🤖 AI 速览

今日关注点从模型规模转向系统可信度：Fyxer 和 Perplexity 展示智能体在邮箱、软件与生产系统中的持续执行能力，Occamy-1.0 则探索更高效的协作式开放模型。与此同时，多项研究指出，工具编排、数据溯源、可复现性与量化推理，正成为高风险场景中衡量 AI 能否真正落地的关键。
📋 文章元数据
发布时间
2026-09-15
类型
ai-daily
字数
5681
阅读时长
27 min

2026-09-15 AI日更 | 当语言模型遇上真实世界：AI落地的下一关是可验证的量化能力 链接到标题

今日关注点从模型规模转向系统可信度：Fyxer 和 Perplexity 展示智能体在邮箱、软件与生产系统中的持续执行能力，Occamy-1.0 则探索更高效的协作式开放模型。与此同时，多项研究指出，工具编排、数据溯源、可复现性与量化推理，正成为高风险场景中衡量 AI 能否真正落地的关键。

📖 本期 Watch List 深度导读 链接到标题

今天最值得关注的第一条主线，是智能体从“会调用模型”走向“可被信任的工作系统”：Fyxer 的 AI 行政助理聚焦跨邮箱、会议与应用的长期上下文，Occamy-1.0 则展示了面向复杂协作流程的开放模型；同时，关于模型能力与 Agent harness 的研究，提醒工程团队重新评估工具、提示词和控制流的真实贡献。第二条主线是可靠性与可验证性：从 R2VC 的检索—验证—置信度校准，到 draft-verify-revise 处理指示语歧义，再到量化推理与事件预测，核心问题都在于如何让模型在高风险场景中少“说得像”，多“证得明”。此外，前沿治理与基础设施的讨论也值得留意：从 Dario Amodei 的 AI 发展节奏之争，到算力、数据中心和软件栈的垂直整合，能力竞赛正在同时演变为治理与产业组织能力的竞争。

🌐 X 平台 AI 热点快讯 链接到标题

话题 1:Anthropic Launches Claude Mods for Custom Terminal Plugins 链接到标题

  • 分类:AI · News
  • 概况:热度时间:3 hours ago,相关帖子数:335
  • 是什么事:Anthropic 推出名为 Claude Mods 的功能，用于创建和定制终端插件。
  • 为什么重要:这表明 AI 助手正从独立对话工具扩展为可嵌入开发者终端工作流的平台，有助于提升自动化能力与生态扩展性。
  • 讨论概况:X 上的讨论焦点预计集中在插件的定制范围、开发体验、与现有终端工具的兼容性，以及安全、权限和维护成本等问题。

话题 2:Yann LeCun Mocks 2019 GPT-2 Release Fears 链接到标题

  • 分类:AI · Other
  • 概况:热度时间:1 day ago,相关帖子数:5800
  • 是什么事:Yann LeCun 回顾并嘲讽了 2019 年 GPT-2 发布时围绕潜在风险和影响的担忧。
  • 为什么重要:这反映了业界对生成式 AI 风险评估、技术进步速度以及开放发布策略的持续反思。
  • 讨论概况:X 上的讨论主要聚焦于当年对 GPT-2 能力的判断是否过度谨慎、后续模型发展是否证明了这些担忧，以及 AI 公司应如何权衡开放共享与安全风险。

话题 3:Meta’s Muse AI Agent Wins Early Praise for Everyday Help 链接到标题

  • 分类:AI · News
  • 概况:热度时间:2 days ago,相关帖子数:1900
  • 是什么事:Meta 的 Muse AI Agent 因在日常任务中的实用表现获得早期好评。
  • 为什么重要:这表明 AI Agent 正从单纯问答进一步走向能够协助处理现实日常事务的应用形态，有助于检验其实际落地价值。
  • 讨论概况:X 上的讨论主要集中在其便利性、任务完成效果与使用体验，同时也存在对可靠性、隐私安全及早期宣传是否过度的质疑。

话题 4:AI Speeds Product Building but Demands Deeper User Insight 链接到标题

  • 分类:AI · News
  • 概况:热度时间:7 hours ago,相关帖子数:804
  • 是什么事:讨论指出，AI 正在加速产品构建，但产品成功仍取决于更深入地理解用户需求。
  • 为什么重要:这表明 AI 可显著提高开发和迭代效率，却不能替代用户研究、需求判断与产品决策，反映出 AI 应用正从单纯提效转向人机协同。
  • 讨论概况:X 上的讨论焦点主要围绕 AI 带来的开发提速与用户洞察不足之间的矛盾：一方强调 AI 降低了产品构建门槛，另一方担心快速产出并不等于解决真实需求。

话题 5:Meta’s Muse AI Agent Tops Download Charts Days After Launch 链接到标题

  • 分类:AI · News
  • 概况:热度时间:21 hours ago,相关帖子数:2100
  • 是什么事:Meta 的 Muse AI Agent 上线数日后登上下载榜首，显示其在消费者市场迅速获得关注。
  • 为什么重要:这表明 AI 应用正从一次性问答转向能够持续执行任务的智能代理，有助于观察用户对代理式交互和产品化 AI 的真实需求。
  • 讨论概况:X 上的讨论集中在“循环才是新的提示词”这一观点，即通过持续迭代任务取代单次提示；同时也有人质疑下载量能否代表实际使用效果，并关注代理的准确性、隐私和自主行动风险。

