Daily Papers

Context-to-LoRA methods encode documents as low-rank parameter updates, allowing language models to answer without retaining the source text in their decoding context. How reliably do these updates preserve facts and support their use under different instructions? Controlled diagnosis of existing context-to-LoRA methods separates fact discrimination, target retention, and response adaptation. Competing sources influence answers, while target-only construction and memory-state optimization improve recovery at fixed update size. Recovering facts also does not ensure that answers follow response instructions. We introduce Cache2LoRA, which uses each query to aggregate evidence from a reusable frozen document cache before generating LoRA updates. Trainable encoding operates only on the query and memory-token suffix, and multi-style supervision teaches both evidence preservation and answer expression. Across 17 tasks, Cache2LoRA improves on existing context-to-LoRA methods and achieves macro F1 of 0.663, approaching full-context in-context learning at 0.671. For 32K documents and 256-token outputs at batch size 16, its decoding throughput is 5.9times that of full-context ICL. At matched memory size, generated LoRA preserves answer accuracy more reliably than the learned-KV variant under long, diverse intervening text and combines more effectively with retained document KV, yielding higher hybrid F1 and EM.

Long contexts enlarge key--value (KV) caches, increasing memory use and attention costs during generation. Compressing these caches must preserve information for unseen questions. We introduce **Latent Compaction**, which learns a compact document-specific cache through test-time training. We initialize cache entries using reconstruction importance, teacher attention, and key-space diversity. Training first fits attention outputs and log-normalizers to those of the full cache, with reconstruction-selected entries held fixed. It then jointly optimizes the entire cache through end-to-end distillation from the full-cache model. Model weights remain frozen throughout, and training uses only generic prompts about the document, without access to downstream questions or answers. Experiments on RULER, LongBench, and Phonebook Lookup demonstrate substantial gains over budget-matched compression baselines. On a RULER, for example, our method achieves 96.7% accuracy at 32times compression, compared with 40.0% for KVZip+. The method also transfers across model scales and families, with retrieval gains on both Qwen3 and MiniCPM5. These results show that treating the KV cache as a neural network and optimizing it with test-time training can preserve retrieval accuracy under aggressive compression. The code is released at https://www.github.com/THUNLP/latent-compaction

Multi-document retrieval-augmented generation (RAG) requires a language model to process multiple retrieved text chunks before answering a question. Precomputing each chunk's KV cache independently and concatenating the caches when the chunks are retrieved can accelerate this step. However, the assembled cache lacks cross-chunk attention information, reducing answer quality. Selective recomputation methods recover the missing cross-chunk context by rerunning the target LLM on selected tokens, incurring substantial online computation. We introduce CacheRepair, a lightweight network that learns the difference between independently computed KV caches and those produced by processing the chunks together. The network combines compressed KV features with token embeddings and uses attention that is bidirectional within each chunk and flows from earlier to later chunks. Each repair block receives the compressed cache features, and the predicted residual is added to every document token's cache. Each repair network is trained for a specific frozen target LLM on a generic retrieval corpus and reused across downstream datasets. Our analysis shows that repair reduces KV errors both near chunk boundaries and throughout chunk interiors. Evaluation across three target LLMs and four downstream datasets places CacheRepair on the measured answer-quality-latency Pareto frontier in eleven of twelve model-dataset combinations. Reported time to first token (TTFT) includes online cache transfer and repair. Across all twelve combinations, the largest repairers achieve 1.69-4.61times speedups in median TTFT over full prefill and improve mean F1 by 2.1-26.1 percentage points over direct cache reuse.

Large language model (LLM) agents repeatedly load reusable content, such as skills, documents, and memory entries, into the current context. Re-encoding this content for every request wastes computation. Position-independent caching (PIC) alleviates this by encoding each artifact independently and reusing its key-value (KV) states at arbitrary positions, but it incurs a quality loss relative to full-context prefill. Existing methods repair this loss by restoring global position IDs or recomputing selected tokens. In this work, we isolate the source of the loss, finding that the positional mismatch has minor effect, and independently cached artifacts retain faithful representations: reading a provided artifact stays largely accurate, and performance degrades only when the model must select among multiple artifacts. Moreover, replacing PIC's attention scores with full-prefill scores recovers performance with the cached KV unchanged, localizing the failure to the attention rather than KV recomputation. Motivated by this, we propose Attuner, a query-side adaptation method that learns to read a frozen artifact cache. Attuner inserts low-rank adapters into the query projections and is trained by distilling full-prefill distribution into the student. It trains fewer than 0.05\% of the model parameters and, at inference, requires neither cache recomputation nor a full-context reference. On Qwen3-4B and Qwen3-8B across seven benchmarks covering skills, documents, memory, and code, Attuner substantially outperforms prior PIC baselines in both in-domain and out-of-domain settings, matches full-context prefill quality while providing up to 3.73times speedup.

Late-interaction retrievers handle exclusion queries poorly. When a user asks for X but not Z, the additive MaxSim score promotes documents covering Z, a problem we call exclusion inversion. We show that no readout of the frozen vectors recovers the constraint, because the difficulty lies in identifying the excluded topic, which depends on the query alone. EXCISE operates at query time and corrects the inversion while leaving the index frozen. Two query-side modules totalling 1.5M parameters identify the topic and re-embed a 100-document shortlist, and a parameter-free rule demotes candidates matching that topic. Across six collections and three backbones, EXCISE is the strongest system in all eighteen backbone-collection cells against that backbone's own frozen and fine-tuned baselines. It raises exclusion success@10 on ExcluIR from 0.058 to 0.691 and raises Boolean NOT accuracy from 0.25-0.29 to 0.90-0.92. Pooled over 1,860 queries, it outperforms every fine-tuned cross-encoder, each of which loses no-harm nDCG@10, whereas EXCISE matches its frozen baseline on its strongest backbone. We release X-BENCH, a tiered benchmark of explicit, implicit, and compound exclusions with no-harm and Boolean controls.

Key-Value (KV) cache plays a crucial role in accelerating inference in large language models (LLMs) by storing intermediate attention states and avoiding redundant computation during autoregressive generation. However, its memory footprint scales linearly with sequence length, often leading to severe memory bottlenecks on resource-constrained hardware. Prior work has explored offloading KV cache to the CPU while retaining only a subset on the GPU, but these approaches often rely on imprecise token selection and suffer performance degradation in long-generation tasks such as chain-of-thought reasoning. In this paper, we propose a novel KV cache management strategy, IceCache, which integrates semantic token clustering with PagedAttention. By organizing semantically related tokens into contiguous memory regions managed by a hierarchical, dynamically updatable data structure, our method enables more efficient token selection and better utilization of memory bandwidth during CPU-GPU transfers. Experimental results on LongBench show that, with a 256-token budget, IceCache maintains 99% of the original accuracy achieved by the full KV cache model. Moreover, compared to other offloading-based methods, IceCache attains competitive or even superior latency and accuracy while using only 25% of the KV cache token budget, demonstrating its effectiveness in long-sequence scenarios. The code is available on our project website at https://yuzhenmao.github.io/IceCache/.

Key-value (KV) caching is a powerful technique for accelerating large language model inference and generation. Inference workloads are large and diverse, which makes them difficult to cache effectively. Existing cache management strategies adopt the least-recently-used policy for evicting cache blocks. However, LRU leads to multiple unrelated workloads flushing each other's caches. To address this, we integrate adaptive caching that dynamically allocates cache space between recently and frequently occurring KV blocks. Evaluations show that it improves the KV cache hit rate by up to 10.8% and reduces time to first token by up to 12.6% over naive vLLM on synthetic document question answering workloads, and 2.1% and 2.0% respectively on real-world conversation workloads. The method generalizes well to batch inference and demonstrates clear interpretability while effectively accommodating diverse workloads.

How much of a running workflow should a language model agent revise? ControlScope compares continuing generated code, editing the next tool call's data arguments, and replacing the unfinished workflow from the same public execution state. The nested permissions separate available repairs from the actions an agent selects. We evaluate one-time and repeated reviews across filesystem tasks, ALFWorld, and AppWorld. Across two source programs per task and three reasoning-reviewer draws on 20 filesystem tasks, FULL completes 15-16 tasks versus 13 for KEEP; across four fast draws it completes 10-13 versus 13. Fresh student-record confirmation reproduces a batch-read repair. ALFWorld fast panels yield KEEP/ARG/FULL scores of 85/86/87 on 87 tasks across 52 scenes and 134/134/127 on 134 tasks across four scenes; reasoning on the 87-task cohort also yields 85/86/87 with substantial review cost. An AppWorld V1 official-test panel of 585 task instances from 195 scenario templates shows small net differences. Frozen replays expose viable agent-written replacements interrupted by later revision in two failed file-organization runs. An offline source-trajectory midpoint comparison shows later reviews completing an insufficient repair. Five-call protection saves 19.4% of logged model output and loses one success across 20 fresh source runs. An argument-only shortcut shows that the broader sampled policy can overlook a cheaper successful edit available in both operation sets. These outcomes tie repair access to actual choices and subsequent execution.

Large language models (LLMs) often incorporate multiple text chunks in their inputs to provide the necessary contexts. To speed up the prefill of the long LLM inputs, one can pre-compute the KV cache of a text and re-use the KV cache when the context is reused as the prefix of another LLM input. However, the reused text chunks are not always the input prefix, and when they are not, their precomputed KV caches cannot be directly used since they ignore the text's cross-attention with the preceding text in the LLM input. Thus, the benefits of reusing KV caches remain largely unrealized. This paper tackles just one question: when an LLM input contains multiple text chunks, how to quickly combine their precomputed KV caches in order to achieve the same generation quality as the expensive full prefill (i.e., without reusing KV cache)? We present CacheBlend, a scheme that reuses the pre-computed KV caches, regardless prefix or not, and selectively recomputes the KV values of a small subset of tokens to partially update each reused KV cache. In the meantime,the small extra delay for recomputing some tokens can be pipelined with the retrieval of KV caches within the same job,allowing CacheBlend to store KV caches in slower devices with more storage capacity while retrieving them without increasing the inference delay. By comparing CacheBlend with the state-of-the-art KV cache reusing schemes on three open-source LLMs of various sizes and four popular benchmark datasets of different tasks, we show that CacheBlend reduces time-to-first-token (TTFT) by 2.2-3.3X and increases the inference throughput by 2.8-5X, compared with full KV recompute, without compromising generation quality or incurring more storage cost.

AI-powered search products such as ChatGPT search, Google's AI Overviews, and Perplexity provide LLM-synthesized answers grounded in live web results. We developed OreoLook (formerly lixSearch), an open-source answer engine using automated browser agents and provider-routed LLM inference. Its local search, caching, session-management, and embedding stack runs on commodity CPU hardware; answer synthesis is performed by a remote inference provider. As usage grew, sessions lost context, equivalent queries triggered redundant work, and URLs were repeatedly embedded across sessions. We present a three-layer caching architecture: (1) a Session Context Window maintaining a rolling window of recent messages in Redis with automatic overflow to Huffman-compressed disk archives; (2) a Semantic Query Cache catches rephrasings via cosine similarity on embedding vectors, eliminating redundant LLM invocations; and (3) a URL Embedding Cache that deduplicates embedding computations across sessions. Deployed on a single 8-vCPU Intel Cascade Lake server (2 GHz, 32 GB RAM) running 30 Hypercorn worker processes across three containerized replicas, the evaluated system reported an 89.3% aggregate Redis keyspace hit rate with 0.1 ms read latency and just 1.38 MB of memory overhead. A background LRU eviction daemon migrates idle sessions from Redis to disk and re-hydrates them on demand, enabling conversations that can be resumed hours or days later under the configured retention policy.

The prefill stage of long-context Retrieval-Augmented Generation (RAG) is severely bottlenecked by computational overhead. To mitigate this, recent methods assemble pre-calculated KV caches of retrieved RAG documents (by a user query) and reprocess selected tokens to recover cross-attention between these pre-calculated KV caches. However, we identify a fundamental "crowding-out effect" in current token selection criteria: globally salient but user-query-irrelevant tokens saturate the limited recomputation budget, displacing the tokens truly essential for answering the user query and degrading inference accuracy. We propose ProphetKV, a user-query-driven KV Cache reuse method for RAG scenarios. ProphetKV dynamically prioritizes tokens based on their semantic relevance to the user query and employs a dual-stage recomputation pipeline to fuse layer-wise attention metrics into a high-utility set. By ensuring the recomputation budget is dedicated to bridging the informational gap between retrieved context and the user query, ProphetKV achieves high-fidelity attention recovery with minimal overhead. Our extensive evaluation results show that ProphetKV retains 96%-101% of full-prefill accuracy with only a 20% recomputation ratio, while achieving accuracy improvements of 8.8%-24.9% on RULER and 18.6%-50.9% on LongBench over the state-of-the-art approaches (e.g., CacheBlend, EPIC, and KVShare).

Despite strong performance on existing benchmarks, it remains unclear whether large language models can reason over genuinely novel scientific information. Most evaluations score end-to-end RAG pipelines, where reasoning is confounded with retrieval and toolchain choices, and the signal is further contaminated by parametric memorization and open-web volatility. We introduce DeR2, a controlled deep-research sandbox that isolates document-grounded reasoning while preserving core difficulties of deep search: multi-step synthesis, denoising, and evidence-based conclusion making. DeR2 decouples evidence access from reasoning via four regimes--Instruction-only, Concepts (gold concepts without documents), Related-only (only relevant documents), and Full-set (relevant documents plus topically related distractors)--yielding interpretable regime gaps that operationalize retrieval loss vs. reasoning loss and enable fine-grained error attribution. To prevent parametric leakage, we apply a two-phase validation that requires parametric failure without evidence while ensuring oracle-concept solvability. To ensure reproducibility, each instance provides a frozen document library (drawn from 2023-2025 theoretical papers) with expert-annotated concepts and validated rationales. Experiments across a diverse set of state-of-the-art foundation models reveal substantial variation and significant headroom: some models exhibit mode-switch fragility, performing worse with the Full-set than with Instruction-only, while others show structural concept misuse, correctly naming concepts but failing to execute them as procedures.

We describe KVLink, an approach for efficient key-value (KV) cache reuse in large language models (LLMs). In many LLM applications, different inputs can share overlapping context, such as the same retrieved document appearing in multiple queries. However, the LLMs still need to encode the entire context for each query, leading to redundant computation. In this paper, we investigate a new strategy to eliminate such inefficiency, where the KV cache of each document is precomputed independently. During inference, the KV caches of retrieved documents are concatenated, allowing the model to reuse cached representations instead of recomputing them. To mitigate the performance degradation when using KV caches computed independently for each document, KVLink introduces two key techniques: adjusting positional embeddings of the KV cache at inference to match the global position after concatenation, and using trainable special tokens to restore self-attention across independently encoded documents. Experiments across 7 datasets demonstrate that KVLink improves question answering accuracy by an average of 4% over state-of-the-art methods. Furthermore, by leveraging precomputed KV caches, our approach reduces time-to-first-token by up to 96% compared to standard LLM inference, making it a scalable and efficient solution for context reuse. Additionally, KVLink can be combined with KV cache compression to further save cache loading and storage overhead while outperforming the baselines.

Retrieval-Augmented Generation (RAG) is often used with Large Language Models (LLMs) to infuse domain knowledge or user-specific information. In RAG, given a user query, a retriever extracts chunks of relevant text from a knowledge base. These chunks are sent to an LLM as part of the input prompt. Typically, any given chunk is repeatedly retrieved across user questions. However, currently, for every question, attention-layers in LLMs fully compute the key values (KVs) repeatedly for the input chunks, as state-of-the-art methods cannot reuse KV-caches when chunks appear at arbitrary locations with arbitrary contexts. Naive reuse leads to output quality degradation. This leads to potentially redundant computations on expensive GPUs and increases latency. In this work, we propose Cache-Craft, a system for managing and reusing precomputed KVs corresponding to the text chunks (we call chunk-caches) in RAG-based systems. We present how to identify chunk-caches that are reusable, how to efficiently perform a small fraction of recomputation to fix the cache to maintain output quality, and how to efficiently store and evict chunk-caches in the hardware for maximizing reuse while masking any overheads. With real production workloads as well as synthetic datasets, we show that Cache-Craft reduces redundant computation by 51% over SOTA prefix-caching and 75% over full recomputation. Additionally, with continuous batching on a real production workload, we get a 1.6X speed up in throughput and a 2X reduction in end-to-end response latency over prefix-caching while maintaining quality, for both the LLaMA-3-8B and LLaMA-3-70B models.

