论文

本文提出 TTKV，一种 KV cache 管理框架，将人类记忆系统映射到具备异构容量与精度的 KV cache 上，在 128K 上下文任务上将跨层流量降低 5.94 倍。TTKV is proposed, a KV cache management framework that maps the human memory system onto the KV cache with heterogeneous capacity and precision, and reduces cross-tier traffic by 5.94x on 128K-context tasks.

本文将 Agentic 检索-生成循环形式化为有限时域部分可观测马尔可夫决策过程，显式建模其控制策略与状态转移，并构建了全面的分类体系与模块化架构分解，按规划机制、检索编排、记忆范式与工具调用行为对系统进行分类。This paper formalizes agentic retrieval-generation loops as finite-horizon partially observable Markov decision processes, explicitly modeling their control policies and state transitions, and develops a comprehensive taxonomy and modular architectural decomposition that categorizes systems by their planning mechanisms, retrieval orchestration, memory paradigms, and tool-invocation behaviors.

本文提出 KV Policy (KVP)，一种仅基于 key 与 value 向量、在预计算生成轨迹上训练的轻量级 per-head RL Agent 框架，证明学习预测未来 token 效用是自适应 KV cache 管理中强大且可扩展的范式。KV Policy (KVP), a framework of lightweight per-head RL agents trained on pre-computed generation traces using only key and value vectors, is introduced, demonstrating that learning to predict future token utility is a powerful and scalable paradigm for adaptive KV cache management.

本文提出 AgenticRAGTracer，这是首个主要由大语言模型自动构建、专为支持逐步验证而设计的 Agentic RAG 基准。AgenticRAGTracer is introduced, the first Agentic RAG benchmark that is primarily constructed automatically by large language models and designed to support step-by-step validation, and is primarily constructed automatically by large language models and designed to support step-by-step validation.

TritonForge 是一个面向自动化 Triton kernel 优化的 profiling 引导框架，融合 kernel 分析、运行时 profiling 与迭代式代码转换以简化优化流程，并为自动化 GPU 性能优化领域的未来研究奠定基础。TritonForge, a profiling-guided framework for automated Triton kernel optimization that integrates kernel analysis, runtime profiling, and iterative code transformation to streamline the optimization process and provides a foundation for future research in automated GPU performance optimization.

本文探讨容器化、微服务、动态调度等 Cloud Native 技术如何从根本上提升 LLM 推理服务，并展示 Cloud Native 系统在高需求场景下实现更高效资源分配、降低延迟与提升吞吐的能力。This article explores how Cloud Native technologies, such as containerization, microservices, and dynamic scheduling, can fundamentally improve LLM inference serving and demonstrates how a Cloud Native system enables more efficient resource allocation, reduces latency, and enhances throughput in high-demand scenarios.

本文首次对 vLLM 启动延迟进行了详细的性能表征，并构建了一个轻量级分析模型，能够针对给定硬件配置准确预测 vLLM 的启动延迟，为大规模推理环境中的资源规划提供了可操作的指导。This paper presents the first detailed performance characterization of vLLM startup latency and develops a lightweight analytical model that accurately predicts vLLM's startup latency for a given hardware configuration, providing actionable guidance for resource planning in large-scale inference environments.

本文将数据安全形式化为 provenance monomials 上的聚合谓词，并提出 Passant——一个无需物化 provenance 即可强制执行 DFC 策略的可移植查询重写层。This paper formalizes data safety as aggregate predicates over provenance monomials and presents Passant, a portable query rewriting layer that enforces DFC policies without materializing provenance.

本工作提出 AlphaEval，一个基于真实生产环境的基准，包含来自七家在其核心业务中部署 AI Agent 的公司的 94 个任务，覆盖六个 O*NET (Occupational Information Network) 领域；并贡献了一套从需求到基准的构建框架，将从需求到评估的完整流程标准化。This work presents AlphaEval, a production-grounded benchmark of 94 tasks sourced from seven companies deploying AI agents in their core business, spanning six O*NET (Occupational Information Network) domains, and contributes a requirement-to-benchmark construction framework that standardizes the entire pipeline from requirement to evaluation.

Continuum，一个通过为 KV cache 保留引入 TTL 机制来优化多轮 Agent 工作负载任务完成时间的服务系统，能保持多轮连续性，并降低 Agent 工作流的延迟。Continuum, a serving system to optimize job completion time for multi-turn agent workloads by introducing time-to-live mechanism for KV cache retention, preserves multi-turn continuity, and reduces delay for agentic workflows.

SarseX 模型无关、无需训练，并与 Prefix Cache 兼容，可为多轮对话、检索增强生成 (RAG) 和 Agent 工作流等常见在线服务场景提供统一支持。SarseX is model-agnostic, training-free, and compatible with Prefix Cache, and it provides unified support for common online serving scenarios including multi-round chat, retrieval-augmented generation (RAG), and agent workflows.