话题 6:Community Ranks Top Multi-Agent Harnesses for AI Development 链接到标题

  • 分类:AI · News
  • 概况:热度时间:13 hours ago,相关帖子数:66
  • 是什么事:X 平台用户正在评选和讨论最适合 AI 开发的多智能体运行框架，同时将其与 AI 规模化部署、开放权重及治理等议题联系起来。
  • 为什么重要:多智能体框架决定了多个 AI 代理如何协作、调用工具和完成复杂开发任务，是推动 AI 从单模型应用走向工程化、规模化落地的重要基础设施。
  • 讨论概况:当前讨论聚焦于不同多智能体框架在开发效率、可靠性、可扩展性和安全治理方面的差异；部分观点关注其商业与国防应用前景，另一些则担忧开放权重、代理失控及监管缺位带来的风险。

今日 X 上的 AI 舆情小结 链接到标题

今天的舆论主线是，AI 正从对话工具加速演变为嵌入终端、能够持续执行任务并支持多智能体协作的工程化平台，Claude Mods、Muse Agent 及相关框架讨论共同体现了这一趋势。整体共识在于，AI 显著降低了产品开发和自动化门槛，并开始展现处理现实事务的价值，但产品成功仍取决于真实用户需求、可靠体验与持续迭代，而非单纯的模型能力或下载量。分歧主要集中在 GPT-2 等早期风险判断是否过度、开放权重与开放发布应走多远，以及代理式产品的宣传热度能否转化为稳定的实际使用效果。潜在风险则包括插件和代理的权限控制、隐私泄露、任务失误与自主行动失控，以及多智能体规模化部署和商业、国防应用可能带来的治理缺位。

💡 大佬观点(Influencer Insights) 链接到标题

今日大佬观点暂缺,推荐阅读 Watch List 深度内容。

📚 附录:今日 Watch List 更新源列表 链接到标题

时间窗口:最近 3 天;覆盖 22 个源;共 35 条更新

Acquired.fm (A_full) 链接到标题

  • The Home Depot
    • 发布时间:2026-09-14 11:35 北京时间
    • 摘要:【待翻译】- The Home Depot’s founding story is like an Avengers movie… if the Avengers got fired, went broke, and stacked empty paint cans ten feet high to look legitimate.
      • After being unceremoniously fired from their previous hardware chain at ages 48 and 35, Bernie Marcus and Arthur Blank took the words of their New York banker Ken Langone (who had also just accidentally caused their firings) to heart: they’d just been “kicked in the ass with a golden horseshoe.” They proceeded to author the greatest compounding story in American retail history, helped by some legendary cameos along the way from Sol Price, Jamie Dimon, and Ross Perot (to name a few).
      • And the ending is as good as any superhero film: from its 1981 IPO to today, Home Depot has been the single highest-returning equity in the entire US stock market — higher than Apple, Microsoft, Berkshire Hathaway, and everything else!
      • Many thanks to our fantastic Fall ‘26 Season partners:.
        • Sign up for email updates , get our takeaways and research photos from each episode, and vote on future topics!
    • EN 要点:
      • The Home Depot’s founding story is like an Avengers movie… if the Avengers got fired, went broke, and stacked empty paint cans ten feet high to look legitimate
      • After being unceremoniously fired from their previous hardware chain at ages 48 and 35, Bernie Marcus and Arthur Blank took the words of their New York banker K…
      • And the ending is as good as any superhero film: from its 1981 IPO to today, Home Depot has been the single highest-returning equity in the entire US stock mark…
      • Many thanks to our fantastic Fall ‘26 Season partners:

All-In Podcast (A_full) 链接到标题

  • Jensen Huang: The Doomer Hoax, Superintelligence Is Here, and The Future of AI (ft. President Trump)
    • 发布时间:2026-09-15 05:24 北京时间
    • 摘要:【待翻译】- IREN is a vertically integrated AI Cloud platform, delivering data centers, compute and software for AI training and inference.
      • Oracle connects the data, applications, and infrastructure that turn AI into business outcomes—with the flexibility, choice, and control to optimize as AI evolves.
      • EY helps tech innovators scale from startup to exit to megacap.
      • We’re focused on giving every person the tools to reach their full potential and making sure the benefits of technology are distributed to all.
      • Keel Infrastructure owns the power, land, and connectivity that HPC and AI run on - backed by secured energy assets and established grid interconnections across North America.
    • EN 要点:
      • (0:00) Jensen Huang joins The Besties
      • (1:39) Thoughts on Dario’s blog, Frontier Labs calling to slow down AI, and Doomer psychology
      • (9:58) Sensible AI regulation and RSI
      • (16:05) Hugging Face acquisition, future of Open Source, and the race with China

Stratechery by Ben Thompson (A_full) 链接到标题

  • Pacing the Frontier, AI’s Digital Limits, AI Commissars
    • 发布时间:2026-09-14 18:00 北京时间
    • 摘要:【待翻译】- Dario Amodei wants to pace the frontier; it’s an unrealistic proposal that seems mostly geared to political control of AI.
      • $15 / month or $150 / year.
      • Substantial analysis of the news of the day delivered via three weekly emails or podcasts.
      • Stratechery Interviews.
      • Interviews with leading public CEOs, private company founders, and discussions with fellow analysts.
    • EN 要点:
      • Dario Amodei wants to pace the frontier; it’s an unrealistic proposal that seems mostly geared to political control of AI.