Large Language Models (LLMs) rely heavily on Key-Value (KV) caching to minimize inference latency. However, standard KV caches are context-dependent: reusing a cached document in a new context requires recomputing KV states to account for shifts in attention distribution. Existing solutions such as CacheBlend, EPIC, and SAM-KV mitigate this issue by selectively recomputing a subset of tokens; however, they still incur non-negligible computational overhead (FLOPs) and increased Time-to-First-Token (TTFT) latency. In this paper, we propose KV Packet, a recomputation-free cache reuse framework that treats cached documents as immutable ``packets'' wrapped in light-weight trainable soft-token adapters, which are trained via self-supervised distillation to bridge context discontinuities. Experiments on Llama-3.1 and Qwen2.5 demonstrate that the proposed KV Packet method achieves near-zero FLOPs and lower TTFT than recomputation-based baselines, while retaining F1 scores comparable to those of the full recomputation baseline.

LLM serving systems process heterogeneous query workloads where different categories exhibit different characteristics. Code queries cluster densely in embedding space while conversational queries distribute sparsely. Content staleness varies from minutes (stock data) to months (code patterns). Query repetition patterns range from power-law (code) to uniform (conversation), producing long tail cache hit rate distributions: high-repetition categories achieve 40-60% hit rates while low-repetition or volatile categories achieve 5-15% hit rates. Vector databases must exclude the long tail because remote search costs (30ms) require 15--20% hit rates to break even, leaving 20-30% of production traffic uncached. Uniform cache policies compound this problem: fixed thresholds cause false positives in dense spaces and miss valid paraphrases in sparse spaces; fixed TTLs waste memory or serve stale data. This paper presents category-aware semantic caching where similarity thresholds, TTLs, and quotas vary by query category. We present a hybrid architecture separating in-memory HNSW search from external document storage, reducing miss cost from 30ms to 2ms. This reduction makes low-hit-rate categories economically viable (break-even at 3-5% versus 15-20%), enabling cache coverage across the entire workload distribution. Adaptive load-based policies extend this framework to respond to downstream model load, dynamically adjusting thresholds and TTLs to reduce traffic to overloaded models by 9-17% in theoretical projections.

Large language model (LLM) based agentic workflows have become a popular paradigm for coordinating multiple specialized agents to solve complex tasks. To improve serving efficiency, existing LLM systems employ prefix caching to reuse key-value (KV) tensors corresponding to agents' fixed prompts, thereby avoiding redundant computation across repeated invocations. However, current systems typically evict KV caches using a Least Recently Used (LRU) policy, which fails to anticipate future agent usage and often discards KV caches shortly before their reuse. This leads to frequent cache misses and substantial recomputation or swapping overhead. We present KVFlow, a workflow-aware KV cache management framework tailored for agentic workloads. KVFlow abstracts the agent execution schedule as an Agent Step Graph and assigns each agent a steps-to-execution value that estimates its temporal proximity to future activation. These values guide a fine-grained eviction policy at the KV node level, allowing KVFlow to preserve entries likely to be reused and efficiently manage shared prefixes in tree-structured caches. Moreover, KVFlow introduces a fully overlapped KV prefetching mechanism, which proactively loads required tensors from CPU to GPU in background threads for agents scheduled in the next step, thereby avoiding cache miss stalls during generation. Compared to SGLang with hierarchical radix cache, KVFlow achieves up to 1.83times speedup for single workflows with large prompts, and up to 2.19times speedup for scenarios with many concurrent workflows.

High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks dynamically. When managed inefficiently, this memory can be significantly wasted by fragmentation and redundant duplication, limiting the batch size. To address this problem, we propose PagedAttention, an attention algorithm inspired by the classical virtual memory and paging techniques in operating systems. On top of it, we build vLLM, an LLM serving system that achieves (1) near-zero waste in KV cache memory and (2) flexible sharing of KV cache within and across requests to further reduce memory usage. Our evaluations show that vLLM improves the throughput of popular LLMs by 2-4times with the same level of latency compared to the state-of-the-art systems, such as FasterTransformer and Orca. The improvement is more pronounced with longer sequences, larger models, and more complex decoding algorithms. vLLM's source code is publicly available at https://github.com/vllm-project/vllm

LLM agents increasingly rely on retrieval buffers to store and reuse past experience, yet the cache management policies governing these buffers remain largely ad-hoc. We formalize this as an online semantic cache replacement problem with switching costs, where items are matched by embedding similarity and hit quality is continuous rather than binary. Through experiments on two datasets from MemoryBench-Full (LoCoMo, DialSim) with 8 replacement policies, we reveal a surprising finding: classic heuristics (LRU, LFU) consistently underperform the naive FIFO baseline on semantic workloads, due to the absence of temporal locality and frequency concentration. We propose SOLAR, a learning-augmented framework that derives modification timing from regret accumulation (achieving sim17\% modification rate) and content selection from Bayesian online learning over implicit retrieval feedback. We prove SOLAR achieves a constant competitive ratio leq 3, independent of cache size and horizon (vs.\ Ω(K) for FIFO), and eviction regret O(KTlog T), matching the Ω(KT) lower bound up to logarithmic factors. Experiments demonstrate 5--75\% relative improvement over FIFO at tight cache sizes, with a clearly characterized phase transition at the working set boundary. Synthetic experiments with 5000-item pools further reveal an inverted-U relationship between pool size and retrieval quality, justifying capacity constraints as a retrieval noise phenomenon rather than a storage limitation.

Techniques enabling large language models (LLMs) to "think more" by generating and attending to intermediate reasoning steps have shown promise in solving complex problems. However, the standard approaches generate sequences of discrete tokens immediately before responding, and so they can incur significant latency costs and be challenging to optimize. In this work, we demonstrate that a frozen LLM can be augmented with an offline coprocessor that operates on the model's key-value (kv) cache. This coprocessor augments the cache with a set of latent embeddings designed to improve the fidelity of subsequent decoding. We train this coprocessor using the language modeling loss from the decoder on standard pretraining data, while keeping the decoder itself frozen. This approach enables the model to learn, in an end-to-end differentiable fashion, how to distill additional computation into its kv-cache. Because the decoder remains unchanged, the coprocessor can operate offline and asynchronously, and the language model can function normally if the coprocessor is unavailable or if a given cache is deemed not to require extra computation. We show experimentally that when a cache is augmented, the decoder achieves lower perplexity on numerous subsequent tokens. Furthermore, even without any task-specific training, our experiments demonstrate that cache augmentation consistently reduces perplexity and improves performance across a range of reasoning-intensive tasks.

The key-value (KV) cache is a primary capacity and bandwidth bottleneck in long-context LLM serving. We present Minima-KV, a retention-preserving hierarchy for mixed-format paged attention. Recent and protected Anchor pages remain in FP8, while older non-anchor pages move to packed TQ3; every live-request page remains addressable. Format-specific kernels compute partial attention states and combine them through a globally normalized online-softmax merge, enabling direct heterogeneous decode without a cache-sized dense shadow. Across separate, configuration-bound Qwen3.6-27B profiles on a single 96-GB NVIDIA RTX PRO 6000 Blackwell GPU, deployment accounting reports 18.3 KiB of attention KV per live token, corresponding to 3.50x compression relative to BF16 and 1.75x relative to FP8. A materializing quality profile matches its dense control on 16K RULER needle-in-a-haystack tasks. On the same 503-question LongBench v2 set, measured deltas are -0.80, -0.60, and -0.40 percentage points at 16K, 32K, and 64K. A separate single-pair direct-decode canary with two 59,008-token requests measures 3.625x active-KV compression and 0.9821x throughput relative to its control, routes all 16 full-attention layers without fallback, and retains no dense shadow. These results establish a practical mixed-format path for compressing long-context state without evicting live-request KV pages.

KV cache compression promises increased throughput and efficiency with negligible loss in performance. While the gains in throughput are indisputable and recent literature has indeed shown minimal degradation on particular benchmarks, in general the consequences of compression in realistic scenarios such as multi-instruction prompting have been insufficiently studied. In this paper, we identify several pitfalls practitioners should be aware of when deploying KV cache compressed LLMs. Importantly, we show that certain instructions degrade much more rapidly with compression, effectively causing them to be completely ignored by the LLM. As a practical example of that, we highlight system prompt leakage as a case study, empirically showing the impact of compression on leakage and general instruction following. We show several factors that play a role in prompt leakage: compression method, instruction order, and KV eviction bias. We then propose simple changes to KV cache eviction policies that can reduce the impact of these factors and improve the overall performance in multi-instruction tasks.

Document understanding in real-world applications often requires processing heterogeneous, multi-page document packets containing multiple documents stitched together. Despite recent advances in visual document understanding, the fundamental task of document packet splitting, which involves separating a document packet into individual units, remains largely unaddressed. We present the first comprehensive benchmark dataset, DocSplit, along with novel evaluation metrics for assessing the document packet splitting capabilities of large language models. DocSplit comprises five datasets of varying complexity, covering diverse document types, layouts, and multimodal settings. We formalize the DocSplit task, which requires models to identify document boundaries, classify document types, and maintain correct page ordering within a document packet. The benchmark addresses real-world challenges, including out-of-order pages, interleaved documents, and documents lacking clear demarcations. We conduct extensive experiments evaluating multimodal LLMs on our datasets, revealing significant performance gaps in current models' ability to handle complex document splitting tasks. The DocSplit benchmark datasets and proposed novel evaluation metrics provide a systematic framework for advancing document understanding capabilities essential for legal, financial, healthcare, and other document-intensive domains. We release the datasets to facilitate future research in document packet processing.

KV cache has traditionally been stored in GPU memory to accelerate the decoding phase of large language model (LLM) inference. However, it is increasingly necessary to move KV caches outside GPU devices, to enable cache reuse across different queries and inference engines. Our real-world usage statistics confirm this trend: over time, the total KV cache stored by users has grown rapidly, far exceeding the capacity of GPU memory. Despite this need, there lacks an efficient solution for offloading and transferring KV caches. We present LMCACHE, the first and so far the most efficient open-source KV caching solution, which extracts and stores KV caches generated by modern LLM engines (vLLM and SGLang) out of the GPU memory and shares them across engines and queries. LMCACHE supports both cache offloading (prefix reuse across queries) and prefill-decode (PD) disaggregation (cross-engine/GPU cache transfer). LMCACHE's high performance and wide adoption stem from the following contributions: (1) highly optimized KV cache data movement powered by batched data movement operations, compute and I/O pipelining; (2) a modular KV cache connector component, decoupling LMCACHE from the rapid evolution of inference engines; (3) a first-class control API for flexible cache orchestration across GPU, CPU, storage, and network layers. Our evaluation shows that combining LMCACHE with vLLM achieves up to 15x improvement in throughput across workloads such as multi-round question answering and document analysis. Large-scale adoption of LMCACHE in enterprise settings provides us valuable insights, for example, fetching KV cache from remote storage has unsurprisingly benefits to prefill delay, and that context truncation, which is a widely applied technique in industry, can greatly reduce prefix cache hit ratio by half. The source code of LMCACHE is at: https://github.com/LMCache/LMCache.

Mainstream visual encoders are pretrained on natural images and cannot be effectively applied to document images without document-oriented adaptation, as dense text and fine-grained character strokes demand character-level visual perception. We present MonkeyOCRv2, a visual-text pretrained model for document AI. First, we construct MonkeyDoc v2, to our knowledge the largest document-image pretraining corpus, comprising 113 million images spanning 17 languages. Second, we propose a pretraining strategy that jointly learns image-to-text generation and pixel-level document reconstruction: the former aligns visual representations with textual content, while the latter preserves character strokes and layout details. Extensive experiments are conducted on five representative document analysis tasks, including text recognition, formula recognition, text detection, document tampering detection, and overlapping text segmentation. Replacing the original encoders with MonkeyOCRv2 consistently improves performance across all five tasks. Finally, we validate its effectiveness as the vision encoder of multimodal large language models on the more challenging tasks of document parsing and document understanding. Kept frozen and paired with a lightweight language model, it yields a 0.7B document parsing model that sets a new open-source state-of-the-art on MDPBench, a recent benchmark spanning digital-born and photographed documents across 17 languages, surpassing the previous best 3B dots.mocr by 2.8% absolute with a vision encoder roughly 11times smaller. The frozen encoder also powers a document understanding model that outperforms counterparts built on CLIP, DINO, and SAM across eight benchmarks under identical training settings. These results suggest that document-oriented visual pretraining can serve as a foundation for document intelligence in its own right.

Coding agents are a fast-growing LLM application, executing as long-running closed-loop sessions in which LLM generations alternate with external tool calls. Yet, unlike chat workloads, their serving behavior has not been studied extensively. We address this gap by collecting a dataset of real-world coding assistant traces. Our analysis shows that coding agent sessions repeatedly reuse large prefixes and create sustained KVCache pressure that conventional LLM serving policies handle poorly. Based on our analysis, we present CacheWise, a KVCache management layer that improves KVCache reuse for coding agent workloads. CacheWise combines prefix-aware scheduling with reuse-aware eviction guided by lightweight predictions from tool call metadata. Implemented in vLLM and evaluated on the collected traces, CacheWise reduces KVCache evictions by up to 2-2.6x and improves total agent session completion time by up to 3.5x.

Long-output reasoning shifts the KV-cache bottleneck from the fixed prompt to the generated trace. Existing reasoning-cache eviction methods largely treat cached entries as routing objects, estimating whether an old key will still be read, will recur, or can be replaced. This routing-only view overlooks two effects: low-attention entries can carry large value payloads whose removal changes future predictions, and newly generated states can appear stale before later queries have had a chance to read them. We introduce AvoKV-E, a training-free eviction policy that first delays eligibility for recent states and then ranks eligible entries using candidate-normalized read pressure, key redundancy, and value-payload potential. According to empirical evaluation across different models and datasets, AvoKV-E matches or exceeds redundancy-aware, recurrence-based, and thought-adaptive eviction baselines at matched active-KV budgets, with its largest gains in the tightest-cache regime. Component and counterfactual analyses further connect these gains to delayed observation, payload-aware scoring, redundancy, and scale-robust normalization. Together, the results show that long-reasoning KV eviction should preserve not only keys that are likely to be read, but also the value payloads that sustain the reasoning trajectory.

Large Language Models (LLMs) have been widely deployed in a variety of applications, and the context length is rapidly increasing to handle tasks such as long-document QA and complex logical reasoning. However, long context poses significant challenges for inference efficiency, including high memory costs of key-value (KV) cache and increased latency due to extensive memory accesses. Recent works have proposed compressing KV cache to approximate computation, but these methods either evict tokens permanently, never recalling them for later inference, or recall previous tokens at the granularity of pages divided by textual positions. Both approaches degrade the model accuracy and output quality. To achieve efficient and accurate recallable KV cache compression, we introduce ClusterKV, which recalls tokens at the granularity of semantic clusters. We design and implement efficient algorithms and systems for clustering, selection, indexing and caching. Experiment results show that ClusterKV attains negligible accuracy loss across various tasks with 32k context lengths, using only a 1k to 2k KV cache budget, and achieves up to a 2times speedup in latency and a 2.5times improvement in decoding throughput. Compared to SoTA recallable KV compression methods, ClusterKV demonstrates higher model accuracy and output quality, while maintaining or exceeding inference efficiency.

Retrieval-Augmented Generation (RAG) has shown significant improvements in various natural language processing tasks by integrating the strengths of large language models (LLMs) and external knowledge databases. However, RAG introduces long sequence generation and leads to high computation and memory costs. We propose RAGCache, a novel multilevel dynamic caching system tailored for RAG. Our analysis benchmarks current RAG systems, pinpointing the performance bottleneck (i.e., long sequence due to knowledge injection) and optimization opportunities (i.e., caching knowledge's intermediate states). Based on these insights, we design RAGCache, which organizes the intermediate states of retrieved knowledge in a knowledge tree and caches them in the GPU and host memory hierarchy. RAGCache proposes a replacement policy that is aware of LLM inference characteristics and RAG retrieval patterns. It also dynamically overlaps the retrieval and inference steps to minimize the end-to-end latency. We implement RAGCache and evaluate it on vLLM, a state-of-the-art LLM inference system and Faiss, a state-of-the-art vector database. The experimental results show that RAGCache reduces the time to first token (TTFT) by up to 4x and improves the throughput by up to 2.1x compared to vLLM integrated with Faiss.

Incorporating external knowledge in large language models (LLMs) enhances their utility across diverse applications, but existing methods have trade-offs. Retrieval-Augmented Generation (RAG) fetches evidence via similarity search, but key information may fall outside top ranked results. Long-context models can process multiple documents but are computationally expensive and limited by context window size. Inspired by students condensing study material for open-book exams, we propose task-aware key-value (KV) cache compression, which compresses external knowledge in a zero- or few-shot setup. This enables LLMs to reason efficiently over a compacted representation of all relevant information. Experiments show our approach outperforms both RAG and task-agnostic compression methods. On LongBench v2, it improves accuracy by up to 7 absolute points over RAG with a 30x compression rate, while reducing inference latency from 0.43s to 0.16s. A synthetic dataset highlights that RAG performs well when sparse evidence suffices, whereas task-aware compression is superior for broad knowledge tasks.

