{
  "title": "2026-10-13 AI Daily | Edit Chains Address Autoregressive Shortcomings, Agent Evaluation Enters Mutual Difficulty-Setting Phase",
  "url": "https://miaok.ong/en/ai-daily/ai-daily-2026-10-13/",
  "date": "2026-10-13T07:00:00+08:00",
  "lastmod": "2026-10-13T07:00:00+08:00",
  "type": "ai-daily",
  "kind": "page",
  "language": "en",
  "description": "Today\u0026rsquo;s core changes are concentrated on reasoning and evaluation: AutoIndexer trains causal models using edit chains to overcome the limitation that autoregressive generation cannot make post-hoc edits; SWE-Duel uses a self-expanding arena where agents set task difficulty for each other, driving evaluation from static to dynamic; Kairos and algebraic integer quantization provide byte-level multimodal unification and error-free multiply-accumulate paths.",
  "keywords": null,
  "tags": [],
  "categories": [],
  "author": "Mark (Miao) Kong",
  "image": "https://miaok.ong/images/avatar.jpg",
  "content": "\u003ch1 id=\"2026-10-13-ai-daily--edit-chains-address-autoregressive-shortcomings-agent-evaluation-enters-mutual-difficulty-setting-phase\"\u003e\n  2026-10-13 AI Daily | Edit Chains Address Autoregressive Shortcomings, Agent Evaluation Enters Mutual Difficulty-Setting Phase\n  \u003ca class=\"heading-link\" href=\"#2026-10-13-ai-daily--edit-chains-address-autoregressive-shortcomings-agent-evaluation-enters-mutual-difficulty-setting-phase\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h1\u003e\n\u003cblockquote\u003e\n\u003cp\u003eToday\u0026rsquo;s core shifts are concentrated across inference and evaluation: AutoIndexer trains causal models on \u0026ldquo;edit chains\u0026rdquo; to compensate for autoregression\u0026rsquo;s inability to modify outputs post-hoc; SWE-Duel introduces a self-expanding arena where agents set task difficulty for each other, driving evaluation from static to dynamic; Kairos and algebraic integer quantization offer byte-level multimodal unification and an error-free multiply-accumulate path.\u003c/p\u003e\n\u003c/blockquote\u003e\n\u003ch2 id=\"-in-depth-guide-to-this-issues-watch-list\"\u003e\n  📖 In-Depth Guide to This Issue\u0026rsquo;s Watch List\n  \u003ca class=\"heading-link\" href=\"#-in-depth-guide-to-this-issues-watch-list\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h2\u003e\n\u003cp\u003eThree sets of updates are worth highlighting today. First, new paradigms in inference and decoding: AutoIndexer trains causal models on \u0026ldquo;edit chains,\u0026rdquo; compensating for autoregression\u0026rsquo;s inability to perform post-hoc modifications; parallel draft context correction in speculative decoding and elastic latent recurrence optimize inference efficiency and recurrent models, respectively. Second, upgrades in Agent evaluation methodology: SWE-Duel proposes a self-expanding arena where tasks are generated during evaluation and difficulty is mutually set by agents; prompt optimization research on Deep-Research Agent measures cross-topic transfer and overfitting; perturber turns semantic-preserving perturbation into a reusable service. Third, underlying foundation models and quantization: Kairos explores byte-level multimodal unification, while algebraic integer quantization provides an error-free multiply-accumulate path, making this section well suited for readers interested in infrastructure.\u003c/p\u003e\n\u003ch2 id=\"-x-platform-ai-highlights\"\u003e\n  🌐 X Platform AI Highlights\n  \u003ca class=\"heading-link\" href=\"#-x-platform-ai-highlights\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h2\u003e\n\u003cp\u003eToday, bird did not detect any qualifying AI topics.\u003c/p\u003e\n\u003ch2 id=\"-influencer-insights\"\u003e\n  💡 Influencer Insights\n  \u003ca class=\"heading-link\" href=\"#-influencer-insights\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h2\u003e\n\u003cblockquote\u003e\n\u003cp\u003eInfluencer insights are currently unavailable today; we recommend reading the in-depth Watch List content.\u003c/p\u003e\n\u003c/blockquote\u003e\n\u003ch2 id=\"-appendix-todays-watch-list-update-source-list\"\u003e\n  📚 Appendix: Today\u0026rsquo;s Watch List Update Source List\n  \u003ca class=\"heading-link\" href=\"#-appendix-todays-watch-list-update-source-list\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h2\u003e\n\u003cblockquote\u003e\n\u003cp\u003eTime window: Last 3 days; Covering 22 sources; Total of 32 updates\u003c/p\u003e\n\u003c/blockquote\u003e\n\u003ch3 id=\"all-in-podcast-a_full\"\u003e\n  All-In Podcast (A_full)\n  \u003ca class=\"heading-link\" href=\"#all-in-podcast-a_full\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003e\u003ca href=\"https://allinchamathjason.libsyn.com/martin-shkreli-big-pharmas-blind-spot-betting-against-quantum-what-prison-taught-him\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eMartin Shkreli: Big Pharma\u0026rsquo;s Blind Spot, Betting Against Quantum \u0026amp; What Prison Taught Him\u003c/a\u003e\u003c/strong\u003e\n\u003cul\u003e\n\u003cli\u003eRelease time: 2026-10-12 23:49 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary:\n\u003cul\u003e\n\u003cli\u003eIREN is a vertically integrated AI cloud platform providing data centers, compute, and software for AI training and inference.\u003c/li\u003e\n\u003cli\u003eOracle connects data, applications, and infrastructure to transform AI into business outcomes, delivering optimized flexibility, choice, and control as AI evolves.\u003c/li\u003e\n\u003cli\u003eEY helps tech innovators from startups to exits to large enterprises.\u003c/li\u003e\n\u003cli\u003eWe are committed to providing tools for everyone to reach their full potential and ensuring that all benefit from technology.\u003c/li\u003e\n\u003cli\u003eKeel Infrastructure possesses the power, land, and connectivity required for high-performance computing and AI operations, backed by secure energy assets and grid interconnections across North America.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Takeaways:\n\u003cul\u003e\n\u003cli\u003e(0:00) Martin Shkreli joins the Besties\u003c/li\u003e\n\u003cli\u003e(0:55) Why quantum computing is a one-trick business and why he\u0026rsquo;s betting against it\u003c/li\u003e\n\u003cli\u003e(7:42) The drugs big pharma sells too cheap and his journey from hedge fund trader to pharma\u003c/li\u003e\n\u003cli\u003e(11:37) AI, drug discovery, and who actually wins\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"stratechery-by-ben-thompson-a_full\"\u003e\n  Stratechery by Ben Thompson (A_full)\n  \u003ca class=\"heading-link\" href=\"#stratechery-by-ben-thompson-a_full\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003e\u003ca href=\"https://stratechery.com/2026/personal-day-4/\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003ePersonal Day\u003c/a\u003e\u003c/strong\u003e\n\u003cul\u003e\n\u003cli\u003eRelease time: 2026-10-12 18:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary:\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003e$15\u003c/strong\u003e/month \u003cem\u003eor\u003c/em\u003e \u003cstrong\u003e$150\u003c/strong\u003e/year.