OpenAI Blog (A_full) 链接到标题

  • How Fyxer built an AI executive assistant people trust

    • 发布时间:2026-09-14 20:00 北京时间
    • 摘要:【待翻译】- For many professionals, work means keeping track of conversations and commitments across inboxes, meetings, messages, and apps.
      • Without rock-solid context, those commitments can fall through the cracks, damaging projects and relationships.
      • Fyxer built an AI executive assistant that follows the thread as work moves across tools.
      • It combines the latest OpenAI models with more than 500,000 hours of executive assistant workflows, dividing the work among dozens of specialized models that improve through real user feedback.
      • Email is one of the clearest places to see it in action.
    • EN 要点:
      • Fyxer uses OpenAI models, fine-tuning, memory, and real user feedback to organize inboxes and draft emails in each user’s voice.

  • Perplexity trusts GPT-6 Astra with end-to-end systems

    • 发布时间:2026-09-14 08:00 北京时间
    • 摘要:【待翻译】- Perplexity uses Astra to write communications, change software, and monitor production systems, and checks in much less frequently than with earlier models.
      • As an AI-powered answer engine, Perplexity is deeply focused on search and accuracy.
      • Its ability to process large amounts of information is critically important.
      • Johnny Ho, Cofounder and Chief Strategy Officer, observes that every time the model gets better at writing code, Perplexity’s search engine improves too.
      • It becomes able to write better programs that search the web and internal information and summarize it very concisely.
    • EN 要点:
      • Perplexity uses Astra to write communications, change software, and monitor production systems, and checks in much less frequently than with earlier models.

ArXiv cs.AI (B_intro+search) 链接到标题

  • Occamy-1.0: Open Pareto-frontier 35B Intelligence for Co-work

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11977v1 Announce Type: new.
      • Abstract: Co-work agents execute complex workflows that combine information gathering, tool use, coding, and file manipulation across many model invocations.
      • Because cost and latency accumulate over the full episode, their practical value depends not only on peak capability but also on how efficiently that capability is delivered.
      • Yet many steps in everyday work emphasize state tracking, coordination, recovery, and follow-through rather than frontier-scale reasoning.
    • EN 要点:
      • arXiv:2609.11977v1 Announce Type: new
      • Abstract: Co-work agents execute complex workflows that combine information gathering, tool use, coding, and file manipulation across many model invocations
      • Because cost and latency accumulate over the full episode, their practical value depends not only on peak capability but also on how efficiently that capability…
      • Yet many steps in everyday work emphasize state tracking, coordination, recovery, and follow-through rather than frontier-scale reasoning

  • Harness or Model? Isolating the Harness Effect in Agentic Coding with a Contamination-Controlled Private Suite

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11987v1 Announce Type: new.
      • Abstract: An agentic coding system couples a language model to a harness: the tools, prompts and control flow that turn a chat model into an autonomous software engineer.
      • Vendors ship harnesses tuned to their own models, and practitioners assume the vendor-native pairing solves more tasks.
      • We measure that assumption with paired same-model contrasts on a private, contamination-controlled suite of 256 repository and post-cutoff contest tasks.
    • EN 要点:
      • arXiv:2609.11987v1 Announce Type: new
      • Abstract: An agentic coding system couples a language model to a harness: the tools, prompts and control flow that turn a chat model into an autonomous software…
      • Vendors ship harnesses tuned to their own models, and practitioners assume the vendor-native pairing solves more tasks
      • We measure that assumption with paired same-model contrasts on a private, contamination-controlled suite of 256 repository and post-cutoff contest tasks

  • Reading the Whole Heart: Latent-Attention Masked Autoencoders for Multimodal Cardiac Representation Learning

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12035v1 Announce Type: new.
      • Abstract: Cardiovascular diagnosis rests on integrating complementary modalities, like ECG, echocardiography, chest radiographs, and clinical variables, each capturing distinct but correlated aspects of cardiac physiology.
      • Yet most medical foundation models remain modality-specific, combining modalities only for finetuning or post-training.
      • This discards the cross-modal evidence clinicians naturally integrate and ignores the structure within each modality.
    • EN 要点:
      • arXiv:2609.12035v1 Announce Type: new
      • Abstract: Cardiovascular diagnosis rests on integrating complementary modalities, like ECG, echocardiography, chest radiographs, and clinical variables, each ca…
      • Yet most medical foundation models remain modality-specific, combining modalities only for finetuning or post-training
      • This discards the cross-modal evidence clinicians naturally integrate and ignores the structure within each modality

  • Competence-Gated Pooling of Language Models and Priors for Event Forecasting

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12101v1 Announce Type: new.
      • Abstract: In hybrid forecasting, a language model is often one of several available signals.
      • A system may already have a market, crowd, or statistical forecast and must decide whether the model adds useful information or should be ignored.
      • The relevant target is therefore not standalone model accuracy, but relative competence, defined as the model’s marginal value beyond the available external forecast.
    • EN 要点:
      • arXiv:2609.12101v1 Announce Type: new
      • Abstract: In hybrid forecasting, a language model is often one of several available signals
      • A system may already have a market, crowd, or statistical forecast and must decide whether the model adds useful information or should be ignored
      • The relevant target is therefore not standalone model accuracy, but relative competence, defined as the model’s marginal value beyond the available external for…

