Daily Papers

Large language models (LLMs) are central to modern natural language processing, delivering exceptional performance in various tasks. However, their intensive computational and memory requirements present challenges, especially for devices with limited DRAM capacity. This paper tackles the challenge of efficiently running LLMs that exceed the available DRAM capacity by storing the model parameters on flash memory but bringing them on demand to DRAM. Our method involves constructing an inference cost model that harmonizes with the flash memory behavior, guiding us to optimize in two critical areas: reducing the volume of data transferred from flash and reading data in larger, more contiguous chunks. Within this flash memory-informed framework, we introduce two principal techniques. First, "windowing'" strategically reduces data transfer by reusing previously activated neurons, and second, "row-column bundling", tailored to the sequential data access strengths of flash memory, increases the size of data chunks read from flash memory. These methods collectively enable running models up to twice the size of the available DRAM, with a 4-5x and 20-25x increase in inference speed compared to naive loading approaches in CPU and GPU, respectively. Our integration of sparsity awareness, context-adaptive loading, and a hardware-oriented design paves the way for effective inference of LLMs on devices with limited memory.

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.

Context-augmented generation (CAG) techniques, including RAG and ICL, require the efficient combination of multiple contexts to generate responses to user queries. Directly inputting these contexts as a sequence introduces a considerable computational burden by re-encoding the combined selection of contexts for every request. To address this, we explore the promising potential of parallel encoding to independently pre-compute and cache each context's KV states. This approach enables the direct loading of cached states during inference while accommodating more contexts through position reuse across contexts. However, due to misalignments in attention distribution, directly applying parallel encoding results in a significant performance drop. To enable effective and efficient CAG, we propose Adaptive Parallel Encoding (APE), which brings shared prefix, attention temperature, and scaling factor to align the distribution of parallel encoding with sequential encoding. Results on RAG and ICL tasks demonstrate that APE can preserve 98% and 93% sequential encoding performance using the same inputs while outperforming parallel encoding by 3.6% and 7.9%, respectively. It also scales to many-shot CAG, effectively encoding hundreds of contexts in parallel. Efficiency evaluation shows that APE can achieve an end-to-end 4.5times speedup by reducing 28times prefilling time for a 128K-length context.

Retrieval-Augmented Generation (RAG) has emerged as the predominant paradigm for grounding Large Language Model outputs in factual knowledge, effectively mitigating hallucinations. However, conventional RAG systems operate under a "retrieve-always" assumption, querying vector databases for every input regardless of query complexity. This static approach incurs substantial computational overhead and inference latency, particularly problematic for high-throughput production deployments. We introduce L-RAG (Lazy Retrieval-Augmented Generation), an adaptive framework that implements hierarchical context management through entropy-based gating. L-RAG employs a two-tier architecture: queries are first processed with a compact document summary, and expensive chunk retrieval is triggered only when the model's predictive entropy exceeds a calibrated threshold, signaling genuine uncertainty. Through experiments on SQuAD 2.0 (N=500) using the Phi-2 model, we demonstrate that L-RAG provides a tunable accuracy-efficiency trade-off: at a conservative threshold (tau=0.5), L-RAG achieves 78.2% accuracy, matching Standard RAG (77.8%), with 8% retrieval reduction; at a balanced threshold (tau=1.0), retrieval reduction increases to 26% with modest accuracy trade-off (76.0%). Latency analysis shows that L-RAG saves 80-210ms per query when retrieval latency exceeds 500ms. Analysis of entropy distributions reveals statistically significant separation (p < 0.001) between correct predictions (H=1.72) and errors (H=2.20), validating entropy as a reliable uncertainty signal. L-RAG offers a practical, training-free approach toward more efficient RAG deployment, providing system architects with a configurable knob to balance accuracy and throughput requirements.

As large language models scale to longer contexts, loading the growing KV cache during attention computation becomes a critical bottleneck. Previous work has shown that attention computation is dominated by a small subset of tokens. This motivates block sparse attention methods that partition the KV cache into fixed-size blocks and selectively compute attention over those blocks exhibiting high importance. However, these methods assign a uniform block size across all attention heads, implicitly assuming homogeneous behavior throughout the model. Our analysis reveals that this assumption is flawed: attention heads exhibit widely varying sensitivity to block granularity, and uniformity leads to suboptimal accuracy. We present AB-Sparse, a training-free algorithm-system co-designed framework that improves accuracy while preserving throughput. AB-Sparse introduces lightweight adaptive block size allocation across attention heads to improve accuracy. To compensate for the additional memory overhead, it further employs lossless block centroid quantization. In addition, custom GPU kernels are developed to support efficient execution with variable block sizes. Evaluation results demonstrate that AB-Sparse achieves an accuracy improvement of up to 5.43% over existing block sparse attention baselines without throughput overhead.

Safety-aligned language models often exhibit fragile and imbalanced safety mechanisms, increasing the likelihood of generating unsafe content. In addition, incorporating new knowledge through editing techniques to language models can further compromise safety. To address these issues, we propose SafeInfer, a context-adaptive, decoding-time safety alignment strategy for generating safe responses to user queries. SafeInfer comprises two phases: the safety amplification phase, which employs safe demonstration examples to adjust the model's hidden states and increase the likelihood of safer outputs, and the safety-guided decoding phase, which influences token selection based on safety-optimized distributions, ensuring the generated content complies with ethical guidelines. Further, we present HarmEval, a novel benchmark for extensive safety evaluations, designed to address potential misuse scenarios in accordance with the policies of leading AI tech giants.

Long-horizon search agents must manage a rapidly growing working context as they reason, call tools, and observe information. Naively accumulating all intermediate content can overwhelm the agent, increasing costs and the risk of errors. We propose that effective context management should be adaptive: parts of the agent's trajectory are maintained at different levels of detail depending on their current relevance to the task. To operationalize this principle, we introduce Context-ReAct, a general agentic paradigm for elastic context orchestration that integrates reasoning, context management, and tool use in a unified loop. Context-ReAct provides five atomic operations: Skip, Compress, Rollback, Snippet and Delete, which allow the agent to dynamically reshape its working context, preserving important evidence, summarizing resolved information, discarding unhelpful branches, and controlling context size. We prove that the Compress operator is expressively complete, while the other specialized operators provide efficiency and fidelity guarantees that reduce generation cost and hallucination risk. Building on this paradigm, we develop LongSeeker, a long-horizon search agent fine-tuned from Qwen3-30B-A3B on 10k synthesized trajectories. Across four representative search benchmarks, LongSeeker achieves 61.5% on BrowseComp and 62.5% on BrowseComp-ZH, substantially outperforming Tongyi DeepResearch (43.2% and 46.7%) and AgentFold (36.2% and 47.3%). These results highlight the potential of adaptive context management, showing that agents can achieve more reliable and efficient long-horizon reasoning by actively shaping their working memory.

Context retrieval systems for LLM inference face a critical challenge: high retrieval latency creates a fundamental tension between waiting for complete context (poor time-to-first-token) and proceeding without it (reduced quality). Streaming context incrementally--overlapping retrieval with inference--can mitigate this latency, but doing so with concurrent requests introduces new challenges: requests contend for GPU compute and memory, and scheduling must adapt to dynamic context arrivals. We present Stream2LLM, a streaming-aware LLM serving system for concurrent prefill-decode disaggregated deployments. Stream2LLM introduces adaptive scheduling and preemption for two distinct retrieval patterns: append-mode (progressive context accumulation) and update-mode (iterative refinement with cache invalidation). It decouples scheduling decisions from resource acquisition, enabling flexible preemption strategies guided by hardware-specific cost models, and uses longest common prefix matching to minimize redundant computation when input changes dynamically. To evaluate Stream2LLM, we collect two large-scale, real-world streaming workloads based on web crawling and approximate nearest neighbor search. Our evaluation demonstrates that streaming architecture delivers up to 11x TTFT improvements, with cost-aware scheduling providing critical benefits under memory pressure, all while maintaining throughput parity with non-streaming baselines. Code: https://github.com/rajveerb/stream2llm/tree/mlsys_artifact

In-context Learning enables training-free adaptation via demonstrations but remains highly sensitive to example selection and formatting. In unified multimodal models spanning understanding and generation, this sensitivity is exacerbated by cross-modal interference and varying cognitive demands. Consequently, In-context Learning efficacy is often non-monotonic and highly task-dependent. To diagnose these behaviors, we introduce a six-level capability-oriented taxonomy that categorizes the functional role of demonstrations from basic perception to high-order discernment. Guided by this cognitive framework, we construct UniICL-760K, a large-scale corpus featuring curated 8-shot In-context Learning episodes across 15 subtasks, alongside UniICL-Bench for rigorous, controlled evaluation. As an architectural intervention to stabilize few-shot adaptation, we propose the Context-Adaptive Prototype Modulator, a lightweight, plug-and-play module. Evaluations on UniICL-Bench show that our approach yields highly competitive unified results, outperforming larger-parameter multimodal large language model baselines on most understanding In-context Learning tasks. Data and code will be available soon at https://github.com/xuyicheng-zju/UniICL.

Retrieval-augmented generation (RAG) improves language-model responses by grounding generation in external passages, which comes with overhead: retrieved context lengthens the prompt, increasing prefill work, KV-cache footprint, memory traffic, latency, and energy. Context compression offers a natural remedy by pruning retrieved text before generation. However, state-of-the-art context-compression methods are typically used with a fixed compression budget, or with the rate selected offline and then applied at inference time. This static view ignores both workload variation and the live state of the edge device. On an edge SoC, compression is not free: the compressor itself runs on the same SoC and consumes latency and energy that can offset any generation savings. This paper proposes a vision for telemetry-informed adaptive compression in edge RAG, grounded in experimental evidence. We characterize the compression tradeoff on the NVIDIA Jetson AGX Thor using Llama and Qwen generators, Natural Questions and HotpotQA datasets, and LLMLingua-2 compression. Our measurements show that generation dominates the RAG budget for larger models, reaching roughly 90% of per-query latency and 91% of GPU energy for 7B-8B generators. Exploring the impact of the compression rate reveals an adaptive operating region: mild compression can miss energy opportunities, and overly aggressive compression can hurt inference quality. Intermediate compression can reduce GPU energy by up to 53.2%, and SoC energy by up to 48.2%, with negligible quality loss. We argue for runtime policies that dynamically manage compression, guided by workload features and edge telemetry.

Recent large language model applications, such as Retrieval-Augmented Generation and chatbots, have led to an increased need to process longer input contexts. However, this requirement is hampered by inherent limitations. Architecturally, models are constrained by a context window defined during training. Additionally, processing extensive texts requires substantial GPU memory. We propose a novel approach, Finch, to compress the input context by leveraging the pre-trained model weights of the self-attention. Given a prompt and a long text, Finch iteratively identifies the most relevant Key (K) and Value (V) pairs over chunks of the text conditioned on the prompt. Only such pairs are stored in the KV cache, which, within the space constrained by the context window, ultimately contains a compressed version of the long text. Our proposal enables models to consume large inputs even with high compression (up to 93x) while preserving semantic integrity without the need for fine-tuning.

Long-context large language models (LC LLMs) combined with retrieval-augmented generation (RAG) hold strong potential for complex multi-hop and large-document tasks. However, existing RAG systems often suffer from imprecise retrieval, incomplete context coverage under constrained windows, and fragmented information from suboptimal context construction. We introduce Multi-scale Adaptive Context RAG (MacRAG), a hierarchical RAG framework that compresses and partitions documents into coarse-to-fine granularities, then adaptively merges relevant contexts through real-time chunk- and document-level expansions. By initiating with finest-level retrieval and progressively incorporating broader, higher-level context, MacRAG constructs effective query-specific long contexts, optimizing both precision and coverage. Evaluations on challenging LongBench expansions of HotpotQA, 2WikiMultihopQA, and Musique confirm MacRAG consistently surpasses baseline RAG pipelines in single- and multi-step generation using Llama-3.1-8B, Gemini-1.5-pro, and GPT-4o. Our results establish MacRAG as an efficient, scalable solution for real-world long-context, multi-hop reasoning. Our code is available at https://github.com/Leezekun/MacRAG.

Retrieval-augmented generation (RAG) enhances large language models (LLMs) with external knowledge but incurs significant inference costs due to lengthy retrieved contexts. While context compression mitigates this issue, existing methods apply fixed compression rates, over-compressing simple queries or under-compressing complex ones. We propose Adaptive Context Compression for RAG (ACC-RAG), a framework that dynamically adjusts compression rates based on input complexity, optimizing inference efficiency without sacrificing accuracy. ACC-RAG combines a hierarchical compressor (for multi-granular embeddings) with a context selector to retain minimal sufficient information, akin to human skimming. Evaluated on Wikipedia and five QA datasets, ACC-RAG outperforms fixed-rate methods and matches/unlocks over 4 times faster inference versus standard RAG while maintaining or improving accuracy.

In-context learning is fundamental to modern Large Language Models (LLMs); however, prevailing architectures impose a rigid and fixed contextual structure by assigning linear or constant positional indices. Drawing on Cognitive Load Theory (CLT), we argue that this uninformative structure increases extraneous cognitive load, consuming finite working memory capacity that should be allocated to deep reasoning and attention allocation. To address this, we propose RePo, a novel mechanism that reduces extraneous load via context re-positioning. Unlike standard approaches, RePo utilizes a differentiable module, f_φ, to assign token positions that capture contextual dependencies, rather than replying on pre-defined integer range. By continually pre-training on the OLMo-2 1B backbone, we demonstrate that RePo significantly enhances performance on tasks involving noisy contexts, structured data, and longer context length, while maintaining competitive performance on general short-context tasks. Detailed analysis reveals that RePo successfully allocate higher attention to distant but relevant information, assign positions in dense and non-linear space, and capture the intrinsic structure of the input context. Our code is available at https://github.com/SakanaAI/repo.