\u003c/li\u003e\n\u003cli\u003eSubstantive analysis of the day\u0026rsquo;s news via three weekly emails or podcasts.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eStrategy Interviews\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eInterviews with leading public company CEOs, private company founders, and discussions with analyst peers.\u003c/li\u003e\n\u003cli\u003eA twice-weekly podcast produced by John Gruber and myself: 15 minutes per episode, not a minute more or less.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Takeaways:\n\u003cul\u003e\n\u003cli\u003ePersonal Day — No Update\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"arxiv-csai-b_introsearch\"\u003e\n  ArXiv cs.AI (B_intro+search)\n  \u003ca class=\"heading-link\" href=\"#arxiv-csai-b_introsearch\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12493\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eFaithfulness Without Anchors: A Position Paper on the Structural Misalignments of Explainable AI Faithfulness Metrics for Feature Importance\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12493v1 Announcement type: new.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAbstract: Explainable Artificial Intelligence (XAI) tools are often presented as methods to increase trust in machine learning models by making various aspects of the algorithm more understandable to humans.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eHowever, evidence accumulated over the past decade points to an overall failure of this objective, instead promoting overreliance on incorrect predictions, which can subsequently lead to algorithm aversion.\u003c/li\u003e\n\u003cli\u003eRather, XAI appears to function primarily as a debugging tool, useful for obtaining anecdotal insights into, for example, features learned by the model.\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12493v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Explainable Artificial Intelligence (XAI) tools have often been presented as methods to increase trust in machine learning models by making aspects of…\u003c/li\u003e\n\u003cli\u003eHowever, evidence accumulated over the last decade points to an overall failure of this objective, instead promoting overreliance on incorrect predictions, whic…\u003c/li\u003e\n\u003cli\u003eRather, XAI appears to function primarily as a debugging tool, useful for obtaining anecdotal insights into, for example, features learned by the model\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12536\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eEliteTable: Self-Improving Program Search for Executable Control Policies\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublished Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12536v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: EliteTable is a self-improving program-search algorithm guided by large language models (LLMs) that uses a fixed-capacity table of high-scoring solutions to guide new proposals, edits to existing solutions, and the remixing of multiple solutions.\u003c/li\u003e\n\u003cli\u003ePerformance was evaluated across six Gymnasium control tasks, each with ten generations and fifty candidate slots per generation.\u003c/li\u003e\n\u003cli\u003eEach task was evaluated across ten independently initialized searches.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12536v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: EliteTable is a self-improving program-search algorithm guided by large language models (LLMs) that uses a fixed-capacity table of high-scoring soluti…\u003c/li\u003e\n\u003cli\u003ePerformance was evaluated across six Gymnasium control tasks, each with ten generations and fifty candidate slots per generation\u003c/li\u003e\n\u003cli\u003eEach task was evaluated across ten independently initialized searches\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12542\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eEqual-Budget Monte Carlo Tree Search versus Randomized-Restart List Scheduling for DAG Workflows: A Controlled Equal-Budget Computational Study\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublished Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12542v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Monte Carlo Tree Search (MCTS) is often proposed for scheduling directed acyclic graph (DAG) workflows, but most studies never perform comparisons that could refute it—by giving a simpler method the same search effort.\u003c/li\u003e\n\u003cli\u003eWe perform comparisons under a controlled protocol.\u003c/li\u003e\n\u003cli\u003eAll arms receive the same budget—building and evaluating the same number of candidate schedules: an MCTS scheduling method pre-registered as the primary arm; a randomized-restart list scheduler without a search tree as a control; and a textbook UCT MCTS branch as a registered follow-up branch on the same domain.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003earXiv:2610.12542v1 Announce Type: new\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eAbstract: Monte Carlo Tree Search (MCTS) is often proposed for scheduling directed-acyclic-graph (DAG) workflows, yet most studies never run the comparison that…\u003c/li\u003e\n\u003cli\u003eWe run that comparison under a controlled protocol\u003c/li\u003e\n\u003cli\u003eAll arms receive an identical budget - the same number of candidate schedules built and evaluated: an MCTS scheduling method, preregistered as the main arm; a r…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12593\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eKnow What You Don\u0026rsquo;t Know: Agentic Data Synthesis by Evolving Generation Space for Closed-Loop Learning\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12593v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: The continued improvement of large language models increasingly depends on synthetic experience, yet effective data synthesis for interactive agents requires not only executable and verifiable trajectories, but also interactions that provide learning value for the current task agent.\u003c/li\u003e\n\u003cli\u003eExisting methods increasingly incorporate environment feedback, data validation, or learner failures, but still face a feedback-to-decision gap: observed weaknesses do not directly specify how future synthesis should change, or whether they can be represented by the current synthesis space.\u003c/li\u003e\n\u003cli\u003eTo address this gap, we propose EvoSyn, a learning-feedback-driven multi-agent framework that organizes synthesis around a revisable state, specifying what interaction factors can be intentionally generated and where synthesis effort should be allocated.