  • Language Is an Insufficient Substrate for Quantitative Reasoning, and Consequential Domains Need Large Quantitative Models

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12105v1 Announce Type: new.
      • Abstract: The prevailing assumption in applied machine learning is that progress on consequential quantitative decisions such as pricing risk, allocating capital, triaging patients, or containing a network intrusion will follow from progress in large language models (LLMs).
      • A language model is trained on a representation of the world that was produced by human description; description is a lossy encoding of the quantitative record, and the loss is irreversible: no downstream model, at any scale, can recover from a description what the description did not encode.
      • We formalize this as a property of the representation on which a model is trained rather than of the model capacity, and we identify three further properties that consequential settings demand of a model and that a language substrate cannot supply by construction: reproducibility, lineage from every output back to the source records that produced.
    • EN 要点:
      • arXiv:2609.12105v1 Announce Type: new
      • Abstract: The prevailing assumption in applied machine learning is that progress on consequential quantitative decisions such as pricing risk, allocating capita…
      • A language model is trained on a representation of the world that was produced by human description; description is a lossy encoding of the quantitative record,…
      • We formalize this as a property of the representation on which a model is trained rather than of the model capacity, and we identify three further properties th…

  • DU-NO: A Parameter-Efficient Double U-Shaped Neural Operator for Phase-Resolving Wave Modeling

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12115v1 Announce Type: new.
      • Abstract: Phase-resolving wave models such as FUNWAVE-TVD are the accuracy standard for nearshore dynamics, resolving the shoaling, refraction, and breaking of individual waves, but their cost rules them out for the ensembles, uncertainty quantification, and real-time warning that operational forecasting demands.
      • Neural operators promise solver-level accuracy at a fraction of that cost, yet on wave-dominated fields the accurate ones are large: hybrid spectral-convolutional operators such as U-FNO (the strongest baseline in our study after DU-NO) buy their fidelity with tens of millions of parameters.
      • We introduce DU-NO (Double U-shaped Neural Operator), a multiscale U-shaped spectral operator that attaches lightweight convolutional U-Net branches only at its two shallowest encoder and decoder levels.
    • EN 要点:
      • arXiv:2609.12115v1 Announce Type: new
      • Abstract: Phase-resolving wave models such as FUNWAVE-TVD are the accuracy standard for nearshore dynamics, resolving the shoaling, refraction, and breaking of…
      • Neural operators promise solver-level accuracy at a fraction of that cost, yet on wave-dominated fields the accurate ones are large: hybrid spectral-convolution…
      • We introduce DU-NO (Double U-shaped Neural Operator), a multiscale U-shaped spectral operator that attaches lightweight convolutional U-Net branches only at its…

  • When Successful Knowledge Graph Edits Displace Correct Answers: Rank-Level Locality beyond Parameter Support

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12116v1 Announce Type: new.
      • Abstract: Editing a knowledge graph embedding (KGE) model to promote a desired answer can displace correct answers from the returned list.
      • Locality tests based only on facts that reuse the edited parameter can miss this ranking effect.
      • We introduce a common rank-displacement audit at three scopes: facts supported by the edited parameter, other correct answers to the target query, and correct answers across queries with the same relation.
    • EN 要点:
      • arXiv:2609.12116v1 Announce Type: new
      • Abstract: Editing a knowledge graph embedding (KGE) model to promote a desired answer can displace correct answers from the returned list
      • Locality tests based only on facts that reuse the edited parameter can miss this ranking effect
      • We introduce a common rank-displacement audit at three scopes: facts supported by the edited parameter, other correct answers to the target query, and correct a…

  • Mined from Scientific Literature: Process Schemas for Atomic Layer Deposition and Etching in Materials Science

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12139v1 Announce Type: new.
      • Abstract: Atomic layer deposition (ALD) and atomic layer etching (ALE) are reported heterogeneously across experimental and simulation literature in materials science, hindering comparison and machine-actionable reuse.
      • We present four domain-expert-reviewed JSON Schemas for ALD and ALE experimental and simulation processes.
      • Curated with schema-miner and grounded in QUDT using schema-miner pro, the schemas structure materials, process conditions, configurations , and measured or predicted results.
    • EN 要点:
      • arXiv:2609.12139v1 Announce Type: new
      • Abstract: Atomic layer deposition (ALD) and atomic layer etching (ALE) are reported heterogeneously across experimental and simulation literature in materials s…
      • We present four domain-expert-reviewed JSON Schemas for ALD and ALE experimental and simulation processes
      • Curated with schema-miner and grounded in QUDT using schema-miner pro, the schemas structure materials, process conditions, configurations , and measured or pre…

  • Can LLMs in Draft-Verify-Revise Pipelines Resolve Deictic Ambiguity?