Large language models (LLMs) rely on key-value cache (KV cache) to accelerate decoding by reducing redundant computations. However, the KV cache memory usage grows substantially with longer text sequences, posing challenges for efficient deployment. Existing KV cache eviction methods prune tokens using prefilling-stage attention scores, causing inconsistency with actual inference queries, especially under tight memory budgets. In this paper, we propose Lookahead Q-Cache (LAQ), a novel eviction framework that generates low-cost pseudo lookahead queries to better approximate the true decoding-stage queries. By using these lookahead queries as the observation window for importance estimation, LAQ achieves more consistent and accurate KV cache eviction aligned with real inference scenarios. Experimental results on LongBench and Needle-in-a-Haystack benchmarks show that LAQ outperforms existing methods across various budget levels, achieving a 1 sim 4 point improvement on LongBench under limited cache budget. Moreover, LAQ is complementary to existing approaches and can be flexibly combined to yield further improvements.

Retrieval-Augmented Generation (RAG) based on Large Language Models (LLMs) is a powerful solution to understand and query the industry's closed-source documents. However, basic RAG often struggles with complex QA tasks in legal and regulatory domains, particularly when dealing with numerous government documents. The top-k strategy frequently misses golden chunks, leading to incomplete or inaccurate answers. To address these retrieval bottlenecks, we explore two strategies to improve evidence coverage and answer quality. The first is a One-SHOT retrieval method that adaptively selects chunks based on a token budget, allowing as much relevant content as possible to be included within the model's context window. Additionally, we design modules to further filter and refine the chunks. The second is an iterative retrieval strategy built on a Reasoning Agentic RAG framework, where a reasoning LLM dynamically issues search queries, evaluates retrieved results, and progressively refines the context over multiple turns. We identify query drift and retrieval laziness issues and further design two modules to tackle them. Through extensive experiments on a dataset of government documents, we aim to offer practical insights and guidance for real-world applications in legal and regulatory domains.

Doc2Query -- the process of expanding the content of a document before indexing using a sequence-to-sequence model -- has emerged as a prominent technique for improving the first-stage retrieval effectiveness of search engines. However, sequence-to-sequence models are known to be prone to "hallucinating" content that is not present in the source text. We argue that Doc2Query is indeed prone to hallucination, which ultimately harms retrieval effectiveness and inflates the index size. In this work, we explore techniques for filtering out these harmful queries prior to indexing. We find that using a relevance model to remove poor-quality queries can improve the retrieval effectiveness of Doc2Query by up to 16%, while simultaneously reducing mean query execution time by 23% and cutting the index size by 33%. We release the code, data, and a live demonstration to facilitate reproduction and further exploration at https://github.com/terrierteam/pyterrier_doc2query.

Large Language Models (LLMs) like ChatGPT and Llama have revolutionized natural language processing and search engine dynamics. However, these models incur exceptionally high computational costs. For instance, GPT-3 consists of 175 billion parameters, where inference demands billions of floating-point operations. Caching is a natural solution to reduce LLM inference costs on repeated queries, which constitute about 31% of the total queries. However, existing caching methods are incapable of finding semantic similarities among LLM queries nor do they operate on contextual queries, leading to unacceptable false hit-and-miss rates. This paper introduces MeanCache, a user-centric semantic cache for LLM-based services that identifies semantically similar queries to determine cache hit or miss. Using MeanCache, the response to a user's semantically similar query can be retrieved from a local cache rather than re-querying the LLM, thus reducing costs, service provider load, and environmental impact. MeanCache leverages Federated Learning (FL) to collaboratively train a query similarity model without violating user privacy. By placing a local cache in each user's device and using FL, MeanCache reduces the latency and costs and enhances model performance, resulting in lower false hit rates. MeanCache also encodes context chains for every cached query, offering a simple yet highly effective mechanism to discern contextual query responses from standalone. Our experiments benchmarked against the state-of-the-art caching method, reveal that MeanCache attains an approximately 17% higher F-score and a 20% increase in precision during semantic cache hit-and-miss decisions while performing even better on contextual queries. It also reduces the storage requirement by 83% and accelerates semantic cache hit-and-miss decisions by 11%.

Large language models (LLMs) have excelled in various applications, yet serving them at scale is challenging due to their substantial resource demands and high latency. Our real-world studies reveal that over 70% of user requests to LLMs have semantically similar counterparts, suggesting the potential for knowledge transfer among requests. However, naively caching and reusing past responses leads to a big quality drop. In this paper, we introduce IC-Cache, a caching system that enables live LLM capability augmentation to improve serving efficiency: by leveraging historical request-response pairs from larger models as in-context examples, IC-Cache empowers small LLMs to imitate and even exceed the compositional abilities (e.g., reasoning) of their larger counterparts, enabling selective offloading of requests to reduce cost and latency. Achieving this live augmentation at scale introduces intricate trade-offs between response quality, latency, and system throughput. For a new request, IC-Cache efficiently selects similar, high-utility examples to prepend them to the new request's input. At scale, it adaptively routes requests across LLMs of varying capabilities, accounting for response quality and serving loads. IC-Cache employs a cost-aware cache replay mechanism that refines example quality offline to maximize online cache utility and efficiency. Evaluations on millions of realistic requests demonstrate that IC-Cache improves LLM serving throughput by 1.4-5.9x and reduces latency by 28-71% without hurting response quality.

This paper presents a novel Chunking-Free In-Context (CFIC) retrieval approach, specifically tailored for Retrieval-Augmented Generation (RAG) systems. Traditional RAG systems often struggle with grounding responses using precise evidence text due to the challenges of processing lengthy documents and filtering out irrelevant content. Commonly employed solutions, such as document chunking and adapting language models to handle longer contexts, have their limitations. These methods either disrupt the semantic coherence of the text or fail to effectively address the issues of noise and inaccuracy in evidence retrieval. CFIC addresses these challenges by circumventing the conventional chunking process. It utilizes the encoded hidden states of documents for in-context retrieval, employing auto-aggressive decoding to accurately identify the specific evidence text required for user queries, eliminating the need for chunking. CFIC is further enhanced by incorporating two decoding strategies, namely Constrained Sentence Prefix Decoding and Skip Decoding. These strategies not only improve the efficiency of the retrieval process but also ensure that the fidelity of the generated grounding text evidence is maintained. Our evaluations of CFIC on a range of open QA datasets demonstrate its superiority in retrieving relevant and accurate evidence, offering a significant improvement over traditional methods. By doing away with the need for document chunking, CFIC presents a more streamlined, effective, and efficient retrieval solution, making it a valuable advancement in the field of RAG systems.

Transformer-based large language models (LLMs) have already achieved remarkable results on long-text tasks, but the limited GPU memory (VRAM) resources struggle to accommodate the linearly growing demand for key-value (KV) cache as the sequence length increases, which has become a bottleneck for the application of LLMs on long sequences. Existing KV cache compression methods include eviction, merging, or quantization of the KV cache to reduce its size. However, compression results in irreversible information forgetting, potentially affecting the accuracy of subsequent decoding. In this paper, we propose SpeCache, which takes full advantage of the large and easily expandable CPU memory to offload the complete KV cache, and dynamically fetches KV pairs back in each decoding step based on their importance measured by low-bit KV cache copy in VRAM. To avoid inference latency caused by CPU-GPU communication, SpeCache speculatively predicts the KV pairs that the next token might attend to, allowing us to prefetch them before the next decoding step which enables parallelization of prefetching and computation. Experiments on LongBench and Needle-in-a-Haystack benchmarks verify that SpeCache effectively reduces VRAM usage while avoiding information forgetting for long sequences without re-training, even with a 10x high KV cache compression ratio.

The linear memory growth of the KV cache poses a significant bottleneck for LLM inference in long-context tasks. Existing static compression methods often fail to preserve globally important information. Although recent dynamic retrieval approaches attempt to address this issue, they typically suffer from coarse-grained caching strategies and incur high I/O overhead. To overcome these limitations, we propose HeteroCache, a training-free dynamic compression framework. Our method is built on two key insights: attention heads exhibit diverse temporal heterogeneity, and there is significant spatial redundancy among heads within the same layer. Guided by these insights, HeteroCache categorizes heads based on stability and similarity, applying a fine-grained weighting strategy that allocates larger cache budgets to heads with rapidly shifting attention to capture context changes. Furthermore, it features a hierarchical storage mechanism where representative heads monitor attention drift to trigger asynchronous, on-demand context retrieval, thereby hiding I/O latency. Experiments demonstrate that HeteroCache achieves state-of-the-art performance on long-context benchmarks and accelerates decoding by up to 3times compared to the original model with a 224K context. Our code is available at https://github.com/ponytaill/HeteroCache.

LLM agents increasingly answer questions over dynamic raw-document collections, where files may change before preprocessing, and relevant evidence (spans, sections, pages, or tables) is query-dependent. Existing retrieval-augmented approaches pre-materialize evidence via fixed chunking, embeddings, or persistent indexes: effective for lookup, yet costly, stale-prone, and committed to a granularity before the query is known. We formulate in-context search as Budgeted Evidence Localization over a latent evidence space induced by dynamic raw documents and propose LENS (Latent Evidence Exploration and Search), an index-free framework. Instead of pre-materializing the evidence space, LENS maintains a query-conditioned belief over candidate units, iteratively selecting candidates via complementary lexical, local, and exploratory proposal policies, updating the belief via an LLM relevance oracle, and narrowing toward high-posterior regions under a controllable budget. Evidence is consolidated into compact, source-grounded regions of interest and compressed into self-organizing knowledge clusters reused across related queries. On a controlled 500-question evaluation with matched corpus snapshots, LENS reaches 62.4% exact match and 84.8% evidence recall vs. 65.2% exact match but 50.4% evidence recall for a ReAct-style baseline. Across scales, LENS gives the strongest supporting-fact localization and answer grounding. On a fixed 150-question fullwiki subset over the raw Wikipedia dump with zero indexing, LENS and ReAct are nearly tied in official answer quality (43.3% vs. 42.7% EM), with LENS grounding more answers in retrieved evidence (84.0% vs. 70.7%). A no-retrieval Closed-Book reference highlights the contribution of model memory. LENS is query-ready after corpus changes, needs no preprocessing or persistent index, and preserves source-grounded evidence localization throughout.

Transformer depth is not used uniformly: lower and middle layers build semantic representations, while upper layers increasingly specialize them for prediction. We turn this division of labor into CoMem (Comprehension Memory), which writes each context chunk only through an intermediate layer, retrieves a fixed number of cached residual states, and recomputes the query-conditioned upper layers over the resulting pack. For a fixed retrieval budget, model-side read compute and memory are independent of stored-context length. We evaluate a continued-trained Qwen3-8B base LM under a unified chat-template-free protocol. The backbone is frozen; the flagship trains only a rank-32 self-distillation LoRA on plain PG19, and we report an adapter-free arm separately. CoMem reaches 97.05 on RULER and 38.27 on LoCoMo versus 34.59 for full-context KV-Direct; the dialogue-memory advantage survives conversation-cluster resampling and an independent judge. Results on additional long-context and long-document tasks expose both the benefits of bounded retrieval and its in-window compression tax. Controlled depth sweeps show that deeper caching lowers per-query recomputation but incurs a fidelity loss that self-distillation substantially repairs. In a separate adapter-free efficiency control on an NVIDIA H20 at 128k, CoMem uses 18.26 GB rather than 89.36 GB and achieves a 7.83x prefill speedup. These results show that long-context memory can be organized along the layer axis, not only the token axis.

Large language models rely on KV caching to reduce repeated prefill computation in long context and interactive applications. As lightweight adapters evolve, cached states reflect earlier versions, so stale reuse distorts current model outputs, while complete affected suffix recomputation restores fidelity at substantial cost. We seek minimal recomputation that recovers current adapter behavior. Existing systems track token, context, or stable adapter identity, but neither represent caches from earlier adapter versions nor distinguish update propagation from the recomputation required for behavioral recovery. To address these gaps, we introduce CacheReforge, which represents stale KV caches as layerwise mixed-version objects. It combines per-layer adapter anchors, calibrated sensitivity, accumulated drift, and executable restart boundaries to select direct reuse, bounded recomputation, or complete affected-suffix recovery. We distinguish dependency depth from the functional recomputation horizon and use cumulative tail influence to characterize when bounded recovery preserves current-model behavior. We evaluate CacheReforge on Qwen2.5-1.5B and Qwen2.5-7B with continual LoRA updates, including 16K HotpotQA and 2WikiMQA workloads. CacheReforge reduces mean KL divergence by 92.4% relative to stale reuse, while recomputing only 5.44% of layers and reducing cache-maintenance time by 93.2% relative to fresh full prefill. These results show that version-aware recovery preserves model fidelity and most KV caching gains.

Long-running LLM agents are stateful: beyond the transcript they accrete compressed summaries, plaintext memory, pending tool plans, and, under every serving API, a KV cache. Yet today's "forget" operations delete a plaintext memory record and stop, leaving every artifact derived from the revoked information intact. We formalize execution-state unlearning: after a forget request, the agent must behave as if it had never observed the target. Modeling the runtime as a deterministic transition system, we prove that the pre-target trajectory prefix is shared with this counterfactual world for free, that the post-target suffix is irreducibly tainted without token-level attribution, and that exact unlearning requires at least T-τ+1 recomputed transitions, where τ is the target's injection step. Provenance-Guided Selective Replay attains this bound as a cross-layer contract spanning prompt, compressed memory, and cache: a provenance graph locates the injection point, checkpoint restoration reduces to cropping the KV cache, and sanitized replay regenerates the counterfactual suffix. Audited with elicitation, stochastic, and string-free behavioral tests across three agent suites, nine baselines, and three model families, memory deletion leaves leakage unchanged, instruction-based forgetting collapses under elicitation (Leak@probes = 1.00), and source redaction still acts on a revoked preference in 80% of episodes, while selective replay is indistinguishable from a full reset at up to 9x fewer recomputed tokens.

Serving large language models at long context is bottlenecked by the key-value (KV) cache, which is read in full at every decode step. Attention keys are locally low-rank though globally high-rank: a fixed low-rank sketch shared across pages is provably blind to page-specific directions, while at the same summary size a page's own basis ranks pages and keeps carriers far better. LOCKS gives every page its own rank-r spectral summary (resident, a tenth of the cache at r{=}8 and a twenty-fifth at r{=}2), reconstructs within-page logits, estimates each page's attention mass by log-sum-exp, and attends only the top pages; selection itself reads no candidate keys or values. Selecting on this summary alone stays within about a point of the full cache on long-document QA (LongBench-v1; Llama-3.1-8B), tracks the read-every-key exact-LSE oracle on retrieval-dense RULER down to the smallest budgets, and holds quality furthest into the small-budget regime on long-form reasoning (AIME26, MATH-500; Qwen3-4B), where selectors and eviction-based reasoning compressors both fall away. At a 2048-token budget LOCKS matches FullKV aggregate quality at 100K+ context (GLM-4-9B-Chat-1M) while attending about 2% of the tokens; since the summary is scanned in full each step, the per-step KV read falls by about 10--25times across that rank range, and this halves per-token decode latency (2.0times at 1M tokens on one H200 NVL, measured at r{=}8) against dense attention. LOCKS ships as a drop-in plugin for unmodified vLLM, with batched decode running in full CUDA graphs.

Hybrid LLMs interleave full-attention layers with linear-attention layers to reduce long-context inference cost, but this structure complicates prefix caching. Full-attention KV caches are token-addressable, whereas linear-attention layers maintain recurrent states that cannot be rolled back to arbitrary prefix boundaries. Existing systems materialize recurrent-state checkpoints, restricting prefix reuse to checkpoint-aligned positions. We present SuffixReplay, the first prefix caching system that lets hybrid LLMs reuse cached prefixes at every cache-supported page boundary without materializing recurrent-state checkpoints. Our key insight is to just let linear states forget the distant past. Modern linear-attention mechanisms use recurrent decay and gating to attenuate the influence of old inputs. Therefore, instead of checkpointing every prefix boundary, SuffixReplay approximates the state at a matched boundary by replaying only a recent suffix of the layer's input hidden states, which we retain as anchors. At the algorithmic level, SuffixReplay combines layer-wise and token-wise anchor sparsity with a bounded replay budget to control storage, computation, and quality. At the system level, it uses an independently managed anchor sidecar and a pipelined replay path to overlap anchor movement and state reconstruction with the native serving pipeline. We evaluate SuffixReplay on three hybrid LLMs: OLMo-Hybrid-7B, Qwen3.5-4B, and Qwen3.6-27B-FP8. Across these models, SuffixReplay retains 91.4-100% of full-prefill quality on average across LongBench and RULER, while using only 0.36-0.51x the amortized per-token storage of SGLang's default 8192-token checkpoint cache. Integrated into SGLang, SuffixReplay reduces median TTFT by 15-70% on branching workloads, sustains 2.3-4.3x SGLang's throughput when the working set exceeds HBM, and matches SGLang on high-hit continuation traffic.