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12593v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: The continued improvement of large language models increasingly depends on synthetic experience, yet effective data synthesis for interactive agents r…\u003c/li\u003e\n\u003cli\u003eExisting methods increasingly incorporate environment feedback, data validation, or learner failures, but still face a feedback-to-decision gap: observed weakne…\u003c/li\u003e\n\u003cli\u003eTo address this gap, we propose EvoSyn, a learning-feedback-driven multi-agent framework that organizes synthesis around a revisable state specifying what inter…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12596\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eSpireBench: Evaluating Long-Horizon Strategic Play of Agentic Models with Slay the Spire\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12596v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Agentic large language models can now handle complex tasks, but long-horizon strategic reasoning remains understudied: most agentic benchmarks are short and episodic, and performance on long-horizon tasks is difficult to disentangle into memorized knowledge, reasoning, and luck.\u003c/li\u003e\n\u003cli\u003eWe introduce SpireBench, where models play the full commercial deck-builder Slay the Spire (StS), in which early decisions dictate long-term success.\u003c/li\u003e\n\u003cli\u003eAcross 20 large language models (LLMs) on the game\u0026rsquo;s highest difficulty, frontier models beat heuristic baselines and match average human players, but still lag far behind experts.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12596v1 Announce Type: new\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAbstract: Agentic large language models now tackle complex tasks, yet strategic reasoning over long horizons is under-studied: most agentic benchmarks are short…\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eWe introduce SpireBench, in which models play the complete commercial deck-builder Slay the Spire (StS), where early decisions decide long-term success\u003c/li\u003e\n\u003cli\u003eAcross 20 large language models (LLMs) at the game\u0026rsquo;s highest difficulty, frontier models beat heuristic baselines and match the average human player but remain…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12598\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eBeyond Binary Verifiable Rewards: Behavioral Self-Rewarding for SWE Agents\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12598v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Reinforcement learning with verifiable rewards (RLVR) provides reliable supervision for software engineering agents through executable tests, but binary rewards cannot distinguish trajectories with significantly different debugging qualities.\u003c/li\u003e\n\u003cli\u003eWe investigate whether current policies can judge their own trajectories and use these judgments to improve task performance without a separately trained reward model.\nTo this end, we introduce Behavioral Self-Rewarding (BSR), a reward-refinement method that reuses the policy to compare observable debugging behavior among trajectories of the same task with identical verifier outcomes.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12598v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Reinforcement learning with verifiable rewards (RLVR) provides reliable supervision for software engineering agents through executable tests, but bina…\u003c/li\u003e\n\u003cli\u003eWe investigate whether the current policy can judge its own trajectories and use these judgments to improve task performance without a separately trained reward…\u003c/li\u003e\n\u003cli\u003eTo this end, we introduce Behavioral Self-Rewarding (BSR), a reward-refinement method that reuses the policy to compare observable debugging behavior among traj…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12655\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eCLEAT: Learning Value-Guided Clarification for Interactive Agents\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12655v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: LLM agents can execute increasingly complex tasks, but effective assistance also requires aligning their actions with goals and constraints that users may not have stated.\u003c/li\u003e\n\u003cli\u003eWe study this problem as interactive intent alignment through clarification.\u003c/li\u003e\n\u003cli\u003eHowever, clarification is only useful when agents can translate answers into better decisions.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12655v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: LLM agents can execute increasingly complex tasks, but effective assistance also requires aligning their actions with goals and constraints that users…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eWe study this problem as interactive intent alignment through clarification\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eYet clarification is useful only when the agent can turn the answer into better decisions\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12659\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eMagicBeanStalk: Escaping the RLVR Zero Advantage Trap with Recursive Data Synthesis\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublication Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12659v1 Announcement Type: New.\n\u003cul\u003e\n\u003cli\u003eAbstract: Data-centric scaling of RLVR has focused on collecting or generating verifiable training problems.\u003c/li\u003e\n\u003cli\u003eHowever, corpus size alone does not determine whether the dataset provides a useful learning signal for a target policy.\u003c/li\u003e\n\u003cli\u003eTo guide the synthesis of effective RLVR data, we experimentally characterize dataset effectiveness as a function of its problems, the target policy, and the training rollout budget.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12659v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Data-centric scaling of RLVR has focused on collecting or generating verifiable training problems\u003c/li\u003e\n\u003cli\u003eHowever, corpus size alone does not determine whether the dataset provides a useful learning signal for a target policy\u003c/li\u003e\n\u003cli\u003eTo guide the synthesis of effective RLVR data, we experimentally characterize dataset effectiveness as a function of its problems, the target policy, and the tr…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12762\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eFaithful and Complete LLM-Assisted Automated Construction of Hierarchical Hardware Knowledge Graphs\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublication Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12762v1 Announcement Type: New.\n\u003cul\u003e\n\u003cli\u003eAbstract: Hardware datasheets contain engineering specifications essential for component selection, system integration, and hardware compatibility verification.\u003c/li\u003e\n\u003cli\u003eHowever, manual inspection is tedious and error-prone because datasheets are lengthy and heterogeneous, with dense information fragmented across tables, figures, and text.