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12162v1 Announce Type: new.
      • Abstract: Draft-verify-revise is a common LLM orchestration pattern for scaling inference-time compute.
      • One LLM drafts, a second critiques the draft and provides feedback, and a third uses that feedback to revise the draft into the final output.
      • As context cascades between stages, LLMs at different stages can resolve a context-dependent expression such as “previous” differently.
    • EN 要点:
      • arXiv:2609.12162v1 Announce Type: new
      • Abstract: Draft-verify-revise is a common LLM orchestration pattern for scaling inference-time compute
      • One LLM drafts, a second critiques the draft and provides feedback, and a third uses that feedback to revise the draft into the final output
      • As context cascades between stages, LLMs at different stages can resolve a context-dependent expression such as “previous” differently

  • GLARE: Generative Learning via Adversarial Reward Estimation For Social Dynamics Forecasting

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12165v1 Announce Type: new.
      • Abstract: Meeting continuation requires tracking the agenda, speaker roles, participant intentions, and disagreement across long multi-party discussions.
      • We introduce the Meeting Dynamic Forecasting Benchmark (MDFB), constructed from 2,207 real-world meetings and 24,794 future-facing queries.
      • Given a transcript prefix and an active question, a model generates a plausible multi-turn continuation in one call.
    • EN 要点:
      • arXiv:2609.12165v1 Announce Type: new
      • Abstract: Meeting continuation requires tracking the agenda, speaker roles, participant intentions, and disagreement across long multi-party discussions
      • We introduce the Meeting Dynamic Forecasting Benchmark (MDFB), constructed from 2,207 real-world meetings and 24,794 future-facing queries
      • Given a transcript prefix and an active question, a model generates a plausible multi-turn continuation in one call

ArXiv cs.CL (B_intro+search) 链接到标题

  • R2VC: Modular Fact-Checking with Retrieval, Verification, and Confidence Calibration

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11955v1 Announce Type: new.
      • Abstract: Large language models are increasingly used for automated fact checking, but end-to-end prompting often entangles evidence retrieval, reasoning, and uncertainty estimation, making failures difficult to diagnose and confidence difficult to trust.
      • We present R2VC, a modular retrieve, reason, verify, calibrate architecture for evidence-grounded fact checking with citations and abstention.
      • R2VC combines hybrid sparse+dense retrieval over Wikipedia, a supervised fine-tuned and DPO-aligned generator that produces diverse structured verdict candidates, an external NLI cross-encoder for evidence-based candidate selection, and a lightweight sequence-level calibrator for confidence estimation and selective abstention.
    • EN 要点:
      • arXiv:2609.11955v1 Announce Type: new
      • Abstract: Large language models are increasingly used for automated fact checking, but end-to-end prompting often entangles evidence retrieval, reasoning, and u…
      • We present R2VC, a modular retrieve, reason, verify, calibrate architecture for evidence-grounded fact checking with citations and abstention
      • R2VC combines hybrid sparse+dense retrieval over Wikipedia, a supervised fine-tuned and DPO-aligned generator that produces diverse structured verdict candidate…

  • What Counts as a Mistake? Annotating Recitation Events in Quran Memorization Transcripts

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12085v1 Announce Type: new.
      • Abstract: Checking Quran recitation from an ASR transcript requires distinguishing unresolved mistakes from repetitions, repairs, opening formulas and accepted spelling differences.
      • We report a completed human annotation of 100 production recording cases: 348 scored units and 162 localized events across ten combined labels.
      • An executable evaluator scores labels and word positions together.
    • EN 要点:
      • arXiv:2609.12085v1 Announce Type: new
      • Abstract: Checking Quran recitation from an ASR transcript requires distinguishing unresolved mistakes from repetitions, repairs, opening formulas and accepted…
      • We report a completed human annotation of 100 production recording cases: 348 scored units and 162 localized events across ten combined labels
      • An executable evaluator scores labels and word positions together

  • Extracting Dataset Mentions in Forced Displacement and FCV Documents: A Weakly Supervised Framework with LLM-Based Label Refinement

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12107v1 Announce Type: new.
      • Abstract: Development and humanitarian organizations produce and support surveys, administrative registries, and other data resources to inform research, policy, and operations, yet systematically identifying where these datasets are referenced remains difficult.
      • Such references are dispersed across research papers, project documents, humanitarian reports, and other unstructured text, limiting both the ability to trace data use and to identify potential gaps in data availability or dissemination.
      • We present a weakly supervised framework for adapting dataset extraction to forced displacement and Fragile, Conflict, and Violence (FCV) documents without first constructing a large manually labeled training corpus.
    • EN 要点:
      • arXiv:2609.12107v1 Announce Type: new
      • Abstract: Development and humanitarian organizations produce and support surveys, administrative registries, and other data resources to inform research, policy…
      • Such references are dispersed across research papers, project documents, humanitarian reports, and other unstructured text, limiting both the ability to trace d…
      • We present a weakly supervised framework for adapting dataset extraction to forced displacement and Fragile, Conflict, and Violence (FCV) documents without firs…

  • The Cost of Compression: A Rate-Distortion Limit on Factual Hallucination

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12111v1 Announce Type: new.
      • Abstract: Factual hallucination in closed-book question answering is often treated as a coverage problem: a model fails because the relevant fact is absent from its internal memory.
      • This view misses a second source of error.
      • Even when a fact has been observed, finite memory may force it to be stored only approximately.
    • EN 要点:
      • arXiv:2609.12111v1 Announce Type: new
      • Abstract: Factual hallucination in closed-book question answering is often treated as a coverage problem: a model fails because the relevant fact is absent from…
      • This view misses a second source of error
      • Even when a fact has been observed, finite memory may force it to be stored only approximately

  • Quantifying Consonant Contributions to Word Intelligibility via Acoustic Masking