Reality may establish that an outcome occurred without identifying which evolving procedure produced it or why. This distinction matters in production ML systems whose code, configuration, and artifacts change while external feedback accumulates. We examine it in a human-directed, agent-engineered quantitative trading system, using oracle to mean an external source of realized outcomes rather than a complete correctness specification. Across one year, the account gained and outperformed a broad market index, while annual alpha was not statistically distinguishable from zero under the main retrospective specification. Retrospectively selected subperiods include adverse relative performance and conditional candidate-level weakness under declared approximate references. Engineering records document changes during the episode, and complete recommendation-to-runtime binding is unavailable. The archive does not establish a common frozen instance or a unique cause. The case motivates an outcome--diagnosis gap: outcome evidence, evaluated-object identity, and causal explanation support distinct claims. We distinguish frozen instances, pre-specified adaptive procedures, and ad-hoc development; organize archive-relative claim identifiability and an evidence hierarchy; and propose a prospective production-binding protocol. An illustrative compatible-history example shows how factual binding can resolve a recommendation's referent without supplying its counterfactual effect. The protocol is proposed rather than prospectively validated. External feedback constrains outcome claims, while provenance and additional identification structure determine the resolution of diagnosis.

The large size of the KV cache has become a major bottleneck for serving LLMs with increasing context lengths. In response, many KV cache compression methods, such as token dropping and quantization, have been proposed. However, almost all of these methods are inherently lossy-despite minimal accuracy degradation for short outputs, their outputs increasingly diverge from full-KV-cache outputs as more tokens are decoded, which leads to catastrophic failures in code generation and tool calling. We present VeriCache, the first inference framework that ensures the same output as full-KV-cache decoding but largely preserves the high decoding throughput of a range of KV cache compression algorithms. VeriCache uses the compressed KV cache to draft tokens, then verifies them against the full KV cache. While it may seem like just speculative decoding, VeriCache requires addressing a key system challenge to work-keeping the full KV cache out of GPU memory and minimizing the overhead of swapping it in for verification. The insight is two-fold: (1) compressed-KV decoding can be parallelized with full-KV swap, because one is HBM-bandwidth-bound and the other is PCIe/network-bound, and (2) the compressed KV cache often produces output similar to the full KV cache, allowing a long drafting horizon to amortize each full-KV swap. VeriCache applies to both long-context decoding and remote prefix caching, supports a broad family of token-dropping and quantization methods through a uniform compressor interface, and composes with traditional speculative decoding. Experimental results show that VeriCache achieves up to 4X higher throughput than full-KV inference while producing identical outputs.

In recent years, Large Language Models (LLMs) have demonstrated remarkable capabilities across a wide array of text-centric tasks. However, their `large' scale introduces significant computational and storage challenges, particularly in managing the key-value states of the transformer, which limits their wider applicability. Therefore, we propose to adaptively release resources from caches and rebuild the necessary key-value states. Particularly, we accomplish this by a lightweight controller module to approximate an ideal top-K sparse attention. This module retains the tokens with the highest top-K attention weights and simultaneously rebuilds the discarded but necessary tokens, which may become essential for future decoding. Comprehensive experiments in natural language generation and modeling reveal that our method is not only competitive with full attention in terms of performance but also achieves a significant throughput improvement of up to 221.8%. The code for replication is available on the https://github.com/WHUIR/ADORE.

Large language model (LLM) agents accumulate heterogeneous context, including system instructions, plans, user turns, retrieved documents, tool outputs, and intermediate reasoning, whose key-value (KV) cache can become a major memory bottleneck. Existing eviction policies generally apply the same attention- or recency-based rule to every token, ignoring semantic structure already available to the agent orchestrator. We introduce MemDecay, a training-free, region-aware KV-cache eviction policy. MemDecay assigns tokens region-specific base priorities and decay rates, refreshes retention scores when tokens receive attention, and evicts the lowest-scoring pages under a fixed cache budget while allowing critical regions to be pinned. We also provide a procedure for calibrating decay rates from measured attention lifetimes. We evaluate MemDecay at approximately 450 and 1,700 token contexts using Qwen2.5-1.5B and 3B. Across all settings, attention lifetimes differ by an order of magnitude across regions: system-token half-lives range from 148 to 189 decoding steps, compared with 14 to 16 for scratchpad tokens. Pinning preserves system-region facts at full-cache accuracy in every setting, while no baseline preserves more than 13 of 24. Region-aware retention remains effective as context grows, whereas recency-based retention collapses. Accumulated-attention retention performs better on unpinned content, however, and ablations identify attention-score normalization as the main limitation of the current formulation. These results establish semantic prompt structure as a robust signal for KV-cache management while clarifying how it should be combined with attention-based importance.

LLM agents that cache recovery suggestions from API errors can skip re-derivation in later episodes, spending fewer tokens and fewer model calls on constraints they have already learned. Server-side data drift turns those cached fixes into silent failures, and the usual remedy, re-deriving on every episode, gives the savings back. We introduce invalidation contracts, a protocol layer that attaches version stamps and cacheability hints to every recovery suggestion so the client can evict stale entries without trial and error, and keep the rest. The contract decomposes realized savings into two independent factors: validity, the fraction of cached suggestions that remain correct after a drift event, and compliance, the fraction the planner applies on the first attempt. Validity depends only on the protocol and is vendor-independent. Compliance depends on the planner model: identical wire bytes yield 100% first-try compliance on Claude Haiku 4.5 and 11% or below on Claude Sonnet 5, which exhibits input-schema conservatism, refusing fixes that add fields the original request did not contain. We evaluate across seven models, three serving paths, two domains, and approximately 9,400 episodes. Row-level invalidation raises compliance by 0 to 66.7 percentage points across the seven models, 55.6 to 66.7 on three, and recovers 29-33% of baseline token cost on four of seven models, while table-level invalidation destroys co-located entries and drops post-drift first-try rates to 0% on five of seven. Eviction precision is 1.00 at row granularity on every model under the row-level oracle of Section 4.1. The contract adds 15% to response payload. Version-stamp validity is deterministic by construction and produced identical results across every model and serving path, with zero contract failures in the entire evaluation.

A personal language agent that acts for its owner across private and shared conversations can learn a fact from one audience and later place it in the context it assembles for another. We study authorization before context across the whole memory lifecycle. Each memory item carries the audience present when it was recorded; derived items are partitioned by audience, receive the intersection of their sources' audiences, or are suppressed; an audience widens only by an explicit, object-specific grant; and an item enters a model attempt only when every current viewer belongs to one of its authorized audiences, with unresolved viewers failing closed to public-only. Under explicit identity, provenance and complete-mediation assumptions, this admission is sound and policy-complete on the exact assembled context, enforced by exclusion rather than by model behavior. We realize it in two independently persisted reference architectures, a flat store and a relationship graph, and, descriptively, in a native agent-memory runtime. In a prospectively frozen confirmation over 10,000 multi-party histories, no forbidden item entered any architecture's context, whereas unscoped retrieval exposed forbidden items in 82% of its contexts. Entitled recall matched policy-equivalent baselines exactly and exceeded unscoped retrieval by 0.30 Recall@5, with a Holm-confirmed advantage that grows with distractors. No architecture produced a wrong-principal substitution, but unscoped substitutions were too rare to establish the prespecified joint decision.

Methods for improving knowledge use in large language models typically fall into two regimes. Non-parametric retrieval offers flexible access to external knowledge, but adds retrieval latency, context overhead, and only shallow integration with the backbone. Parametric adaptation is efficient at inference time, but entangles knowledge with model weights and can be hard to update, audit, or transfer. Engram-style hashed memory occupies a middle regime: it stores learned information in an external, addressable table, yet consumes that table through a small learned reader. This raises a basic question: when such a memory is moved across backbones, what matters more, the frozen memory itself or the target-side reader? We study this question through cross-model frozen-memory extraction, in which a memory trained on a source model is frozen and attached to a different target model, with only a lightweight reader trained. Ablations show that learned memory content and correct addressing both matter, but the transferred table becomes useful only through a reader aligned to the target model. In downstream question answering tasks, a dual-layer, four-branch reader nearly closes the gap between same-model and cross-model reuse, achieving an average score of 38.8 under our controlled evaluation protocol. Moreover, when the provider reader is directly compatible with the target interface, the frozen artifact can provide substantial utility without target-side training, while optional reader adaptation yields further improvement. These results suggest that Engram can serve as a reusable external knowledge artifact, provided that the target has access to a compatible reader interface; target-side adaptation can further improve alignment when direct reader reuse is insufficient.

Post-hoc context erasing over the KV cache is challenging because a local edit has a global consequence: once a span has been processed, its influence propagates into the cached states of all subsequent tokens. This issue arises naturally in long-context LLM applications, where stale retrieved facts, incorrect tool observations, retracted user preferences, or harmful prompt injections may be identified only after prefill. Exact erasing must then recompute all tokens after the deleted span, making its computational cost depend on suffix length rather than erased-span length. We introduce KVEraser, a learned KV-cache editing method for efficient localized context erasing. Given a processed context and a span to remove, KVEraser replaces only the KV states of the erased interval with learned steering states while reusing the remaining cache unchanged. To learn a transferable erasing mechanism, we build a two-stage training pipeline: generic span-neighbor pre-training teaches the eraser to suppress the influence of the erased span, while task-specific fine-tuning adapts this capability to downstream scenarios. Experiments show that KVEraser nearly matches full recomputation in post-erasure performance on in-domain tasks across 1K--32K context lengths, while its latency increases by only 24% compared with a 17.6x increase for full recomputation. KVEraser also generalizes to unseen long-document QA tasks with harmful factual distractors, achieving the best performance among approximate baselines with a 3--4x speedup over full recomputation.

Current Retrieval-Augmented Generation (RAG) systems concatenate and process numerous retrieved document chunks for prefill which requires a large volume of computation, therefore leading to significant latency in time-to-first-token (TTFT). To reduce the computation overhead as well as TTFT, we introduce TurboRAG, a novel RAG system that redesigns the inference paradigm of the current RAG system by first pre-computing and storing the key-value (KV) caches of documents offline, and then directly retrieving the saved KV cache for prefill. Hence, online computation of KV caches is eliminated during inference. In addition, we provide a number of insights into the mask matrix and positional embedding mechanisms, plus fine-tune a pretrained language model to maintain model accuracy of TurboRAG. Our approach is applicable to most existing large language models and their applications without any requirement in modification of models and inference systems. Experimental results across a suite of RAG benchmarks demonstrate that TurboRAG reduces TTFT by up to 9.4x compared to the conventional RAG systems (on an average of 8.6x), but reserving comparable performance to the standard RAG systems.

Retrieval-augmented generation (RAG) has gained traction as a powerful approach for enhancing language models by integrating external knowledge sources. However, RAG introduces challenges such as retrieval latency, potential errors in document selection, and increased system complexity. With the advent of large language models (LLMs) featuring significantly extended context windows, this paper proposes an alternative paradigm, cache-augmented generation (CAG) that bypasses real-time retrieval. Our method involves preloading all relevant resources, especially when the documents or knowledge for retrieval are of a limited and manageable size, into the LLM's extended context and caching its runtime parameters. During inference, the model utilizes these preloaded parameters to answer queries without additional retrieval steps. Comparative analyses reveal that CAG eliminates retrieval latency and minimizes retrieval errors while maintaining context relevance. Performance evaluations across multiple benchmarks highlight scenarios where long-context LLMs either outperform or complement traditional RAG pipelines. These findings suggest that, for certain applications, particularly those with a constrained knowledge base, CAG provide a streamlined and efficient alternative to RAG, achieving comparable or superior results with reduced complexity.

When re-finding items, users who forget or are uncertain about identifying details often rely on creative strategies for expressing their information needs -- complex queries that describe content elements (e.g., book characters or events), information beyond the document text (e.g., descriptions of book covers), or personal context (e.g., when they read a book). This retrieval setting, called tip of the tongue (TOT), is especially challenging for models heavily reliant on lexical and semantic overlap between query and document text. In this work, we introduce a simple yet effective framework for handling such complex queries by decomposing the query into individual clues, routing those as sub-queries to specialized retrievers, and ensembling the results. This approach allows us to take advantage of off-the-shelf retrievers (e.g., CLIP for retrieving images of book covers) or incorporate retriever-specific logic (e.g., date constraints). We show that our framework incorportating query decompositions into retrievers can improve gold book recall up to 7% relative again for Recall@5 on a new collection of 14,441 real-world query-book pairs from an online community for resolving TOT inquiries.

Large language models increasingly operate over long contexts, where the KV cache becomes a dominant memory bottleneck: its size grows linearly with sequence length and must be retained throughout decoding, making full GPU caching prohibitively expensive without compression. Existing KV cache compression methods struggle to balance efficiency with faithful context preservation. Token eviction discards information, while semantic grouping fixes compression decisions at prefill time; neither can recover token-level detail from a compressed span once it becomes relevant during generation. As a solution, we propose SeKV, a resolution-adaptive semantic KV cache that organizes context into entropy-guided semantic spans and stores them across a GPU-CPU memory hierarchy without discarding information. Each span keeps a lightweight summary vector on GPU for coarse routing and a low-rank SVD basis on CPU for on-demand token-level reconstruction. A trained zoom-in mechanism selectively expands query-relevant spans during decoding, enabling precise retrieval without materializing the full KV cache on GPU. SeKV enables adaptive token-level reconstruction while keeping the base LLM fully frozen and adding fewer than 0.05% trainable parameters. Across four benchmarks, SeKV improves over the strongest semantic compression baseline by 5.9% on average while reducing GPU memory by 53.3% versus full KV caching at 128K context. Code is available on https://github.com/AmirAbaskohi/SeKV.

Retrieval-augmented generation (RAG) is increasingly deployed in enterprise search and document-centric assistants, where responses must be grounded in long and complex source materials. In practice, verifying that generated answers faithfully reflect retrieved documents is difficult: large language models can check long contexts but are too slow and costly for interactive services, while lightweight classifiers operate within strict context limits and frequently miss evidence outside truncated passages. We present the design of a real-time verification component integrated into a production RAG pipeline that enables full-document grounding under latency constraints. The system processes documents up to 32K tokens and employs adaptive inference strategies to balance response time and verification coverage across workloads. We describe the architectural decisions, operational trade-offs, and evaluation methodology used to deploy the verifier, and show that full-context verification substantially improves detection of unsupported responses compared with truncated validation. Our experience highlights when long-context verification is necessary, why chunk-based checking often fails in real documents, and how latency budgets shape model design. These findings provide practical guidance for practitioners building reliable large-scale retrieval-augmented applications. (Model, benchmark, and code: https://huggingface.co/llm-semantic-router)

Paging is a prototypical problem in the area of online algorithms. It has also played a central role in the development of learning-augmented algorithms -- a recent line of research that aims to ameliorate the shortcomings of classical worst-case analysis by giving algorithms access to predictions. Such predictions can typically be generated using a machine learning approach, but they are inherently imperfect. Previous work on learning-augmented paging has investigated predictions on (i) when the current page will be requested again (reoccurrence predictions), (ii) the current state of the cache in an optimal algorithm (state predictions), (iii) all requests until the current page gets requested again, and (iv) the relative order in which pages are requested. We study learning-augmented paging from the new perspective of requiring the least possible amount of predicted information. More specifically, the predictions obtained alongside each page request are limited to one bit only. We consider two natural such setups: (i) discard predictions, in which the predicted bit denotes whether or not it is ``safe'' to evict this page, and (ii) phase predictions, where the bit denotes whether the current page will be requested in the next phase (for an appropriate partitioning of the input into phases). We develop algorithms for each of the two setups that satisfy all three desirable properties of learning-augmented algorithms -- that is, they are consistent, robust and smooth -- despite being limited to a one-bit prediction per request. We also present lower bounds establishing that our algorithms are essentially best possible.