\u003c/li\u003e\n\u003cli\u003eAlthough large language models (LLMs) can automate this process, naive one-shot processing omits relevant specifications even when the document fits within the model\u0026rsquo;s context window.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12762v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Hardware datasheets contain engineering specifications essential for component selection, system integration, and hardware compatibility verification\u003c/li\u003e\n\u003cli\u003eHowever, manual inspection is tedious and error-prone because datasheets are lengthy and heterogeneous, with dense information fragmented across tables, figures…\u003c/li\u003e\n\u003cli\u003eAlthough large language models (LLMs) can automate this process, na\u0026quot;ive one-shot processing omits relevant specifications even when the document fits within th…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12781\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eRepresentation Alignment from a Few Paired Data via Distilled Optimal Transport\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eSummary: - arXiv:2610.12781v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Aligning two independently pretrained unimodal encoders typically demands millions of paired examples.\u003c/li\u003e\n\u003cli\u003eIn this paper, we study the regime where only $N \\sim 100$\u0026ndash;$500$ pairs are available, while unpaired data on either side is abundant.\u003c/li\u003e\n\u003cli\u003eOur method expresses every representation by its similarities to the $N$ paired anchors, which places both modalities in a common coordinate system; matches the two point clouds with a single entropic optimal transport problem solved over the entire pool, whose marginal constraints counteract centrality that defeats nearest-neighbor matching; and distills the resulting plan into two lightweight encoders, thereby making the correspondence inductive.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12781v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Aligning two independently pretrained unimodal encoders typically demands millions of paired examples\u003c/li\u003e\n\u003cli\u003eIn this paper, we study the regime where only $N \\sim 100$\u0026ndash;$500$ pairs are available, while unpaired data on either side is abundant\u003c/li\u003e\n\u003cli\u003eOur method expresses every representation by its similarities to the $N$ paired anchors, which places both modalities in a common coordinate system; matches the…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"arxiv-cscl-b_introsearch\"\u003e\n  ArXiv cs.CL (B_intro+search)\n  \u003ca class=\"heading-link\" href=\"#arxiv-cscl-b_introsearch\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12545\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eAutoIndexer: Flexible-Order Decoding by Training Causal Models on Chains of Edits\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12545v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Autoregressive decoding is less suited to non-causally ordered data, e.g., cross-references or future dependencies, since unlike iterative human reasoning, their outputs cannot be revised post hoc.\u003c/li\u003e\n\u003cli\u003eDiffusion LLMs partially mitigate this limitation, but still cannot flexibly insert or remove arbitrary numbers of tokens from a pre-generated context.\u003c/li\u003e\n\u003cli\u003eWe design AutoIndexer as a generative model with a novel training and decoding mechanism that simulates a Chain of Edits, allowing the model to flexibly modify its previous output via token insertion, replacement, and deletion.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12545v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Autoregressive decoding is less suited to non-causally ordered data, e.g., cross-references or future dependencies, since their outputs cannot be revi…\u003c/li\u003e\n\u003cli\u003eDiffusion LLMs partially mitigate this limitation but still cannot flexibly insert or remove arbitrary numbers of tokens from a pre-generated context\u003c/li\u003e\n\u003cli\u003eWe design AutoIndexer as a generative model with a novel training and decoding mechanism that simulates a Chain of Edits, allowing the model to flexibly modify…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12569\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eSWE-Duel: Towards a Self-Scaling Arena for Coding Agents\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12569v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Static coding benchmarks are fixed at release.\u003c/li\u003e\n\u003cli\u003eOver time, their tasks leak into training data, and more capable agents eventually saturate them.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eWe argue that a benchmark should instead aim to be self-scaling: its tasks and oracles are created at evaluation time, so no content exists in advance to leak, and its difficulty is set by participating agents rather than fixed at release.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12569v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Static coding benchmarks are fixed at release\u003c/li\u003e\n\u003cli\u003eOver time their tasks leak into training data, and stronger agents eventually saturate them\u003c/li\u003e\n\u003cli\u003eWe argue that a benchmark should instead aim to be self-scaling: its tasks and oracles are created at evaluation time, so no content exists in advance to leak,…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12570\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eContextual Correction of Parallel Draftsfor Speculative Decoding\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12570v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Speculative decoding accelerates large language model inference by using a lightweight drafter to propose multiple tokens that are subsequently verified in parallel by a target model.\u003c/li\u003e\n\u003cli\u003eRecent diffusion-based drafters make this paradigm particularly attractive: deep attention-based neural backbones provide strong predictions, while drafting multiple future positions in parallel avoids the sequential cost of autoregressive generation.\u003c/li\u003e\n\u003cli\u003eThis parallelism, however, comes at a cost.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12570v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Speculative decoding accelerates large language model inference by using a lightweight drafter to propose multiple tokens that are subsequently verifi…\u003c/li\u003e\n\u003cli\u003eRecent diffusion-based drafters make this paradigm particularly attractive: deep attention-based neural backbones provide strong predictions, while drafting mul…\u003c/li\u003e\n\u003cli\u003eThis parallelism, however, comes at a cost\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12580\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eCLUE: A Contrastive Learning Framework for Unsupervised Event Type Induction and Annotation\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12580v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Event ontology is the foundational framework for defining events in a dataset.