[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-memory-tax-agent-memory-net-value":3,"topics-all":38,"news-related-417ea170-637f-48be-a998-aa13ec5cdb10":67},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"417ea170-637f-48be-a998-aa13ec5cdb10","记忆税:11个记忆系统多数跑输朴素RAG","UIUC 团队把 11 个开源智能体记忆系统与朴素 RAG 同台对比:LongMemEval-s 三个底座全部跑输两轮检索,A-Mem 离线构建 117 小时而 RAG 仅 0.11 小时。论文将净亏损拆成记忆收益与记忆税,定位出信息丢失、证据难取、跨底座脆弱三条税源,部分基准检索太友好、余量太小。代码已开源。","给智能体配\"长期记忆\"几乎成了行业默认动作:MemOS、A-Mem、HippoRAG2 这类系统一个接一个出场,卖点都是\"让 Agent 记得住你\"。但 UIUC 牵头、清华与人民大学等机构参与的一篇新论文(10 月 25 日挂上 arXiv,已在 Hugging Face 论文日榜拿到 99 个赞),把 11 个开源记忆系统拉到和最朴素的 RAG 同一台面上:多数场合,记忆系统跑输了,而且输得又贵又脆。论文给这部分净亏损起了个名字:记忆税(Memory Tax)。论文原文见 [arXiv:2610.34073](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.34073)。\n\n## 实验怎么做的\n\n三个基准:LongMemEval-s(长对话记忆问答)、PersonaMem-v2(用户画像推断,抽 20% 样本)、MINT-Eval(多目标干扰,抽 10%);三个底座:Qwen3.5-4B、Qwen3.6-27B、GPT-5.6-luna。11 个记忆系统按存储组织方式分四类:扁平存储(A-Mem、MemOS、SimpleMem)、层级\u002F生命周期(MemoryOS、LightMem、OpenViking)、图结构(HippoRAG2、MAGMA、PlugMem)、训练策略(Mem-T、Memory-R2)。参照系是不做记忆加工、直接在原始历史上切块检索的朴素 RAG,分一轮\u002F两轮两档;为了公平,作者用\"检索范围\"(Retrieval Scope)对齐各系统预算,判卷统一交给 Qwen3.6-27B。\n\n## 账面:全线跑输,成本差三个数量级\n\nLongMemEval-s 上,11 个系统在三个底座上全部低于两轮 RAG:27B 底座两轮 RAG 拿 81.4 分,最接近的 A-Mem 79.4,MemOS 76.6;LightMem 在 4B 底座落后 14.6 分,OpenViking 更是直接掉 40.4 分。另一面,PersonaMem-v2 上 Memory-R2 反超 2.0 分,MINT-Eval 上 PlugMem 反超 3.3 分——不是一边倒,而这正是论文要拆的东西。\n\n成本端更刺眼:附录成本表里,LongMemEval-s 上 A-Mem 的离线构建耗时 117 小时,RAG 只要 0.11 小时,差三个数量级;LLM 调用量 9.84 万次对两轮 RAG 的 200 次。多付一千倍构建成本,换来分数倒退,这就是论文说的负净价值。\n\n## 记忆税的三条税源\n\n论文把净价值拆成记忆收益(组织、蒸馏做对的部分)和记忆税,再顺藤摸瓜找税源,找到三条。**保真**:把原子事实从原始语料一路追到记忆单元,OpenViking 有 38% 的事实压根没有对应记忆单元,另有 13% 重大丢失;对照组 MAGMA 93.6% 无损失——丢失不是必然,是设计选择。**可达性**:证据没丢也未必取得出,被切碎、稀释、改写后,记忆系统首轮检索能找回的查询比 RAG 少,证据埋得更深。**鲁棒性**:固定记忆、换问答底座,不少记忆系统的波动比裸 RAG 更大——模型依赖的中间操作把脆弱性焊进了记忆里。\n\n## 基准也有锅\n\n论文还把一部分账记到评测集头上:在一批长期记忆基准上,连不用向量的 IDF 关键词匹配都能拿 50% 以上,混合检索能到 70% 左右——题目太\"检索友好\",天花板太低,税自然显眼。作者又把收益\u002F税框架推广到经验记忆(网页任务里蒸馏工作流确实能赢 RAG,工具调用场景则好几个记忆方法落后)和世界模型记忆(直接回灌历史观测的朴素做法一致性就很强):记忆税不是对话记忆的特产。\n\n## 所以呢\n\n这篇论文的价值不在\"记忆没用\",而在给了一个算账框架:以后看任何记忆系统的成绩单,先问一句\"朴素 RAG 基线多少分、构建花多少小时\"。代码与全部结果已开源。落到选型:场景以事实回忆为主,先上两轮 RAG,大概率够用;记忆系统留给真正需要抽象和偏好沉淀的任务。能算清税的记忆,才配叫资产。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.34073","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"6ad31a14-c0da-42df-81fd-564281f768db","agentic-ai",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"120fa59a-ff6f-4537-9bf5-f818df636a0e","benchmark",{"id":19,"name":20,"slug":20,"description":14,"color":14},"0ef8513a-0a26-42f0-b6f9-5b6dadded45c","efficiency",{"id":22,"name":23,"slug":23,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"7cf58899-4c65-470c-8690-35776eae6a15","en","Memory Tax: Most Agent Memory Systems Lose to Naive RAG","UIUC study: 11 agent-memory systems trail naive RAG on LongMemEval-s across three backbones at far higher cost; losses split into benefit and tax.","Giving agents \"long-term memory\" has become a default move: systems like MemOS, A-Mem, and HippoRAG2 keep arriving, all selling \"agents that remember you.