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12122v1 Announce Type: new.
      • Abstract: Consonants contribute unequally to whether a word is understood.
      • Given the limited time available for therapy, ranking consonants by contribution to intelligibility helps prioritize intervention targets in motor speech disorders.
      • However, measuring this contribution relies on perceptual studies that are difficult to scale.
    • EN 要点:
      • arXiv:2609.12122v1 Announce Type: new
      • Abstract: Consonants contribute unequally to whether a word is understood
      • Given the limited time available for therapy, ranking consonants by contribution to intelligibility helps prioritize intervention targets in motor speech disord…
      • However, measuring this contribution relies on perceptual studies that are difficult to scale

  • Local Edits, Global Ripples: Replay-Informed Policy Adaptation for Workflow Synthesis

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12127v1 Announce Type: new.
      • Abstract: Prompt-policy editing offers a practical way to improve agents that synthesize executable workflows without updating the underlying model.
      • However, persistent prompt editing has two coupled properties.
      • First, edit locality does not imply effect locality: an edit confined to one policy segment can ripple through downstream execution, altering behavior beyond the edited segment.
    • EN 要点:
      • arXiv:2609.12127v1 Announce Type: new
      • Abstract: Prompt-policy editing offers a practical way to improve agents that synthesize executable workflows without updating the underlying model
      • However, persistent prompt editing has two coupled properties
      • First, edit locality does not imply effect locality: an edit confined to one policy segment can ripple through downstream execution, altering behavior beyond th…

  • Population-level measures of perceived food access reveal barriers beyond geographic proximity

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12132v1 Announce Type: new.
      • Abstract: Food access is multidimensional, but population-level measurement still relies heavily on geography because perceived dimensions of access are difficult to measure at scale.
      • Here, we use 25,125 Google Maps reviews from 49 grocery stores in Raleigh, North Carolina, to measure five dimensions of food access: availability, accessibility, affordability, accommodation, and acceptability.
      • We identify review topics with unsupervised topic modeling and assign them to access dimensions using zero-shot classification, with 85.4% agreement against manual coding.
    • EN 要点:
      • arXiv:2609.12132v1 Announce Type: new
      • Abstract: Food access is multidimensional, but population-level measurement still relies heavily on geography because perceived dimensions of access are difficu…
      • Here, we use 25,125 Google Maps reviews from 49 grocery stores in Raleigh, North Carolina, to measure five dimensions of food access: availability, accessibilit…
      • We identify review topics with unsupervised topic modeling and assign them to access dimensions using zero-shot classification, with 85.4% agreement against man…

  • GAUGE: When Not to Trust LLM-as-a-Judge in User-Simulated Evaluation of Task-Oriented Agents

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12191v1 Announce Type: new.
      • Abstract: Comparing and selecting task-oriented LLM agents increasingly relies on a low-cost offline evaluation gate: persona-driven LLM user-simulators converse with each candidate, an LLM-as-a-judge scores the transcripts, and the higher-scoring agent is promoted.
      • We introduce GAUGE, a reusable offline protocol that measures whether this gate’s ranking matches a grounded verifiable reward across 25 agents from six providers on the $\tau^2$-bench and SimulatorArena benchmarks, separating two kinds of evaluation validity that release practices conflate: ranking validity and construct validity.
      • First, a satisfaction-success gap: satisfaction carries essentially no information about task success, as conversations rated satisfied by our blind panel are decorrelated from actual success, with 57.5% of them failing the customer’s task, a pattern consistent across five rater populations, both benchmarks, and every subjective dimension we rated.
    • EN 要点:
      • arXiv:2609.12191v1 Announce Type: new
      • Abstract: Comparing and selecting task-oriented LLM agents increasingly relies on a low-cost offline evaluation gate: persona-driven LLM user-simulators convers…
      • We introduce GAUGE, a reusable offline protocol that measures whether this gate’s ranking matches a grounded verifiable reward across 25 agents from six provide…
      • First, a satisfaction-success gap: satisfaction carries essentially no information about task success, as conversations rated satisfied by our blind panel are d…

  • Repair Before Reinforce: Context-Augmented Knowledge Graph Reasoning for Multi-Hop Question Answering

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12230v1 Announce Type: new.
      • Abstract: Question-answering often requires reasoning across multiple connected facts rather than retrieving a single isolated relation.
      • Knowledge graphs (KGs) provide a structured way to represent such facts, but training large language models (LLMs) only on isolated KG head-relation-tail triples may limit their ability to learn the surrounding context needed for multi-hop reasoning.
      • In this work, we propose a context-augmented training framework for multi-hop question-answering.
    • EN 要点:
      • arXiv:2609.12230v1 Announce Type: new
      • Abstract: Question-answering often requires reasoning across multiple connected facts rather than retrieving a single isolated relation
      • Knowledge graphs (KGs) provide a structured way to represent such facts, but training large language models (LLMs) only on isolated KG head-relation-tail triple…
      • In this work, we propose a context-augmented training framework for multi-hop question-answering

  • Chopthin-Consensus Power Sampling: A Diversity-Preserving Approach to LLM Decoding