\u003c/li\u003e\n\u003cli\u003eEvent type induction generally involves domain experts who define guidelines for creating event types and their names based on the dataset\u0026rsquo;s domain and scope.\u003c/li\u003e\n\u003cli\u003eHowever, this process is resource-intensive in terms of annotation costs and is prone to human biases.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12580v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Event ontology is the foundational framework for defining events in a dataset\u003c/li\u003e\n\u003cli\u003eEvent type induction generally involves domain experts who define guidelines for creating event types and their names based on the dataset\u0026rsquo;s domain and scope\u003c/li\u003e\n\u003cli\u003eHowever, this process is resource-intensive in terms of annotation costs and is prone to human biases\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12582\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eOne Model to Loop Them All: Elastic Latent Recurrence for Language Models\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12582v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: A latent recurrent language model repeatedly applies a shared block of layers, so a single checkpoint can trade inference compute for accuracy by changing the loop count.\u003c/li\u003e\n\u003cli\u003eHowever, at one loop, such models typically fall well short of a non-recurrent model of the same size—a one-loop tax that makes a separate non-recurrent model a better choice when latency is important.\u003c/li\u003e\n\u003cli\u003eWe present Elastic Latent Recurrence (ELR), a method for training looped language models without this tax.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12582v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: A latent recurrent language model applies a shared block of layers repeatedly, so one checkpoint can trade inference compute for accuracy by changing…\u003c/li\u003e\n\u003cli\u003eAt one loop, however, such models typically fall well short of a non-recurrent model of the same size, a one-loop tax that makes a separate non-recurrent model…\u003c/li\u003e\n\u003cli\u003eWe present Elastic Latent Recurrence (ELR), a recipe for training looped language models without this tax\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12601\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eReflective Prompt Optimization of a Deep-Research Agent: A Topic-Disjoint, Multi-Seed Protocol for Measuring Transferable Gains and Prompt-Level Overfitting\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12601v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Prompt optimization is an attractive alternative to reinforcement learning for adapting language model agents, since it is low-cost and produces human-readable artifacts.\u003c/li\u003e\n\u003cli\u003eHowever, on factual tasks, it carries a failure mode that scalar-reward reinforcement learning does not.\u003c/li\u003e\n\u003cli\u003eAn optimizer that reads training examples can write their answers into the prompt, so in-distribution gains may be memorization rather than transferable behavior.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12601v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Prompt optimization is an attractive alternative to reinforcement learning for adapting language-model agents, since it is cheap and produces human-re…\u003c/li\u003e\n\u003cli\u003eOn factual tasks, however, it carries a failure mode that scalar-reward reinforcement learning does not\u003c/li\u003e\n\u003cli\u003eAn optimizer that reads the training examples can write their answers into the prompt, so an in-distribution gain may be memorization rather than transferable b…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12602\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eSelf-Preservation in Language Models\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12602v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Continual learning (CL) seeks to adapt pre-trained models while preserving previous capabilities.\u003c/li\u003e\n\u003cli\u003eHowever, tasks provided for preservation cover only a portion of what the model knows, and forgetting may extend beyond these boundaries.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eRecent work shows that language models can supplement this limited evidence by generating useful replay without access to their original training data.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12602v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Continual learning (CL) seeks to adapt pretrained models while preserving prior capabilities\u003c/li\u003e\n\u003cli\u003eYet the tasks supplied for preservation cover only part of what a model knows, while forgetting can extend beyond them\u003c/li\u003e\n\u003cli\u003eRecent work shows that language models can supplement this limited evidence by generating useful replay without access to their original training data\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12617\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eGeopoliticalQA: Benchmarking Retrieval and Generation over Intelligence Reports\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublication Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12617v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Geopolitical and strategic intelligence reports constitute a rich source of expert knowledge and have increasing value in the current political climate, but their heterogeneity and analytical complexity pose challenges to retrieval-augmented question answering.\u003c/li\u003e\n\u003cli\u003eWe introduce a benchmark for grounded question answering over 415 open-source geopolitical and strategic intelligence reports, comprising 5,525 document-grounded question-answer pairs covering factual, reasoning, comparative, and risk-oriented questions.\u003c/li\u003e\n\u003cli\u003eWe propose a document-aware generation pipeline that profiles reports before generating questions and validates dataset quality through expert annotation and LLM-as-a-judge agreement.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12617v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Geopolitical and strategic intelligence reports constitute a rich source of expert knowledge and have increasing value in the current political climat…\u003c/li\u003e\n\u003cli\u003eWe introduce a benchmark for grounded question answering over 415 open-source geopolitical and strategic intelligence reports, comprising 5,525 document-grounde…\u003c/li\u003e\n\u003cli\u003eWe propose a document-aware generation pipeline that profiles reports before generating questions and validate dataset quality through expert annotation and an…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12667\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eperturber: A Stateless Service for Semantics-Preserving Text Perturbations\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublication Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12667v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Evaluating large language models (LLMs) on clean benchmarks overestimates model robustness: real-world inputs are perturbed, and these meaning-preserving variations often shift model answers.