NetKV，一种使用该 oracle 信息的 O(|D|) 每请求贪心策略，其层级排序被证明对过时遥测数据具有鲁棒性；并证明随着上下文长度增长，忽略网络项会使仅缓存感知的调度任意次优。NetKV, the O(|D|) per-request greedy that consumes this oracle, has tier rankings that are provably robust to stale telemetry, and it is proved that ignoring the network term renders cache-aware-only scheduling arbitrarily suboptimal as context length grows.

本工作提出 LMCACHE，首个也是目前最高效的开源 KV 缓存方案，可将现代 LLM 引擎生成的 KV 缓存从 GPU 显存中提取并存储，并跨引擎和查询共享。This work presents LMCACHE, the first and so far the most efficient open-source KV caching solution, which extracts and stores KV caches generated by modern LLM engines out of the GPU memory and shares them across engines and queries.

构建了 RoboGenesis，一个基于仿真的工作流与数据引擎，可从原子技能组合配置好的实验工作流，对 rollout 进行验证与过滤，并跨支持的机器人配置导出结构化演示数据。RoboGenesis is built, a simulation-based workflow and data engine that composes configured laboratory workflows from atomic skills, validates and filters rollouts, and exports structured demonstrations across supported robot profiles.

本文提出 DashboardQA，这是首个明确设计用于评估视觉-语言 GUI Agent 对真实世界仪表板理解与交互能力的基准，结果表明交互式仪表板推理对所有受评估的 VLM 而言都是一项具有挑战性的任务。DashboardQA is introduced, the first benchmark explicitly designed to assess how vision-language GUI agents comprehend and interact with real-world dashboards, and indicates that interactive dashboard reasoning is a challenging task overall for all the VLMs evaluated.

本工作提出 12 个具体指标，从一致性、鲁棒性、可预测性和安全性四个关键维度分解 Agent 可靠性，可与传统评估互补，并提供用于分析 Agent 表现、退化与失败方式的工具。This work proposes twelve concrete metrics that decompose agent reliability along four key dimensions: consistency, robustness, predictability, and safety, which complement traditional evaluations while offering tools for reasoning about how agents perform, degrade, and fail.

本文提出一个 Agentic RAG 框架，使 LLM 能够使用逻辑表达式构建检索意图，同时将检索后端简化为基于倒排索引的系统，并表明将检索过程锚定在逻辑查询上可显著降低生成响应中的幻觉。This paper proposes an agentic RAG framework that enables LLMs to formulate retrieval intents using logical expressions while simplifying the retrieval backend to an inverted-index-based system, and shows that anchoring the retrieval process in logical queries substantially reduces hallucinations in generated responses.

本文对一个自 2026 年 3 月起持续运行的个人助理 Agent 运行时中的静默失败进行纵向研究，该系统包含约 40 个定时任务、8 个 LLM 提供商、一个工具治理代理以及一个知识库记忆层，由 4,286 个单元测试和 827 项治理检查守护。A longitudinal study of silent failures in a personal-assistant agent runtime in continuous production since March 2026, with roughly 40 scheduled jobs, 8 LLM providers, a tool-governance proxy, and a knowledge-base memory plane, defended by 4,286 unit tests and 827 governance checks is presented.

结果表明，纯视觉设置会降低准确率并增加 token 成本，因为 Agent 缺乏足够的符号化细节，需通过重复的视觉查询进行补偿；研究指向一种面向下一代编码 Agent 的实用文本与视觉混合设计。The results show that a strictly vision-only setup degrades accuracy and increases token cost, because agents lack sufficient symbolic detail and compensate with repeated visual queries, and point to a practical hybrid text-and-vision design for next-generation coding agents.

认为（该早期版本的）GPT-4 属于新一代具备更通用智能的 LLM（与 ChatGPT、谷歌 PaLM 等并列），并讨论了这些模型不断增强的能力及其影响。It is argued that (this early version of) GPT-4 is part of a new cohort of LLMs (along with ChatGPT and Google's PaLM for example) that exhibit more general intelligence than previous AI models, and the rising capabilities and implications of these models are discussed.

发现对 GPT 语言模型进行重复采样，是为困难 prompt 生成可用解的一种出乎意料的有效策略，并讨论了部署强大代码生成技术在安全性、安全保障与经济等方面的潜在更广泛影响。It is found that repeated sampling from the GPT language model is a surprisingly effective strategy for producing working solutions to difficult prompts, and the potential broader impacts of deploying powerful code generation technologies, covering safety, security, and economics are discussed.

提出 Flamingo，一个 VLM 系列，能够桥接强大的纯视觉与纯语言预训练模型，处理任意交错排列的图文序列，并无缝接收图像或视频作为输入。This work introduces Flamingo, a family of Visual Language Models (VLM) with this ability to bridge powerful pretrained vision-only and language-only models, handle sequences of arbitrarily interleaved visual and textual data, and seamlessly ingest images or videos as inputs.

