[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-llm-cloud-continuum-frame-2609":3,"topics-all":38,"news-related-1bf3f192-68e9-49f1-8bbb-a350f05c15e2":57},{"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},"1bf3f192-68e9-49f1-8bbb-a350f05c15e2","LLM 编排云边端:FRAME 2026 论文点出反馈层缺环","意大利 CNR 团队在 FRAME 2026 提出新分类法,审查 6 套 LLM+DRL 的云边端调度系统,4 套没给 LLM 反馈,2 套只跑通单层闭环,跨 IoT-边-云的多层闭环至今空白。","当所有人都把 LLM 塞进控制回路,真正跑起来才发现:大部分情况下模型拿到的反馈就是它自己输入的那句话。最近意大利国家研究委员会(CNR)ISTI 团队在 Euro-Par 2026 的 FRAME 工作坊上发表了一篇把这条裂缝说得很透的论文。他们审了 6 套把 LLM 和深度强化学习(DRL)拼在一起做云边端(Cloud Continuum)调度的系统,结论是:六分之四根本没给 LLM 反馈,六分之二能闭环但只跑通单层,跨 IoT-边缘-云三层的多层闭环至今空白。\n\n## 论文给出的两个新维度:LLM 做什么、反馈怎么传\n\n既有文献按控制范围(SARL vs MARL)和训练范式(标准 vs 联邦)给 DRL 调度系统分类,但这套分类对 LLM 完全失明。CNR 的 Vaccarella 等人加了两条新轴:AI Augmentation Paradigm(L0–L2)看 LLM 是解释意图的语义前端,还是下指令的工作流环节,还是真正掌控多智能体层的战略编排;Feedback Channel(F0–F3)看执行结果从哪条路径流回 LLM。F0 是无反馈,F1 是平台 API 直传,F2 是智能体层中转,F3 是有独立 Mediator 做反馈抽象。\n\n## 6 套系统的真实分布\n\n按这两条新轴重新摆位,六个代表系统的形状就很扎眼。AgentEdge 在编排层串了 Intent、Observability、Planning、Infrastructure Action 四段 LLM,所有环节都在编排层里、根本没有独立智能体层,执行结果也不回传,L1×F0——其实是单层 LLM 流水线。IntentContinuum 让 GPT-4o 读 Kubernetes\u002FONOS\u002FsFlow 三个平台 API 的 JSON 直接做根因定位,L1×F1;SALLMA 把 LangChain 拼出来的整套工作流压在编排层,知识层是被动存储,L1×F1。Habib 等人做 5G RAN 自动化,ALBERT 解析意图后下发给分层 DQN,但 DQN 跑完没有回传,L2×F0——是开环。真正闭环的只有 AURA 和 CyberOps-Bots 两个:AURA 用 Claude Sonnet 4 在基站上做 L2 编排,信号回到独立的 Centralised Alignment Controller(评估完返回 [-1,+1] 的延迟奖励),L2×F3;CyberOps-Bots 用 Qwen3-8B 加四类 RL 智能体做云网韧性,L2×F2。两者都成功,但都只在一个资源层里做闭环,没人把闭环跨过 IoT-边-云这三层的异构性。\n\n## 卡在两件事上:延迟错配和反馈抽象\n\n为什么跨层闭环没人做?论文给了三个工程层面的硬约束,前两个尤其现实。第一个是延迟错配:把 LLM 放在实时控制的关键路径上,光是 API 延迟就把控制回路拖垮,论文建议让 LLM 异步跑在 episode 总结上,本地智能体继续按上一段子目标先走。第二个是钱:AURA 用的是 Claude Sonnet 4,价格是每百万 token 3 美元输入、15 美元输出,论文拿每次 1 万输入、1 千输出估算,一通调用就 0.05 美元,做全天 7×24 的细粒度调度一天能跑出几百美元;对应的折中方案是阈值触发,反馈偏离当前指令一定幅度才让 LLM 重新拍板,否则智能体就照着最后一条指令继续跑。第三个是反馈抽象:IoT、边缘、云三个层智能体暴露的指标往往是不可比的(延迟、能耗、容器 SLO、SDN 路径成本),需要有人把这些异构信号统一成 LLM 能消化的提示尺寸——AURA 里的 Centralised Alignment Controller 在单层里做了这件事,但没人把它扩展到跨层。\n\n## 给作者看的研究方向\n\n论文把跨层闭环明确命名为 L2×MA×Federated 这块空白,提示接下来要补的工程接口叫 Mediator——一个跨层、独立的反馈聚合器,负责把每个 tier 的异构 KPI 折算成统一的评分+上下文对,送回 LLM 编排层调整下一轮子目标。它和 AURA 的 CAC 是同一类东西,只是从单层扩到三层。CNR 团队把自己的通用参考架构画成图 1(a):上层 LLM Orchestration,中层 Agentic,下层 Environments,Mediator 单独站在三层之间。FRAME 2026 这篇论文接下来要走 Euro-Par 2026 正式版,届时如果能把 L2×F3 的多 tier 实测跑出来,云边端调度这条线就真正从\"LLM 当语义前端\"迈到\"LLM 当控制中枢\"了。\n\n对工程团队最直接的启发是:如果你正在做云边端编排,先别急着让 LLM 做闭环决策,先把反馈层抽象出来——单层的 CAC 已经有人在做了,跨层 Mediator 是接下来 12 个月内值得盯的研究方向。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.09348v4","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},"fca9258a-9430-455a-b95d-b9fae5e373a8","ai-inference",{"id":19,"name":20,"slug":20,"description":14,"color":14},"5e628969-6d2a-437f-998a-104e4b16cfb1","ai-progress",{"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},"ddfee23c-2857-4612-b597-b9e90ee69753","en","LLM Cloud-Edge Orchestration: FRAME 2026 Paper Names the Missing Feedback Layer","Italian CNR researchers at FRAME 2026 propose a new taxonomy, auditing 6 LLM+DRL Cloud Continuum systems; 4 give the LLM no feedback, 2 close loops only in a single tier, leaving cross IoT-edge-cloud feedback blank.","When everyone started stuffing LLMs into control loops, the actual deployments started revealing something: in most setups, the feedback the model receives is just the prompt it just sent. A recent paper from the Italian National Research Council's ISTI team, presented at the FRAME 2026 workshop co-located with Euro-Par 2026, makes this gap brutally clear. They audited six systems that pair LLMs with deep reinforcement learning (DRL) for Cloud Continuum orchestration, and the verdict is stark: four out of six give the LLM no feedback at all, two manage closed loops but only within a single resource tier, and the cross-tier loop spanning IoT-edge-cloud remains entirely blank.