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.12243v1 Announce Type: new.
      • Abstract: Inference-time power sampling via Sequential Monte Carlo (SMC) can substantially improve large language model (LLM) reasoning without requiring post-training.
      • However, many existing SMC approaches rely on equal-weight resampling, which can aggressively prune low-weight trajectories, discarding potentially correct reasoning paths and degrading the genealogical diversity of the search space.
      • To address this, we introduce Chopthin-Consensus Power Sampling (CCPS).
    • EN 要点:
      • arXiv:2609.12243v1 Announce Type: new
      • Abstract: Inference-time power sampling via Sequential Monte Carlo (SMC) can substantially improve large language model (LLM) reasoning without requiring post-t…
      • However, many existing SMC approaches rely on equal-weight resampling, which can aggressively prune low-weight trajectories, discarding potentially correct reas…
      • To address this, we introduce Chopthin-Consensus Power Sampling (CCPS)

ArXiv cs.LG (B_intro+search) 链接到标题

  • Fundamental Dynamical Units for Physics-Informed Structural Inference from Perturbation Time-Series in Networked Systems

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11934v1 Announce Type: new.
      • Abstract: In networked dynamical systems, the parameter of primary mechanistic interest is signed interaction structure.
      • Recovering this structure from perturbation time-series data is a fundamental identification problem, compounded by three coupled obstacles: the combinatorial complexity of interaction architectures, ambiguity of causal attribution under limited interventions, and state-dependent dynamics that confound structural inference.
      • Each obstacle is structural in origin and calls for a structural solution.
    • EN 要点:
      • arXiv:2609.11934v1 Announce Type: new
      • Abstract: In networked dynamical systems, the parameter of primary mechanistic interest is signed interaction structure
      • Recovering this structure from perturbation time-series data is a fundamental identification problem, compounded by three coupled obstacles: the combinatorial c…
      • Each obstacle is structural in origin and calls for a structural solution

  • Physics-Informed Conformal Prediction: Embedding PDE Consistency into Distribution-Free Uncertainty Quantification for Neural Operators

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11935v1 Announce Type: new.
      • Abstract: Neural operators such as the Fourier Neural Operator (FNO) achieve remarkable accuracy in approximating solutions to partial differential equations (PDEs).
      • However, providing rigorous uncertainty estimates remains an open challenge.
      • We propose Physics-Informed Conformal Prediction (PI-CP), a framework that embeds PDE residuals into the nonconformity score of split conformal prediction, producing prediction intervals that are (i) distribution-free with provable coverage guarantees, and (ii) spatially adaptive when the PDE residual correlates with prediction error – tighter where physics is well-satisfied, wider where it is violated.
    • EN 要点:
      • arXiv:2609.11935v1 Announce Type: new
      • Abstract: Neural operators such as the Fourier Neural Operator (FNO) achieve remarkable accuracy in approximating solutions to partial differential equations (P…
      • However, providing rigorous uncertainty estimates remains an open challenge
      • We propose Physics-Informed Conformal Prediction (PI-CP), a framework that embeds PDE residuals into the nonconformity score of split conformal prediction, prod…

  • Fed-Equilibrium Framework for Topological Pareto Control in Robust and Fair Clinical Federated Learning

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11937v1 Announce Type: new.
      • Abstract: The deployment of Federated Learning (FL) in multi-center clinical networks faces the challenge of “knowledge dominance,” where high-volume hubs naturally overwhelm minority community nodes, implicitly treating the distinct clinical patterns of smaller cohorts as outliers.
      • Existing geometric defenses provide a security baseline but leave this efficiency-fairness dilemma unresolved.
      • To bridge this gap, we propose Fed-Equilibrium, a framework that advances the paradigm from simple defense to topological equilibrium.
    • EN 要点:
      • arXiv:2609.11937v1 Announce Type: new
      • Abstract: The deployment of Federated Learning (FL) in multi-center clinical networks faces the challenge of “knowledge dominance,” where high-volume hubs natur…
      • Existing geometric defenses provide a security baseline but leave this efficiency-fairness dilemma unresolved
      • To bridge this gap, we propose Fed-Equilibrium, a framework that advances the paradigm from simple defense to topological equilibrium

  • Efficient AI Model Deployment Using Quantization Analysis Tool

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11954v1 Announce Type: new.
      • Abstract: As deep learning models are increasingly deployed on resource constrained devices, the demand for efficient model optimization techniques continues to grow.
      • Effective deployment of AI models on edge and low power platforms requires optimization methods that reduce model size and computational cost while maintaining high accuracy.
      • This paper presents Quantization Analysis Tool, a practical system designed to streamline quantization workflows and support performance efficient model deployment.
    • EN 要点:
      • arXiv:2609.11954v1 Announce Type: new
      • Abstract: As deep learning models are increasingly deployed on resource constrained devices, the demand for efficient model optimization techniques continues to…
      • Effective deployment of AI models on edge and low power platforms requires optimization methods that reduce model size and computational cost while maintaining…
      • This paper presents Quantization Analysis Tool, a practical system designed to streamline quantization workflows and support performance efficient model deploym…

  • Performance, Efficiency and Collapse – Advantages and Challenges in Offline Post-training of Code LLMs

