# Research Papers — 2026-10 · bundle page 13 of 156 > Full-text markdown of papers #601–650 of 7799 published in 2026-10, newest first. AI-readable mirrors of the full 7799-paper 2026-10 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-10/page/13 [← Previous bundle](https://aipod.dev/papers/archive/2026-10/page/12.md) [Next bundle →](https://aipod.dev/papers/archive/2026-10/page/14.md) ### StochLOB: Heavy-Tailed Score Matching for Robust Limit Order Book Trend Prediction Authors: Arshia Abolghasemi, Seyed Reza Tavakoli, Parsa Naderi, Nazanin Mirzaei, Behrad Jahani, Saba Yousefzadeh, Tahereh Javaheri, Mohammad Hosein Rohban, Gholamali Aminian, Ali Habibnia Published: 2026-10-15 Categories: cs.LG, q-fin.TR, stat.ML arXiv: https://arxiv.org/abs/2610.21081v1 Score-based diffusion methods usually perturb data with Brownian Gaussian noise, corresponding to a light-tailed forward stochastic differential equation (SDE). For market time series this SDE can be a poor abstraction: limit order book (LOB) order flow is heavy-tailed, bursty, and occasionally jump-like. We study STOCHLOB, a training-only score-matching framework that changes the forward diffusion corruption law while leaving the deployed predictor deterministic. Instead of Brownian-only Gaussian corruption, the noising process uses two non-Gaussian SDE-inspired alternatives: jump-Lévy noise through a compound-Poisson variance mixture, and clipped $α$-stable noise through a positive-stable Gaussian scale mixture. These corruptions inject finite jumps and heavy-tailed residuals into training without requiring reverse-time diffusion sampling at inference. The resulting score target has a useful interpretation: its radial factor $h_t(r)$ is the posterior expected corruption precision, so large residuals are weighted according to the latent corruption scale that most plausibly generated them. On chronologically held-out Coinbase LOB data, the proposed variants remain competitive on uncorrupted BTC and altcoin test horizons and improve several realized tail-window stress diagnostics. An exploratory Feishu A-share transfer study further tests whether the probability outputs remain usable under transaction costs and trading-rule constraints. Code and configuration files are available at https://github.com/ArshiAbolghasemi/Penny. --- ### Adaptive Randomized Neural Networks for PDEs: Structure-Aware Feature Redistribution and Residual-Driven Block Enrichment Authors: You Yang, Fei Wang Published: 2026-10-15 Categories: math.NA, cs.LG arXiv: https://arxiv.org/abs/2610.21077v1 Randomized neural networks retain an output-linear least-squares solve, but globally sampled random features may fail to capture the spatial and scale structure of PDE solutions. We propose SFRBE--RaNN, a two-stage adaptive trial-space construction. SFR redistributes a fixed feature budget according to regional indicator masses and adapts regional weight scales through low-dimensional residual-based optimization. RBE constructs nonredundant enrichments by screening multiple activation families, removing represented response directions, and selecting the smallest admissible prefix. For solution-dependent operators, candidates are screened with a frozen operator and accepted only after nonlinear residual verification. For fixed linear problems, we establish an exact discrete residual-reduction identity and a conditional solution-error estimate. Building on this identity, we combine safeguarded SFR with RBE capture factors to derive a finite-horizon error bound involving baseline best approximation and empirical consistency, together with a conditional random-search success-probability bound. Numerical experiments on localized, multiscale, and nonlinear space--time problems demonstrate the method's effectiveness. --- ### OmicsBench: a long-horizon multi-turn benchmark of AI agents on research-study-scale bioinformatics Authors: Gideon Shaked, Siddharth Nagisetty Published: 2026-10-15 Categories: cs.AI, q-bio.GN, q-bio.QM arXiv: https://arxiv.org/abs/2610.21074v1 Computational biologists increasingly delegate substantial analyses to AI agents in long-horizon, multi-turn sessions, yet most benchmarks of agents' bioinformatics capabilities pose a single short task with a single-shot prompt. We present OmicsBench, a long-horizon, multi-turn bioinformatics benchmark for AI agents. Each of its 97 tasks links dependent subtasks into a directed acyclic graph, for 662 underlying subtasks in all. Each subtask is posed as one user turn and graded deterministically against an answer key, with optional in-context feedback on incorrect answers. Across 12 models on three harnesses, the slowest agent averages 3.6 hours per task, and no combination comes near saturating the benchmark, with mean scores of 0.43 to 0.57 out of a possible 1 without in-context feedback and 0.54 to 0.67 with it. Multi-turn evaluation also uncovers behavior that single-shot prompts cannot, as agents score lower on subtasks that depend on earlier subtasks, and in-context feedback helps agents recover some of that loss. Because deterministic scoring gives no qualitative account of how an agent failed, we also introduce an unsupervised failure mode discovery pipeline that describes, embeds, and clusters agents' failed attempts to surface emergent failure modes. Applied to OmicsBench, this