\u003c/li\u003e\n\u003cli\u003eSystematically studying this requires applying one perturbation at a time, repeatably and identically regardless of where they run, but existing tools couple them to benchmark stacks rather than packaging them as a reusable transformation layer.\u003c/li\u003e\n\u003cli\u003eWe introduce perturber, a stateless service and Python library that applies only one semantics-preserving perturbation per call, drawn from a first-party catalog of 68 perturbations across 11 families.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12667v1 Announce Type: new\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAbstract: Large language model (LLM) evaluations on clean benchmarks overstate model robustness: real inputs arrive perturbed, and these meaning-preserving vari…\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eStudying this systematically requires applying perturbations one at a time, reproducibly, and identically wherever they run, yet existing tools couple them to a…\u003c/li\u003e\n\u003cli\u003eWe present perturber, a stateless service and Python library that applies exactly one semantics-preserving perturbation per call, drawn from a first-party catal…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12669\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eGeoFitBench-LLM: A Benchmark for Assessing Large Language Models\u0026rsquo; Capabilities in Geospatial Data Requirements Elicitation for Geo-Analytic Tasks\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12669v1 Announce Type: new\n\u003cul\u003e\n\u003cli\u003eAbstract: Large language models are becoming central to GeoAI systems, yet their evaluations rarely consider the elicitation of geospatial data requirements.\u003c/li\u003e\n\u003cli\u003eThis omission is critical, as retrieval-oriented geo-analytic tasks cannot be reliably supported without first identifying the necessary datasets and constraints.\u003c/li\u003e\n\u003cli\u003eGeoFitBench-LLM is introduced as a benchmark for assessing whether LLMs can infer the datasets, supporting resources, and constraints required by natural language geo-analytic requests.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12669v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Large language models are becoming central to GeoAI systems, yet their evaluation rarely considers the elicitation of geospatial data requirements\u003c/li\u003e\n\u003cli\u003eThis omission is critical as retrieval oriented geo-analytic tasks cannot be supported reliably without first identifying the necessary datasets and constraints\u003c/li\u003e\n\u003cli\u003eGeoFitBench-LLM is introduced as a benchmark for assessing whether LLMs can infer the datasets, the supporting resources and the constraints required by geo-ana…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"arxiv-cslg-b_introsearch\"\u003e\n  ArXiv cs.LG (B_intro+search)\n  \u003ca class=\"heading-link\" href=\"#arxiv-cslg-b_introsearch\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"Link to heading\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003eLink to heading\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12494\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eAlgebraic Integer Quantization for Machine Learning Applications\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12494v1 Announce Type: new\n\u003cul\u003e\n\u003cli\u003eAbstract: In this paper, we propose Algebraic Integer Format (AINF) as a framework for highly configurable and precise compressed computation.\u003c/li\u003e\n\u003cli\u003eA key feature of AINF is that it can be directly reduced to pure integer arithmetic, thereby achieving error-free multiply-accumulate operations.\u003c/li\u003e\n\u003cli\u003eWe develop a matrix multiplication engine that leverages this feature to achieve performance within a predictable factor of its most efficient numerical counterparts on modern hardware accelerators.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12494v1 Announce Type: new\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAbstract: In this article, we propose algebraic integer number formats (AINF) as a framework for highly configurable and accurate compressed computation\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eA key property of AINF is its straightforward reduction to purely integer arithmetic, enabling error-free multiply-accumulate operations\u003c/li\u003e\n\u003cli\u003eWe develop a matrix multiplication engine that exploits this property, obtaining performance within a predictable factor of the most efficient numerical counter…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12495\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eKairos: Toward a Byte-Level Foundation Model for Every Modality\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublication Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12495v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Small language models trained on a single local machine leave little room for architectural trial and error.\u003c/li\u003e\n\u003cli\u003eKairos combines bidirectional gated-DeltaNet linear attention and sliding-window attention over shared projections (LiZAttention2), block-wise attention-residual aggregation, sparse Mixture-of-Experts (MoE) feed-forward layers, and multi-scale byte-level codecs, trained with a unified noise-denoising objective on a single byte space shared by text, images, audio, video, LiDAR, and control streams; the report asks whether the combination is fully optimized before asking whether it is useful.\u003c/li\u003e\n\u003cli\u003eWe provide code-derived algorithms for each component, including a cross-session recurrent-state memory gate, and then run convergence sanity checks: overfitting a single 440-byte sentence, text-only, with three seeds for codec and attention selection, and a single sweep over seven other factors.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Highlights:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12495v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Small language models trained on a single local machine leave little room for architectural trial and error\u003c/li\u003e\n\u003cli\u003eKairos combines bidirectional gated-DeltaNet linear attention and sliding-window attention over shared projections (LiZAttention2), block-wise Attention-Residua…\u003c/li\u003e\n\u003cli\u003eWe give code-derived algorithms for each component, including a cross-session recurrent-state memory gate, then run convergence sanity checks: overfittinga sing…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12496\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eAn Input-Dependent Readout Phase for Resonant Sequence Models\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublication Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eAbstract: - arXiv:2610.12496v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Sequence models with a fixed-size state avoid the quadratic cost of attention, but face a design choice.