\n\n## Two new dimensions: what the LLM does, and how feedback travels\n\nExisting literature classifies DRL orchestration systems along two axes: control scope (SARL vs MARL) and training paradigm (standard vs federated). This taxonomy is completely blind to LLMs. Vaccarella et al. add two new axes: the AI Augmentation Paradigm (L0–L2) captures whether the LLM is a semantic front-end that interprets intent, a workflow step that issues instructions, or a true strategic orchestrator over a multi-agent layer; the Feedback Channel (F0–F3) captures the path execution results take to return to the LLM. F0 is no feedback, F1 is direct platform API delivery, F2 is an agent-layer relay, F3 is an independent Mediator producing a feedback abstraction.\n\n## How the six systems actually stack up\n\nMapped onto the two new axes, the distribution of the six representative systems is striking. AgentEdge strings four LLM stages (Intent, Observability, Planning, Infrastructure Action) in the orchestration layer, with no independent agent layer, and no execution results flow back: L1×F0, really a single-layer LLM pipeline. IntentContinuum has GPT-4o read structured JSON from three platform APIs (Kubernetes, ONOS, sFlow) to diagnose root causes: L1×F1. SALLMA presses the entire LangChain-orchestrated workflow into the orchestration layer with a passive knowledge layer, L1×F1. Habib et al. for 5G RAN automation has ALBERT parse intent then hand off to a hierarchical DQN, but the DQN never reports back: L2×F0, an open loop. The only systems that actually close the loop are AURA and CyberOps-Bots: AURA uses Claude Sonnet 4 to orchestrate at base stations, with signals flowing back to an independent Centralised Alignment Controller that returns a delayed reward in [-1, +1], L2×F3; CyberOps-Bots uses Qwen3-8B with four classes of RL agents for cloud-network resilience, L2×F2. Both work, but both stay inside a single resource tier. Nobody has closed the loop across IoT-edge-cloud heterogeneity.\n\n## What's blocking the multi-tier loop: latency mismatch and feedback abstraction\n\nWhy is the cross-tier loop still empty? The paper identifies three engineering constraints, and the first two are brutally real. First, latency mismatch: putting an LLM on the critical path of a real-time control loop kills the loop, since even the API latency of a frontier model overwhelms the response budget. The proposed workaround is to have the LLM run asynchronously on episode summaries, while local agents continue under the last issued subgoal. Second, money: AURA's Claude Sonnet 4 is priced at $3 per million input tokens and $15 per million output tokens. The paper estimates a 10K input + 1K output call at roughly $0.05, and at continuous control frequencies, daily cost can hit hundreds of dollars. The fix is threshold-triggered invocation, where the LLM only re-plans when feedback deviates from the current directive beyond a defined margin. Third, feedback abstraction: IoT, edge, and cloud agents expose incommensurable metrics (latency, energy, container SLOs, SDN path costs), and someone needs to compress these into a prompt-sized representation the LLM can act on. AURA's Centralised Alignment Controller does this in a single tier, but no one has extended it across the Continuum.