The overwhelming volume of data generated and indexed by search engines poses a significant challenge in retrieving documents from the index efficiently and effectively. Even with a well-crafted query, several relevant documents often get buried among a multitude of competing documents, resulting in reduced accessibility or `findability' of the desired document. Consequently, it is crucial to develop a robust methodology for assessing this dimension of Information Retrieval (IR) system performance. While previous studies have focused on measuring document accessibility disregarding user queries and document relevance, there exists no metric to quantify the findability of a document within a given IR system without resorting to manual labor. This paper aims to address this gap by defining and deriving a metric to evaluate the findability of documents as perceived by end-users. Through experiments, we demonstrate the varying impact of different retrieval models and collections on the findability of documents. Furthermore, we establish the findability measure as an independent metric distinct from retrievability, an accessibility measure introduced in prior literature.

Visual Document Retrieval (VDR), the task of retrieving visually-rich document pages using queries that combine visual and textual cues, is crucial for numerous real-world applications. Recent state-of-the-art methods leverage Large Vision-Language Models (LVLMs) in a multi-vector paradigm, representing each document as patch-level embeddings to capture fine-grained details. While highly effective, this approach introduces a critical challenge: prohibitive storage overhead, as storing hundreds of vectors per page makes large-scale deployment costly and impractical. To address this, we introduce DocPruner, the first framework to employ adaptive patch-level embedding pruning for VDR to effectively reduce the storage overhead. DocPruner leverages the intra-document patch attention distribution to dynamically identify and discard redundant embeddings for each document. This adaptive mechanism enables a significant 50-60% reduction in storage for leading multi-vector VDR models with negligible degradation in document retrieval performance. Extensive experiments across more than ten representative datasets validate that DocPruner offers a robust, flexible, and effective solution for building storage-efficient, large-scale VDR systems.

Query-agnostic KV cache eviction compresses a context once and reuses the resulting cache for arbitrary future queries, but performance can collapse under tight budgets. Existing methods primarily improve which original KV pairs are retained. We introduce RestoreKV, which complements this selection-based formulation with learned restoration under the same total KV budget. Our key insight is that, although the information lost through eviction is context-specific, the mechanism for generating its compact complement can be shared across contexts. After context prefill, a few restore tokens attend to the full KV cache in a single LoRA-adapted pass, generating a compact, context-conditioned restore cache. The base importance scorer and eviction rule remain unchanged, and the adapters are disabled for all subsequent queries and decoding. RestoreKV is trained through parameter-efficient self-distillation from the frozen full-cache model, optimizing only 0.4% of the parameters and requiring no task-specific tuning. Across four backbones and four long-context benchmarks, RestoreKV substantially reduces compression-induced degradation. On Qwen3-4B, it improves 59 of 60 paired, budget-matched settings across five base eviction methods; at a 5% budget, it raises KVzip from 38.2 to 73.2 on RULER-4K. Applied to KVzip+, RestoreKV reaches 86.4 RULER accuracy at 16times compression on the KVPress Benchmark, while adding less than 0.5% one-time cache-construction overhead in a 32K-context evaluation. Our project page is available at https://paper.pnu-cvsp.com/RestoreKV/

The evaluation of Retrieval-Augmented Generation (RAG) systems typically examines retrieval quality and generation parameters like temperature in isolation, overlooking their interaction. This work presents a systematic investigation of how text perturbations (simulating noisy retrieval) interact with temperature settings across multiple LLM runs. We propose a comprehensive RAG Perturbation-Temperature Analysis Framework that subjects retrieved documents to three distinct perturbation types across varying temperature settings. Through extensive experiments on HotpotQA with both open-source and proprietary LLMs, we demonstrate that performance degradation follows distinct patterns: high-temperature settings consistently amplify vulnerability to perturbations, while certain perturbation types exhibit non-linear sensitivity across the temperature range. Our work yields three key contributions: (1) a diagnostic benchmark for assessing RAG robustness, (2) an analytical framework for quantifying perturbation-temperature interactions, and (3) practical guidelines for model selection and parameter tuning under noisy retrieval conditions.

As LLM capabilities advance, agents are tackling increasingly complex tasks over longer horizons. Their growing interaction histories make memory compaction essential for staying within context windows and reducing prefill cost. Existing methods summarize the history or compress its KV cache, often adding model computation to preserve information for future requests. A new user request can change which history matters, but reassessing that history with the model requires re-encoding it if the KV cache has expired. Past attention provides signals of historical importance and dependencies between messages, while relevance to the current task must be assessed using the new user request. We introduce ReCAP, a memory compaction method that stores attention-derived importance scores and dependency links in a lightweight, persistent context graph. For each new request, ReCAP combines stored importance with relevance cues from the request and follows dependency links to select messages and their supporting context, without additional model calls for selection. Compared with Codex's default summarization-based compaction, ReCAP reduces estimated latency for compaction and cold restoration by approximately 95% on both Qwen3-Coder and gpt-oss. It also roughly halves the historical context per call on SWE-Together at comparable task quality and improves accuracy on the code tasks of Lost-in-Conversation over full history by 19.8 and 41.2 points.

Intelligent document processing (IDP) encompasses a broad range of tasks, including optical character recognition (OCR), document question answering (DocQA), and key information extraction (KIE). Despite their distinct objectives, these tasks share a common need to perceive document content, acquire task-relevant information, and progressively refine intermediate results. However, they are typically formulated as separate prediction problems and addressed by task-specific models or processing pipelines. We introduce DocClaw, a unified agentic system that formulates diverse intelligent document processing tasks as a shared process of interaction between an agent and a document. Given a document and a task-specific query, DocClaw follows an appropriate document skill to iteratively identify the information required, invoke relevant tools, and integrate the resulting observations into the desired output. Throughout this process, a structured document state organizes reusable document knowledge and task-specific interaction context, allowing the agent to accumulate, revisit, and progressively refine information as the interaction proceeds. Under this formulation, task-specific requirements are captured by the agent's interpretation of the query objective and the corresponding document skill, while the underlying interaction loop, tool space, and document state are shared across tasks. Extensive experiments across multiple intelligent document processing benchmarks demonstrate that DocClaw effectively handles diverse tasks within a single agentic framework and achieves competitive performance compared with both general-purpose VLMs and task-specific methods.

Recent advances in large language models (LLMs) have significantly boosted long-context processing. However, the increasing key-value (KV) cache size poses critical challenges to memory and execution efficiency. Most KV cache compression methods rely on heuristic token eviction using all attention heads in Grouped Query Attention (GQA)-based LLMs. This method ignores the different functionalities of attention heads, leading to the eviction of critical tokens and thus degrades the performance of LLMs. To address the issue above, instead of using all the attention heads in GQA-based LLMs to determine important tokens as in the previous work, we first identify the attention heads in each layer that are not only capable of retrieving the initial and final tokens of a prompt, but also capable of retrieving important tokens within the text and attending to their surrounding semantic context. Afterwards, we exploit such heads to determine the important tokens and retain their corresponding KV cache pairs. Furthermore, we analyze the cache eviction error of each layer individually and introduce a layer-adaptive KV cache allocation strategy. Experimental results demonstrate the proposed CompressKV consistently outperforms state-of-the-art approaches under various memory budgets on LongBench and Needle-in-a-Haystack benchmarks. Our code is publicly available at: https://github.com/TUDa-HWAI/CompressKV.git.

Large Language Models (LLMs), such as OpenAI-o1 and DeepSeek-R1, have demonstrated strong reasoning capabilities. To further enhance LLM capabilities, recent agentic systems, such as Deep Research, incorporate web interactions into LLM reasoning to mitigate uncertainties and reduce potential errors. However, existing research predominantly focuses on reasoning performance, often neglecting the efficiency of agentic systems. In this work, we present a comprehensive empirical study that identifies efficiency bottlenecks in web-interactive agentic systems. We decompose end-to-end latency into two primary components: LLM API latency and web environment latency. We conduct a comprehensive empirical study across 15 models and 5 providers to demonstrate high variability in API-based agentic systems. We observe that web environment latency can contribute as much as 53.7% to the overall latency in a web-based agentic system. To improve latency, we propose SpecCache, a caching framework augmented with speculative execution that can reduce web environment overhead. Extensive evaluations on two standard benchmarks show that our approach improves the cache hit rate by up to 58x compared to a random caching strategy, while reducing web environment overhead by up to 3.2x, without degrading agentic system performance.

The personal archive of Konstantin Tsiolkovsky (1857-1935) is held as fond 555 of the Archive of the Russian Academy of Sciences. The archive scanned the fond and published the images, but with no queryable catalogue, no full-text search and no dataset: the holdings can only be browsed one page at a time. This paper describes a machine-readable catalogue of all 2,019 files and 51,008 scans, a dating for 1,969 files taken from the archive's own descriptions, a page-level classification of every scan into handwriting and typescript, and a growing corpus of machine transcriptions (currently 322 files, 5,454 scans). It also reports a way to measure handwritten-text-recognition accuracy in an archive with no ground truth. Archives of the typewriter era often preserve one text twice, as manuscript and as a typed copy; transcribing both and comparing isolates the reading error, since source and pipeline are identical and only page difficulty differs. Across 294 such pairs from 27 files, two readings of a handwritten page agree on a median 37% of words. On two files that also have a published edition the estimate can be checked against ground truth: it is unbiased to within a percentage point and ranks pages as the truth does (rank correlation 0.92 where the edition is a faithful witness). This bounds use: two variants of one work here share 19% of words, below the rate at which two readings of a single page agree, so the redactions cannot be collated word by word at this quality. That negative result is reported as such, and the constraint is built into the tool.

Emerging LLM services increasingly host many sparse MoE models, yet most models receive sparse requests and remain cold. This creates a GPU memory problem: model weights are stable and model-determined, while KV-cache is transient and demand-determined. Because cold models rarely reach peak KV-cache demand at the same time, reserving worst-case KV capacity per model wastes memory; a shared KV-cache pool can instead provision aggregate active demand. However, KV-cache sharing is not sufficient when weights and KV-cache remain in a monolithic GPU memory pool. Static weights compete with dynamic KV-cache, and KV-head-limited attention under cold, low-concurrency traffic exposes only a fraction of replicated KV capacity, leading to low GPU memory utilization and weak long-context support. We present CrossPool, a serving engine for cold MoE models that separates FFN weights and KV-cache into two GPU memory pools: a weights pool that consolidates FFN weights across cold models, and a KV-cache pool that dynamically serves active requests while keeping attention local to KV-cache. CrossPool combines a KV-cache planner and virtualizer, a layer-wise pipeline scheduler that hides hidden-state transfers, and persistent kernels with control lowering to reduce CPU-GPU control overhead. With efficient GPU memory pooling, CrossPool underpins bursty long-context requests and outperforms the state-of-the-art kvcached-based multi-LLM serving system, reducing P99 TBT by up to 10.4x.

Transformer-based large language models (LLMs) cache context as key-value (KV) pairs during inference. As context length grows, KV cache sizes expand, leading to substantial memory overhead and increased attention latency. This paper introduces KVzip, a query-agnostic KV cache eviction method enabling effective reuse of compressed KV caches across diverse queries. KVzip quantifies the importance of a KV pair using the underlying LLM to reconstruct original contexts from cached KV pairs, subsequently evicting pairs with lower importance. Extensive empirical evaluations demonstrate that KVzip reduces KV cache size by 3-4times and FlashAttention decoding latency by approximately 2times, with negligible performance loss in question-answering, retrieval, reasoning, and code comprehension tasks. Evaluations include various models such as LLaMA3.1-8B, Qwen2.5-14B, and Gemma3-12B, with context lengths reaching up to 170K tokens. KVzip significantly outperforms existing query-aware KV eviction methods, which suffer from performance degradation even at a 90% cache budget ratio under multi-query scenarios.

Large Language Models(LLMs) have had a profound impact on AI applications, particularly in the domains of long-text comprehension and generation. KV Cache technology is one of the most widely used techniques in the industry. It ensures efficient sequence generation by caching previously computed KV states. However, it also introduces significant memory overhead. We discovered that KV Cache is not necessary and proposed a novel KCache technique to alleviate the memory bottleneck issue during the LLMs inference process. KCache can be used directly for inference without any training process, Our evaluations show that KCache improves the throughput of popular LLMs by 40% with the baseline, while keeping accuracy.

Recent advances in large language models (LLMs) have extended context lengths, enabling assistants to sustain long histories for coherent, personalized responses. This ability, however, hinges on Key-Value (KV) caching, whose memory grows linearly with dialogue length and quickly dominates under strict resource constraints. An active line of research for reducing this overhead is KV cache compression, which seeks to limit cache size while preserving accuracy. Yet existing methods face two major limitations: (i) evicting entries after full-context prefill causes unbounded peak memory, and (ii) query-dependent eviction narrows the cache to a single query, leading to degraded accuracy in multi-turn conversations. We introduce EpiCache, a training-free KV cache management framework for long conversational question answering (LongConvQA) under fixed memory budgets. EpiCache bounds cache growth through block-wise prefill and preserves topic-relevant context via episodic KV compression, which clusters conversation history into coherent episodes and applies episode-specific KV cache eviction. We further design an adaptive layer-wise budget allocation strategy that measures each layer's sensitivity to eviction and distributes the memory budget across layers accordingly. Across three LongConvQA benchmarks, EpiCache improves accuracy by up to 40% over recent baselines, sustains near-full KV accuracy under 4-6x compression, and reduces latency and memory by up to 2.4x and 3.5x, thereby enabling efficient multi-turn interaction under strict resource constraints.

Complex question-answering (QA) systems face significant challenges in retrieving and reasoning over information that addresses multi-faceted queries. While large language models (LLMs) have advanced the reasoning capabilities of these systems, the bounded-recall problem persists, where procuring all relevant documents in first-stage retrieval remains a challenge. Missing pertinent documents at this stage leads to performance degradation that cannot be remedied in later stages, especially given the limited context windows of LLMs which necessitate high recall at smaller retrieval depths. In this paper, we introduce SUNAR, a novel approach that leverages LLMs to guide a Neighborhood Aware Retrieval process. SUNAR iteratively explores a neighborhood graph of documents, dynamically promoting or penalizing documents based on uncertainty estimates from interim LLM-generated answer candidates. We validate our approach through extensive experiments on two complex QA datasets. Our results show that SUNAR significantly outperforms existing retrieve-and-reason baselines, achieving up to a 31.84% improvement in performance over existing state-of-the-art methods for complex QA.

Multimodal agents repeatedly re-examine the same video frames, UI screenshots, and rendered artifacts as their context window slides and reasoning iterates, yet every look-back re-encodes from scratch, because prefix caches serve reuse only at a fixed leading position. We show this recompute is avoidable, and identify exactly what naive KV reuse loses: the cross-chunk conditioning a chunk absorbs from its neighbours. This loss is asymmetric. The direct readout of a cached chunk is recovered exactly and for free by the standard state-merge. What remains is a diffuse, low-rank residue concentrated in deep layers, invisible to single-hop retrieval but precisely what multi-hop reasoning binds on. Blind reuse therefore leaves single-hop recall intact while halving multi-hop accuracy; this is the failure mode prior position-independent caches, designed for single-context or single-image reuse, do not address. We repair it with a small, training-free low-rank conditioning patch stored alongside each position-free chunk. Reuse reduces to one operator across MLA, GQA, and MHA: exact RoPE re-rotation to any target position, plus the patch that restores cross-chunk binding. This makes three window operations cheap: reorder (one patch serves every ordering of a cached set), sliding-window survival (surviving chunks relocate via rotation only, zero re-encode), and recall (an evicted chunk is rehydrated by its patch, never re-encoded). A rank-m patch recovers full task accuracy on cross-chunk-binding benchmarks, MM-NIAH across two attention families and two-page doc-QA, at a fraction of the KV footprint, and reconstructs re-prefill KV to within bf16 rounding in a production SGLang kernel across six backbones. The conditioning signal is strongest in redundant vision and video streams, making our solution most impactful where multimodal agents spend their recompute budget.

Recent advances in Large Language Model (LLM)-based agents have been propelled by Retrieval-Augmented Generation (RAG), which grants the models access to vast external knowledge bases. Despite RAG's success in improving agent performance, agent-level cache management, particularly constructing, maintaining, and updating a compact, relevant corpus dynamically tailored to each agent's need, remains underexplored. Therefore, we introduce ARC (Agent RAG Cache Mechanism), a novel, annotation-free caching framework that dynamically manages small, high-value corpora for each agent. By synthesizing historical query distribution patterns with the intrinsic geometry of cached items in the embedding space, ARC automatically maintains a high-relevance cache. With comprehensive experiments on three retrieval datasets, our experimental results demonstrate that ARC reduces storage requirements to 0.015% of the original corpus while offering up to 79.8% has-answer rate and reducing average retrieval latency by 80%. Our results demonstrate that ARC can drastically enhance efficiency and effectiveness in RAG-powered LLM agents.