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11956v1 Announce Type: new.
      • Abstract: Post-training with reinforcement learning (RL) is a critical phase in the development of code-generating large language models (LLMs), as it ensures adherence to instructions and the production of functionally correct code.
      • This process typically requires computationally intensive code sample generation from Transformer-based LLMs and substantial GPU-CPU communication for sequence verification.
      • To address these computational challenges, this work examines whether RL-based post-training can be performed entirely offline by leveraging existing datasets rather than generating new samples.
    • EN 要点:
      • arXiv:2609.11956v1 Announce Type: new
      • Abstract: Post-training with reinforcement learning (RL) is a critical phase in the development of code-generating large language models (LLMs), as it ensures a…
      • This process typically requires computationally intensive code sample generation from Transformer-based LLMs and substantial GPU-CPU communication for sequence…
      • To address these computational challenges, this work examines whether RL-based post-training can be performed entirely offline by leveraging existing datasets r…

  • Look Before You Leap: Pre-Action Verification for LLM Agents

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11957v1 Announce Type: new.
      • Abstract: An LLM agent acts on the world by emitting actions: shell commands to run, edits to apply.
      • A wrong action does not always fail loudly; it can fail silently, producing a plausible but incorrect effect that raises no error.
      • We argue that a cheap deterministic check, run before an action takes effect, is an effective and underused form of agent oversight, and we study it across two action modalities in one framework.
    • EN 要点:
      • arXiv:2609.11957v1 Announce Type: new
      • Abstract: An LLM agent acts on the world by emitting actions: shell commands to run, edits to apply
      • A wrong action does not always fail loudly; it can fail silently, producing a plausible but incorrect effect that raises no error
      • We argue that a cheap deterministic check, run before an action takes effect, is an effective and underused form of agent oversight, and we study it across two…

  • Decoding Mixture Perception through Computational Modeling of Component Interactions

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11958v1 Announce Type: new.
      • Abstract: Olfaction played an indispensable role throughout human evolution and civilization.
      • Even in the contemporary era of advanced technology, olfaction remains a critical channel for person to conduct danger discrimination, emotional experience, and memory formation.
      • However, most substances in nature exist as multi-molecule mixtures.
    • EN 要点:
      • arXiv:2609.11958v1 Announce Type: new
      • Abstract: Olfaction played an indispensable role throughout human evolution and civilization
      • Even in the contemporary era of advanced technology, olfaction remains a critical channel for person to conduct danger discrimination, emotional experience, and…
      • However, most substances in nature exist as multi-molecule mixtures

  • Space as an Interventional Invariant: Cross-Modal Predictive Geometry for Stratified Cities and Em-Spaced Intelligence

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11959v1 Announce Type: new.
      • Abstract: Space is a foundational concept across mathematics, physics, spatial cognition, urban science, and embodied intelligence, yet these fields often treat spatial structure either as a shared geometric container or as a collection of disconnected representations.
      • Such approaches struggle to explain how heterogeneous sensory and urban processes can jointly reveal a common spatial structure, particularly when different modalities do not share the same metric or representation.
      • This paper addresses this gap by defining space as an interventional invariant: the minimal relational structure that preserves local compatibility and the conditional laws of future observations under admissible actions.
    • EN 要点:
      • arXiv:2609.11959v1 Announce Type: new
      • Abstract: Space is a foundational concept across mathematics, physics, spatial cognition, urban science, and embodied intelligence, yet these fields often treat…
      • Such approaches struggle to explain how heterogeneous sensory and urban processes can jointly reveal a common spatial structure, particularly when different mod…
      • This paper addresses this gap by defining space as an interventional invariant: the minimal relational structure that preserves local compatibility and the cond…

  • On-Device Language Models for Privacy-Preserving Stress Prediction: A Multimodal Evaluation on Mobile Health

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11961v1 Announce Type: new.
      • Abstract: Stress is a pervasive determinant of mental health and a key target for mobile health interventions.
      • On-device language models (ODLMs) offer privacy-preserving inference without cloud dependency, yet their feasibility for health prediction under mobile resource constraints remains underexplored.
      • We evaluate ODLMs for multi-modal stress prediction using zero-shot prompting, measuring predictive accuracy alongside latency and throughput.
    • EN 要点:
      • arXiv:2609.11961v1 Announce Type: new
      • Abstract: Stress is a pervasive determinant of mental health and a key target for mobile health interventions
      • On-device language models (ODLMs) offer privacy-preserving inference without cloud dependency, yet their feasibility for health prediction under mobile resource…
      • We evaluate ODLMs for multi-modal stress prediction using zero-shot prompting, measuring predictive accuracy alongside latency and throughput

  • FINESSE: An Agent-Based Simulator and Benchmark Dataset for Multimodal Financial Event Sequences

    • 发布时间:2026-09-14 12:00 北京时间
    • 摘要:【待翻译】- arXiv:2609.11993v1 Announce Type: new.
      • Abstract: Machine learning research in financial services is limited by the scarcity of representative open-source datasets.
      • Existing resources are often narrowly focused on a single modality or task and fail to reflect the structured, multimodal, and dynamic nature inherent to many problems in financial services.
      • In this paper, we introduce FINESSE, a Financial Event Sequence Simulation Environment, an agent-based simulation framework for generating synthetic, structured datasets composed of multiple interdependent event streams.
    • EN 要点:
      • arXiv:2609.11993v1 Announce Type: new
      • Abstract: Machine learning research in financial services is limited by the scarcity of representative open-source datasets
      • Existing resources are often narrowly focused on a single modality or task and fail to reflect the structured, multimodal, and dynamic nature inherent to many p…
      • In this paper, we introduce FINESSE, a Financial Event Sequence Simulation Environment, an agent-based simulation framework for generating synthetic, structured…