pipeline exposes qualitative shortcomings in how agents carry out bioinformatics analyses, ranging from agents applying their own statistical filters in place of the specified ones to agents writing whole answers from memory. --- ### Approve What, Exactly? Effect Binding and Trusted Rendering for Agent Step-Up Authorization Authors: Sai Varun Kodathala, Prashanth Pollishetty, Jaylen Cargill Published: 2026-10-15 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2610.21063v1 AI agents now pause for approval before they send money, delete files or email outsiders, and that approval is usually treated as a second factor. We argue it is one only when compromising the agent cannot produce it and when it names exactly what it approves, and we test both conditions. We define twelve attacks on the approval path, including decoy recipients, amount cancellation, card padding, label spoofing, enrollment squatting and threshold splitting, and run them against the Agent Payments Protocol (AP2), the Microsoft action-bound approval design and Model Context Protocol (MCP) elicitation. An audit of ten agent runtimes finds no third-party verifiable approval record, no flood limits where measurable, and post-approval argument changes in four of eight runtimes with measurable binding. As a reference defense we build ASU (Agent Step-Up), a gateway that renders the approval card from a canonical effect descriptor, has a passkey on the phone of the user sign its digest, and executes only the stored call. Tamarin proves its binding, single-use and key-level non-repudiation properties under three stated hypotheses (53 lemmas). Across 27,108 scripted trials and 609 AgentDojo attack pairs, ASU stops substitution, replay and grant stretching, and the trusted card cuts forged-narration attack success from 0.54 to 0.03. With 15 current language models as automated approvers, the trusted card raises the attack catch rate from 0.30 to 0.98, at the cost of refusing 16 percent of harmless near misses, mostly due to a placeholder artifact in our stimuli. A second device adds nothing measurable over a trusted on-host confirmation in these experiments, flooding still succeeds 0.45 to 0.60 of the time against a modeled approver, and approval cannot stop effects that policy already allows. Code, data and proofs: https://github.com/Sports-Vision-Inc/agent-step-up --- ### Multimodal AI Detection for Math-Heavy Scientific Text Authors: Yixuan Huang, Danish Pruthi, Bhuwan Dhingra Published: 2026-10-15 Categories: cs.CL arXiv: https://arxiv.org/abs/2610.21062v1 As large language models become increasingly capable of generating, and assisting with, scientific writing, there is a need for greater transparency and disclosure around the use of AI in research. Consequently, a veritable subfield of research and a number of commercial services aim to flag AI-generated content automatically. However, existing detectors which rely only on text extracted from PDFs fail to benefit from the rich visual information contained in the formatting of math-heavy scientific content. To study this, we select human-written proof and prose passages from pre-2022 arXiv LaTeX sources and re-generate them using 15 modern LLMs within the context of the same paper. This leaves us with coupled human and AI outputs in three different representations: the LaTeX source, a rendered image of the PDF, and the text extracted from the PDF. We find that the accuracy of the state-of-the-art Pangram 4 drops significantly as the mathematical content in a passage increases, especially when evaluated on text as opposed to LaTeX. However, this is not an inherent limitation of detecting AI-generated math: finetuning Qwen-3.5-27B based detectors on this domain recovers the gap, with LaTeX inputs performing significantly better than text since they preserve more formatting information. Since LaTeX sources are rarely available in practice, we also train SnapJudge: an image-based detector operating on screenshots of both prose and proofs from papers. At 0.1% false positive rate, SnapJudge achieves a detection rate of 92% on proofs and 93% on prose generated by frontier models, substantially outperforming both Pangram 4 and our text-based detector. We release our data, models and code at https://nlplab.cs.duke.edu/snapjudge. --- ### Score Mismatch and Set-Constrained Sampling in Zero-Shot Conditional Discrete Diffusion Authors: Renxiang Huang, Yuchen Liang, Yingbin Liang, Ness Shroff, Lifeng Lai Published: 2026-10-15 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2610.21061v1 Discrete diffusion models have become a powerful framework for modeling complex discrete distributions across a range of applications. While theoretical convergence guarantees depend on consistency between the target and training distributions, such a case rarely holds in practice, making score mismatch a central challenge in discrete diffusion sampling. This is particularly relevant to zero-shot conditional generation, where a model trained on the unconditional distribution is directly reused at inference time to sample from a conditional target distribution. In this paper, we provide the first theoretical analysis for score-mismatched discrete diffusion models and derive a general KL divergence upper bound that clearly identifies the additional bias caused by such a