ValueDiff introduces a value‑geometric KV cache eviction strategy for large language models that suppress attention sinks. It ranks tokens by the L2 deviation of their value vectors from the cache mean, a score that aligns with minimal‑disturbance eviction under a max‑entropy assumption. Across several benchmarks—RULER, LongBench, and MATH‑500—ValueDiff consistently retains a higher proportion of useful tokens than prior methods, especially under tight cache budgets.
By Junyoung Park, Jungwook Choi, Mingu Lee
arXiv:2609.27981v1 Announce Type: cross
Abstract: KV-cache eviction is typically evaluated through average quality-memory trade-offs, yet a small average loss can hide requests whose utility degrades...
By Beomgu Kang, SoJin Yun, Hojoon Kim, Hyunseok Seo
arXiv:2602. 08686v3 Announce Type: replace-cross Abstract: Prefill-only KV compression freezes a token subset at the end of prefill and decodes from it without further eviction.
By Ning Yang, Chengzhi Wang, Yibo Liu, Baoliang Tian, Haijun Zhang
arXiv:2607. 21475v1 Announce Type: cross Abstract: Deterministic KV-cache eviction keeps the top-$k$ tokens under an importance score and deletes the rest.
By Peng Xie
arXiv:2608. 05863v1 Announce Type: new Abstract: Modern models no longer keep a plain KV cache: latent caches, learned sparse selectors and recurrent states each carry the model's memory in a different form, and each fails differently under compression.
By Fanzhe Wei, Li Liu, Ziyang Wang, Chenyu Wang
The paper introduces Random Attention, a method that evicts KV cache entries uniformly at random within each attention head while preserving the prompt. Experiments on four models and six reasoning tasks show that this simple strategy matches the performance of the best existing eviction methods and achieves 32‑43% higher throughput in vLLM deployments. The authors explain that the prompt is the most fragile cache component and that reasoning traces are redundantly stored across text and attention heads, making a selection score unnecessary.
By Heng Wang, Jielin Qiu, Wenting Zhao, Cheng Qian, Liangwei Yang, Jiawei Han, Heng Ji, Silvio Savarese, Shelby Heinecke, Huan Wang
arXiv:2606. 17872v1 Announce Type: cross Abstract: Large language models (LLMs) outperform earlier architectures on generative inference and long-context tasks, but their large size introduces significant challenges in memory usage, energy cost, and on-device deployment.
By Ning Ni, Yingjie Lao
arXiv:2607. 04668v1 Announce Type: cross Abstract: On-device LLM decoding is a hard-barriered CPU-SIMD computation that wants every core for milliseconds per token, while the rest of the OS wants those same cores continuously.
By Daeyeon Son
The paper investigates KV‑cache eviction strategies for sparse‑attention models, showing that selecting the largest attention weights is nearly optimal—closing only a median 2–5 % of the gap to full attention. It further demonstrates that differences in performance between eviction methods largely stem from memory usage, with the new training‑free ContourKV allocator outperforming state‑of‑the‑art methods in most pairwise comparisons while matching their byte‑efficiency.
By Jack Shi, Jerry Gu
The paper introduces an elastic key‑value (KV) cache for large language model (LLM) serving that dynamically reclaims a pre‑allocated reserve during decode‑heavy phases and restores it before prefill, using a userspace CUDA virtual‑memory trick that requires no driver changes. The authors implement this mechanism, test it under realistic workloads, and find that it offers only marginal benefits—about a 1 % difference in time‑to‑first‑token for large prefill chunks—and that simpler strategies such as lowering the maximum batch size can achieve similar results. The study also notes that the reserve’s impact diminishes with higher tensor‑parallelism levels.
whyItMatters":"The work demonstrates that a dynamic KV cache reclamation strategy can be implemented without driver patches and that its practical benefits are limited, guiding future LLM serving optimizations toward simpler approaches."
By Sathishkumar Sivashanmugam
TwinKV is a training‑free, attention‑free repair pass that identifies and swaps orphaned and redundant tokens in a KV cache, improving long‑context inference for small models. It works by detecting near‑duplicate keys and can be composed with existing eviction policies without altering their scoring rules. Experiments on Qwen3‑4B and Llama‑3.2‑1B across LongBench, LooGLE, RULER, and MMLU‑Pro show that TwinKV consistently improves performance for most configurations, especially at tighter compression ratios.
By Hong Chen, Yudong Zeng, Yongwei Huang, Zuhao Ouyang, Junyan Zhang, Xuming Hu
arXiv:2607. 27281v1 Announce Type: new Abstract: A capability appears in a language model when the last parts of its circuit align in one stochastic attempt, and getting all but one right is worth nothing.
By Lei Dong