Long-form video generation presents a dual challenge: models must capture long-range dependencies while preventing the error accumulation inherent in autoregressive decoding. To address these challenges, we make two contributions. First, for dynamic context modeling, we propose MemoryPack, a learnable context-retrieval mechanism that leverages both textual and image information as global guidance to jointly model short- and long-term dependencies, achieving minute-level temporal consistency. This design scales gracefully with video length, preserves computational efficiency, and maintains linear complexity. Second, to mitigate error accumulation, we introduce Direct Forcing, an efficient single-step approximating strategy that improves training-inference alignment and thereby curtails error propagation during inference. Together, MemoryPack and Direct Forcing substantially enhance the context consistency and reliability of long-form video generation, advancing the practical usability of autoregressive video models.

We introduce Context Tuning, a simple and effective method to significantly enhance few-shot adaptation of language models (LLMs) without fine-tuning model parameters. While prompt-based adaptation techniques have demonstrated the effectiveness of lightweight adaptation methods for large language models (LLMs), they typically initialize a trainable prompt or prefix with irrelevant tokens for the task at hand. In contrast, Context Tuning initializes the trainable prompt or prefix with task-specific demonstration examples, leveraging the model's inherent In-Context Learning (ICL) ability to extract relevant information for improved few-shot learning performance. Extensive evaluations on benchmarks such as CrossFit, UnifiedQA, MMLU, BIG-Bench Hard, and ARC demonstrate that Context Tuning outperforms traditional prompt-based adaptation methods and achieves competitive accuracy to Test-Time Training with significantly higher training efficiency.

Recent video models increasingly support generation, reference conditioning, and editing within a single model, yet typically expose them as separate operations over fixed inputs. Practical creation unfolds across multiple shots, requiring one model to generate from text, follow a reference, or edit source footage while maintaining shared history. We formalize this setting as interactive multi-shot video creation (IMVC) and introduce ContextMaster, a unified model with a role-aware context representation for these operations. An interactive model must retain access to an expanding history without allowing the context read cost at each denoising step to grow. ContextMaster combines reusable clean context states with fixed budget sparse context routing and uses ConstraintSink to keep task constraints visible. To address the dual challenges of sparse context access and inference with few denoising steps, we propose a two-stage privileged context distillation framework, which transfers full context behavior from a dense teacher through consistency distillation and then refines deployment rollouts with distribution matching. Experiments on the three primitive tasks demonstrate improved task fulfillment and consistency across shots over specialized baselines. User studies further validate flexibly composed workflows, while the model reaches 16 FPS on a single GPU.

Recent advances in interactive video generation have shown promising results, yet existing approaches struggle with scene-consistent memory capabilities in long video generation due to limited use of historical context. In this work, we propose Context-as-Memory, which utilizes historical context as memory for video generation. It includes two simple yet effective designs: (1) storing context in frame format without additional post-processing; (2) conditioning by concatenating context and frames to be predicted along the frame dimension at the input, requiring no external control modules. Furthermore, considering the enormous computational overhead of incorporating all historical context, we propose the Memory Retrieval module to select truly relevant context frames by determining FOV (Field of View) overlap between camera poses, which significantly reduces the number of candidate frames without substantial information loss. Experiments demonstrate that Context-as-Memory achieves superior memory capabilities in interactive long video generation compared to SOTAs, even generalizing effectively to open-domain scenarios not seen during training. The link of our project page is https://context-as-memory.github.io/.

Recent approaches to real-time long video generation typically employ streaming tuning strategies, attempting to train a long-context student using a short-context (memoryless) teacher. In these frameworks, the student performs long rollouts but receives supervision from a teacher limited to short 5-second windows. This structural discrepancy creates a critical student-teacher mismatch: the teacher's inability to access long-term history prevents it from guiding the student on global temporal dependencies, effectively capping the student's context length. To resolve this, we propose Context Forcing, a novel framework that trains a long-context student via a long-context teacher. By ensuring the teacher is aware of the full generation history, we eliminate the supervision mismatch, enabling the robust training of models capable of long-term consistency. To make this computationally feasible for extreme durations (e.g., 2 minutes), we introduce a context management system that transforms the linearly growing context into a Slow-Fast Memory architecture, significantly reducing visual redundancy. Extensive results demonstrate that our method enables effective context lengths exceeding 20 seconds -- 2 to 10 times longer than state-of-the-art methods like LongLive and Infinite-RoPE. By leveraging this extended context, Context Forcing preserves superior consistency across long durations, surpassing state-of-the-art baselines on various long video evaluation metrics.

LLM agents increasingly face long-horizon tasks such as web search and deep research in real-world applications, where accumulated context can cause long-context degradation and reasoning failures. Prior work mitigates this through context management with agent-side context control or fixed strategies such as summarization, which require training the agent itself for adaptation - making it impractical for closed-source agents and ignoring that different agents may require different strategies. We introduce Adaptive Context Management (AdaCoM), which trains an external LLM to manage the context of a frozen agent through flexible modification actions and end-to-end reinforcement learning. Across diverse agents on web search and deep research benchmarks, AdaCoM substantially improves performance by preserving task constraints and progress while pruning stale content. The learned strategies reveal a Fidelity-Reliability Trade-off: agents with higher vanilla ReAct performance benefit from higher-fidelity context preservation, whereas lower-performing agents require more aggressive compression to stay within a reliable reasoning regime. Transfer experiments show that AdaCoM generalizes most effectively across agents with similar capability (measured by vanilla ReAct performance), suggesting a practical path toward reusable context managers for agent systems.

The learning and deployment of long-LLMs remains a challenging problem despite recent progresses. In this work, we argue that the long-LLMs are not a necessity to solve long-context tasks, as common long-context tasks are short-context solvable, i.e. they can be solved by purely working with oracle short-contexts within the long-context tasks' inputs. On top of this argument, we propose a framework called LC-Boost (Long-Context Bootstrapper), which enables a short-LLM to address the long-context tasks in a bootstrapping manner. In our framework, the short-LLM prompts itself to reason for two critical decisions: 1) how to access to the appropriate part of context within the input, 2) how to make effective use of the accessed context. By adaptively accessing and utilizing the context based on the presented tasks, LC-Boost can serve as a general framework to handle diversified long-context processing problems. We comprehensively evaluate different types of tasks from popular long-context benchmarks, where LC-Boost is able to achieve a substantially improved performance with a much smaller consumption of resource.

Long-horizon LLM agents act on an append-only interaction history that is re-sent to the model at every step, so the context and its cost grow with steps until the sessions exceed the context window. Existing methods manage the context through context requirement prediction, relying on additional model calls, heuristic rules, or trained policies. However, these predictive approaches introduce runtime overhead, invalidate prefix caches, and permanently discard content with no guarantee of recovery. To overcome these limitations, we introduce ReFold: a training-free rendering layer that preserves the underlying interaction history while compressing only the model's rendered context. It removes two kinds of inter-turn redundancy without an auxiliary predictor: content an earlier turn already displayed, replaced by a stub, and turns the agent itself reports finished, folded into a one-line note. Both operators use chunked rendering, rewriting the cached prefix once every few steps rather than at every step. Every removal is strictly reversible, a wrong removal costs one restore from the history rather than permanent content loss. Because it operates at the rendering layer, ReFold is plug-and-play across standard ReAct-style harnesses. Evaluations across five long-horizon benchmarks and two frontier LLMs demonstrate that ReFold reduces token consumption by up to 2.5x and halves the KV-cache memory per session without degrading task success rates. Under capped context budgets, it avoids up to 92% of forced compactions. Under concurrent serving workloads, it reduces request queuing delays by up to 100%, accelerating inference by up to 1.7x, while cutting inference costs by up to 3.4x.

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.

Large language model (LLM) applications such as agents and domain-specific reasoning increasingly rely on context adaptation -- modifying inputs with instructions, strategies, or evidence, rather than weight updates. Prior approaches improve usability but often suffer from brevity bias, which drops domain insights for concise summaries, and from context collapse, where iterative rewriting erodes details over time. Building on the adaptive memory introduced by Dynamic Cheatsheet, we introduce ACE (Agentic Context Engineering), a framework that treats contexts as evolving playbooks that accumulate, refine, and organize strategies through a modular process of generation, reflection, and curation. ACE prevents collapse with structured, incremental updates that preserve detailed knowledge and scale with long-context models. Across agent and domain-specific benchmarks, ACE optimizes contexts both offline (e.g., system prompts) and online (e.g., agent memory), consistently outperforming strong baselines: +10.6% on agents and +8.6% on finance, while significantly reducing adaptation latency and rollout cost. Notably, ACE could adapt effectively without labeled supervision and instead by leveraging natural execution feedback. On the AppWorld leaderboard, ACE matches the top-ranked production-level agent on the overall average and surpasses it on the harder test-challenge split, despite using a smaller open-source model. These results show that comprehensive, evolving contexts enable scalable, efficient, and self-improving LLM systems with low overhead.

Long-context inference remains challenging for large language models due to attention dilution and out-of-distribution degradation. Context selection mitigates this limitation by attending to a subset of key-value cache entries, yet most methods allocate a fixed context budget throughout decoding despite highly non-uniform token-level contextual demands. To address this issue, we propose Uncertainty-Triggered Adaptive Context Allocation (UT-ACA), an inference-time framework that dynamically adjusts the context window based on token-wise uncertainty. UT-ACA learns an uncertainty detector that combines semantic embeddings with logit-based confidence while accounting for uncertainty accumulation across decoding steps. When insufficient evidence is indicated, UT-ACA selectively rolls back, expands the context window, and regenerates the token with additional support. Experiments show that UT-ACA substantially reduces average context usage while preserving generation quality in long-context settings.

Most existing large language models (LLMs) are expensive to adapt after deployment, especially when a task requires newly produced information or niche domain knowledge. Recent work has shown that, by manipulating and optimizing their context, LLMs can be tailored to downstream tasks without updating their weights. However, most existing methods remain closed-loop, relying solely on the model's intrinsic knowledge. In this paper, we equip these context optimizers with Wikipedia search and browser tools for active information seeking. We show that naively adding these tools to a standard sequential context optimization pipeline can actually degrade performance compared to baselines. However, when paired with a search-based training procedure that maintains and prunes multiple candidate contexts, active information seeking delivers consistent and substantial gains. We demonstrate these improvements across diverse domains, including low-resource translation (Flores+), health scenarios (HealthBench), and reasoning-heavy tasks (LiveCodeBench and Humanity's Last Exam). Furthermore, our method proves to be data-efficient, robust across different hyperparameters, and capable of generating effective textual contexts that generalize well across different models.

Extending the context length (i.e., the maximum supported sequence length) of LLMs is of paramount significance. To facilitate long context training of LLMs, sequence parallelism has emerged as an essential technique, which scatters each input sequence across multiple devices and necessitates communication to process the sequence. In essence, existing sequence parallelism methods assume homogeneous sequence lengths (i.e., all input sequences are equal in length) and therefore leverages a single, static scattering strategy for all input sequences. However, in reality, the sequence lengths in LLM training corpora exhibit substantial variability, often following a long-tail distribution, which leads to workload heterogeneity. In this paper, we show that employing a single, static strategy results in inefficiency and resource under-utilization, highlighting the need for adaptive approaches to handle the heterogeneous workloads across sequences. To address this, we propose a heterogeneity-adaptive sequence parallelism method. For each training step, our approach captures the variability in sequence lengths and assigns the optimal combination of scattering strategies based on workload characteristics. We model this problem as a linear programming optimization and design an efficient and effective solver to find the optimal solution. Furthermore, we implement our method in a high-performance system that supports adaptive parallelization in distributed LLM training. Experimental results demonstrate that our system outperforms state-of-the-art training frameworks by up to 1.98x.

Retrieval-augmented generation improves the factual accuracy of Large Language Models (LLMs) by incorporating external context, but often suffers from irrelevant retrieved content that hinders effectiveness. Context compression addresses this issue by filtering out irrelevant information from context before LLM generation. However, existing methods struggle to adaptively adjust compression rates for different context, maintain low latency and integrate information across multiple documents. To overcome these limitations, We introduce AttnComp, an adaptive, efficient and context-aware compression framework. By leveraging the attention mechanism of LLMs to identify relevant information, AttnComp employs a Top-P compression algorithm to retain the minimal set of documents whose cumulative attention weights exceeds a predefined threshold. In addition to compression, AttnComp estimates response confidence by assessing the overall relevance of the retrieved content, enabling users to gauge response reliability. Experiments demonstrate that AttnComp outperforms existing compression methods and uncompressed baselines, achieving higher accuracy with substantial compression rates and lower latency.

The performance of Large Language Models (LLMs) is fundamentally determined by the contextual information provided during inference. This survey introduces Context Engineering, a formal discipline that transcends simple prompt design to encompass the systematic optimization of information payloads for LLMs. We present a comprehensive taxonomy decomposing Context Engineering into its foundational components and the sophisticated implementations that integrate them into intelligent systems. We first examine the foundational components: context retrieval and generation, context processing and context management. We then explore how these components are architecturally integrated to create sophisticated system implementations: retrieval-augmented generation (RAG), memory systems and tool-integrated reasoning, and multi-agent systems. Through this systematic analysis of over 1300 research papers, our survey not only establishes a technical roadmap for the field but also reveals a critical research gap: a fundamental asymmetry exists between model capabilities. While current models, augmented by advanced context engineering, demonstrate remarkable proficiency in understanding complex contexts, they exhibit pronounced limitations in generating equally sophisticated, long-form outputs. Addressing this gap is a defining priority for future research. Ultimately, this survey provides a unified framework for both researchers and engineers advancing context-aware AI.