Large Language Model (LLM) agents can handle increasingly complex tasks, navigating long trajectories of observations and actions. As these trajectories grow, key-value (KV) cache I/O and storage become the dominant cost. Trajectories can also exceed the LLM's maximum context length, requiring control over active cache growth. Prior work uses context engineering to reduce cache size by trimming or summarizing history. However, modifying the trajectory prefix invalidates subsequent cached states and triggers costly re-prefilling of retained context. Task accuracy can also suffer when too little context is retained to limit re-prefill costs. We propose cache engineering as a new scheme. It allows agents to directly control their active KV cache without modifying the trajectory or re-prefilling retained context. Specifically, agents can proactively use drop to exclude selected cache from future attention and repos to adjust retained keys' position encodings. We formalize these cache operations and identify history dependence as a serving challenge: identical token sequences can yield different cache states under different cache operation sequences, making prefix matching insufficient for correct cache reuse. Thus, we co-design the AKV serving engine. It provides a stateless interface compatible with existing serving APIs, while internally managing stateful cache reuse, eviction, and restoration. Experiments show that a tool-response rolling-drop cache policy improves average benchmark scores by 1.7 points and reduces average cost by 27.7%, compared with conventional threshold-triggered summarization. With AKV, it achieves up to 1.2x the output throughput and a 1.5x speedup in P95 request latency. Recursive self-improvement of cache policies further reduces cost by 7.1% over expert design. Our code is at https://github.com/thu-nics/AKV.

The KV cache used in large language models has linearly growing time complexity, so LLMs face memory blow-up and reduced decoding efficiency when they process long contexts. Current KV Cache eviction has become an important research direction; however, existing methods based on fixed Soft Tokens (e.g., Judge Q) rely on a static parameter set as the query to evaluate the importance of KV pairs, so they cannot adapt dynamically to different input prompts, and they cannot precisely capture complex and changing task relevance. Also, evicted KV pairs are discarded permanently, so this causes irreversible information loss and context breaks. To address this problem, we propose Meta-Soft, a dynamic compression framework based on probe-driven context integration. Specifically, we build a meta-library with a learnable orthogonal basis matrix L, and we use a selector network with Gumbel-Softmax to produce differentiable sparse combination weights, so we dynamically synthesize the most targeted k Soft Tokens from the input prompt features. We append these Soft Tokens to the end of the input sequence to probe key information. We also introduce an attention-flow based integration mechanism, which redistributes the semantic information of removed tokens into retained tokens, and this keeps the dropped context information effectively. Experiments on multiple datasets show that our method outperforms existing state-of-the-art eviction methods and provides a new solution for KV Cache compression.

Text-to-image diffusion models can be fine-tuned in custom domains to adapt to specific user preferences, but such unconstrained adaptability has also been utilized for illegal purposes, such as forging public figures' portraits and duplicating copyrighted artworks. Most existing work focuses on detecting the illegally generated contents, but cannot prevent or mitigate illegal adaptations of diffusion models. Other schemes of model unlearning and reinitialization, similarly, cannot prevent users from relearning the knowledge of illegal model adaptation with custom data. In this paper, we present FreezeAsGuard, a new technique that addresses these limitations and enables irreversible mitigation of illegal adaptations of diffusion models. The basic approach is that the model publisher selectively freezes tensors in pre-trained diffusion models that are critical to illegal model adaptations, to mitigate the fine-tuned model's representation power in illegal domains but minimize the impact on legal model adaptations in other domains. Such tensor freezing can be enforced via APIs provided by the model publisher for fine-tuning, can motivate users' adoption due to its computational savings. Experiment results with datasets in multiple domains show that FreezeAsGuard provides stronger power in mitigating illegal model adaptations of generating fake public figures' portraits, while having the minimum impact on model adaptation in other legal domains. The source code is available at: https://github.com/pittisl/FreezeAsGuard/

Reusing the attention KV cache across requests is a standard way to cut prefill cost in long-context LLM serving, and production systems increasingly back the per-engine cache with a remote tier shared by every engine. We study such a tier in a disaggregated deployment of GLM-5.3, an FP8 mixture-of-experts model with multi-head latent attention (MLA), on 8xB300 nodes running prefill context parallelism (PCP8). Enabling the remote tier (LMCache with a Valkey store) left median time-to-first-token unchanged but raised the 16k-token p99 from 3.96 s to 17.3 s. Replaying a production request stream on an isolated engine, we find that a remote hit took 0.88 s against 0.31 s to recompute the prompt, and that synchronous loading turned a hit-heavy stream into an unstable queue: p99 83.32 s with the GPUs 7% busy. A two-term cost model explains both: a hit pays off only if the tier delivers more than B* = sR (2.53 GB/s for this model and GPU), and a fixed per-hit engine cost (T0 ~ 0.33 s) means that hits on prompts shorter than about 17k tokens cannot beat recompute at any bandwidth. Guided by the model, we (i) shard the MLA KV store across PCP ranks, with a proof that the existing minimum-over-ranks lookup returns the longest cached prefix and a load protocol that can never desynchronize collectives; (ii) partition the remote tier by rank, one server instance per rank with no coordination, which raises GET bandwidth 6.3x to 10.76 GB/s; (iii) add asynchronous loading to the sharded store; and (iv) fix three client bugs that caused unbounded stalls. On the hit-heavy stream, with 98.5% of prompt tokens served from the remote tier, p99 falls from 83.32 s to 0.67 s. The sharded store raises prefill throughput by 14%, and by 34% together with an upstream kernel fix. All changes are submitted upstream to LMCache.

Memory-augmented LLM agents store and retrieve information from prior interactions, yet the relative importance of how memories are written versus how they are retrieved remains unclear. We introduce a diagnostic framework that analyzes how performance differences manifest across write strategies, retrieval methods, and memory utilization behavior, and apply it to a 3x3 study crossing three write strategies (raw chunks, Mem0-style fact extraction, MemGPT-style summarization) with three retrieval methods (cosine, BM25, hybrid reranking). On LoCoMo, retrieval method is the dominant factor: average accuracy spans 20 points across retrieval methods (57.1% to 77.2%) but only 3-8 points across write strategies. Raw chunked storage, which requires zero LLM calls, matches or outperforms expensive lossy alternatives, suggesting that current memory pipelines may discard useful context that downstream retrieval mechanisms fail to compensate for. Failure analysis shows that performance breakdowns most often manifest at the retrieval stage rather than at utilization. We argue that, under current retrieval practices, improving retrieval quality yields larger gains than increasing write-time sophistication. Code is publicly available at https://github.com/boqiny/memory-probe.

Mainstream LLM serving systems reuse prefix work mainly through paged or radix key-value (KV) caches. This is highly effective for high-throughput, high-concurrency serving, but it manages only one positional fragment of execution state: the KV cache. We study the opposite regime: low-latency, small-batch, on-device physical-AI serving, where interactive LLM agents, speech systems, and robot policies repeatedly branch, reset, interrupt, and re-enter under tight responsiveness budgets. We introduce execution-state capsules, a graph-bound checkpoint and restore mechanism for the complete restorable state at a committed boundary. FlashRT is a white-box, backend-facing kernel runtime whose evaluated NVIDIA CUDA backend runs captured graph plans over contiguous static buffers with no block-table indirection. Because the live state is a closed set of named buffers, a capsule can snapshot, restore, fork, or roll back the whole execution boundary, including KV, recurrent state, convolution state, MTP state, and metadata. This moves reuse from token-addressed KV fragments to graph-bound execution-state boundaries. On an RTX 5090, capsule restore is byte-exact at the stored-state level and token-identical under greedy decode. A KV-only ablation diverges, showing that recurrent state is load-bearing. GPU-resident snapshot and restore are sub-millisecond, and TTFT speedup over cold prefill grows from 3.9x at 2k tokens to 27x at 16k tokens. On Jetson AGX Thor and DGX Spark, the same correctness and structural properties hold. Capsules are not a replacement for high-throughput KV-cache serving; they define a complementary latency-first serving point for explicit execution-state reuse.

LLM-based agentic applications have shown increasingly remarkable capabilities in complex workflows but incur substantial costs due to extensive planning and reasoning requirements. Existing LLM caching techniques (like context caching and semantic caching), primarily designed for serving chatbots, are insufficient for agentic applications where outputs depend on external data or environmental contexts. We propose agentic plan caching, a novel approach that extracts, stores, adapts, and reuses structured plan templates from planning stages of agentic applications across semantically similar tasks to reduce the cost of serving. Unlike traditional semantic caching, our system extracts plan templates from completed agent executions at test-time, employs keyword extraction to match new requests against cached plans, and utilizes lightweight models to adapt these templates to task-specific plans with contexts. Evaluation across multiple real-world agentic applications shows that our system can reduce costs by 46.62% on average while maintaining performance, offering a more efficient solution for serving LLM-based agents that complements existing LLM serving infrastructures.

Generating texts with a large language model (LLM) consumes massive amounts of memory. Apart from the already-large model parameters, the key/value (KV) cache that holds information about previous tokens in a sequence can grow to be even larger than the model itself. This problem is exacerbated in one of the current LLM serving frameworks which reserves the maximum sequence length of memory for the KV cache to guarantee generating a complete sequence as they do not know the output sequence length. This restricts us to use a smaller batch size leading to lower GPU utilization and above all, lower throughput. We argue that designing a system with a priori knowledge of the output sequence can mitigate this problem. To this end, we propose S^{3}, which predicts the output sequence length, schedules generation queries based on the prediction to increase device resource utilization and throughput, and handle mispredictions. Our proposed method achieves 6.49times throughput over those systems that assume the worst case for the output sequence length.

KV-cache memory is a major bottleneck in real-world LLM serving, where systems must simultaneously support latency-sensitive small-batch requests and high-throughput concurrent workloads. Although many KV-cache compression methods improve offline accuracy or compression ratio, they often violate practical serving constraints such as paged memory layouts, regular memory access, and fused attention execution, limiting their effectiveness in deployment. In this work, we identify the minimal set of 4-bit KV-cache quantization methods that remain viable under these constraints. Our central finding is that a simple design--token-wise INT4 quantization with block-diagonal Hadamard rotation--consistently achieves the best accuracy-efficiency trade-off. Across multiple models and benchmarks, this approach recovers nearly all of the accuracy lost by naive INT4, while more complex methods such as vector quantization and Hessian-aware quantization provide only marginal additional gains once serving compatibility is taken into account. To make this practical, we implement a fused rotation-quantization kernel that integrates directly into paged KV-cache layouts and introduces zero measurable end-to-end overhead, matching plain INT4 throughput across concurrency levels. Our results show that effective KV-cache compression is fundamentally a systems co-design problem: under real serving constraints, lightweight block-diagonal Hadamard rotation is a viable method that delivers near-lossless accuracy without sacrificing serving efficiency.

As the demand for long-context large language models (LLMs) increases, models with context windows of up to 128K or 1M tokens are becoming increasingly prevalent. However, long-context LLM inference is challenging since the inference speed decreases significantly as the sequence length grows. This slowdown is primarily caused by loading a large KV cache during self-attention. Previous works have shown that a small portion of critical tokens will dominate the attention outcomes. However, we observe the criticality of a token highly depends on the query. To this end, we propose Quest, a query-aware KV cache selection algorithm. Quest keeps track of the minimal and maximal Key values in KV cache pages and estimates the criticality of a given page using Query vectors. By only loading the Top-K critical KV cache pages for attention, Quest significantly speeds up self-attention without sacrificing accuracy. We show that Quest can achieve up to 2.23x self-attention speedup, which reduces inference latency by 7.03x while performing well on tasks with long dependencies with negligible accuracy loss. Code is available at http://github.com/mit-han-lab/Quest .

Due to the rapid generation and dissemination of information, large language models (LLMs) quickly run out of date despite enormous development costs. Due to this crucial need to keep models updated, online learning has emerged as a critical necessity when utilizing LLMs for real-world applications. However, given the ever-expanding corpus of unseen documents and the large parameter space of modern LLMs, efficient adaptation is essential. To address these challenges, we propose Memory of Amortized Contexts (MAC), an efficient and effective online adaptation framework for LLMs with strong knowledge retention. We propose an amortized feature extraction and memory-augmentation approach to compress and extract information from new documents into compact modulations stored in a memory bank. When answering questions, our model attends to and extracts relevant knowledge from this memory bank. To learn informative modulations in an efficient manner, we utilize amortization-based meta-learning, which substitutes the optimization process with a single forward pass of the encoder. Subsequently, we learn to choose from and aggregate selected documents into a single modulation by conditioning on the question, allowing us to adapt a frozen language model during test time without requiring further gradient updates. Our experiment demonstrates the superiority of MAC in multiple aspects, including online adaptation performance, time, and memory efficiency. Code is available at: https://github.com/jihoontack/MAC.

Agent memory is a systems problem for long-horizon agents. Practical deployments require retention of task state across extended conversations, recovery of user-specific facts and preferences across sessions, and accumulation of procedural knowledge from prior outcomes. These requirements extend beyond document retrieval: a memory layer must determine which interactions become durable state, how that state is scoped, how it is retrieved under latency constraints, and how it is revised or removed over time. This report studies Oracle Agent Memory as a database-native memory substrate built on Oracle Database. Three themes organize the discussion: memory as a lifecycle spanning ingestion, extraction, consolidation, retrieval, summarization, and revision or removal; a layered architecture that separates an active memory core from a passive memory-store interface with explicit scope control across users, agents, and threads; and evaluation methodology in which downstream task accuracy is complemented by memory-centric measures such as evidence retrieval, recall, latency, and estimated token use. The report summarizes LongMemEval results, reaching 93.8% accuracy, compares Oracle Agent Memory against flat-history baselines, using about 10.7x fewer tokens, and published or reported external baselines where available, and closes with implementation-oriented appendix material covering setup, thread lifecycle, and search semantics.

Large Language Model (LLM)-based search agents have shown remarkable capabilities in solving complex tasks by dynamically decomposing problems and addressing them through interleaved reasoning and retrieval. However, this interleaved paradigm introduces substantial efficiency bottlenecks. First, we observe that both highly accurate and overly approximate retrieval methods degrade system efficiency: exact search incurs significant retrieval overhead, while coarse retrieval requires additional reasoning steps during generation. Second, we identify inefficiencies in system design, including improper scheduling and frequent retrieval stalls, which lead to cascading latency -- where even minor delays in retrieval amplify end-to-end inference time. To address these challenges, we introduce SearchAgent-X, a high-efficiency inference framework for LLM-based search agents. SearchAgent-X leverages high-recall approximate retrieval and incorporates two key techniques: priority-aware scheduling and non-stall retrieval. Extensive experiments demonstrate that SearchAgent-X consistently outperforms state-of-the-art systems such as vLLM and HNSW-based retrieval across diverse tasks, achieving up to 3.4times higher throughput and 5times lower latency, without compromising generation quality. SearchAgent-X is available at https://github.com/tiannuo-yang/SearchAgent-X.

Large Language Models (LLMs) play powerful, black-box readers in the retrieve-then-read pipeline, making remarkable progress in knowledge-intensive tasks. This work introduces a new framework, Rewrite-Retrieve-Read instead of the previous retrieve-then-read for the retrieval-augmented LLMs from the perspective of the query rewriting. Unlike prior studies focusing on adapting either the retriever or the reader, our approach pays attention to the adaptation of the search query itself, for there is inevitably a gap between the input text and the needed knowledge in retrieval. We first prompt an LLM to generate the query, then use a web search engine to retrieve contexts. Furthermore, to better align the query to the frozen modules, we propose a trainable scheme for our pipeline. A small language model is adopted as a trainable rewriter to cater to the black-box LLM reader. The rewriter is trained using the feedback of the LLM reader by reinforcement learning. Evaluation is conducted on downstream tasks, open-domain QA and multiple-choice QA. Experiments results show consistent performance improvement, indicating that our framework is proven effective and scalable, and brings a new framework for retrieval-augmented LLM.

Enterprise repositories are large, heteroge- neous, and continuously updated, making re- trieval difficult when efficient access, source- faithful evidence, and cross-query adaptation must be supported together. Enterprise mem- ory extends retrieval beyond the model con- text, but existing systems do not jointly address collection-specific hierarchy construction, low- cost routing, detailed evidence loading, and workload-aware memory updates at this scale. We introduce MEMONDEMAND, short for On- Demand Memory, a memory management sys- tem with three coordinated mechanisms: a dy- namic multi-level hierarchy that determines the abstraction structure and depth for each col- lection, dual memory at every hierarchy level that separates distilled routing from detailed evidence, and on-demand memory promotion that updates node priority under a bounded active-state budget. On EnterpriseRAG-Bench, MEMONDEMAND outperforms the strongest published LB#1 result at every evaluated scale from 10M tokens through the complete 618M- token collection, with gains of 12.23% at 10M and 4.66% at 618M. Results on FinanceBench, HotpotQA, and FRAMES further show strong performance across financial, multi-hop, and fact-retrieval settings. Together, these results establish MEMONDEMAND as an accurate, ef- ficient, and scalable memory solution for very large enterprise repositories across data scales, domains, and evidence requirements. Our code is available at https://github.com/ xfab-xinyuansong/MemOnDemand.git.

