross-Layer Residual Factorization for Long-Context KV Cache Compression

arXiv: https://arxiv.org/abs/2607.06523 · HTML: https://mlsanity.com/paper/2607.06523/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in KV cache optimization, one of 12 directions in Efficient LLM inference and serving.
- Builds on 50 of its 70 references that are in the map.
- First posted on arXiv 7 Jul 2026.

## Placed in

- Efficient LLM inference and serving / [KV cache optimization](https://mlsanity.com/inference/kv-cache-optimization/)

## Builds on (most cited first, 30 of 50)

- [A Survey on LLM-as-a-Judge](https://mlsanity.com/paper/2411.15594/) (2024)
- [Mooncake: A KVCache-centric Disaggregated Architecture for LLM Serving](https://mlsanity.com/paper/2407.00079/) (2024)
- [A Survey on Large Language Model Acceleration based on KV Cache Management](https://mlsanity.com/paper/2412.19442/) (2024)
- [MiniCache: KV Cache Compression in Depth Dimension for Large Language Models](https://mlsanity.com/paper/2405.14366/) (2024)
- [Hymba: A Hybrid-head Architecture for Small Language Models](https://mlsanity.com/paper/2411.13676/) (2024)
- [Model Tells You Where to Merge: Adaptive KV Cache Merging for LLMs on Long-Context Tasks](https://mlsanity.com/paper/2407.08454/) (2024)
- [Eigen Attention: Attention in Low-Rank Space for KV Cache Compression](https://mlsanity.com/paper/2408.05646/) (2024)
- [The Modern Mathematics of Deep Learning](https://mlsanity.com/paper/2105.04026/) (2021)
- [ChunkKV: Semantic-Preserving KV Cache Compression for Efficient Long-Context LLM Inference](https://mlsanity.com/paper/2502.00299/) (2025)
- [R-KV: Redundancy-aware KV Cache Compression for Reasoning Models](https://mlsanity.com/paper/2505.24133/) (2025)
- [KVSharer: Efficient Inference via Layer-Wise Dissimilar KV Cache Sharing](https://mlsanity.com/paper/2410.18517/) (2024)
- [Deep learning observables in computational fluid dynamics](https://mlsanity.com/paper/1903.03040/) (2019)
- [Attention Is All You Need for KV Cache in Diffusion LLMs](https://mlsanity.com/paper/2510.14973/) (2025)
- [Learn to Accumulate Evidence from All Training Samples: Theory and Practice](https://mlsanity.com/paper/2306.11113/) (2023)
- [SwiftKV: Fast Prefill-Optimized Inference with Knowledge-Preserving Model Transformation](https://mlsanity.com/paper/2410.03960/) (2024)
- [The Rise of Small Language Models in Healthcare: A Comprehensive Survey](https://mlsanity.com/paper/2504.17119/) (2025)
- [The Pitfalls of KV Cache Compression](https://mlsanity.com/paper/2510.00231/) (2025)
- [FastKV: Decoupling of Context Reduction and KV Cache Compression for Prefill-Decoding Acceleration](https://mlsanity.com/paper/2502.01068/) (2025)
- [Efficient Many-Shot In-Context Learning with Dynamic Block-Sparse Attention](https://mlsanity.com/paper/2503.08640/) (2025)
- [Make Each Token Count: Towards Improving Long-Context Performance with KV Cache Eviction](https://mlsanity.com/paper/2605.09649/) (2026)
- [Compressing KV Cache for Long-Context LLM Inference with Inter-Layer Attention Similarity](https://mlsanity.com/paper/2412.02252/) (2024)
- [More Tokens, Lower Precision: Towards the Optimal Token-Precision Trade-off in KV Cache Compression](https://mlsanity.com/paper/2412.12706/) (2024)
- [TailorKV: A Hybrid Framework for Long-Context Inference via Tailored KV Cache Optimization](https://mlsanity.com/paper/2505.19586/) (2025)
- [EvolKV: Evolutionary KV Cache Compression for LLM Inference](https://mlsanity.com/paper/2509.08315/) (2025)
- [SpindleKV: A Novel KV Cache Reduction Method Balancing Both Shallow and Deep Layers](https://mlsanity.com/paper/2507.06517/) (2025)
- [AI Agents as Universal Task Solvers](https://mlsanity.com/paper/2510.12066/) (2025)
- [Q-KVComm: Efficient Multi-Agent Communication Via Adaptive KV Cache Compression](https://mlsanity.com/paper/2512.17914/) (2025)
- [ReST-KV: Robust KV Cache Eviction with Layer-wise Output Reconstruction and Spatial-Temporal Smoothing](https://mlsanity.com/paper/2605.08840/) (2026)
- [Lossless KV Cache Compression to 2%](https://mlsanity.com/paper/2410.15252/) (2024)
- [Residual vector quantization for KV cache compression in large language model](https://mlsanity.com/paper/2410.15704/) (2024)

## Built on by (most cited first, 1 of 1)

- [CORAL-KV: Compact Cross-Layer Prediction and Residual Coding for Autoregressive Video KV Caches](https://mlsanity.com/paper/2610.12514/) (2026)

## Abstract (arXiv)

Long-context language model inference is increasingly limited by the memory bandwidth and capacity required to store key-value caches, yet existing compression methods often apply uniform budgets across layers or tokens and degrade retrieval when lexical cues and semantic states require different preservation. We introduce DepthWeave-KV, a token-adaptive cache compression method that factorizes key and value states across neighboring transformer layers using shared low-rank channel bases while retaining lightweight token-specific residuals where attention behavior is sensitive. DepthWeave-KV combines cross-depth residual factorization with a token-conditional depth router that allocates higher reconstruction rank to instruction-bearing and retrieval-critical tokens, and uses calibration-free online error tracking from attention-output probes to adapt compression during generation without retraining the base model. A fused CUDA implementation jointly performs basis lookup, residual dequantization, and attention projection to reduce decode-time memory traffic. Across LongBench, Needle-in-a-Haystack, L-Eval, and long-form QA and summarization benchmarks, DepthWeave-KV achieves near-full-cache task quality with substantially lower memory use, improving average score and retrieval accuracy over prior compressed caches while reaching 8.3x KV memory reduction and 72.8 tokens per second at 64K context.