The autoregressive video diffusion model has recently gained considerable research interest due to its causal modeling and iterative denoising. In this work, we identify that the multi-head self-attention in these models under-utilizes historical frames: approximately 25% heads attend almost exclusively to the current frame, and discarding their KV caches incurs only minor performance degradation. Building upon this, we propose Dummy Forcing, a simple yet effective method to control context accessibility across different heads. Specifically, the proposed heterogeneous memory allocation reduces head-wise context redundancy, accompanied by dynamic head programming to adaptively classify head types. Moreover, we develop a context packing technique to achieve more aggressive cache compression. Without additional training, our Dummy Forcing delivers up to 2.0x speedup over the baseline, supporting video generation at 24.3 FPS with less than 0.5% quality drop. Project page is available at https://csguoh.github.io/project/DummyForcing/.

Retrieval-augmented language models (RALMs) improve performance by accessing long-tail and up-to-date knowledge from external data stores, but are challenging to build. Existing approaches require either expensive retrieval-specific modifications to LM pre-training or use post-hoc integration of the data store that leads to suboptimal performance. We introduce Retrieval-Augmented Dual Instruction Tuning (RA-DIT), a lightweight fine-tuning methodology that provides a third option by retrofitting any LLM with retrieval capabilities. Our approach operates in two distinct fine-tuning steps: (1) one updates a pre-trained LM to better use retrieved information, while (2) the other updates the retriever to return more relevant results, as preferred by the LM. By fine-tuning over tasks that require both knowledge utilization and contextual awareness, we demonstrate that each stage yields significant performance improvements, and using both leads to additional gains. Our best model, RA-DIT 65B, achieves state-of-the-art performance across a range of knowledge-intensive zero- and few-shot learning benchmarks, significantly outperforming existing in-context RALM approaches by up to +8.9% in 0-shot setting and +1.4% in 5-shot setting on average.

Large Language Models (LLMs) have demonstrated exceptional capabilities across diverse tasks. However, their deployment in long context scenarios remains hindered by computational inefficiency and information redundancy. Context compression methods address these challenges by significantly reducing input length and eliminating redundancy. We propose COMI, a coarse-to-fine adaptive context compression framework that jointly optimizes for semantic relevance and diversity under high compression rates. We introduce Marginal Information Gain (MIG), a metric defined as the relevance of a unit to the input query minus its semantic redundancy with other units, guiding the compression process to prioritize information that is both relevant and low redundant. The framework operates in two stages: (1) Coarse-Grained Group Reallocation, where the context is partitioned into groups and dynamically assigned compression rates based on inter-group MIG, ensuring compression budgets align with information value distribution; and (2) Fine-Grained Token Merging, where tokens within each group are fused via an intra-group MIG-based weighting mechanism, thereby preserving key semantics while avoiding the accumulation of redundancy. Extensive experiments across question-answering (e.g., NaturalQuestions, 2WikiMQA, HotpotQA and NarrativeQA), summarization (e.g., MultiNews) with various backbones (e.g., LLaMA-2-7B, Qwen2-7B) show that COMI outperforms existing baselines by a large margin, e.g., approximately 25-point Exact Match (EM) improvement under 32x compression constraint with Qwen2-7B on NaturalQuestions.

Large language models (LLMs) have revolutionized AI, but are constrained by limited context windows, hindering their utility in tasks like extended conversations and document analysis. To enable using context beyond limited context windows, we propose virtual context management, a technique drawing inspiration from hierarchical memory systems in traditional operating systems that provide the appearance of large memory resources through data movement between fast and slow memory. Using this technique, we introduce MemGPT (Memory-GPT), a system that intelligently manages different memory tiers in order to effectively provide extended context within the LLM's limited context window, and utilizes interrupts to manage control flow between itself and the user. We evaluate our OS-inspired design in two domains where the limited context windows of modern LLMs severely handicaps their performance: document analysis, where MemGPT is able to analyze large documents that far exceed the underlying LLM's context window, and multi-session chat, where MemGPT can create conversational agents that remember, reflect, and evolve dynamically through long-term interactions with their users. We release MemGPT code and data for our experiments at https://memgpt.ai.

Transformer-based long context generative models power emerging AI applications like hour-long video understanding and project-level coding agent. Deploying long context transformers (e.g., 100K to 10M tokens) is prohibitively expensive compared to short context (e.g., 4K tokens) model variants. Reducing the cost of long-context transformers is becoming a pressing research and engineering challenge starting from the year of 2024. This work describes a concurrent programming framework for quantitatively analyzing the efficiency challenges in serving multiple long-context requests under limited size of GPU high-bandwidth memory (HBM) regime. We give a detailed analysis of how all additional computational costs, compared to 4K context, trace back to one single source: the large size of the KV cache. We use a 34B GPT-3.5 level model of 50K context on A100 NVLink as a running example, and describe how its large KV cache causes four types of deployment challenges: (1) prefilling long inputs takes much longer compute time and GPU memory than short inputs; (2) after prefilling, the large KV cache residing on the GPU HBM substantially restricts the number of concurrent users being served; (3) during decoding, repeatedly reading the KV cache from HBM to SM largely increases latency; (4) when KV cache memory overflows, swapping it from HBM to DDR causes significant context switching latency. We use this framework to analyze existing works and identify possibilities of combining them to build end-to-end systems. Overall, this work offers a foundational framework for analyzing long context transformer deployment and identifies directions towards reducing the inference cost of 1M context to be as cheap as 4K.

In-context learning (ICL) allows a language model to improve its problem-solving capability when provided with suitable information in context. Since the choice of in-context information can be determined based on the problem itself, in-context learning is analogous to human learning from teachers in a classroom. Recent works (Didolkar et al., 2024a; 2024b) show that ICL performance can be improved by leveraging a frontier large language model's (LLM) ability to predict required skills to solve a problem, popularly referred to as an LLM's metacognition, and using the recommended skills to construct necessary in-context examples. While this skill-based strategy boosts ICL performance in larger models, its gains on small language models (SLMs) have been minimal, highlighting a performance gap in ICL capabilities. We investigate this gap and show that skill-based prompting can hurt SLM performance on easy questions by introducing unnecessary information, akin to cognitive overload. To address this, we introduce AdaptMI, an adaptive approach to selecting skill-based in-context Math Instructions for SLMs. Inspired by cognitive load theory from human pedagogy, our method only introduces skill-based examples when the model performs poorly. We further propose AdaptMI+, which adds examples targeted to the specific skills missing from the model's responses. On 5-shot evaluations across popular math benchmarks and five SLMs (1B--7B; Qwen, Llama), AdaptMI+ improves accuracy by up to 6% over naive skill-based strategies.

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.

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.

Data selection is a key component of efficient instruction tuning for large language models, as recent work has shown that data quality often matters more than data quantity. Accordingly, prior studies have introduced various multi-dimensional heuristics to evaluate and filter instruction data. However, most existing methods rely on static task-agnostic and model-agnostic weighting schemes, which overlook the varying requirements of specific downstream tasks and the differing pre-existing capabilities of models. In this paper, we propose a framework for learning multi-indicator weights that jointly adapts data selection to both the downstream task and the specific model. Our method identifies optimal weight configurations without full-scale fine-tuning by utilizing in-context learning (ICL) signals on compact tiny-validation sets. These signals serve as efficient performance proxies that ensure high-fidelity evaluation at minimal computational cost. Experiments across multiple benchmarks and model families, including Mistral, Qwen, and Llama, show that the approach achieves performance comparable to or exceeding full-dataset tuning while using only 30\% of the training samples on GSM8K. Furthermore, our analysis reveals a trade-off between semantic diversity and logical complexity in reasoning tasks, highlighting the necessity of joint task-model adaptation.

Large language models have shown remarkable performance across a wide range of language tasks, owing to their exceptional capabilities in context modeling. The most commonly used method of context modeling is full self-attention, as seen in standard decoder-only Transformers. Although powerful, this method can be inefficient for long sequences and may overlook inherent input structures. To address these problems, an alternative approach is parallel context encoding, which splits the context into sub-pieces and encodes them parallelly. Because parallel patterns are not encountered during training, naively applying parallel encoding leads to performance degradation. However, the underlying reasons and potential mitigations are unclear. In this work, we provide a detailed analysis of this issue and identify that unusually high attention entropy can be a key factor. Furthermore, we adopt two straightforward methods to reduce attention entropy by incorporating attention sinks and selective mechanisms. Experiments on various tasks reveal that these methods effectively lower irregular attention entropy and narrow performance gaps. We hope this study can illuminate ways to enhance context modeling mechanisms.

We introduce Context Language Models (CLMs), language models that natively manage their own context. We implement this by treating the context as a file and allowing the model to make unrestricted updates to this file. This allows the model to learn what is most important to maintain in context, and naturally extends to multi-agent systems where multiple agent contexts coexist as files. Building CLMs zero-shot with existing models outperforms SOTA context management strategies across a variety of tasks: 11.4% higher accuracy with 21.5% fewer FLOPs on BrowseComp-Plus, 5% higher scores with 59% fewer FLOPs on 12-hour EdgeBench, and 65% greater improvement with the same compute on a 24-hour multi-repository agent-swarm task. Moreover, by shifting context management from external harness control to intrinsic model behavior, CLMs naturally enable both in-context and parametric learning of context-management strategies. We show that CLMs can be steered with natural-language instructions evolved through a standard skill-optimization loop, improving held-out accuracy by up to 35.9 points on a context-management task while reducing compute. We also introduce an online reinforcement learning method for CLMs, improving Qwen3.5-9B performance on BrowseComp-Plus by 47.6% while using 12% fewer FLOPs. Finally, we co-design Suffix Cache Reuse for CLM serving, further reducing server-side compute by 35% relative to standard SGLang at matched performance.

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.

Recent advancements in Large Language Model (LLM)-based frameworks have extended their capabilities to complex real-world applications, such as interactive web navigation. These systems, driven by user commands, navigate web browsers to complete tasks through multi-turn dialogues, offering both innovative opportunities and significant challenges. Despite the introduction of benchmarks for conversational web navigation, a detailed understanding of the key contextual components that influence the performance of these agents remains elusive. This study aims to fill this gap by analyzing the various contextual elements crucial to the functioning of web navigation agents. We investigate the optimization of context management, focusing on the influence of interaction history and web page representation. Our work highlights improved agent performance across out-of-distribution scenarios, including unseen websites, categories, and geographic locations through effective context management. These findings provide insights into the design and optimization of LLM-based agents, enabling more accurate and effective web navigation in real-world applications.

Large language models can solve complex problems across a variety of subjects, such as mathematics and coding. For such complex analytical tasks, models with a long context perform better. Serving models with long context is currently mostly done on powerful systems with multiple GPUs, which are too expensive for most users. An alternative is local inference, or serving a model on a user's own machine, which removes per-token costs and rate limits, besides keeping sensitive data private and giving users full control over the models they run. The common understanding, however, is that local LLMs need compression (e.g., quantization) to run on a single GPU. In this paper, we set out to study the limitations that prevent serving long-context models for single users, besides the capacity issue. Rather than characterizing a single model on one GPU, we analyze how different models behave across several GPU generations. This lets us uncover not just individual bottlenecks but broader trends in where inefficiencies arise. Our analysis confirms that, as expected, self-attention dominates the runtime for long-context inference. More unexpectedly, our evaluation reveals that the current structure of self-attention leaves key GPU resources significantly underutilized, and the newest GPU is often the least well served. Many of these findings are fundamentally data management problems, from paging the KV cache to improving cache utilization and synchronization in the GPU, which is why we bring this study to the Data Systems community. To engage this community, we present our findings as a series of experimental observations and formulate recommendations for future work.

In-context learning (ICL) approaches typically leverage prompting to condition decoder-only language model generation on reference information. Just-in-time processing of a context is inefficient due to the quadratic cost of self-attention operations, and caching is desirable. However, caching transformer states can easily require almost as much space as the model parameters. When the right context isn't known in advance, caching ICL can be challenging. This work addresses these limitations by introducing models that, inspired by the encoder-decoder architecture, use cross-attention to condition generation on reference text without the prompt. More precisely, we leverage pre-trained decoder-only models and only train a small number of added layers. We use Question-Answering (QA) as a testbed to evaluate the ability of our models to perform conditional generation and observe that they outperform ICL, are comparable to fine-tuned prompted LLMs, and drastically reduce the space footprint relative to standard KV caching by two orders of magnitude.

Interactive video generation has significant potential for scene simulation and video creation. However, existing methods often struggle with maintaining scene consistency during long video generation under dynamic camera control due to limited contextual information. To address this challenge, we propose MemCam, a memory-augmented interactive video generation approach that treats previously generated frames as external memory and leverages them as contextual conditioning to achieve controllable camera viewpoints with high scene consistency. To enable longer and more relevant context, we design a context compression module that encodes memory frames into compact representations and employs co-visibility-based selection to dynamically retrieve the most relevant historical frames, thereby reducing computational overhead while enriching contextual information. Experiments on interactive video generation tasks show that MemCam significantly outperforms existing baseline methods as well as open-source state-of-the-art approaches in terms of scene consistency, particularly in long video scenarios with large camera rotations.

Long-context capabilities are essential for a wide range of applications, including document and video understanding, in-context learning, and inference-time scaling, all of which require models to process and reason over long sequences of text and multimodal data. In this work, we introduce a efficient training recipe for building ultra-long context LLMs from aligned instruct model, pushing the boundaries of context lengths from 128K to 1M, 2M, and 4M tokens. Our approach leverages efficient continued pretraining strategies to extend the context window and employs effective instruction tuning to maintain the instruction-following and reasoning abilities. Our UltraLong-8B, built on Llama3.1-Instruct with our recipe, achieves state-of-the-art performance across a diverse set of long-context benchmarks. Importantly, models trained with our approach maintain competitive performance on standard benchmarks, demonstrating balanced improvements for both long and short context tasks. We further provide an in-depth analysis of key design choices, highlighting the impacts of scaling strategies and data composition. Our findings establish a robust framework for efficiently scaling context lengths while preserving general model capabilities. We release all model weights at: https://ultralong.github.io/.