Chunked KV-cache compression reduces the memory and attention costs of long-context inference by compressing windows of consecutive tokens into fewer cache entries at a fixed stride. Such compression also introduces a new positional coordinate: a token's phase, or its position relative to compression-window boundaries. We uncover a systematic asymmetry in models using such compression: the same information can be easy to retrieve at one phase and difficult at another. We call this periodic variation in retrieval performance phase sensitivity. In large open-weight models with such compression, long-context retrieval accuracy can differ by up to 40 percentage points across phases, revealing periodic weak spots that average benchmark scores can conceal. To investigate this behavior, we pretrain a family of transformers from scratch across multiple KV-compression designs, reproducing phase sensitivity across the variants. Mechanistic analysis using causal interventions in these models reveals phase specialization: different attention components contribute asymmetrically to retrieving information at different source phases. We further analyze idealized retrieval models, showing how gradient flow dynamics may favor sharp phase specialization. Evaluating models with chunked KV-cache compression thus requires measuring across compression phases: high average accuracy can coexist with systematic positional failures.

We present AutoIndex, a framework for learning representation programs: executable transformations that map raw documents into the representations exposed to a retrieval system. Rather than tuning retrievers, rerankers, or a small set of preprocessing hyperparameters, AutoIndex searches over programs that slice, enrich, normalize, reweight, or reorganize documents before indexing. At each iteration, AutoIndex performs validation-guided program search, in which agents diagnose failures of the current program and synthesize candidate updates, retaining only updates that improve retrieval quality under the resulting index. We evaluate AutoIndex on CRUMB, a benchmark of heterogeneous retrieval tasks, with BM25 held fixed across all experiments. The learned programs improve recall over a static full-document BM25 baseline on all 8 tasks, with average gains of +8.4% in Recall@100 and +8.3% in nDCG@10, and largest gains of +30.5% in Recall@100 and +43.6% in nDCG@10. These results suggest that document representation should not be treated as a fixed preprocessing choice made before retrieval begins, but as an explicit optimization target. Code to reproduce our results is available at https://github.com/auto-index/autoindex.

LLM serving relies on prefix caching to improve inference performance. As growing contexts push key-value (KV) cache footprint far beyond GPU HBM and CPU DRAM capacity, KV cache is increasingly offloaded to NVMe SSDs. Unfortunately, restoring KV cache from SSDs suffers from poor I/O performance and incurs significant GPU stalls. This is primarily because the fragmented GPU memory layout results in a massive number of tiny random I/Os, rendering the low-parallelism CPU a severe bottleneck even with GPU Direct Storage (GDS), which still relies on CPU intervention to initiate each I/O and thus remains CPU-centric. This paper presents Tutti, an efficient SSD-backed KV caching solution that eliminates CPU intervention from the critical data and I/O control paths between HBM and SSDs. At the core of Tutti is a GPU-centric KV cache object store, in which the CPU is only responsible for asynchronously loading I/O kernels once per layer to the GPU. Tutti saturates NVMe SSD bandwidth and reduces GPU stalls to near zero through the following designs: (i) we provide a GPU-native object abstraction that enables bulk KV cache transfers and management; (ii) we re-architect the GPU storage stack by introducing GPU io_uring to support asynchronous GPU direct object I/O; and (iii) we propose slack-aware I/O scheduling to avoid GPU resource contention. We have implemented Tutti and integrated it to vLLM. Extensive evaluation shows that compared to the state-of-the-art GDS-enabled, SSD-backed LMCache, Tutti reduces TTFT by 78.3% under strict SLO constraints and improves the achievable request rate by 2x. The serving cost is reduced by 27%. Tutti achieves nearly the same inference performance as DRAM-backed LMCache, while providing almost infinite capacity.

Text-to-image retrieval with frozen dual encoders degrades when the query names a small, visually subordinate object in a cluttered scene: a single global image embedding underrepresents the localized visual evidence. We present MINER, a training-free inference framework that augments a frozen dual encoder's global image embedding with a small bank of region-level embeddings and a hubness-correcting similarity rescoring, recovering visual evidence that global pooling underweights. To evaluate this setting, we introduce ROCS, a benchmark built from high-clutter subsets of Flickr30K and MS COCO whose images are re-captioned to name a single low-salience object. Experiments on CLIP, SigLIP, and SigLIP 2 show that MINER improves retrieval on every backbone, on ROCS and on the standard splits. Analyses show that these gains come primarily from broader spatial coverage rather than precise crop placement, revealing a simple and general way to recover localized evidence from frozen representations. Code: https://github.com/aalquwayfili/MINER. Dataset: https://huggingface.co/datasets/aalquwayfili/ROCS.

Semantic caching cuts LLM inference costs by serving a cached response to semantically similar queries. Standard practice evaluates these systems using PR-AUC, a metric that only measures how well scores rank and ignores whether they are usable at a fixed threshold. We show this mismatch leads to systematically poor deployment choices, as models with the highest PR-AUC are often the worst in operation. We introduce Precision--Cache Hit Ratio (P-CHR) AUC, a cache-aware metric that measures precision across cache utilization levels, and Operational Retention Rate (ORR), which captures how much offline ranking quality survives at deployment. We decompose the operational gap between offline and deployed quality into a recoverable threshold-utility component and an irreducible structural component fixed by the dataset's positive rate. Our experiments show that the threshold-utility gap is governed by the training objective rather than data scale, and yields only to re-normalizing scores over the candidate pool or changing the training objective. Ultimately, model selection for semantic caching is a threshold-utility problem, not a ranking one, and measuring it is the first step to closing the gap.

Efficient key-value (KV) cache management is crucial for the practical deployment of large language models (LLMs), yet existing compression techniques often incur a trade-off between performance degradation and computational overhead. We propose a novel gating-based KV cache eviction method for frozen-weight LLMs that achieves high compression ratios with negligible computational cost. Our approach introduces lightweight sink-attention gating modules to identify and retain critical KV pairs, and integrates seamlessly into both the prefill and decoding stages. The proposed gate training algorithm relies on forward passes of an LLM, avoiding expensive backpropagation, while achieving strong task generalization through a task-agnostic reconstruction objective. Extensive experiments across the Qwen2.5-1M, Qwen3, and Gemma3 families show that our method maintains near-lossless performance while evicting up to 70% of the KV cache. The results are consistent across a wide range of tasks, including long-context understanding, code comprehension, and mathematical reasoning, demonstrating the generality of our approach.

Long-horizon LLM agents accumulate tool results, files, plans, and user constraints that are too structured to be treated as a disposable text suffix. Current systems mostly rely on in-run heuristics such as chronological pruning and tool-output masking, or on final self-summary near a context limit. Heuristics are cheap but blind to future dependencies; summaries preserve narrative state but often hide exact evidence, locators, and editable artifacts. We present Self-GC, where GC denotes self-governing context while deliberately echoing garbage collection: the system does not merely reclaim unused tokens, but governs the lifecycle of agent context objects. Self-GC turns user turns, tool spans, and skill state into indexed objects; asks a side-channel planner to propose fold, mask, and prune actions; and lets the harness enforce recoverable sidecars, safe commit boundaries, and cache-aware commit. On a 33-session Hard Set, Self-GC prunes 43.95% of prefix tokens while leaving 84.85% of future continuations unaffected, compared with no-impact rates of 54.55% to 69.70% for heuristic baselines. On a 332-session production-derived suite, three planner backbones reach no-impact rates of 91.27% to 94.58%, while baselines remain at 77.71% to 87.46%. In production, an online account-level split reduces daytime average input tokens by 10% to 15%, with peak reductions near 20%. These results point to context management as runtime lifecycle control over indexed, recoverable objects rather than post hoc text cleanup.

We report a way to make a frozen small language model both more capable and dramatically cheaper at once, without changing any weights. Verified knowledge is deposited once as a byte-exact key-value (KV) state artifact and later restored, by graft, into a fresh inference context. The restore is bit-exact: under a pinned deterministic configuration, the grafted logits are byte-for-byte identical to a fresh computation (SHA-256 equality), with zero KL divergence and 100% argmax agreement over fifty samples. We show that own-position graft is the unique numerically exact operating point on a model with floating-point rotary encoding, and we verify byte-exactness on two model scales (12B, 31B) and two GPU targets, one through a pre-registered replay. On AIME 2025, a frozen Gemma-4-12B moves from 80.0% to 93.3% once a verified solution library is grafted, above its own 77.5% and its 31B sibling's 89.2% published anchors. On the recurring case, eight problems the base model never solves within a 401,026-token budget are answered from cached verified solutions in 61 total decode tokens, a factor of 6,574 fewer tokens and about 8,700x less energy; the capability claim proper rests on held-out transfer (7 of 7 at 31B). The same byte-exact store widens usable context from 32,768 to 2,854,766 tokens at zero extra accelerator memory, and moves byte-identical between machines of the same architecture. We describe the system at the behavior level; the engine is proprietary, and every reported number is backed by committed input and output hashes so the scoring can be re-checked without it.

Long-horizon agents increasingly reuse their KV cache as memory: a serving system keeps a subset of cached entries and drops the rest. Eviction and episodic-memory schemes therefore rest on a premise rarely tested directly, that a retained event is still informative once the observations that produced it are gone. We test it by omitting one earlier observation from what is served, across otherwise identical agent histories. Among items sensitive to that observation, the answer overwhelmingly follows the omitted value, though no served span says which value is correct. We call this semantic materialization: a downstream event's cached rows act as an independently servable view of computation whose inputs are gone. It can also be written on purpose. A deliberately phrased, answer-free event raises donor-aligned recovery from 6% to 51% on Qwen3-8B without ever naming the value, whereas passively harvesting natural mentions from long-term dialog yields no detected advantage. What such a row carries is specific and bounded. Compact state survives, larger payloads decay toward chance, and whether a construction writes at all turns on phrasing rather than on meaning alone, so two phrasings the model comprehends equally well can diverge sharply. The result is a memory contract for sparse event-KV serving: what to write, where it lands, and what survives once the source is gone. For anyone who evicts the corollary is that dropping a source event and observing no accuracy loss does not show the source was unnecessary.

The emergence of long-context text applications utilizing large language models (LLMs) has presented significant scalability challenges, particularly in memory footprint. The linear growth of the Key-Value (KV) cache responsible for storing attention keys and values to minimize redundant computations can lead to substantial increases in memory consumption, potentially causing models to fail to serve with limited memory resources. To address this issue, we propose a novel approach called Cache Sparse Representation (CSR), which converts the KV cache by transforming the dense Key-Value cache tensor into sparse indexes and weights, offering a more memory-efficient representation during LLM inference. Furthermore, we introduce NeuralDict, a novel neural network-based method for automatically generating the dictionary used in our sparse representation. Our extensive experiments demonstrate that CSR achieves performance comparable to state-of-the-art KV cache quantization algorithms while maintaining robust functionality in memory-constrained environments.

Industrial asset operations workflows are latency-sensitive because a single user query may require coordination over sensor data, work orders, failure modes, forecasting tools, and domain-specific agents. We evaluate this problem on AssetOpsBench (AOB), an industrial agent benchmark whose plan-execute pipeline exposes repeated overhead from tool discovery, LLM planning, MCP tool execution, and final summarization. Existing LLM caching techniques such as KV-cache reuse and embedding-based semantic caching were designed for chatbot serving and break down when output validity depends on time, asset, or sensor parameters. We propose two complementary optimization layers for AOB plan-execute pipelines: a temporal semantic cache and a set of MCP workflow optimizations combining disk-backed tool-discovery caching and dependency-aware parallel step execution. MCP workflow optimizations corresponded to a 1.67x speedup and reduced median end-to-end latency by about 40.0% while the temporal-cache benchmark achieved a median of 30.6x speedup on cache hits. Beyond the speedup, our results expose a concrete failure mode of pure semantic caching for parameter-rich industrial queries, providing a critical analysis of how caching choices interact with evaluation correctness in MCP-backed agent benchmarks.

Hybrid models that combine the language modeling capabilities of Attention layers with the efficiency of Recurrent layers (e.g., State Space Models) have gained traction in practically supporting long contexts in Large Language Model serving. Yet, the unique properties of these models complicate the usage of complementary efficiency optimizations such as prefix caching that skip redundant computations across requests. Most notably, their use of in-place state updates for recurrent layers precludes rolling back cache entries for partial sequence overlaps, and instead mandates only exact-match cache hits; the effect is a deluge of (large) cache entries per sequence, most of which yield minimal reuse opportunities. We present Marconi, the first system that supports efficient prefix caching with Hybrid LLMs. Key to Marconi are its novel admission and eviction policies that more judiciously assess potential cache entries based not only on recency, but also on (1) forecasts of their reuse likelihood across a taxonomy of different hit scenarios, and (2) the compute savings that hits deliver relative to memory footprints. Across diverse workloads and Hybrid models, Marconi achieves up to 34.4times higher token hit rates (71.1% or 617 ms lower TTFT) compared to state-of-the-art prefix caching systems.

This report investigates enhancing semantic caching effectiveness by employing specialized, fine-tuned embedding models. Semantic caching relies on embedding similarity rather than exact key matching, presenting unique challenges in balancing precision, query latency, and computational efficiency. We propose leveraging smaller, domain-specific embedding models, fine-tuned with targeted real-world and synthetically generated datasets. Our empirical evaluations demonstrate that compact embedding models fine-tuned for just one epoch on specialized datasets significantly surpass both state-of-the-art open-source and proprietary alternatives in precision and recall. Moreover, we introduce a novel synthetic data generation pipeline for the semantic cache that mitigates the challenge of limited domain-specific annotated data, further boosting embedding performance. Our approach effectively balances computational overhead and accuracy, establishing a viable and efficient strategy for practical semantic caching implementations.

Long-context reasoning is essential for complex and long-horizon tasks, yet the performance of large language models (LLMs) degrades as context length increases. Recent approaches address this by processing input chunk by chunk while maintaining a bounded textual memory in model context. However, premature information compression can discard critical details essential for subsequent reasoning. In this paper, we introduce Commit-on-Evidence Memory (CoEM), which learns when to convert source evidence into compact memory facts. Specifically, under a fixed context-memory budget, CoEM preserves potentially useful source excerpts verbatim in a pending set, allowing subsequent context to clarify their relevance before irreversible compression. As new context arrives, a learned policy revisits each pending excerpt and decides whether to promote it to the committed memory, retain it for further consideration, or discard it. A frozen verifier ensures proposed facts are accepted only if supported by retained excerpts and current context. To further guide effective memory management, we train this policy using reinforcement learning by combining fine-grained, step-level evidence rewards with final answer rewards. Extensive experiments demonstrate that CoEM consistently improves long-context reasoning. When evaluated on 6,400 documents long-context input, CoEM outperforms the strongest memory baseline by 10.4-11.4 F1 points on Qwen3.5-9B. Code repository: https://github.com/benmagnifico/CoEM.

Context compression improves long-context efficiency but may introduce faithfulness issues through an information bottleneck. Existing evaluations rely on QA metrics, but a reader can still answer correctly from an unfaithful compression, so these metrics underestimate such loss. To bridge this gap, we introduce KEEP-Bench, a diagnostic benchmark that directly assesses compression faithfulness with 4,102 cases across three challenging settings that require keeping evidence for plausible reference resolutions, explicit qualifiers, and minority perspectives. Experiments show that even with tailored prompting, the strongest models remain unreliable: despite context windows exceeding 100K tokens and inputs averaging only 3.1K tokens, the best open-weight model (31B) preserves all required evidence in only 66.8% of cases, and the best proprietary model in only 61.6%. Performance also varies sharply across evidence structures, so conclusions from limited structures may be biased. We further introduce a metric, Evidence Preservation Return (EPR), to relate evidence recall to inference FLOPs, and find that it peaks at moderate scales (4B-8B), where recall remains low. Additional output budget mainly benefits already strong models and widens capability gaps, yet the stronger preservation of these models comes at a steep computational cost. These findings highlight the importance for future work to investigate compressors with strong information preservation under a small computation budget, rather than relying on a large output budget or model scaling.