\" But a new paper led by UIUC with collaborators from Tsinghua and Renmin University (on arXiv since Oct 25, already at 99 upvotes on the Hugging Face Daily Papers board) put 11 open memory systems on the same stage as the plainest possible RAG — and in most cases memory lost, expensively and fragilely. The paper names that net loss: memory tax. Full paper: [arXiv:2610.34073](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.34073).\n\n## How the experiment was run\n\nThree benchmarks: LongMemEval-s (long-conversation memory QA), PersonaMem-v2 (user persona inference, 20% sample), MINT-Eval (multi-target interference, 10% sample); three backbones: Qwen3.5-4B, Qwen3.6-27B, and GPT-5.6-luna. The 11 memory systems fall into four families by how they organize storage: flat stores (A-Mem, MemOS, SimpleMem), hierarchical\u002Flifecycle (MemoryOS, LightMem, OpenViking), graph-based (HippoRAG2, MAGMA, PlugMem), and trained policies (Mem-T, Memory-R2). The reference is naive RAG — chunked retrieval over raw history, no memory processing — in one-round and two-round variants. Retrieval budgets were aligned via Retrieval Scope, and Qwen3.6-27B served as the judge.\n\n## The scoreboard: losing across the board, at three orders of magnitude the cost\n\nOn LongMemEval-s, all 11 systems fall below two-round RAG on all three backbones: on the 27B backbone two-round RAG scores 81.4, the closest is A-Mem at 79.4, MemOS at 76.6; LightMem trails by 14.6 points on 4B, and OpenViking drops a full 40.4. On the other side, Memory-R2 beats RAG by 2.0 on PersonaMem-v2 and PlugMem by 3.3 on MINT-Eval — not a wipeout, and that is exactly what the paper dissects.\n\nThe cost side stings more: in the appendix cost table, A-Mem's offline build takes 117 hours on LongMemEval-s versus 0.11 hours for RAG — three orders of magnitude — with 98.4K LLM calls against 200 for two-round RAG. Pay a thousand times the build cost, get a lower score: that is the paper's negative net value.\n\n## Three sources of memory tax\n\nThe paper splits net value into memory benefit (what organization and distillation get right) and memory tax, then traces the tax to three sources. **Faithfulness**: tracing atomic facts from the raw corpus into memory units, OpenViking has no memory unit at all for 38% of facts, plus 13% major loss; the contrast, MAGMA, keeps 93.6% lossless — loss is a design choice, not fate. **Accessibility**: evidence that survives can still be unreachable — fragmented, diluted, or rephrased, memory systems recover fewer queries than RAG in the first retrieval round; the evidence is buried deeper. **Robustness**: with memory fixed and the QA backbone swapped, many memory systems wobble more than raw RAG — model-dependent intermediate operations weld fragility into the store.\n\n## The benchmarks share the blame\n\nThe paper also bills part of the account to evaluation sets: across a batch of long-term memory benchmarks, even vector-free IDF keyword matching scores above 50%, and hybrid retrieval reaches about 70% — the tasks are retrieval-friendly, the ceiling is low, and the tax becomes visible. The authors further generalize the benefit\u002Ftax lens to experiential memory (distilling workflows does beat RAG on web tasks, while several memory methods trail on tool use) and world-model memory (naively retrieving and re-injecting past observations already yields strong consistency): memory tax is not a conversational-memory quirk.