Although existing model editing methods perform well in recalling exact edit facts, they often struggle in complex scenarios that require deeper semantic understanding rather than mere knowledge regurgitation. Leveraging the strong contextual reasoning abilities of large language models (LLMs), in-context learning (ICL) becomes a promising editing method by comprehending edit information through context encoding. However, this method is constrained by the limited context window of LLMs, leading to degraded performance and efficiency as the number of edits increases. To overcome this limitation, we propose InComeS, a flexible framework that enhances LLMs' ability to process editing contexts through explicit compression and selection mechanisms. Specifically, InComeS compresses each editing context into the key-value (KV) cache of a special gist token, enabling efficient handling of multiple edits without being restricted by the model's context window. Furthermore, specialized cross-attention modules are added to dynamically select the most relevant information from the gist pools, enabling adaptive and effective utilization of edit information. We conduct experiments on diverse model editing benchmarks with various editing formats, and the results demonstrate the effectiveness and efficiency of our method.

Large language model (LLM) agents often struggle in long-context interactions. As the agent accumulates more interaction history, context management approaches such as sliding window and prompt compression may omit earlier structured information that later steps rely on. Recent retrieval-based memory systems surface relevant content but still overlook the causal and logical structure needed for multi-step reasoning. We introduce ContextWeaver, a selective and dependency-structured memory framework that organizes an agent's interaction trace into a graph of reasoning steps and selects the relevant context for future actions. Unlike prior context management approaches, ContextWeaver supports: (1) dependency-based construction and traversal that link each step to the earlier steps it relies on; (2) compact dependency summarization that condenses root-to-step reasoning paths into reusable units; and (3) a lightweight validation layer that incorporates execution feedback. On the SWE-Bench Verified and Lite benchmarks, ContextWeaver improves performance over a sliding-window baseline in pass@1, while reducing reasoning steps and token usage. Our observations suggest that modeling logical dependencies provides a stable and scalable memory mechanism for LLM agents that use tools.

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.

Edge deployment of large language models (LLMs) increasingly relies on libraries of lightweight LoRA adapters, yet GPU/DRAM can keep only a small resident subset at a time. Serving a request through a non-resident adapter requires paging its weights from storage, incurring measurable latency. This creates a two-timescale online control problem: on a slow timescale, the system selects which adapters remain resident in fast memory, while on a fast timescale it routes each request to an adapter whose context-dependent utility is unknown a priori. The two decisions are tightly coupled: the cache determines the cost of exploration, and the router determines which adapters receive informative feedback. We formulate this joint caching-and-routing problem as a two-timescale contextual bandit and propose POLAR (Paging and Online Learning for Adapter Routing). POLAR pairs a cache-aware LinUCB router with an epoch-based cache controller. We study two variants. A fixed-epoch version provides a robust baseline with worst-case regret guarantees under arbitrary contexts. An epoch-doubling version, POLAR+, adds forced exploration and improved cache optimization to achieve mathcal{O}(dNT+KT) sublinear regret under stochastic regularity and cacheability conditions, where N is the adapter count, K the cache size, d the context dimension, and T the horizon. The routing term matches the standard contextual-bandit rate up to logarithmic factors, showing that the memory hierarchy does not fundamentally slow routing learning. Experiments using 15 real LoRA adapters for Qwen2.5-7B together with measured GPU paging latencies show that adaptive cache control substantially outperforms non-adaptive baselines and exhibits scaling trends consistent with the theory.

The success of expanded context windows in Large Language Models (LLMs) has driven increased use of broader context in retrieval-augmented generation. We investigate the use of LLMs for retrieval augmented question answering. While longer contexts make it easier to incorporate targeted knowledge, they introduce more irrelevant information that hinders the model's generation process and degrades its performance. To address the issue, we design an adaptive prompting strategy which involves splitting the retrieved information into smaller chunks and sequentially prompting a LLM to answer the question using each chunk. Adjusting the chunk size allows a trade-off between incorporating relevant information and reducing irrelevant information. Experimental results on three open-domain question answering datasets demonstrate that the adaptive strategy matches the performance of standard prompting while using fewer tokens. Our analysis reveals that when encountering insufficient information, the LLM often generates incorrect answers instead of declining to respond, which constitutes a major source of error. This finding highlights the need for further research into enhancing LLMs' ability to effectively decline requests when faced with inadequate information.

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.

Soft context compression reduces the computational workload of processing long contexts in LLMs by encoding long context into a smaller number of latent tokens. However, existing frameworks apply uniform compression ratios, failing to account for the extreme variance in natural language information density. While adopting a density-aware dynamic compression ratio seems intuitive, empirical investigations reveal that models struggle intrinsically with operations parameterized by input dependent, continuous structural hyperparameters. To resolve this pitfall, we introduce Semi-Dynamic Context Compression framework. Our approach features a Discrete Ratio Selector, which predicts a compression target based on intrinsic information density and quantizes it to a predefined set of discrete compression ratios. It is efficiently jointly trained with the compressor on synthetic data, with the summary lengths as a proxy to create labels for compression ratio prediction. Extensive evaluations confirm that our density-aware framework, utilizing mean pooling as the backbone, consistently outperforms static baselines, establishing a robust Pareto frontier for context compression techniques. Our code, data and model weights are available at https://github.com/yuyijiong/semi-dynamic-context-compress

Efficient long-context LLM deployment is stalled by a dichotomy between amortized compression, which struggles with out-of-distribution generalization, and Test-Time Training, which incurs prohibitive synthetic data costs and requires modifying model weights, creating stateful parameters that complicate concurrent serving. We propose Latent Context Compilation, a framework that fundamentally shifts context processing from adaptation to compilation. By utilizing a disposable LoRA module as a compiler, we distill long contexts into compact buffer tokens -- stateless, portable memory artifacts that are plug-and-play compatible with frozen base models. Crucially, we introduce a self-aligned optimization strategy that eliminates the need for synthetic context-relevant QA pairs. By regularizing context reconstruction task with context-agnostic random queries, we force compressed tokens to reside within the model's existing instruction-following manifold. Experiments with Llama-3.1-8B demonstrate that Latent Context Compilation preserves fine-grained details and reasoning capabilities where prior methods falter, effectively decoupling memory density from model parameters even at a 16x compression ratio.

The quadratic complexity of self-attention constrains Large Language Models (LLMs) in processing long contexts, a capability essential for many advanced applications. Context compression aims to alleviate this computational bottleneck while retaining critical semantic information. However, existing approaches often fall short: explicit methods may compromise local detail, whereas implicit methods can suffer from positional biases, information degradation, or an inability to capture long-range semantic dependencies. We propose AdmTree, a novel framework for adaptive, hierarchical context compression with a central focus on preserving high semantic fidelity while maintaining efficiency. AdmTree dynamically segments input based on information density, utilizing gist tokens to summarize variable-length segments as the leaves of a semantic binary tree. This structure, together with a lightweight aggregation mechanism and a frozen backbone LLM (thereby minimizing new trainable parameters), enables efficient hierarchical abstraction of the context. By preserving fine-grained details alongside global semantic coherence, mitigating positional bias, and dynamically adapting to content, AdmTree robustly retains the semantic information of long contexts.

Retrieval-augmented generation (RAG) and long-context language models (LCLMs) both address context limitations of LLMs in open-domain question answering (QA). However, optimal external context to retrieve remains an open problem: fixing the retrieval size risks either wasting tokens or omitting key evidence. Existing adaptive methods like Self-RAG and Self-Route rely on iterative LLM prompting and perform well on factoid QA, but struggle with aggregation QA, where the optimal context size is both unknown and variable. We present Adaptive-k retrieval, a simple and effective single-pass method that adaptively selects the number of passages based on the distribution of the similarity scores between the query and the candidate passages. It does not require model fine-tuning, extra LLM inferences or changes to existing retriever-reader pipelines. On both factoid and aggregation QA benchmarks, Adaptive-k matches or outperforms fixed-k baselines while using up to 10x fewer tokens than full-context input, yet still retrieves 70% of relevant passages. It improves accuracy across five LCLMs and two embedding models, highlighting that dynamically adjusting context size leads to more efficient and accurate QA.

Multimodal Agentic Retrieval (MAR) requires agents to solve complex information-seeking tasks by iteratively invoking external tools. Typical frameworks such as ReAct maintain raw multimodal inputs and the accumulating interaction history in a single, ever-growing context, leading to the context explosion problem. While existing methods alleviate this issue by compressing redundant text, effective strategies for managing token-intensive visual content remain largely underexplored. To address this gap, we first conduct a systematic empirical study of approximately 10,000 trajectories. The results show that as visual cues are progressively extracted through external tools and textualized into the context, raw images become increasingly redundant. Continued image retention is associated with higher output entropy and can even degrade task accuracy. Motivated by these findings, we propose MM-ContextFold, a training-free framework that loads raw images only when needed. It maintains a persistent, text-only main context for high-level planning and spawns ephemeral branch contexts for image-dependent subtasks. Within each branch, the agent loads the relevant images, completes the subtask, and folds the result back into the main context as a concise textual summary; the images and branch trace are then discarded. Experiments on seven MAR benchmarks across five backbone models show that MM-ContextFold improves average accuracy by 6.3 percentage points over ReAct while reducing the working context length by 27.5\%.

The Model Context Protocol (MCP) (MCP Community, 2025) has emerged as a widely used framework for enabling LLM-based agents to communicate with external tools and services. The original MCP implementation (Anthropic, 2024) relies on a Large Language Model (LLM) to decompose tasks and issue instructions to servers. In particular, the agents, models, and servers are stateless and do not have access to a global context. However, in tasks involving LLM-driven coordination, it is natural that a Shared Context Store (SCS) could improve the efficiency and coherence of multi-agent workflows by reducing redundancy and enabling knowledge transfer between servers. Thus, in this work, we design and assess the performance of a Context-Aware MCP (CA-MCP) that offloads execution logic to specialized MCP servers that read from and write to a shared context memory, allowing them to coordinate more autonomously in real time. In this design, context management serves as the central mechanism that maintains continuity across task executions by tracking intermediate states and shared variables, thereby enabling persistent collaboration among agents without repeated prompting. We present experiments showing that the CA-MCP can outperform the traditional MCP by reducing the number of LLM calls required for complex tasks and decreasing the frequency of response failures when task conditions are not satisfied. In particular, we conducted experiments on the TravelPlanner (Yang et al., 2024) and REALM-Bench (Geng & Chang, 2025) benchmark datasets and observed statistically significant results indicating the potential advantages of incorporating a shared context store via CA-MCP in LLM-driven multi-agent systems.

Retrieval-Augmented Generation (RAG) allows overcoming the limited knowledge of LLMs by extending the input with external information. As a consequence, the contextual inputs to the model become much longer which slows down decoding time directly translating to the time a user has to wait for an answer. We address this challenge by presenting COCOM, an effective context compression method, reducing long contexts to only a handful of Context Embeddings speeding up the generation time by a large margin. Our method allows for different compression rates trading off decoding time for answer quality. Compared to earlier methods, COCOM allows for handling multiple contexts more effectively, significantly reducing decoding time for long inputs. Our method demonstrates a speed-up of up to 5.69 times while achieving higher performance compared to existing efficient context compression methods.

Large Language Models (LLMs) are highly sensitive to their input contexts, motivating the development of automated context engineering. However, existing methods predominantly treat this as a global search problem, seeking a single context strategy that maximizes average performance across a dataset. This restrictive assumption overlooks the fact that different inputs often require distinct guidance, leaving substantial instance-level performance gains untapped. In this paper, we propose a paradigm shift by formulating context engineering as a recommendation problem. We introduce Neural Collaborative Context Engineering (NCCE), a framework that transitions optimization from a static global search to dynamic, instance-wise routing. NCCE first bootstraps a diverse catalog of anchor contexts and then employs a novel Context-CF Co-Evolution mechanism. This stage establishes a synergistic feedback loop: a lightweight Neural Collaborative Filtering (NCF) model learns instance-context preferences to guide the generation of specialized context variants, while the newly evaluated contexts continuously refine the NCF model's understanding of latent preferences. At inference time, the trained NCF model acts as a context router, dynamically assigning the most suitable context strategy to each unseen instance. Theoretical Proofs and comprehensive experiments demonstrate that by matching individual inputs with their optimal contexts, NCCE significantly improves task accuracy, highlighting the critical importance of personalization in LLM context engineering.

Conventional recommendation systems (RSs) are typically optimized to enhance performance metrics uniformly across all training samples, inadvertently overlooking the needs of diverse user populations. The performance disparity among various populations can harm the model's robustness to sub-populations due to the varying user properties. While large language models (LLMs) show promise in enhancing RS performance, their practical applicability is hindered by high costs, inference latency, and degraded performance on long user queries. To address these challenges, we propose a hybrid task allocation framework designed to promote social good by equitably serving all user groups. By adopting a two-phase approach, we promote a strategic assignment of tasks for efficient and responsible adaptation of LLMs. Our strategy works by first identifying the weak and inactive users that receive a suboptimal ranking performance by RSs. Next, we use an in-context learning approach for such users, wherein each user interaction history is contextualized as a distinct ranking task. We evaluate our hybrid framework by incorporating eight different recommendation algorithms and three different LLMs -- both open and close-sourced. Our results on three real-world datasets show a significant reduction in weak users and improved robustness to subpopulations without disproportionately escalating costs.