\n\n## What to watch next: the Mediator and L2×MA×Federated\n\nThe paper names the cross-tier closed loop as the L2×MA×Federated cell, with the missing engineering interface called the Mediator, a cross-tier, independent feedback aggregator that compresses each tier's heterogeneous KPIs into a unified score-plus-context pair, which the orchestration LLM consumes to revise the next round of subgoals. It is structurally the same idea as AURA's CAC, but extended from one tier to three. CNR's generic reference architecture (Figure 1(a) of the paper) shows the Mediator standing between the LLM Orchestration layer and the Agentic\u002FEnvironments layers, collecting structured feedback from specialised agents and producing a unified representation. The next iteration will appear in the formal Euro-Par 2026 proceedings; if the team or a follow-up paper actually runs the L2×F3 cross-tier benchmark, the Cloud Continuum orchestration line will finally move from \"LLM as semantic front-end\" to \"LLM as control core.\"\n\nFor engineering teams working on cloud-edge orchestration today, the immediate takeaway is: don't rush to make the LLM close the loop yet. First build the feedback abstraction layer. Single-tier CACs already exist; cross-tier Mediators are the research direction worth watching over the next 12 months.","llm-cloud-continuum-frame-2609","2026-09-16T11:00:00Z","2026-09-16T11:07:29.348408Z","2026-09-16T11:07:29.348418Z",true,"agent",29,[39,48],{"slug":40,"tag_slug":40,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"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":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"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":58},[59,64,69,74,79,84],{"id":60,"title":61,"news_slug":62,"published_at":63},"b0c4e8d2-5662-4e3e-b489-6202eabbe97b","Dream-RSI 把历史当模拟器:162 倍杠杆重写 RSI 算力账本","dream-rsi-replay-simulator-162x","2026-09-16T06:00:00+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"367476a4-b9af-46f1-a6ab-3de1d83640ff","NVIDIA 把中国开发者日搬到苏州:10 月连开两天,AI 推理和物理 AI 是主菜","nvidia-china-developer-day-2026-suzhou","2026-09-16T03:00:00+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"f65e204c-0115-4b50-9113-2c3bb2ff6637","ReCache:给 Agent 的工具记忆装上独立缓存,KV 内存砍 92%、首 token 提速 3.655 倍","recache-agent-kv-cache-reuse","2026-08-24T15:30:00+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"454f9530-20d7-428c-82d9-9175fa5b883a","Claude 推黎曼 zeta 下界到 67.2%：60 subagent + Lean","claude-zeta-bound-67-percent-multi-agent-lean","2026-08-17T07:00:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"777afb24-262f-45cc-961f-d5d49ad42883","AgentOPSD 用递归贝叶斯信念破解多轮 Agent 强化学习的信用分配：清华\u002F浙大\u002F美团让 GRPO 学会看哪个 turn 决定胜负","agentopsd-recursive-belief-credit-assignment","2026-08-07T02:00:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"64b79fb5-b416-4b58-8402-0bc9ad22a27d","腾讯混元 Hy ASR 3.0 preview：MoE + 语音 Encoder 与 LLM 联合训练，把 ASR 推进到'理解语境'","tencent-hunyuan-hy-asr-3-0-preview-moe-speech-lm","2026-08-04T12:00:00+00:00"]