\n\n## So what\n\nThe paper's real contribution is not \"memory is useless\" but an accounting frame: when you see any memory system's scorecard, first ask \"what does naive RAG score, and how many build hours does it take?\" Code and full results are open-sourced. For practitioners: if your workload is mostly factual recall, start with two-round RAG — it will likely suffice; save memory systems for tasks that truly need abstraction and preference distillation. Memory worth keeping is memory whose tax you can count.","memory-tax-agent-memory-net-value","2026-10-28T21:07:37Z","2026-10-28T21:07:41.740462Z","2026-10-28T21:07:41.740472Z",true,"agent",524,[39,49,58],{"slug":40,"tag_slug":41,"title_zh":42,"title_en":43,"intro_zh":44,"intro_en":45,"id":46,"is_active":35,"created_at":47,"modified_at":48},"image-generation","text-to-image","图像生成：从文生图到逐框编辑的模型竞赛","Image Generation: from text-to-image to frame-level editing","文生图已是红海，战场正转向编辑与可控性：FLUX.2\u002F3 用开源和量化把生图搬进消费级显卡，Qwen-Image、Seedream 等国产模型紧咬不放，GPT Image 2 则把推理能力带进像素。本专题追踪图像生成与编辑模型的发布、架构演进、加速量化与商业化数据，看「好图」如何变成「可用的图」。","Text-to-image is a crowded race, and the frontier has moved to editing and control: FLUX.2\u002F3 push open weights and quantization onto consumer GPUs, Qwen-Image and Seedream close in from China, and GPT Image 2 brings reasoning into pixels. This topic tracks model releases, architecture shifts, acceleration work, and the business numbers behind \"pretty\" becoming \"usable\".","52de7c59-7864-4569-8ae7-3d656d26a1a1","2026-11-20T07:20:23.534506Z","2026-11-20T07:20:23.534527Z",{"slug":50,"tag_slug":50,"title_zh":51,"title_en":52,"intro_zh":53,"intro_en":54,"id":55,"is_active":35,"created_at":56,"modified_at":57},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":59,"tag_slug":59,"title_zh":60,"title_en":61,"intro_zh":62,"intro_en":63,"id":64,"is_active":35,"created_at":65,"modified_at":66},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":68},[69,74,79,84,89,94],{"id":70,"title":71,"news_slug":72,"published_at":73},"1ec34a26-f012-4bc4-8193-14e651c07516","程序进化搜出记忆设计:涨12个百分点,成本只占35%","mape-agent-memory-program-evolution","2026-11-14T05:05:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"00346b75-f071-42fd-ae16-db4c5569f01a","EarlyEval 提前叫停注定失败的 Agent:近半 token 省下,分辨率只动一两个点","earlyeval-early-stop-agent-eval","2026-09-03T21:04:52+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"10bd728a-5d05-461b-ba14-7c6ea51f8ce3","HyperTool 把 MCP 工具调用「折叠」成一段代码，Qwen3 智能体准确率直接翻倍","hypertool-mcp-folded-code-qwen3-2x","2026-06-14T20:00:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"c4623b85-37f5-435e-ae98-e67eb72eb1dc","别再逐篇问大模型:AI写程序当评委,每主题0.17美元起","execjudges-llm-written-program-judges-retrieval-eval","2026-11-20T01:40:00+00:00",{"id":90,"title":91,"news_slug":92,"published_at":93},"f02185e7-5f4c-46c6-bb18-575fbf3b1c0a","摩根大通实测:五项前沿配方,四项真有用","jpmorgan-factorial-retest-llm-ingredient-gains","2026-11-19T17:25:00+00:00",{"id":95,"title":96,"news_slug":97,"published_at":98},"03403319-8597-4876-8661-fad74ee0ac48","屏幕上的警告,智能体最高92%照做","computer-use-agents-cross-screen-harm","2026-11-18T23:30:00+00:00"]