Large Language Models (LLMs) present a critical trade-off between inference quality and computational cost: larger models offer superior capabilities but incur significant latency, while smaller models are faster but less powerful. Existing serving strategies often employ fixed model scales or static two-stage speculative decoding, failing to dynamically adapt to the varying complexities of user requests or fluctuations in system performance. This paper introduces , a novel framework that reimagines LLM inference as an adaptive routing problem solved through multi-level speculative decoding. dynamically constructs and optimizes inference "paths" (chains of models) based on real-time feedback, addressing the limitations of static approaches. Our contributions are threefold: (1) An adaptive model chain scheduling mechanism that leverages performance profiling (execution times) and predictive similarity metrics (derived from token distribution divergence) to continuously select the optimal sequence of draft and verifier models, minimizing predicted latency per generated token. (2) A multi-level collaborative verification framework where intermediate models within the selected chain can validate speculative tokens, reducing the verification burden on the final, most powerful target model. (3) A synchronized state management system providing efficient, consistent KV cache handling across heterogeneous models in the chain, including precise, low-overhead rollbacks tailored for asynchronous batch processing inherent in multi-level speculation. Preliminary experiments demonstrate the validity of our method.

We introduce a dynamic benchmarking system for conversational agents that evaluates their performance through a single, simulated, and lengthy userleftrightarrowagent interaction. The interaction is a conversation between the user and agent, where multiple tasks are introduced and then undertaken concurrently. We context switch regularly to interleave the tasks, which constructs a realistic testing scenario in which we assess the Long-Term Memory, Continual Learning, and Information Integration capabilities of the agents. Results from both proprietary and open-source Large-Language Models show that LLMs in general perform well on single-task interactions, but they struggle on the same tasks when they are interleaved. Notably, short-context LLMs supplemented with an LTM system perform as well as or better than those with larger contexts. Our benchmark suggests that there are other challenges for LLMs responding to more natural interactions that contemporary benchmarks have heretofore not been able to capture.

Many-Shot In-Context Learning (ICL) has emerged as a promising paradigm, leveraging extensive examples to unlock the reasoning potential of Large Language Models (LLMs). However, existing methods typically rely on a predetermined, fixed number of shots. This static approach often fails to adapt to the varying difficulty of different queries, leading to either insufficient context or interference from noise. Furthermore, the prohibitive computational and memory costs of long contexts severely limit Many-Shot's feasibility. To address the above limitations, we propose AdapShot, which dynamically optimizes shot counts and leverages KV cache reuse for efficient inference. Specifically, we design a probe-based evaluation mechanism that utilizes output entropy to determine the optimal number of shots. To bypass the redundant prefilling computation during both the probing and inference phases, we incorporate a semantics-aware KV cache reuse strategy. Within this reuse strategy, to address positional encoding incompatibilities, we introduce a decoupling and re-encoding method that enables the flexible reordering of cached key-value pairs. Extensive experiments demonstrate that AdapShot achieves an average performance gain of around 10% and a 4.64x speedup compared to state-of-the-art DBSA.

As large language models increasingly gain popularity in real-world applications, processing extremely long contexts, often exceeding the model's pre-trained context limits, has emerged as a critical challenge. While existing approaches to efficient long-context processing show promise, recurrent compression-based methods struggle with information preservation, whereas random access approaches require substantial memory resources. We introduce REFORM, a novel inference framework that efficiently handles long contexts through a two-phase approach. First, it incrementally processes input chunks while maintaining a compressed KV cache, constructs cross-layer context embeddings, and utilizes early exit strategy for improved efficiency. Second, it identifies and gathers essential tokens via similarity matching and selectively recomputes the KV cache. Compared to baselines, REFORM achieves over 50% and 27% performance gains on RULER and BABILong respectively at 1M context length. It also outperforms baselines on Infinite-Bench and MM-NIAH, demonstrating flexibility across diverse tasks and domains. Additionally, REFORM reduces inference time by 30% and peak memory usage by 5%, achieving both efficiency and superior performance.

The rapid evolution of large language models (LLMs) has intensified the demand for effective personalization techniques that can adapt model behavior to individual user preferences. Despite the non-parametric methods utilizing the in-context learning ability of LLMs, recent parametric adaptation methods, including personalized parameter-efficient fine-tuning and reward modeling emerge. However, these methods face limitations in handling dynamic user patterns and high data sparsity scenarios, due to low adaptability and data efficiency. To address these challenges, we propose a fine-grained and instance-tailored steering framework that dynamically generates sample-level interference vectors from user data and injects them into the model's forward pass for personalized adaptation. Our approach introduces two key technical innovations: a fine-grained steering component that captures nuanced signals by hooking activations from attention and MLP layers, and an input-aware aggregation module that synthesizes these signals into contextually relevant enhancements. The method demonstrates high flexibility and data efficiency, excelling in fast-changing distribution and high data sparsity scenarios. In addition, the proposed method is orthogonal to existing methods and operates as a plug-in component compatible with different personalization techniques. Extensive experiments across diverse scenarios--including short-to-long text generation, and web function calling--validate the effectiveness and compatibility of our approach. Results show that our method significantly enhances personalization performance in fast-shifting environments while maintaining robustness across varying interaction modes and context lengths. Implementation is available at https://github.com/KounianhuaDu/Fints.

In this paper, we point out that the objective of the retrieval algorithms is to align with the LLM, which is similar to the objective of knowledge distillation in LLMs. We analyze the similarity in information focus between the distilled language model(DLM) and the original LLM from the perspective of information theory, and thus propose a novel paradigm that leverages a DLM as the retrieval algorithm. Based on the insight, we present SpeContext, an algorithm and system co-design for long-context reasoning. (1) At the algorithm level, SpeContext proposes lightweight retrieval head based on the head-level attention weights of DLM, achieving > 90% parameters reduction by pruning the redundancy. (2) At the system level, SpeContext designs an asynchronous prefetch dataflow via the elastic loading strategy, effectively overlapping KV cache retrieval with the LLM computation. (3) At the compilation level, SpeContext constructs the theoretical memory model and implements an adaptive memory management system to achieve acceleration by maximizing GPU memory utilization. We deploy and evaluate SpeContext in two resourceconstrained environments, cloud and edge. Extensive experiments show that, compared with the Huggingface framework, SpeContext achieves up to 24.89x throughput improvement in cloud and 10.06x speedup in edge with negligible accuracy loss, pushing the Pareto frontier of accuracy and throughput.

Modern object detectors achieve strong performance on standard benchmarks, yet their robustness to contextual variation remains insufficiently understood. Prior evaluations largely rely on aggregate metrics such as AP on uncontrolled distribution shifts, which can obscure how performance degrades under context change. We introduce ContextShift, a controlled benchmark that systematically manipulates object--context relationships while preserving object appearance. Built on COCO 2017, it isolates context as an independent variable through geometric transformations and synthetic and natural background substitutions, including a continuous compatibility axis based on normalized pointwise mutual information (NPMI). Across diverse detector architectures, we observe a consistent degradation pattern: false negatives increase by up to 227% and prediction volume decreases by up to 44%, while false positives remain stable or decline. This suppression behavior is not captured by aggregate metrics such as AP, which can mask substantial recall loss and changes in prediction dynamics. Further analysis suggests that degradation is driven less by reduced confidence than by a reduced formation of valid detection candidates. Moreover, performance along the statistical compatibility axis is non-monotonic, peaking at intermediate NPMI and degrading toward both extremes, indicating that statistical co-occurrence does not correlate linearly with effective visual context. Finally, we show that context-aware augmentation improves robustness: every augmented variant outperforms the dataset-only baseline on both original and manipulated test images, partially recovering performance lost to prediction-suppression failures by exposing models to object--context decoupling during training.

The quadratic complexity of the attention module makes it gradually become the bulk of compute in Transformer-based LLMs during generation. Moreover, the excessive key-value cache that arises when dealing with long inputs also brings severe issues on memory footprint and inference latency. In this work, we propose a plug-and-play approach that is able to incrementally compress the intermediate activation of a specified span of tokens into compact ones, thereby reducing both memory and computational cost when processing subsequent context. Experiments on both in-domain language modeling and zero-shot open-ended document generation demonstrate the advantage of our approach over sparse attention baselines in terms of fluency, n-gram matching, and semantic similarity. At last, we comprehensively profile the benefit of context compression on improving the system throughout. Code is available at https://github.com/DRSY/KV_Compression.

Retrieval-Augmented Generation (RAG) helps LLMs stay accurate, but feeding long documents into a prompt makes the model slow and expensive. This has motivated context compression, ranging from token pruning and summarization to embedding-based compression. While researchers have tried ''compressing'' these documents into smaller summaries or mathematical embeddings, there is a catch: the more you compress the data, the more the LLM struggles to understand it. To address this challenge, we propose ArcAligner (Adaptive recursive context *Aligner*), a lightweight module integrated into the language model layers to help the model better utilize highly compressed context representations for downstream generation. It uses an adaptive ''gating'' system that only adds extra processing power when the information is complex, keeping the system fast. Across knowledge-intensive QA benchmarks, ArcAligner consistently beats compression baselines at comparable compression rates, especially on multi-hop and long-tail settings. The source code is publicly available.

Augmenting LLMs with context leads to improved performance across many applications. Despite much research on Retrieval Augmented Generation (RAG) systems, an open question is whether errors arise because LLMs fail to utilize the context from retrieval or the context itself is insufficient to answer the query. To shed light on this, we develop a new notion of sufficient context, along with a way to classify instances that have enough information to answer the query. We then use sufficient context to analyze several models and datasets. By stratifying errors based on context sufficiency, we find that proprietary LLMs (Gemini, GPT, Claude) excel at answering queries when the context is sufficient, but often output incorrect answers instead of abstaining when the context is not. On the other hand, open-source LLMs (Llama, Mistral, Gemma) hallucinate or abstain often, even with sufficient context. We further categorize cases when the context is useful, and improves accuracy, even though it does not fully answer the query and the model errs without the context. Building on our findings, we explore ways to reduce hallucinations in RAG systems, including a new selective generation method that leverages sufficient context information for guided abstention. Our method improves the fraction of correct answers among times where the model responds by 2-10% for Gemini, GPT, and Gemma.

Large language models are increasingly deployed as research agents for deep search and long-horizon information seeking, yet their performance often degrades as interaction histories grow. This degradation, known as context rot, reflects a failure to maintain coherent and task-relevant internal states over extended reasoning horizons. Existing approaches primarily manage context through raw accumulation or passive summarization, treating it as a static artifact and allowing early errors or misplaced emphasis to persist. Motivated by this perspective, we propose ARC, which is the first framework to systematically formulate context management as an active, reflection-driven process that treats context as a dynamic internal reasoning state during execution. ARC operationalizes this view through reflection-driven monitoring and revision, allowing agents to actively reorganize their working context when misalignment or degradation is detected. Experiments on challenging long-horizon information-seeking benchmarks show that ARC consistently outperforms passive context compression methods, achieving up to an 11% absolute improvement in accuracy on BrowseComp-ZH with Qwen2.5-32B-Instruct.

As large language models (LLMs) evolve into autonomous agents for long-horizon information-seeking, managing finite context capacity has become a critical bottleneck. Existing context management methods typically commit to a single fixed strategy throughout the entire trajectory. Such static designs may work well in some states, but they cannot adapt as the usefulness and reliability of the accumulated context evolve during long-horizon search. To formalize this challenge, we introduce a probabilistic framework that characterizes long-horizon success through two complementary dimensions: search efficiency and terminal precision. Building on this perspective, we propose AgentSwing, a state-aware adaptive parallel context management routing framework. At each trigger point, AgentSwing expands multiple context-managed branches in parallel and uses lookahead routing to select the most promising continuation. Experiments across diverse benchmarks and agent backbones show that AgentSwing consistently outperforms strong static context management methods, often matching or exceeding their performance with up to 3times fewer interaction turns while also improving the ultimate performance ceiling of long-horizon web agents. Beyond the empirical gains, the proposed probabilistic framework provides a principled lens for analyzing and designing future context management strategies for long-horizon agents.

Multi-agent large language model (LLM) systems are increasingly adopted for complex language processing tasks that require communication and coordination among agents. However, these systems often suffer substantial overhead from repeated reprocessing of overlapping contexts across agents. In typical pipelines, once an agent receives a message from its predecessor, the full context-including prior turns-must be reprocessed from scratch, leading to inefficient processing. While key-value (KV) caching is an effective solution for avoiding redundant computation in single-agent settings where prefixes remain unchanged, it cannot be directly reused in multi-agent scenarios due to diverging prefixes introduced by agent-specific context extensions. We identify that the core challenge lies in the offset variance of KV-caches across agents. To address this, we propose KVCOMM, a training-free framework that enables efficient prefilling in multi-agent inference by reusing KV-caches and aligning cache offsets of overlapping contexts under diverse prefix contexts. KVCOMM estimates and adjusts KV-caches for shared content by referencing a pool of cached examples-termed anchors-that store observed cache deviations under varying prefixes. The anchor pool is maintained and updated online, allowing dynamic adaptation to distinct user requests and context structures. KVCOMM achieves over 70% reuse rate across diverse multi-agent workloads, including retrieval-augmented generation, math reasoning, and collaborative coding tasks, all without quality degradation. Particularly, when each fully-connected agent receives 1K input tokens with 512 prefix tokens and 512 output tokens under a five-agent setting, KVCOMM achieves up to 7.8x speedup compared to the standard prefill pipeline, reducing TTFT from ~430 ms to ~55 ms.