\u003c/li\u003e\n\u003cli\u003eInput-dependent state transitions provide content-dependent memory at the expense of FFT-based training, whereas fixed transitions retain the O(T log T) convolution path but read memory in the same way for every input.\u003c/li\u003e\n\u003cli\u003eResonatorLM, a recent model of the second class, reads a set of damped complex oscillators through a fixed learned phase per head.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Highlights:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12496v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Sequence models with a fixed-size state avoid the quadratic cost of attention, but face a design choice\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAn input-dependent state transition gives content-dependent memory at the cost of FFT-based training, while a fixed transition keeps the O(T log T) convolutiona…\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eResonatorLM, a recent model of the second kind, reads a bank of damped complex oscillators through a fixed learned phase per head\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12520\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eCircuitTree: Composing Circuits into A Syllable Transcriber\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12520v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: The goal of circuit discovery is to decompose a monolithic neural network into tiny modular subnetworks, each of which is interpretable and performs a recognizable subtask.\u003c/li\u003e\n\u003cli\u003eWe develop a practical method for composing arbitrarily many circuits into a single model.\u003c/li\u003e\n\u003cli\u003eWe propose a greedy, density-scaled composition algorithm that overcomes the limitations of previous attempts to compose circuits via naive union, then generalize it to $N$ circuits via CircuitTree, a novel data structure built on a $\\Theta(N\\log N)$ tree composition scheme.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12520v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: The goal of circuit discovery is to decompose a monolithic neural network into tiny modular subnetworks which are interpretable in the sense that each…\u003c/li\u003e\n\u003cli\u003eWe develop a practical method for composing arbitrarily many circuits into a single model\u003c/li\u003e\n\u003cli\u003eWe propose a greedy, density-scaled composition algorithm that overcomes limitations of previous attempts to compose circuits by naive union, then generalize it…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12527\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eBehind-the-Meter Photovoltaic Visibility from Net Smart Meters: Deployment Regimes and Operational Insights\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12527v1 Announcement Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Distribution operators need visibility of behind-the-meter photovoltaic (PV) generation, yet most residential customers are observed only through net smart meters.\u003c/li\u003e\n\u003cli\u003eDeep-learning-based PV disaggregation can restore this visibility, but its operational value depends on whether the model serves known customers, new customers, future periods, or new regions.\u003c/li\u003e\n\u003cli\u003eWe quantify this dependence using four deployment regimes that separate household transfer from temporal extrapolation, applied to Australian grid smart meter data with cross-dataset adaptation on Pecan Street data.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12527v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Distribution operators need visibility of behind-the-meter photovoltaic (PV) generation, yet most residential customers are observed only through net…\u003c/li\u003e\n\u003cli\u003eDeep-learning-based PV disaggregation can restore this visibility, but its operational value depends on whether the model serves known customers, new customers,…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eWe quantify this dependence using four deployment regimes that separate household transfer from temporal extrapolation, applied to Ausgrid smart-meter data with…\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12528\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eGlucoPRISM: Protocol-Supervised Pretraining for Cross-Cohort Glucose Transfer\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublication Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12528v1 Announcement Type: New.\n\u003cul\u003e\n\u003cli\u003eAbstract: Continuous glucose monitoring (CGM) foundation models encode each day into a single vector that mixes who the person is, what happened that day, and how the sensor measured it, so measurement discrepancies shift as representations enter new cohorts.\u003c/li\u003e\n\u003cli\u003eWe ask whether pretraining can place the measurement components that harm transfer into known coordinates, so that deployment can discard them with fixed projections and without target data.\u003c/li\u003e\n\u003cli\u003eGlucoPRISM supervises the GlucoFM backbone through response contracts and rules governing how each part of the representation should change when only the day or only the sensor changes: the 64-dimensional feature band should remain invariant to both, the 48-dimensional state band changes only across days, and the 16-dimensional sensor band tracks the device.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12528v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Continuous glucose monitoring (CGM) foundation models encode each day as one vector that mixes who the person is, what happened that day and how the s…\u003c/li\u003e\n\u003cli\u003eWe ask whether pretraining can place the measurement component that harms transfer in known coordinates, so that deployment drops them with a fixed projection a…\u003c/li\u003e\n\u003cli\u003eGlucoPRISM supervises a GlucoFM backbone with response contracts, rules for how each part of the representation should change when only the day or only the sens…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12530\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eCacheScope: Exact-Compute Allocation for Training-Free Diffusion Acceleration\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003ePublication Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12530v1 Announcement Type: New.\n\u003cul\u003e\n\u003cli\u003eAbstract: Training-free feature reuse accelerates diffusion transformer inference by replacing full denoising computation at selected steps with cached approximations.\u003c/li\u003e\n\u003cli\u003eExisting methods mainly improve the reconstruction of skipped features, but we find that once a stable first-order reconstructor is used, increasing reconstruction complexity yields limited gains, whereas the placement of exact-compute steps has a much greater impact on fidelity.