mismatch. Our analysis covers partial-observation constraints, including text infilling, sentence completion, and more general coordinatewise set constraints. Specifically, we analyze the classical Conditional SEDD sampler and show that it can incur a bias term that scales with the dimension, vocabulary size, and compatibility. Based on this analysis, we propose a Randomized Hard Clamping (RHC) sampler for coordinatewise set constraints. We show that the assignment probabilities can be expressed exactly by the unconditional scores with absorbing rate. Finally, numerical and real-data experiments demonstrate the effectiveness of the proposed RHC sampler. --- ### Bellman-Consistent Trajectory RL for LLM Arithmetic Reasoning Authors: Lianfa Li Published: 2026-10-15 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2610.21060v1 To address sparse terminal rewards in LLM arithmetic reasoning, we propose a Bellman-consistent trajectory RL method. Our framework introduces Trajectory Bellman Residual Minimization (TBRM) to transform sparse rewards into dense token-level signals. We integrate a Reinforce Leave-One-Out (RLOO) group strategy to reduce gradient variance of Bellman residuals and use hindsight goal relabeling to accelerate intermediate learning. Evaluated on the Countdown task, our method alleviates optimization bottlenecks and stabilizes gradient variance significantly better than GRPO. The TBRM-RLOO framework boosts Pass@1 to 0.55--0.64 and achieves a state-of-the-art Pass@16 accuracy of 76.0--84.0\%, outperforming the SFT, IPO/DPO, and RLOO baselines. Our approach demonstrates exceptional single-sample efficiency and successfully circumvents standard performance plateaus via test-time scaling. --- ### Athanor: An Agent-Native System for Autonomous Post-Training Research Authors: Yufa Zhou, Yichen Wang, Jiakun Fan, Ziqi Shu, Zhichao Liu, Anru R. Zhang, Neiwen Ling, Zijun Wei Published: 2026-10-15 Categories: cs.AI, cs.CL, cs.LG, cs.MA arXiv: https://arxiv.org/abs/2610.21055v1 Recursive self-improvement, in which AI systems build stronger AI systems, is one promising route to superintelligence. We study a weaker form that can be measured today: frontier agents post-training a small raw base model with public data, as a test of their research potential. We build Athanor, an open agent-native system for autonomous post-training research (AutoPostTrain) on real GPUs. We give the agents full research autonomy, low-friction engineering tools, and efficient starter code. The agents carry out the research themselves: they read datasets and evaluation results, build their training data, write and modify training algorithms, tune hyperparameters, and analyze the results to decide what to try next. They work in a sandbox and cannot use external teacher models. Athanor's services supply frozen evaluation, managed jobs, decontaminated data, a persistent research record, and a held-out suite the agents are unaware of. In a 104-hour campaign, a team of four Claude Opus 5.5 agents, each on an eight-GPU node and given no research advice, improves Qwen3-4B-Base from 15.9 to 67.3 on a six-task development suite and from 9.2 to 48.2 on a held-out suite, against 66.5 and 47.9 for the official Qwen3-4B-Instruct-2507. To our knowledge, this is the first demonstration that a team of frontier agents can autonomously post-train a raw base model into a cross-domain generalist model that matches or surpasses the official instruct version on average. The campaign's full record shows that the agents reach this level by recombining known methods and each other's models: SFT on public reasoning data, RL with a rollout change one agent introduced, weight merges across agents, and rounds of self-distillation. With execution handled, today's agents already have the engineering and research ability to conquer this task. We hope Athanor provides a foundation for scaling autonomous AI research. --- ### SALTO: Solver-Anchored Lifted Transfer Operators for Nonlinear Event Inference Authors: Qi Wang, Rylie Foster Published: 2026-10-15 Categories: cs.LG, math.DS arXiv: https://arxiv.org/abs/2610.21049v1 Learned dynamical models can accurately predict trajectories while producing unreliable sensitivities when differentiated or transposed, limiting their use in inverse problems and scientific decision-making. Classical adjoints provide exact local derivatives, but a single adjoint cannot represent the nonlinear response of a system to a family of finite-amplitude perturbations. We introduce SALTO (Solver-Anchored Lifted Transfer Operators), a Koopman-based framework for inferring finite-amplitude sources. In the lifted space, the nonlinear dynamics are approximated by a linear propagator, and its transpose carries each sensor backward in a single solve. The result is a reusable lifted adjoint that predicts the measurement produced by any candidate localized event with one encoder pass and an inner product. To give the model the simulator's exact local behavior, SALTO imposes three structural constraints: it anchors the physical-state dynamics to the simulator's linearized propagator, centers the learned observables so that they vanish to second order at the reference state, and uses a block-triangular lifted propagator. These constraints guarantee that the model reduces to the simulator's tangent