Large Language Models (LLMs) have ignited an innovative surge of AI applications, marking a new era of exciting possibilities equipped with extended context windows. However, hosting these models is cost-prohibitive mainly due to the extensive memory consumption of KV Cache involving long-context modeling. Despite several works proposing to evict unnecessary tokens from the KV Cache, most of them rely on the biased local statistics of accumulated attention scores and report performance using unconvincing metric like perplexity on inadequate short-text evaluation. In this paper, we propose NACL, a general framework for long-context KV cache eviction that achieves more optimal and efficient eviction in a single operation during the encoding phase. Due to NACL's efficiency, we combine more accurate attention score statistics in PROXY TOKENS EVICTION with the diversified random eviction strategy of RANDOM EVICTION, aiming to alleviate the issue of attention bias and enhance the robustness in maintaining pivotal tokens for long-context modeling tasks. Notably, our method significantly improves the performance on short- and long-text tasks by 80% and 76% respectively, reducing KV Cache by up to 50% with over 95% performance maintenance. The code is available at https://github.com/PaddlePaddle/Research/tree/master/NLP/ACL2024-NACL.

This paper introduces Neurocache, an approach to extend the effective context size of large language models (LLMs) using an external vector cache to store its past states. Like recent vector retrieval approaches, Neurocache uses an efficient k-nearest-neighbor (kNN) algorithm to retrieve relevant past states and incorporate them into the attention process. Neurocache improves upon previous methods by (1) storing compressed states, which reduces cache size; (2) performing a single retrieval operation per token which increases inference speed; and (3) extending the retrieval window to neighboring states, which improves both language modeling and downstream task accuracy. Our experiments show the effectiveness of Neurocache both for models trained from scratch and for pre-trained models such as Llama2-7B and Mistral-7B when enhanced with the cache mechanism. We also compare Neurocache with text retrieval methods and show improvements in single-document question-answering and few-shot learning tasks. We made the source code available under: https://github.com/alisafaya/neurocache

In the field of multimodal large language models (MLLMs), common methods typically involve unfreezing the language model during training to foster profound visual understanding. However, the fine-tuning of such models with vision-language data often leads to a diminution of their natural language processing (NLP) capabilities. To avoid this performance degradation, a straightforward solution is to freeze the language model while developing multimodal competencies. Unfortunately, previous works have not attained satisfactory outcomes. Building on the strategy of freezing the language model, we conduct thorough structural exploration and introduce the Inner-Adaptor Architecture (IAA). Specifically, the architecture incorporates multiple multimodal adaptors at varying depths within the large language model to facilitate direct interaction with the inherently text-oriented transformer layers, thereby enabling the frozen language model to acquire multimodal capabilities. Unlike previous approaches of freezing language models that require large-scale aligned data, our proposed architecture is able to achieve superior performance on small-scale datasets. We conduct extensive experiments to improve the general multimodal capabilities and visual grounding abilities of the MLLM. Our approach remarkably outperforms previous state-of-the-art methods across various vision-language benchmarks without sacrificing performance on NLP tasks. Code and models are available at https://github.com/360CVGroup/Inner-Adaptor-Architecture.

Semantic caching has emerged as a pivotal technique for scaling LLM applications, widely adopted by major providers including AWS and Microsoft. By utilizing semantic embedding vectors as cache keys, this mechanism effectively minimizes latency and redundant computation for semantically similar queries. In this work, we conceptualize semantic cache keys as a form of fuzzy hashes. We demonstrate that the locality required to maximize cache hit rates fundamentally conflicts with the cryptographic avalanche effect necessary for collision resistance. Our conceptual analysis formalizes this inherent trade-off between performance (locality) and security (collision resilience), revealing that semantic caching is naturally vulnerable to key collision attacks. While prior research has focused on side-channel and privacy risks, we present the first systematic study of integrity risks arising from cache collisions. We introduce CacheAttack, an automated framework for launching black-box collision attacks. We evaluate CacheAttack in security-critical tasks and agentic workflows. It achieves a hit rate of 86\% in LLM response hijacking and can induce malicious behaviors in LLM agent, while preserving strong transferability across different embedding models. A case study on a financial agent further illustrates the real-world impact of these vulnerabilities. Finally, we discuss mitigation strategies.

As large language models (LLMs) process increasingly longer prompts, computation and KV-cache memory costs have emerged as major bottlenecks in inference systems. Existing input-level prompt compression methods address this, but rank each sentence by a scalar relevance score, treating the document as an unstructured pool of words and sentences. Under tight budgets, this causes theme collapse, where the dominant theme(s) of a document consumes the budget, discarding less-frequent yet task-relevant themes. Preserving thematic coverage instead requires allocating the budget across recurring themes rather than scoring sentences in isolation. To this end, we propose SALT, a model-agnostic extractive framework that organizes per-sentence keywords into a trie ordered by sentence frequency (SF), a lightweight, reusable proxy for document thematic structure. This trie-based organization smooths memory allocation and prevents dominant themes from monopolizing the budget. Multi-anchor retrieval activates trie nodes labeled by query keywords at any depth, and the trie persists across dialogue turns, supporting multi-turn use without re-encoding the document. By preserving document themes, SALT reduces the prefill computation and memory cost of long-context prompts while remaining composable with KV-cache methods that target decoding-time latency and memory.

A search engine that embeds text, code, documents, images, audio and video into the same representation space has to run its encoder and keep its index somewhere, and almost every component built for the purpose assumes a server. We present omni-macos, which runs its encoder, index and store on the Mac that already holds the files, so no file, query or vector ever leaves the machine. It keeps a background indexer and an interactive search box inside one memory budget the user sets: it re-encodes only the chunks an edit changes, hands the GPU smaller units while the user is typing, answers queries from a quantized replica with exact rescoring, and propagates that budget to the allocators that draw on unified memory. We measure every mechanism on five Macs spanning an eightfold range of accelerator width and a thirty-twofold range of memory, each one indexing its own local files.

Retrieval-augmented generation degrades when scaled to large, heterogeneous document collections, where dense similarity loses discriminative power, and top-k retrieval increasingly returns semantically similar but contextually incorrect chunks. We refer to this failure mode as vector search dilution. Even when using hybrid dense+sparse retrieval, we observed this firsthand in a deployed Wyoming Department of Transportation corpus, where scaling from 54 to 1,128 documents (88,907 chunks) reduced accuracy from 75% to below 40%. To address this dilution, we propose MASDR-RAG ( Multi-Agent Scoped Domain Retrieval for RAG) and evaluate it on 200 expert-validated queries across five LLM backbones, six corpora, and two index stacks. Our results indicate that domain scoping using organizational metadata is the key fix, significantly improving P@10 from 0.77 to 0.86 (p < 0.05). Furthermore, our investigation of multi-agent orchestration revealed that a high degree of configuration dependence results --creating what we call the precision-faithfulness paradox. Based on these varied outcomes, our practical recommendation is simple: scope first, then perform a single synthesis call, reserving full multi-agent orchestration for genuinely multi-domain corpora paired with native-tool-call backbones. Code and Data will be made public upon acceptance.

Long-document understanding requires models to find and combine evidence across many pages, layouts, tables, figures, and charts. Existing retrieval-augmented systems usually select evidence from a static index before generation, while recent agentic systems add multi-turn tool use but often rely on frozen proprietary backbones whose behavior is set by prompts. We present DocAtlas, a system that treats long-document understanding as a mutable-state information-seeking process. We instantiate DocAtlas as a mutable document harness: an external environment that determines what document information is searched, read, stored, reviewed, and shown to the model at each step. Given a document and question, the harness exposes search, reading, note-taking, and review tools, maintains a hierarchical tree and note store, and updates both as the agent records evidence. DocAtlas combines self-improving retrieval, selective evidence access, and active working memory under a fixed context budget. The same harness supports inference-time use with large VLMs and end-to-end reinforcement learning for compact VLM agents. With GPT-5.4, DocAtlas reaches 71.4\% on MMLongBench-Doc, exceeding the human-expert reference of 65.8\%. A Qwen3.5-4B VLM trained with end-to-end RL in the DocAtlas environment reaches 63.7\%, compared with a 54.4\% direct-input baseline, showing that mutable document-harness design can improve compact document agents by a large margin.

As LLM agents are deployed in long-horizon sessions, context accumulation drives up inference costs. Existing approaches utilize text pruning or dynamic memory eviction to minimize token footprints; however, their unconstrained sequence mutations alter layouts, introducing prefix mismatches and cache invalidation. This reveals a critical trade-off between text sparsity and prompt cache continuity. To address this, we present TokenPilot, a dual-granularity context management framework. Globally, Ingestion-Aware Compaction acts as a framework harness to stabilize prompt prefixes and eliminate open-world environmental noise at the ingestion gate. Locally, Lifecycle-Aware Eviction monitors the ongoing residual utility of context segments, enforcing a conservative batch-turn schedule to offload content segments only when task relevance expires. Experiments on PinchBench and Claw-Eval under both isolated and continuous modes demonstrate that TokenPilot reduces costs by 61% and 56% in isolated mode, and 61% and 87% in continuous mode, while maintaining competitive performance compared to prior systems. TokenPilot has been integrated into LightMem2 at https://github.com/zjunlp/LightMem2.

Text-based person retrieval under a sim-to-real gap (synthetic training data, a real-image gallery) is usually tackled with costly fine-tuned cross-encoders. We ask whether a frozen-encoder system can compete. We present SCOUT, which casts cross-modal retrieval as prediction in embedding space. A trainable predictor maps the patch tokens of a frozen video encoder into the embedding space of a frozen text encoder under a bidirectional InfoNCE objective, and no encoder is fine-tuned in the base model. The video encoder is V-JEPA, the text encoder is EmbeddingGemma, and the predictor is initialized from a Qwen3.5-0.8B decoder. We make three findings. First, the best frozen text encoder is simply the one whose geometry best matches the video features. A training-free alignment score ranks three candidate text encoders in the same order as their retrieval accuracy on our held-out split (Spearman ρ= 1.0); a fourth, LLM-based encoder shows the rule is metric-dependent, holding for a neighborhood-overlap score (ρ= 0.8) but not for a linear probe (ρ= -0.2). Second, two precision-targeted levers, parameter-efficient ExPLoRA adaptation of the video encoder and a training-free attribute-decomposed reranker built on a vision-language model, improve the top-rank precision that otherwise limits the frozen system, adding 2.2 points of leaderboard R@1. Third, a local-versus-public calibration study explains which interventions transfer to the real domain. On AI City Challenge 2026 Track 4 the full retrieve-fuse-rerank system reaches 84.25 mAP@10 on the final leaderboard, while a single frozen model submitted alone reaches 60.63. Our trained components cost about 95 GPU-hours. CMP, the dataset authors' fine-tuned cross-encoder that trains for sixteen GPU-days, is one fusion member of the full system, not an alternative. Code and annotations: https://github.com/abtraore/SCOUT-ECCV

Large Language Model (LLM) agents now take part in organizational work, where many authors record decisions across documents over months. Because a revised decision arrives as a new document rather than an edit, answering a question requires knowing which version held at a given time. However, most memory systems compress the record at write time. By distilling each document into facts, notes or graph edges, these methods fix what can be answered before any question is asked. To address this, we propose Mem++, a non-destructive memory framework shifting from write-time distillation to read-time selection. Mem++ stores every document whole with its date and author, and it calls no generative model at write time. At read time, it retrieves only documents dated up to the time a question asks about and fuses lexical and semantic rankings. Unlike systems that overwrite older versions, Mem++ keeps them and leaves the choice to the answering model. Evaluations on the organizational benchmark OrgMemBench demonstrate that Mem++ surpasses the strongest memory system baseline by 8.0 to 13.1 points across two answering models. With gpt-4.1-mini, it also achieves the best overall score, 2.6 points above RAG. In addition, Mem++ achieves the best average LLM-judge score on LoCoMo and ranks second on LongMemEval-S, behind only its entity-graph variant. Code for benchmark evaluation is available at https://github.com/AIDAChip-Inc/mem-plus-plus.

Many computational factors limit broader deployment of large language models. In this paper, we focus on a memory bottleneck imposed by the key-value (KV) cache, a computational shortcut that requires storing previous KV pairs during decoding. While existing KV cache methods approach this problem by pruning or evicting large swaths of relatively less important KV pairs to dramatically reduce the memory footprint of the cache, they can have limited success in tasks that require recollecting a majority of previous tokens. To alleviate this issue, we propose LESS, a simple integration of a (nearly free) constant sized cache with eviction-based cache methods, such that all tokens can be queried at later decoding steps. Its ability to retain information throughout time shows merit on a variety of tasks where we demonstrate LESS can help reduce the performance gap from caching everything, sometimes even matching it, all while being efficient.

The vast portion of workloads employing LLMs involves answering questions grounded on PDF or scan content. We introduce the Arctic-TILT achieving accuracy on par with models 1000times its size on these use cases. It can be fine-tuned and deployed on a single 24GB GPU, lowering operational costs while processing Visually Rich Documents with up to 400k tokens. The model establishes state-of-the-art results on seven diverse Document Understanding benchmarks, as well as provides reliable confidence scores and quick inference, which are essential for processing files in large-scale or time-sensitive enterprise environments.

As large language models (LLMs) take on more complex tasks, their inputs incorporate longer contexts to respond to questions that require domain knowledge or user-specific conversational histories. Yet, using long contexts poses a challenge for responsive LLM systems, as nothing can be generated until all the contexts are fetched to and processed by the LLM. Existing systems optimize only the computation delay in context processing (e.g., by caching intermediate key-value features of the text context) but often cause longer network delays in context fetching (e.g., key-value features consume orders of magnitude larger bandwidth than the text context). This paper presents CacheGen to minimize the delays in fetching and processing contexts for LLMs. CacheGen reduces the bandwidth needed for transmitting long contexts' key-value (KV) features through a novel encoder that compresses KV features into more compact bitstream representations. The encoder combines adaptive quantization with a tailored arithmetic coder, taking advantage of the KV features' distributional properties, such as locality across tokens. Furthermore, CacheGen minimizes the total delay in fetching and processing a context by using a controller that determines when to load the context as compressed KV features or raw text and picks the appropriate compression level if loaded as KV features. We test CacheGen on three models of various sizes and three datasets of different context lengths. Compared to recent methods that handle long contexts, CacheGen reduces bandwidth usage by 3.7-4.3x and the total delay in fetching and processing contexts by 2.7-3x while maintaining similar LLM performance on various tasks as loading the text contexts.

Efficient use of GPU memory is essential for high throughput LLM inference. Prior systems reserved memory for the KV-cache ahead-of-time, resulting in wasted capacity due to internal fragmentation. Inspired by OS-based virtual memory systems, vLLM proposed PagedAttention to enable dynamic memory allocation for KV-cache. This approach eliminates fragmentation, enabling high-throughput LLM serving with larger batch sizes. However, to be able to allocate physical memory dynamically, PagedAttention changes the layout of KV-cache from contiguous virtual memory to non-contiguous virtual memory. This change requires attention kernels to be rewritten to support paging, and serving framework to implement a memory manager. Thus, the PagedAttention model leads to software complexity, portability issues, redundancy and inefficiency. In this paper, we propose vAttention for dynamic KV-cache memory management. In contrast to PagedAttention, vAttention retains KV-cache in contiguous virtual memory and leverages low-level system support for demand paging, that already exists, to enable on-demand physical memory allocation. Thus, vAttention unburdens the attention kernel developer from having to explicitly support paging and avoids re-implementation of memory management in the serving framework. We show that vAttention enables seamless dynamic memory management for unchanged implementations of various attention kernels. vAttention also generates tokens up to 1.97x faster than vLLM, while processing input prompts up to 3.92x and 1.45x faster than the PagedAttention variants of FlashAttention and FlashInfer.

Diffusion-based Large Language Models(DLLMs) enable parallel generation via Semi-Autoregressive (SAR) decoding in text generation. However, current methods suffer from severe operator-level redundancy: they recompute the entire sequence during denoising steps, ignoring that the prefix and masked suffix remain invariant within a block. We propose LaCache, a training-free acceleration framework that alleviates this redundancy through lossless caching and mixed precision. Specifically, LaCache employs Lossless State Memoization (LSM) by caching three types of intermediate results: (i) EmbedCache for embedding outputs, (ii) RoPECache for token-wise pre-attention states, and (iii) FACache for the online softmax statistics within FlashAttention. These caches allow the model to skip redundant computation on unchanged tokens without altering the output. To further alleviate memory-bandwidth bottlenecks, LaCache inegrates a per-group FP8 quantization strategy for FFN layers, tailored to step-dependent activation distributions across the diffusion process. Experiments demonstrate that LaCache alone achieves approximately 1.3X end-to-end speedup over vanilla DLLM. When combined with existing acceleration methods, LaCache reaches up to 40.2X end-to-end speedup while maintaining comparable task accuracy.