Recent techniques such as retrieval-augmented generation or chain-of-thought reasoning have led to longer contexts and increased inference costs. Context compression techniques can reduce these costs, but the most effective approaches require fine-tuning the target model or even modifying its architecture. This can degrade its general abilities when not used for this specific purpose. Here we explore an alternative approach: an encoder that compresses the context into continuous representations which replace token embeddings in decoder LLMs. First, we perform a systematic study of training strategies and architecture choices for the encoder. Our findings led to the design of an Adaptable text Representations Compressor, named ARC-Encoder, which outputs x-times fewer continuous representations (typically x!in!{4,8}) than text tokens. We evaluate ARC-Encoder across a variety of LLM usage scenarios, ranging from in-context learning to context window extension, on both instruct and base decoders. Results show that ARC-Encoder achieves state-of-the-art performance on several benchmarks while improving computational efficiency at inference. Finally, we demonstrate that our models can be adapted to multiple decoders simultaneously, allowing a single encoder to generalize across different decoder LLMs. This makes ARC-Encoder a flexible and efficient solution for portable encoders that work seamlessly with multiple LLMs. We release a training code at https://github.com/kyutai-labs/ARC-Encoder , fine-tuning dataset and pretrained models are available at https://huggingface.co/collections/kyutai/arc-encoders-68ee18787301407d60a57047 .

We present context parallelism for long-context large language model inference, which achieves near-linear scaling for long-context prefill latency with up to 128 H100 GPUs across 16 nodes. Particularly, our method achieves 1M context prefill with Llama3 405B model in 77s (93% parallelization efficiency, 63% FLOPS utilization) and 128K context prefill in 3.8s. We develop two lossless exact ring attention variants: pass-KV and pass-Q to cover a wide range of use cases with the state-of-the-art performance: full prefill, persistent KV prefill and decode. Benchmarks on H100 GPU hosts inter-connected with RDMA and TCP both show similar scalability for long-context prefill, demonstrating that our method scales well using common commercial data center with medium-to-low inter-host bandwidth.

Large Language Models (LLMs) have demonstrated impressive quality when applied to predictive tasks such as relevance ranking and semantic search. However, deployment of such LLMs remains prohibitively expensive for industry applications with strict latency and throughput requirements. In this work, we present lessons and efficiency insights from developing a purely text-based decoder-only Small Language Model (SLM) for a semantic search application at LinkedIn. Particularly, we discuss model compression techniques such as pruning that allow us to reduce the model size by up to 40% while maintaining the accuracy. Additionally, we present context compression techniques that allow us to reduce the input context length by up to 10x with minimal loss of accuracy. Finally, we present practical lessons from optimizing the serving infrastructure for deploying such a system on GPUs at scale, serving millions of requests per second. Taken together, this allows us to increase our system's throughput by 10x in a real-world deployment, while meeting our quality bar.

Training large language models (LLMs) is resource-intensive, and adapting them for diverse deployment scenarios with varying computational constraints remains challenging. While elastic architectures enable flexible model deployment and sparsely activated models allow input-adaptive computation, existing approaches treat these dimensions independently. Moreover, models catered towards on-device edge inference need to conform to the memory and compute limitations of the serving devices. In this paper, we introduce a unified framework that combines elastic structures with sparsely gated architectures to create models that adapt simultaneously to both deployment constraints and task requirements. Our approach employs a model backbone that conditions on both the context and target efficiency specifications, enabling fine-grained control over the accuracy-efficiency trade-off at inference time. The model learns to activate task-relevant parameters within elastically-nested sub-networks, allowing a single model to span multiple capacity points while maintaining input-adaptive routing. Through experiments we demonstrate that we can create a model that allows the flexibility to use 1,2,3,4 billion parameters while being more accurate than their dense counter-parts (2-5\% on knowledge-intensive benchmarks) and at par with their static versions while delivering similar latency metrics as dense models. Overall, we save on device disk space by sharing the model parameters, allow flexibility of serving based on DRAM and compute available while delivering more accurate results.

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.

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.

Few-shot in-context learning (ICL) enables pre-trained language models to perform a previously-unseen task without any gradient-based training by feeding a small number of training examples as part of the input. ICL incurs substantial computational, memory, and storage costs because it involves processing all of the training examples every time a prediction is made. Parameter-efficient fine-tuning (PEFT) (e.g. adapter modules, prompt tuning, sparse update methods, etc.) offers an alternative paradigm where a small set of parameters are trained to enable a model to perform the new task. In this paper, we rigorously compare few-shot ICL and PEFT and demonstrate that the latter offers better accuracy as well as dramatically lower computational costs. Along the way, we introduce a new PEFT method called (IA)^3 that scales activations by learned vectors, attaining stronger performance while only introducing a relatively tiny amount of new parameters. We also propose a simple recipe based on the T0 model called T-Few that can be applied to new tasks without task-specific tuning or modifications. We validate the effectiveness of T-Few on completely unseen tasks by applying it to the RAFT benchmark, attaining super-human performance for the first time and outperforming the state-of-the-art by 6% absolute. All of the code used in our experiments is publicly available.

Large Language Model (LLM) agents are increasingly deployed in environments that generate massive, dynamic contexts. However, a critical bottleneck remains: while agents have access to this context, their static prompts lack the mechanisms to manage it effectively, leading to recurring Corrective and Enhancement failures. To address this capability gap, we introduce SCOPE (Self-evolving Context Optimization via Prompt Evolution). SCOPE frames context management as an online optimization problem, synthesizing guidelines from execution traces to automatically evolve the agent's prompt. We propose a Dual-Stream mechanism that balances tactical specificity (resolving immediate errors) with strategic generality (evolving long-term principles). Furthermore, we introduce Perspective-Driven Exploration to maximize strategy coverage, increasing the likelihood that the agent has the correct strategy for any given task. Experiments on the HLE benchmark show that SCOPE improves task success rates from 14.23\% to 38.64\% without human intervention. We make our code publicly available at https://github.com/JarvisPei/SCOPE.

Retrieval-augmented generation (RAG) greatly benefits language models (LMs) by providing additional context for tasks such as document-based question answering (DBQA). Despite its potential, the power of RAG is highly dependent on its configuration, raising the question: What is the optimal RAG configuration? To answer this, we introduce the RAGGED framework to analyze and optimize RAG systems. On a set of representative DBQA tasks, we study two classic sparse and dense retrievers, and four top-performing LMs in encoder-decoder and decoder-only architectures. Through RAGGED, we uncover that different models suit substantially varied RAG setups. While encoder-decoder models monotonically improve with more documents, we find decoder-only models can only effectively use < 5 documents, despite often having a longer context window. RAGGED offers further insights into LMs' context utilization habits, where we find that encoder-decoder models rely more on contexts and are thus more sensitive to retrieval quality, while decoder-only models tend to rely on knowledge memorized during training.

Modern Large Language Models (LLMs) are increasingly trained to support very large context windows. We present Compactor, a training-free, query-agnostic KV compression strategy that uses approximate leverage scores to determine token importance. We show that Compactor can achieve the same performance as competing methods while retaining 20% fewer tokens in both synthetic and real-world context tasks, while being more task-robust. We further introduce a procedure for context-calibrated compression: inferring the maximum compression a given context supports before significant performance loss. Using context-calibrated compression, we show that Compactor achieves full KV performance on Longbench while reducing the KV memory burden by 68%, on average. To demonstrate the efficacy and generalizability of our approach, we apply Compactor to 27 synthetic and real-world tasks from RULER and Longbench, with models from both the Qwen 2.5 and Llama 3.1 families. Finally, we release compactor-vllm, an inference engine and suite of optimized Triton kernels designed to efficiently support the sparse, non-contiguous memory access patterns inherent to compressed KV caches. This work demonstrates that Compactor offers a practical, high-performance solution for alleviating the memory bottleneck in modern LLM deployment.

Long-context LLMs have enabled numerous downstream applications but also introduced significant challenges related to computational and memory efficiency. To address these challenges, optimizations for long-context inference have been developed, centered around the KV cache. However, existing benchmarks often evaluate in single-request, neglecting the full lifecycle of the KV cache in real-world use. This oversight is particularly critical, as KV cache reuse has become widely adopted in LLMs inference frameworks, such as vLLM and SGLang, as well as by LLM providers, including OpenAI, Microsoft, Google, and Anthropic. To address this gap, we introduce SCBench(SharedContextBench), a comprehensive benchmark for evaluating long-context methods from a KV cachecentric perspective: 1) KV cache generation, 2) KV cache compression, 3) KV cache retrieval, 4) KV cache loading. Specifically, SCBench uses test examples with shared context, ranging 12 tasks with two shared context modes, covering four categories of long-context capabilities: string retrieval, semantic retrieval, global information, and multi-task. With it, we provide an extensive KV cache-centric analysis of eight categories long-context solutions, including Gated Linear RNNs, Mamba-Attention hybrids, and efficient methods such as sparse attention, KV cache dropping, quantization, retrieval, loading, and prompt compression. The evaluation is conducted on 8 long-context LLMs. Our findings show that sub-O(n) memory methods suffer in multi-turn scenarios, while sparse encoding with O(n) memory and sub-O(n^2) pre-filling computation perform robustly. Dynamic sparsity yields more expressive KV caches than static patterns, and layer-level sparsity in hybrid architectures reduces memory usage with strong performance. Additionally, we identify attention distribution shift issues in long-generation scenarios. https://aka.ms/SCBench.

Drag-based image editing aims to modify visual content followed by user-specified drag operations. Despite existing methods having made notable progress, they still fail to fully exploit the contextual information in the reference image, including fine-grained texture details, leading to edits with limited coherence and fidelity. To address this challenge, we introduce ContextDrag, a new paradigm for drag-based editing that leverages the strong contextual modeling capability of editing models, such as FLUX-Kontext. By incorporating VAE-encoded features from the reference image, ContextDrag can leverage rich contextual cues and preserve fine-grained details, without the need for finetuning or inversion. Specifically, ContextDrag introduced a novel Context-preserving Token Injection (CTI) that injects noise-free reference features into their correct destination locations via a Latent-space Reverse Mapping (LRM) algorithm. This strategy enables precise drag control while preserving consistency in both semantics and texture details. Second, ContextDrag adopts a novel Position-Consistent Attention (PCA), which positional re-encodes the reference tokens and applies overlap-aware masking to eliminate interference from irrelevant reference features. Extensive experiments on DragBench-SR and DragBench-DR demonstrate that our approach surpasses all existing SOTA methods. Code will be publicly available.

Large language models (LLMs) achieved remarkable performance across various tasks. However, they face challenges in managing long documents and extended conversations, due to significantly increased computational requirements, both in memory and inference time, and potential context truncation when the input exceeds the LLM's fixed context length. This paper proposes a method called Selective Context that enhances the inference efficiency of LLMs by identifying and pruning redundancy in the input context to make the input more compact. We test our approach using common data sources requiring long context processing: arXiv papers, news articles, and long conversations, on tasks of summarisation, question answering, and response generation. Experimental results show that Selective Context significantly reduces memory cost and decreases generation latency while maintaining comparable performance compared to that achieved when full context is used. Specifically, we achieve a 50\% reduction in context cost, resulting in a 36\% reduction in inference memory usage and a 32\% reduction in inference time, while observing only a minor drop of .023 in BERTscore and .038 in faithfulness on four downstream applications, indicating that our method strikes a good balance between efficiency and performance.

We present CIFLEX (Contextual Instruction Flow for Sub-task Execution), which is a novel execution system for efficient sub-task handling in multi-turn interactions with a single on-device large language model (LLM). As LLMs become increasingly capable, a single model is expected to handle diverse sub-tasks that more effectively and comprehensively support answering user requests. Naive approach reprocesses the entire conversation context when switching between main and sub-tasks (e.g., query rewriting, summarization), incurring significant computational overhead. CIFLEX mitigates this overhead by reusing the key-value (KV) cache from the main task and injecting only task-specific instructions into isolated side paths. After sub-task execution, the model rolls back to the main path via cached context, thereby avoiding redundant prefill computation. To support sub-task selection, we also develop a hierarchical classification strategy tailored for small-scale models, decomposing multi-choice decisions into binary ones. Experiments show that CIFLEX significantly reduces computational costs without degrading task performance, enabling scalable and efficient multi-task dialogue on-device.

Context-heavy agents place unusual pressure on the key-value (KV) cache: long prefixes are reused across many short turns, while concurrency determines whether the serving system can keep GPUs utilized. We study 4-bit KV-cache compression for this setting, using TurboQuant-style rotation and codebook quantization as a quality anchor and vLLM FP8 KV caching as the deployment anchor. We report three contributions. First, we frame 4-bit KV caching around multi-round agent workloads where task quality, cache residency, and serving throughput must be measured jointly. Second, we describe the practical design choices needed to make the 4-bit path robust, including asymmetric K/V treatment, Walsh-Hadamard rotation, QJL removal, and block-scale variants. Third, we present serving optimizations on AMD GPUs, including optimized decode-attention kernels and UltraQuant, an FP4 approximation path that uses FP8 queries, FP4 KV tensors, UE8M0 group scales, and native scaled-MFMA support on CDNA4. On a long-context, multi-turn agentic workload, UltraQuant cuts P50 time-to-first-token by 3.47x in the cache-pressured late rounds (2.3x across all rounds) and raises output throughput by 1.63x over the FP8 KV baseline.

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.

Despite the successes of large language models (LLMs), they exhibit significant drawbacks, particularly when processing long contexts. Their inference cost scales quadratically with respect to sequence length, making it expensive for deployment in some real-world text processing applications, such as retrieval-augmented generation (RAG). Additionally, LLMs also exhibit the "distraction phenomenon," where irrelevant context in the prompt degrades output quality. To address these drawbacks, we propose a novel RAG prompting methodology, superposition prompting, which can be directly applied to pre-trained transformer-based LLMs without the need for fine-tuning. At a high level, superposition prompting allows the LLM to process input documents in parallel prompt paths, discarding paths once they are deemed irrelevant. We demonstrate the capability of our method to simultaneously enhance time efficiency across a variety of question-answering benchmarks using multiple pre-trained LLMs. Furthermore, our technique significantly improves accuracy when the retrieved context is large relative the context the model was trained on. For example, our approach facilitates an 93x reduction in compute time while improving accuracy by 43\% on the NaturalQuestions-Open dataset with the MPT-7B instruction-tuned model over naive RAG.

