CacheWeaver is a lightweight prompt‑layer technique that orders evidence for Retrieval‑Augmented Generation (RAG) to improve cache reuse in serving engines like vLLM. By maintaining a prefix tree of recently served evidence sequences and greedily placing the most reusable prefix first, it reduces median time‑to‑first‑token by 20‑33 % across three vLLM configurations without harming answer quality. The greedy policy achieves 97.5 % of the gain possible with oracle ordering, showing that most reusable prefix locality can be recovered with a simple scheduling layer.
arXiv:2606. 05875v1 Announce Type: new Abstract: Retrieval-augmented generation (RAG) improves large language model (LLM) answer quality by grounding generation in external evidence, but processing retrieved contexts makes the prefill stage a dominant serving cost.
Retrieval-augmented generation (RAG) improves large language model (LLM) answer quality by grounding generation in external evidence, but processing retrieved contexts makes the prefill stage a dominant serving cost. RAG cache fusion reduces this cost by reusing precomputed key-value (KV) caches for retrieved chunks and selectively recomputing tokens under the current prompt.
The paper investigates cache replacement strategies for large language model (LLM) prefix reuse, analyzing production traces from two companies and testing 14 eviction algorithms in both high-bandwidth memory (HBM) and large memory-pool environments. It finds that sophisticated policies designed for traditional caches offer little advantage over simple LRU, because prefix reuse is largely driven by the regular pacing of active sessions, making recency a strong predictor. The study also highlights new challenges such as heavy-tailed session footprints and variable miss costs, and proposes a compute-savings ratio along with two offline oracles to better quantify these effects, suggesting that effective prefix-cache management should combine recency with selective quick demotion, compute-aware partial eviction, and capacity-dependent granularity.
arXiv:2607. 04281v1 Announce Type: cross Abstract: Semantic caching reduces the latency and cost of retrieval-augmented generation (RAG) by serving cached answers to semantically similar queries, but most existing methods do not model the time-varying freshness of open-web evidence.
arXiv:2608. 03276v1 Announce Type: new Abstract: Long-context inference with large language models is constrained by the linear growth of the key-value cache to sequence length.
arXiv:2608.21362v1 Announce Type: new Abstract: Transformer-based large language models (LLMs) incur high prefill latency because key-value (KV) tensors must be recomputed for each request. Existing...
arXiv:2606. 24467v1 Announce Type: new Abstract: Long-context large language model (LLM) inference is increasingly constrained by the memory footprint and decoding cost of key-value (KV) caches, limiting sustainable deployment on resource-constrained hardware.
StepKV introduces a step-aware approach to compressing the key-value cache used during large language model inference, treating reasoning steps as primary units of retention rather than individual tokens. By linking cache entries to the steps that generated them and estimating each step’s utility from trajectory signals, StepKV assigns a combined token‑ and step‑level score to guide pruning. Experiments on multi‑hop question answering and long‑horizon web reasoning show that StepKV maintains accuracy even under tight cache budgets, outperforming token‑level baselines that suffer sharp performance drops.
arXiv:2606. 04302v1 Announce Type: cross Abstract: Key-value (KV) caching accelerates inference of large language models (LLMs) by reusing past computations for generated tokens.
BeaconKV is a training‑free key‑value cache compression technique for Large Reasoning Models that uses beacon queries—compact representatives of query clusters—to predict which KV pairs will be revisited during long‑horizon reasoning. By focusing on Thought Revisiting Tokens that re‑attend distant context, BeaconKV reduces memory usage up to 5.8× and improves throughput by over 4.3× while largely preserving cache accuracy across multiple open‑source LRMs and reasoning benchmarks.
Long-context large language model (LLM) inference is increasingly constrained by the memory footprint and decoding cost of key-value (KV) caches, limiting sustainable deployment on resource-constrained hardware. Existing KV cache eviction methods typically apply heuristic token scoring over all heads in GQA-based LLMs.
The paper introduces REVA, a method for compressing retrieval-augmented generation (RAG) prompts by aggregating historical query–document–model interactions into reusable evidence views. REVA mines attention traces from the target generator, maps token-level attention to readable words, aggregates importance across repeated document accesses, and produces budget‑specific plain‑text views that maintain document order and the standard RAG interface. Experiments on four benchmarks with modern LLMs show that REVA improves generation quality by 1.0–5.8 points over existing compressors while reducing compression overhead by 5.3 to 15.6 times and adding less than 40 ms of latency.