\u003c/li\u003e\n\u003cli\u003eMore importantly, absolute shallow probing changes are jointly influenced by input trajectory displacement and network response strength, and thus cannot directly reflect caching risk.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12530v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Training-free feature reuse accelerates diffusion transformer inference by replacing full denoising computation at selected steps with cached approxim…\u003c/li\u003e\n\u003cli\u003eExisting methods mainly improve the reconstruction of skipped features, but we find that once a stable first-order reconstructor is used, increasing reconstruct…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eMore importantly, absolute shallow-probe variation is jointly affected by input-trajectory displacement and network response strength, and therefore does not di…\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12537\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eMarket Impact Estimation via Causal Foundation Models\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12537v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: When a trader executes a large order, the relevant question is not \\emph{what will the price be?}, but \\emph{what would the price have been had I not traded?}. Market impact is thus a causal quantity, yet classical models (Almgren\u0026ndash;Chriss, propagators) regress price on flow, which is biased whenever informed flow affects price through information and trading.\u003c/li\u003e\n\u003cli\u003eWe propose a proof-of-concept \\emph{Causal Market Simulator}: a continuous-time Causal Prior-data Fitted Network (PFN) trained on synthetic market priors, where hidden information flow confounds the flow-price relationship by construction, with the confounding strength being an exact per-episode knob.\u003c/li\u003e\n\u003cli\u003eIn this prior, the bias of the naive estimator follows a closed form that we empirically confirm (up to 68% overestimation), while PFN tracks the unbiased ground-truth parameter oracle under high confounding.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12537v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: When a trader executes a large order, the relevant question is not \\emph{what will the price be\u003c/li\u003e\n\u003cli\u003e} but \\emph{what would the price have been had I not traded\u003c/li\u003e\n\u003cli\u003e} Market impact is thus a causal quantity, yet classical models (Almgren\u0026ndash;Chriss, propagators) regress price on flow, which is biased whenever informed flow mov…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12538\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eEfficient Learning of Survival Analysis Models Under Outcome-Dependent Truncation and Censoring\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12538v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Modern machine learning models are typically trained on data assumed to be representative of the predictive tasks the model will face in the future, but actual datasets violate this assumption.\u003c/li\u003e\n\u003cli\u003eWe focus on outcome-dependent truncation, where observations appear in the training data depending on their labels.\u003c/li\u003e\n\u003cli\u003eOur work connects two largely distinct literatures: efficient learning under truncation and survival analysis.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12538v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Modern machine learning models are typically trained on data assumed to be representative of the predictive tasks the model will face in the future, b…\u003c/li\u003e\n\u003cli\u003eWe focus on outcome-dependent truncation in which an observation appears in the training data depending on its label\u003c/li\u003e\n\u003cli\u003eOur work connects two largely distinct literatures: efficient learning under truncation and survival analysis\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2610.12541\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eEnriched Holomorphic Neural Networks for 2D and 3D Laplace Problems\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eRelease Time: 2026-10-12 12:00 Beijing Time\u003c/li\u003e\n\u003cli\u003eSummary: - arXiv:2610.12541v1 Announce Type: new.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eAbstract: Boundary-value problems for the Laplace equation produce singular behavior near corners, edges, and junctions, which smooth approximators struggle to resolve.\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003eWe combine holomorphic neural networks with singular functions derived from local geometry and boundary conditions.\u003c/li\u003e\n\u003cli\u003eThe representation is harmonic by construction, so no interior residual is needed: the network fits the smooth remainder from boundary data alone, and enrichment terms capture singular behavior.\u003c/li\u003e\n\u003cli\u003eEN Key Points:\n\u003cul\u003e\n\u003cli\u003earXiv:2610.12541v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Boundary-value problems for the Laplace equation develop singular behaviour near corners, edges and junctions that smooth approximators resolve poorly\u003c/li\u003e\n\u003cli\u003eWe combine holomorphic neural networks with singular functions derived from the local geometry and boundary conditions\u003c/li\u003e\n\u003cli\u003eThe representation is harmonic by construction, so no interior residual is needed: the network fits the smooth remainder from boundary data alone, and the enric…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n",
  "wordCount": 5545,
  "readingTime": 27,
  "tableOfContents": "\u003cnav id=\"TableOfContents\"\u003e\n  \u003cul\u003e\n    \u003cli\u003e\u003ca href=\"#-in-depth-guide-to-this-issues-watch-list\"\u003e📖 In-Depth Guide to This Issue\u0026rsquo;s Watch List\u003c/a\u003e\u003c/li\u003e\n    \u003cli\u003e\u003ca href=\"#-x-platform-ai-highlights\"\u003e🌐 X Platform AI Highlights\u003c/a\u003e\u003c/li\u003e\n    \u003cli\u003e\u003ca href=\"#-influencer-insights\"\u003e💡 Influencer Insights\u003c/a\u003e\u003c/li\u003e\n    \u003cli\u003e\u003ca href=\"#-appendix-todays-watch-list-update-source-list\"\u003e📚 Appendix: Today\u0026rsquo;s Watch List Update Source List\u003c/a\u003e\n      \u003cul\u003e\n        \u003cli\u003e\u003ca href=\"#all-in-podcast-a_full\"\u003eAll-In Podcast (A_full)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#stratechery-by-ben-thompson-a_full\"\u003eStratechery by Ben Thompson (A_full)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#arxiv-csai-b_introsearch\"\u003eArXiv cs.AI (B_intro+search)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#arxiv-cscl-b_introsearch\"\u003eArXiv cs.CL (B_intro+search)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#arxiv-cslg-b_introsearch\"\u003eArXiv cs.LG (B_intro+search)\u003c/a\u003e\u003c/li\u003e\n      \u003c/ul\u003e\n    \u003c/li\u003e\n  \u003c/ul\u003e\n\u003c/nav\u003e",
  "isDraft": false
}