Recent advances demonstrate that multimodal large language models (MLLMs) exhibit strong multimodal in-context learning (ICL) capabilities, enabling them to adapt to novel vision-language tasks from a few contextual examples. However, existing ICL approaches face challenges in reconciling scalability with robustness across diverse tasks and noisy contextual examples: manually selecting examples produces clean contexts but is labor-intensive and task-specific, while similarity-based retrieval improves scalability but could introduce irrelevant or structurally inconsistent samples that degrade ICL performance. To address these limitations, we propose ContextNav, the first agentic framework that integrates the scalability of automated retrieval with the quality and adaptiveness of human-like curation, enabling noise-robust and dynamically optimized contextualization for multimodal ICL. ContextNav unifies context management and noise-robust contextualization within a closed-loop workflow driven by graph-based orchestration. Specifically, it builds a resource-aware multimodal embedding pipeline, maintains a retrievable vector database, and applies agentic retrieval and structural alignment to construct noise-resilient contexts. An Operational Grammar Graph (OGG) further supports adaptive workflow planning and optimization, enabling the agent to refine its operational strategies based on downstream ICL feedback. Experimental results demonstrate that ContextNav achieves state-of-the-art performance across various datasets, underscoring the promise of agentic workflows for advancing scalable and robust contextualization in multimodal ICL.

Users may strive to formulate an adequate textual query for their information need. Search engines assist the users by presenting query suggestions. To preserve the original search intent, suggestions should be context-aware and account for the previous queries issued by the user. Achieving context awareness is challenging due to data sparsity. We present a probabilistic suggestion model that is able to account for sequences of previous queries of arbitrary lengths. Our novel hierarchical recurrent encoder-decoder architecture allows the model to be sensitive to the order of queries in the context while avoiding data sparsity. Additionally, our model can suggest for rare, or long-tail, queries. The produced suggestions are synthetic and are sampled one word at a time, using computationally cheap decoding techniques. This is in contrast to current synthetic suggestion models relying upon machine learning pipelines and hand-engineered feature sets. Results show that it outperforms existing context-aware approaches in a next query prediction setting. In addition to query suggestion, our model is general enough to be used in a variety of other applications.

Modern LLMs are increasingly deployed in long-context applications such as retrieval-augmented generation, repository-level coding, and agentic workflows whose accumulated reasoning and tool traces routinely push the input an order of magnitude past the pretraining window, making zero-shot context extension the dominant deployment path for open-weight checkpoints. Most existing zero-shot methods fix a single rescaling factor up front, so an aggressive factor sacrifices short-context fidelity while a conservative one breaks down at long contexts. We propose Jet-Long, a tuning-free zero-shot method that pairs a local RoPE-faithful window with a long-range window whose rescaling factor adapts dynamically to the current sequence length, recovering the base model exactly at short inputs while extrapolating cleanly at long ones. An inclusion-exclusion attention merge and an on-the-fly RoPE correction rotation make the bifocal construction essentially free at inference; fused into a single CuTe kernel, long-context prefill reaches up to 1.39times FA2 throughput on H100 (approaching the Hopper-only FA4), and single-batch generation incurs le 4% overhead at every length. On Qwen3-1.7B/4B/8B up to 128K context, Jet-Long leads RULER by +4.79/+2.18/+2.03~pp over the strongest baseline at 1.7B/4B/8B, achieves the best overall accuracy on HELMET-RAG (a benchmark identified by HELMET as the most efficient predictor of downstream long-context performance) and attains the lowest PG-19 perplexity. Jet-Long also generalizes to hybrid attention architectures such as Jet-Nemotron for further long-context improvement without retraining, and remains hyperparameter-resilient for ease of deployment.

In-context learning (ICL) involves reasoning from given contextual examples. As more modalities comes, this procedure is becoming more challenging as the interleaved input modalities convolutes the understanding process. This is exemplified by the observation that multimodal models often struggle to effectively extrapolate from contextual examples to perform ICL. To address these challenges, we introduce MultiModal In-conteXt Tuning (M^2IXT), a lightweight module to enhance the ICL capabilities of multimodal unified models. The proposed M^2IXT module perceives an expandable context window to incorporate various labeled examples of multiple modalities (e.g., text, image, and coordinates). It can be prepended to various multimodal unified models (e.g., OFA, Unival, LLaVA) of different architectures and trained via a mixed-tasks strategy to enable rapid few-shot adaption on multiple tasks and datasets. When tuned on as little as 50K multimodal data, M^2IXT can boost the few-shot ICL performance significantly (e.g., 18\% relative increase for OFA), and obtained state-of-the-art results across an array of tasks including visual question answering, image captioning, visual grounding, and visual entailment, while being considerably small in terms of model parameters (e.g., sim20times smaller than Flamingo or MMICL), highlighting the flexibility and effectiveness of M^2IXT as a multimodal in-context learner.

Large Language Models cannot reliably acquire new knowledge post-deployment -- even when relevant text resources exist, models fail to transform them into actionable knowledge without retraining. Retrieval-Augmented Generation attempts to bridge this gap by surfacing relevant documents at inference time, yet similarity-based retrieval often fails to identify context that actually improves task performance. We introduce Evolutionary Context Search (ECS), an evolutionary method that searches context combinations using accuracy on a small development set, requiring only inference calls without weight updates. ECS moves beyond semantic similarity to discover non-obvious context pairings that significantly boost performance. Our empirical results show that ECS improves BackendBench by 27\% and τ-bench airline by 7\%. The evolved contexts are model-agnostic, as those evolved with Gemini-3-Flash transfer effectively to Claude Sonnet and DeepSeek. This suggests that ECS opens a path toward automated context discovery for skill acquisition -- an efficient alternative to manual prompt engineering or costly fine-tuning.

Scaling laws have transformed our understanding of large language models by linking upstream metrics like cross-entropy loss to design factors such as model size, training data, and compute. However, these conventional laws fail to capture downstream task performance, where context plays a critical role. In this work, we propose a straightforward, interpretable framework that jointly models downstream performance as a function of the training compute and the provided context. We empirically validate our framework by fitting it on the observed downstream performance of extended-context variants of Llama-2-7B and Llama-2-13B across 65,500 unique instances spanning three tasks: arithmetic reasoning, common sense reasoning, and machine translation. Our results demonstrate that our framework accurately models in-distribution downstream performance, generalizes across three orders of magnitude in training compute, and reliably extrapolates performance as the amount of context increases. These findings offer valuable insights into the interplay between training compute and context utilization, providing guidance for designing more efficient long-context LLMs for diverse downstream tasks. Our code is available at https://github.com/wang-research-lab/context-scaling.

Large language models (LLMs) increasingly receive information as streams of passages, conversations, and long-context workflows. While longer context windows expose more evidence, they do not ensure that useful information is preserved and reused. We study continual context consolidation: writing current context into model weights while limiting interference with previously consolidated information. We propose Self-Consolidating Language Models (SCoL), a post-training framework in which, given current context, an LLM learns to generate textual update instructions specifying which of its own Transformer layers should be updated. Because committed updates change the model that later generates future selections, we train SCoL with meta-reinforcement learning over an evolving model state. We instantiate SCoL with supervised QA rewards on SQuAD knowledge incorporation and intrinsic likelihood-based rewards for LongBench v2 long-context consolidation. Across both settings, SCoL improves acquisition and retention over prompting, summarization, batch test-time training, and sequential finetuning baselines. Analysis of learned selection patterns shows that SCoL encourages the LLM to generate sparse update locations that align with layers of high Fisher information, suggesting that the model learns to route plasticity toward loss-sensitive regions while limiting interference. Moreover, SCoL transfers from shorter meta-training streams to longer LongBench v2 streams at evaluation, suggesting that our framework supports scalable streaming consolidation.

In-context learning (ICL) allows large language models (LLMs) to adapt to new tasks directly from the given demonstrations without requiring gradient updates. While recent advances have expanded context windows to accommodate more demonstrations, this approach increases inference costs without necessarily improving performance. To mitigate these issues, We propose StreamAdapter, a novel approach that directly updates model parameters from context at test time, eliminating the need for explicit in-context demonstrations. StreamAdapter employs context mapping and weight absorption mechanisms to dynamically transform ICL demonstrations into parameter updates with minimal additional parameters. By reducing reliance on numerous in-context examples, StreamAdapter significantly reduce inference costs and allows for efficient inference with constant time complexity, regardless of demonstration count. Extensive experiments across diverse tasks and model architectures demonstrate that StreamAdapter achieves comparable or superior adaptation capability to ICL while requiring significantly fewer demonstrations. The superior task adaptation and context encoding capabilities of StreamAdapter on both language understanding and generation tasks provides a new perspective for adapting LLMs at test time using context, allowing for more efficient adaptation across scenarios and more cost-effective inference

Which forms of test-time compute improve the predictions of strong pretrained tabular foundation models (TFMs)? We systematically study this along three axes: adaptation, aggregation, and context construction. Our evaluation spans modern TFMs across the TabArena benchmark, supplemented by experiments on wide and large-scale tables from OpenML. For adaptation, we introduce DiagScale, a diagonal query-key similarity update. It trains only 0.003-0.03% of model parameters and achieves gains comparable to full fine-tuning across three independently pretrained backbones. For aggregation, both pool composition and selection strategy matter. TabPFN-3 already averages predictions from different preprocessing variants of the same data, and adding more such predictions yields diminishing returns. With a broader pool of 96 configurations, greedy selection reduces error by 2.4% relative to the default predictor, but uniform averaging increases error. For context construction, attention-guided retrieval improves TabPFN-3's predictions on some large tables and supports source pools beyond the full context memory limit. The context expansion methods we test yield no consistent improvement. Taken together, our results suggest that adaptation and selective aggregation yield consistent benchmark-level gains. The benefits of context construction depend more on the task and data regime. Adaptation and aggregation over the same backbone yield further gains when combined, but require substantially more computation than default inference. These trade-offs motivate choosing strategies according to the available computation budget. Code is available at https://github.com/kanghui-learning/test-time-compute-for-tabular-foundation-models.

Human interaction with the external world fundamentally involves the exchange of personal memory, whether with other individuals, websites, applications, or, in the future, AI agents. A significant portion of this interaction is redundant, requiring users to repeatedly provide the same information across different contexts. Existing solutions, such as browser-stored credentials, autofill mechanisms, and unified authentication systems, have aimed to mitigate this redundancy by serving as intermediaries that store and retrieve commonly used user data. The advent of large language models (LLMs) presents an opportunity to redefine memory management through an AI-native paradigm: SECOND ME. SECOND ME acts as an intelligent, persistent memory offload system that retains, organizes, and dynamically utilizes user-specific knowledge. By serving as an intermediary in user interactions, it can autonomously generate context-aware responses, prefill required information, and facilitate seamless communication with external systems, significantly reducing cognitive load and interaction friction. Unlike traditional memory storage solutions, SECOND ME extends beyond static data retention by leveraging LLM-based memory parameterization. This enables structured organization, contextual reasoning, and adaptive knowledge retrieval, facilitating a more systematic and intelligent approach to memory management. As AI-driven personal agents like SECOND ME become increasingly integrated into digital ecosystems, SECOND ME further represents a critical step toward augmenting human-world interaction with persistent, contextually aware, and self-optimizing memory systems. We have open-sourced the fully localizable deployment system at GitHub: https://github.com/Mindverse/Second-Me.

Long-context LLMs demand accurate inference at low latency, yet decoding becomes primarily constrained by KV cache as context grows. Prior pruning methods are largely context-agnostic: their token selection ignores step-wise relevance and local semantics, which undermines quality. Moreover, their irregular accesses and selection overheads yield only limited wall-clock speedups. To address this, we propose CHESS, an algorithm-system co-design KV-cache management system. Algorithmically, CHESS introduces a context-aware, hierarchical selection policy that dynamically reconstructs a coherent context for the current decoding. System-wise, coarse granularity selection eliminates expensive data movement, fully realizing practical acceleration from theoretical sparsity. Extensive evaluations demonstrate that CHESS surpasses Full-KV quality using only 1\% of the KV cache, delivers low-latency stable inference with up to 4.56times higher throughput, and consistently outperforms other strong baselines. Code is available at https://anonymous.4open.science/r/CHESS-9958/{https://anonymous.4open.science/r/CHESS/}.

Large language models (LLMs) achieve strong performance when all task-relevant information is available upfront, as in static prediction and instruction-following problems. However, many real-world decision-making tasks are inherently online: crucial information must be acquired through interaction, feedback is delayed, and effective behavior requires balancing information collection and exploitation over time. While in-context learning enables adaptation without weight updates, existing LLMs often struggle to reliably leverage in-context interaction experience in such settings. In this work, we show that this limitation can be addressed through training. We introduce ORBIT, a multi-task, multi-episode meta-reinforcement learning framework that trains LLMs to learn from interaction in context. After meta-training, a relatively small open-source model (Qwen3-14B) demonstrates substantially improved in-context online learning on entirely unseen environments, matching the performance of GPT-5.2 and outperforming standard RL fine-tuning by a large margin. Scaling experiments further reveal consistent gains with model size, suggesting significant headroom for learn-at-inference-time decision-making agents. Code reproducing the results in the paper can be found at https://github.com/XiaofengLin7/ORBIT.