该综述细致梳理了 RAG 范式的演进，涵盖 Naive RAG、Advanced RAG 与 Modular RAG，并对 RAG 框架的三大基础支柱——检索、生成与增强技术——进行了深入审视。This comprehensive review paper offers a detailed examination of the progression of RAG paradigms, encompassing the Naive RAG, the Advanced RAG, and the Modular RAG, and meticulously scrutinizes the tripartite foundation of RAG frameworks, which includes the retrieval, the generation and the augmentation techniques.

该综述全面回顾了生成模型的历史与基本组件，以及 AIGC 在单模态交互与多模态交互方向的最新进展，并介绍了文本与图像生成任务及相关模型。This survey provides a comprehensive review on the history of generative models, and basic components, recent advances in AIGC from unimmodal interaction and multimodal interaction, and introduces the generation tasks and relative models of text and image.

提出 BloombergGPT，一个 500 亿参数的语言模型，在广泛的金融数据上训练而成，并基于 Bloomberg 丰富的数据源构建了包含 3630 亿 token 的数据集，可能是迄今最大的领域专用数据集。This work presents BloombergGPT, a 50 billion parameter language model that is trained on a wide range of financial data, and constructs a 363 billion token dataset based on Bloomberg's extensive data sources, perhaps the largest domain-specific dataset yet.

一篇关于基于 LLM 的 Agent 的全面综述，追溯了 Agent 概念从其哲学起源到在 AI 中的发展历程，解释了为何 LLM 适合作为 Agent 的基础，并提出一个包含三个核心组件的通用框架：大脑、感知与行动。A comprehensive survey on LLM-based agents, tracing the concept of agents from its philosophical origins to its development in AI, and explaining why LLMs are suitable foundations for agents, and presenting a general framework, comprising three main components: brain, perception, and action.

综合评估表明，DeepSeek-V3 优于其他开源模型，并达到与领先闭源模型相当的性能。Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models.

本文综述了一些最具代表性的 LLM，包括三大主流 LLM 家族（GPT、LLaMA、PaLM），讨论其特性、贡献与局限性，并概述了构建与增强 LLM 的相关技术。This paper reviews some of the most prominent LLMs, including three popular LLM families (GPT, LLaMA, PaLM), and discusses their characteristics, contributions and limitations, and gives an overview of techniques developed to build, and augment LLMs.

本文对 LLM 的评估方法进行了全面综述，围绕三个关键维度展开：评估什么、在何处评估、如何评估，并为 LLM 评估领域的研究者提供了宝贵洞见。This paper presents a comprehensive review of these evaluation methods for LLMs, focusing on three key dimensions: what to evaluate, where to evaluate, and how to evaluate, and offers invaluable insights to researchers in the realm of LLMs evaluation.

分析表明，借助 LLM，美国约 15% 的工作任务可在保持同等质量的前提下显著提速完成，意味着 LLM 驱动的软件将对底层模型经济影响的规模化产生实质性作用。The analysis suggests that, with access to an LLM, about 15% of all worker tasks in the US could be completed significantly faster at the same level of quality, implying that LLM-powered software will have a substantial effect on scaling the economic impacts of the underlying models.

研究发现，在上述多个维度上进行的指令微调可显著提升多种模型类别（PaLM、T5、U-PaLM）、多种提示设定以及多种评测基准（MMLU、BBH、TyDiQA、MGSM、开放式生成）上的表现。It is found that instruction finetuning with the above aspects dramatically improves performance on a variety of model classes (PaLM, T5, U-PaLM), prompting setups, and evaluation benchmarks (MMLU, BBH, TyDiQA, MGSM, open-ended generation).

本文描述了一份以模式形式呈现的 prompt 工程技巧目录，这些技巧已被用于解决与 LLMs 对话时的常见问题，以改进 LLM 对话的输出。A catalog of prompt engineering techniques presented in pattern form that have been applied to solve common problems when conversing with LLMs to improve the outputs of LLM conversations is described.

研究发现，ChatGPT 最成功地被用作数学助手，用于查询事实、充当数学搜索引擎和知识库接口；GPT-4 还可被用于本科水平的数学问题，但在研究生难度的题目上表现不佳。It is found that ChatGPT can be used most successfully as a mathematical assistant for querying facts, acting as a Mathematical search engine and knowledge base interface, and GPT-4 can additionally be used for undergraduate-level mathematics but fails on graduate-level difficulty.

本文进行了迄今为止对 Code LLMs 最全面的评估，结果显示 StarCoderBase 在支持多编程语言的开放 Code LLMs 中表现最优，并且能够匹敌或超越 OpenAI code-cushman-001 模型。This work performs the most comprehensive evaluation of Code LLMs to date and shows that StarCoderBase outperforms every open Code LLM that supports multiple programming languages and matches or outperforms the OpenAI code-cushman-001 model.

本文显著推进了针对已对齐语言模型的对抗攻击 SOTA，并提出了关于如何防止此类系统生成不良信息的重要问题。This work significantly advances the state-of-the-art in adversarial attacks against aligned language models, raising important questions about how such systems can be prevented from producing objectionable information.