A high-performance image compression algorithm is crucial for real-time information transmission across numerous fields. Despite rapid progress in image compression, computational inefficiency and poor redundancy modeling still pose significant bottlenecks, limiting practical applications. Inspired by the effectiveness of state space models (SSMs) in capturing long-range dependencies, we leverage SSMs to address computational inefficiency in existing methods and improve image compression from multiple perspectives. In this paper, we integrate the advantages of SSMs for better efficiency-performance trade-off and propose an enhanced image compression approach through refined context modeling, which we term MambaIC. Specifically, we explore context modeling to adaptively refine the representation of hidden states. Additionally, we introduce window-based local attention into channel-spatial entropy modeling to reduce potential spatial redundancy during compression, thereby increasing efficiency. Comprehensive qualitative and quantitative results validate the effectiveness and efficiency of our approach, particularly for high-resolution image compression. Code is released at https://github.com/AuroraZengfh/MambaIC.

Key-Value (KV) cache has become a bottleneck of LLMs for long-context generation. Despite the numerous efforts in this area, the optimization for the decoding phase is generally ignored. However, we believe such optimization is crucial, especially for long-output generation tasks based on the following two observations: (i) Excessive compression during the prefill phase, which requires specific full context impairs the comprehension of the reasoning task; (ii) Deviation of heavy hitters occurs in the reasoning tasks with long outputs. Therefore, SCOPE, a simple yet efficient framework that separately performs KV cache optimization during the prefill and decoding phases, is introduced. Specifically, the KV cache during the prefill phase is preserved to maintain the essential information, while a novel strategy based on sliding is proposed to select essential heavy hitters for the decoding phase. Memory usage and memory transfer are further optimized using adaptive and discontinuous strategies. Extensive experiments on LongGenBench show the effectiveness and generalization of SCOPE and its compatibility as a plug-in to other prefill-only KV compression methods.

Many-shot in-context learning has recently shown promise as an alternative to finetuning, with the major advantage that the same model can be served for multiple tasks. However, this shifts the computational burden from training-time to inference-time, making deployment of many-shot ICL challenging to justify in-practice. This cost is further increased if a custom demonstration set is retrieved for each inference example. We present Dynamic Block-Sparse Attention, a training-free framework for retrieval-based many-shot in-context learning. By combining carefully designed block-sparse attention and retrieval of cached groups of demonstrations, we achieve comparable per-example latency to finetuning while maintaining on average >95% of the best method's accuracy across strong ICL and finetuning baselines. We hope that this will further enable the deployment of many-shot ICL at scale.

As video content creation shifts toward long-form narratives, composing short clips into coherent storylines becomes increasingly important. However, prevailing retrieval formulations remain context-agnostic at inference time, prioritizing local semantic alignment while neglecting state and identity consistency. To address this structural limitation, we formalize the task of Consistent Video Retrieval (CVR) and introduce a diagnostic benchmark spanning YouCook2, COIN, and CrossTask. We propose CAST (Context-Aware State Transition), a lightweight, plug-and-play adapter compatible with diverse frozen vision-language embedding spaces. By predicting a state-conditioned residual update (Δ) from visual history, CAST introduces an explicit inductive bias for latent state evolution. Extensive experiments show that CAST improves performance on YouCook2 and CrossTask, remains competitive on COIN, and consistently outperforms zero-shot baselines across diverse foundation backbones. Furthermore, CAST provides a useful reranking signal for black-box video generation candidates (e.g., from Veo), promoting more temporally coherent continuations.

Recent work has demonstrated the potential of large language models (LLMs) for program optimization, a key challenge in programming languages. We propose a blackbox adaptation method called Retrieval Augmented Search (RAS) that performs beam search over candidate optimizations; at each step, it retrieves in-context examples from a given training dataset of slow-fast program pairs to guide the LLM. Critically, we find that performing contextual retrieval based on an LLM-generated natural language description significantly outperforms retrieval based on the source code. We also propose AEGIS, a method for improving interpretability by decomposing training examples into ''atomic edits'' that are significantly more incremental in nature. We show that RAS performs up to 2.06times better than prior state-of-the-art blackbox adaptation strategies on optimizing C++ programs, and that AEGIS performs up to 1.37times better while making significantly smaller edits. We also show that using RAS improves the mean runtime percentile of Python programs by 10.27 compared to baselines.

This paper presents WorldPlay, a streaming video diffusion model that enables real-time, interactive world modeling with long-term geometric consistency, resolving the trade-off between speed and memory that limits current methods. WorldPlay draws power from three key innovations. 1) We use a Dual Action Representation to enable robust action control in response to the user's keyboard and mouse inputs. 2) To enforce long-term consistency, our Reconstituted Context Memory dynamically rebuilds context from past frames and uses temporal reframing to keep geometrically important but long-past frames accessible, effectively alleviating memory attenuation. 3) We also propose Context Forcing, a novel distillation method designed for memory-aware model. Aligning memory context between the teacher and student preserves the student's capacity to use long-range information, enabling real-time speeds while preventing error drift. Taken together, WorldPlay generates long-horizon streaming 720p video at 24 FPS with superior consistency, comparing favorably with existing techniques and showing strong generalization across diverse scenes. Project page and online demo can be found: https://3d-models.hunyuan.tencent.com/world/ and https://3d.hunyuan.tencent.com/sceneTo3D.

Test-time scaling improves model performance by allocating additional compute during inference. Using this compute effectively across multiple context windows requires deciding how to allocate fresh contexts and what information to carry between them. We call a model's ability to make these decisions contextual reasoning. Existing approaches largely prescribe these decisions through their harness; we instead shift them to the model. We introduce 1) Hermes, a family of simple, configurable harnesses that progressively varies model control over context allocation and reuse, and 2) Hermes-Learn, a two-stage framework for learning these capabilities. We find that capable models can exploit this flexibility to scale with additional inference-time compute, while smaller open-source models initially struggle to do so. Training with Hermes-Learn closes this gap, inducing adaptive contextual reasoning strategies that vary with both the problem and the progress of reasoning. These gains generalize across benchmarks and models, extrapolate beyond the inference-time compute seen during training, and transfer to complementary test-time scaling methods beyond Hermes.

This paper presents a novel context compression method for Transformer language models in online scenarios such as ChatGPT, where the context continually expands. As the context lengthens, the attention process requires more memory and computational resources, which in turn reduces the throughput of the language model. To this end, we propose a compressed context memory system that continually compresses the growing context into a compact memory space. The compression process simply involves integrating a lightweight conditional LoRA into the language model's forward pass during inference. Based on the compressed context memory, the language model can perform inference with reduced memory and attention operations. Through evaluations on conversation, personalization, and multi-task learning, we demonstrate that our approach achieves the performance level of a full context model with 5times smaller context memory space. Codes are available at https://github.com/snu-mllab/context-memory.

Large Language Models (LLMs) possess impressive capabilities to generate meaningful code snippets given natural language intents in zero-shot, i.e., without the need for specific fine-tuning. In the perspective of unleashing their full potential, prior work has demonstrated the benefits of fine-tuning the models to task-specific data. However, fine-tuning process demands heavy computational costs and is intractable when resources are scarce, especially for models with billions of parameters. In light of these challenges, previous studies explored In-Context Learning (ICL) as an effective strategy to generate contextually appropriate code without fine-tuning. However, it operates at inference time and does not involve learning task-specific parameters, potentially limiting the model's performance on downstream tasks. In this context, we foresee that Parameter-Efficient Fine-Tuning (PEFT) techniques carry a high potential for efficiently specializing LLMs to task-specific data. In this paper, we deliver a comprehensive study of LLMs with the impact of PEFT techniques under the automated code generation scenario. Our experimental results reveal the superiority and potential of such techniques over ICL on a wide range of LLMs in reducing the computational burden and improving performance. Therefore, the study opens opportunities for broader applications of PEFT in software engineering scenarios.

Large Language Models (LLMs) are known to have limited extrapolation ability beyond their pre-trained context window, constraining their application in downstream tasks with lengthy inputs. Recent studies have sought to extend LLMs' context window by modifying rotary position embedding (RoPE), a popular position encoding method adopted by well-known LLMs such as LLaMA, PaLM, and GPT-NeoX. However, prior works like Position Interpolation (PI) and YaRN are resource-intensive and lack comparative experiments to assess their applicability. In this work, we identify the inherent need for LLMs' attention entropy (i.e. the information entropy of attention scores) to maintain stability and introduce a novel extension to RoPE which combines adjusting RoPE's base frequency and scaling the attention logits to help LLMs efficiently adapt to a larger context window. We validate the superiority of our method in both fine-tuning performance and robustness across different context window sizes on various context-demanding tasks. Notably, our method extends the context window of LLaMA-2-7B-Chat to 16,384 with only 100 samples and 6 training steps, showcasing extraordinary efficiency. Finally, we also explore how data compositions and training curricula affect context window extension for specific downstream tasks, suggesting fine-tuning LLMs with lengthy conversations as a good starting point. We release our code and SFT data at https://github.com/GAIR-NLP/Entropy-ABF.

Context adaptation automates prompt engineering in LLM-based systems by iteratively revising tunable prompts from task feedback, without modifying model weights. Extending this paradigm to multi-LLM agentic systems is crucial: existing methods suffer from inaccurate credit assignment and lack convergence guarantees. We propose Graph-based Target Back-Propagation (GTBP), a context adaptation framework for agentic workflows modeled as directed acyclic graphs. GTBP propagates local target outputs backward through the workflow graph and uses target--output discrepancies to guide a stage-wise prompt update mechanism. Theoretically, we show that GTBP's stage-wise prompt updates become stable over iterations, and that a sufficiently capable LLM optimizer can decrease the overall objective. Empirically, GTBP consistently outperforms strong baselines across three benchmarks while maintaining comparable computational cost.

Retrieval-augmented generation (RAG) methods can enhance the performance of LLMs by incorporating retrieved knowledge chunks into the generation process. In general, the retrieval and generation steps usually have different requirements for these knowledge chunks. The retrieval step benefits from comprehensive information to improve retrieval accuracy, whereas excessively long chunks may introduce redundant contextual information, thereby diminishing both the effectiveness and efficiency of the generation process. However, existing RAG methods typically employ identical representations of knowledge chunks for both retrieval and generation, resulting in suboptimal performance. In this paper, we propose a heterogeneous RAG framework (\myname) that decouples the representations of knowledge chunks for retrieval and generation, thereby enhancing the LLMs in both effectiveness and efficiency. Specifically, we utilize short chunks to represent knowledge to adapt the generation step and utilize the corresponding chunk with its contextual information from multi-granular views to enhance retrieval accuracy. We further introduce an adaptive prompt tuning method for the retrieval model to adapt the heterogeneous retrieval augmented generation process. Extensive experiments demonstrate that \myname achieves significant improvements compared to baselines.

Recent advancements in LLM pretraining have featured ever-expanding context windows to process longer sequences. However, our pilot study reveals that models pretrained with shorter context windows consistently outperform their long-context counterparts under a fixed token budget. This finding motivates us to explore an optimal context window scheduling strategy to better balance long-context capability with pretraining efficiency. To this end, we propose SkyLadder, a simple yet effective approach that implements a short-to-long context window transition. SkyLadder preserves strong standard benchmark performance, while matching or exceeding baseline results on long context tasks. Through extensive experiments, we pre-train 1B-parameter models (up to 32K context) and 3B-parameter models (8K context) on 100B tokens, demonstrating that SkyLadder yields consistent gains of up to 3.7% on common benchmarks, while achieving up to 22% faster training speeds compared to baselines. The code is at https://github.com/sail-sg/SkyLadder.

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.

Large Language Models (LLMs) have transformed code auto-completion by generating context-aware suggestions. Yet, deciding when to present these suggestions remains underexplored, often leading to interruptions or wasted inference calls. We propose an adaptive timing mechanism that dynamically adjusts the delay before offering a suggestion based on real-time developer feedback. Our suggested method combines a logistic transform of recent acceptance rates with a bounded delay range, anchored by a high-level binary prediction of the developer's cognitive state. In a two-month deployment with professional developers, our system improved suggestion acceptance from 4.9% with no delay to 15.4% with static delays, and to 18.6% with adaptive timing-while reducing blind rejections (rejections without being read) from 8.3% to 0.36%. Together, these improvements increase acceptance and substantially reduce wasted inference calls by 75%, making LLM-based code assistants more efficient and cost-effective in practice.

Large language models (LLMs) have demonstrated exceptional capabilities in generating text, images, and video content. However, as context length grows, the computational cost of attention increases quadratically with the number of tokens, presenting significant efficiency challenges. This paper presents an analysis of various Key-Value (KV) cache compression strategies, offering a comprehensive taxonomy that categorizes these methods by their underlying principles and implementation techniques. Furthermore, we evaluate their impact on performance and inference latency, providing critical insights into their effectiveness. Our findings highlight the trade-offs involved in KV cache compression and its influence on handling long-context scenarios, paving the way for more efficient LLM implementations.

Visual Language Models (VLMs) demonstrate impressive capabilities in processing multimodal inputs, yet applications such as visual agents, which require handling multiple images and high-resolution videos, demand enhanced long-range modeling. Moreover, existing open-source VLMs lack systematic exploration into extending their context length, and commercial models often provide limited details. To tackle this, we aim to establish an effective solution that enhances long context performance of VLMs while preserving their capacities in short context scenarios. Towards this goal, we make the best design choice through extensive experiment settings from data curation to context window extending and utilizing: (1) we analyze data sources and length distributions to construct ETVLM - a data recipe to balance the performance across scenarios; (2) we examine existing position extending methods, identify their limitations and propose M-RoPE++ as an enhanced approach; we also choose to solely instruction-tune the backbone with mixed-source data; (3) we discuss how to better utilize extended context windows and propose hybrid-resolution training. Built on the Qwen-VL series model, we propose Giraffe, which is effectively extended to 128K lengths. Evaluated on extensive long context VLM benchmarks such as VideoMME and Viusal Haystacks, our Giraffe achieves state-of-the-art performance among similarly sized open-source long VLMs and is competitive with commercial model GPT-4V. We will open-source the code, data, and models.