dynamics and classical adjoint as the amplitude vanishes, while the learned observables carry only the finite-amplitude correction. On three nonlinear PDE systems, SALTO predicts the measurements of finite-amplitude sources to within 2-8% where the classical adjoint is in error by 41-165%, recovers source position and intensity from sparse sensors nearly as accurately as inversion through the solver, and samples the joint posterior of two interacting sources, matching the solver's posterior on reaction-diffusion at a twentieth of its cost. One trained model serves new sensor layouts and measurement times with a single backward solve. --- ### SHIP: Structured and Hierarchical Insight with Privacy Authors: Kai Chen, Guisi Ni, Albert Cheu, Kassem Fawaz, Yuzheng Hu, Ryan McKenna, Peter Kairouz Published: 2026-10-15 Categories: cs.CL, cs.CR, cs.LG arXiv: https://arxiv.org/abs/2610.21048v1 Extracting insights from textual data while protecting user privacy is increasingly important. Existing methods produce private insights in diverse forms for different analytical purposes, making their utility difficult to assess and compare. In this work, we formulate differentially private (DP) insights as reusable representations for answering analytical queries, providing a common basis for utility assessment. Answering these queries, however, requires fine-grained details and cross-attribute statistics that existing representations may omit. To capture this information, we introduce Structured and Hierarchical Insight with Privacy (SHIP), which automatically maps queries to attributes, builds a hierarchical hybrid taxonomy, and releases a DP synthetic table for further analysis. We also develop an evaluation protocol based primarily on QA performance and structured fidelity. Our experiments demonstrate SHIP's strong performance on both measures relative to existing private insight baselines, while identifying queries that remain challenging for structured representations. --- ### PremiseCheck: Efficient Fact-Checking of Human and AI Peer Reviews Authors: Dingjie Song, Zhiling Yan, Ziqian Zhang, Yuxuan Zhang, Lichao Sun Published: 2026-10-15 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2610.21046v1 Peer review is under strain from the growth of submissions, and review comments written by humans or AI can misstate what a paper says. We introduce ReviewFact, a benchmark that labels such factual errors in review comments under one definition across NLP, machine learning, and the natural sciences, and that measures cost and latency alongside accuracy. A frontier LLM checks comments well but must read the whole paper for each check, while cheap decision models, which score whether a statement holds in a text, fall far short on whole comments that mix several implicit assertions. PremiseCheck is a framework that combines the two. An LLM rewrites each comment as explicit premises without reading the paper, and a decision model checks each premise against the paper. When each comment is checked on its own, PremiseCheck reaches accuracy comparable to GPT-5.5 across all three fields at about half its cost, and on NLP reviews at a quarter of its cost and less than half its latency. --- ### Learning When to Withdraw: Curriculum Coverage and Rule-Specific Failures in Fine-Tuned Language Agents Authors: Mohamed Abouzahra Published: 2026-10-15 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.21045v1 Tool-using language agents must revise prepared actions when evidence changes or authority is revoked. Selective withdrawal means suspending affected actions, preserving supported work, and resuming only after sufficient repair. We study supervised fine-tuning in NAQD-Env, a synthetic environment with explicit dependencies and a deterministic reference policy. Four LoRA checkpoints of Qwen2.5-3B-Instruct compare an original curriculum with an enriched one, the latter under three system-prompt framings with identical views and targets. Three instruction-tuned baselines and all four checkpoints are evaluated under three inference prompts on fresh validation cases, held-out dependency families, and held-out structural combinations. Under the naqd prompt, SFT checkpoints reach decision agreement of 0.77-0.92 and withdrawal recall of 0.46-0.86, versus 0.45-0.63 and at most 0.06 for the baselines. Rule-level diagnostics qualify these gains. The original-curriculum checkpoint wrongly withdraws 26 of 34 actions with surviving alternative support; enriched checkpoints withdraw 0-1. Conversely, all SFT checkpoints narrow prohibited combinations less often than the 7B and 8B baselines, a case absent from both curricula. Delegation results show more withdrawal but poorer preservation of independently authorized actions, and all four checkpoints omit valid event reports. Framing contrasts vary across metrics and splits and do not establish a reproducible advantage or equivalence. One training run per arm, jointly changed curriculum properties, prior evaluation exposure, and baseline-SFT quantization differences limit causal attribution. The contribution is a method for detecting rule-specific regressions that aggregate accuracy can conceal, not evidence that fine-tuning alone ensures agent safety. --- ### ReviewPilot: A Multi-Agent System for Screening and Data Extraction in Systematic Reviews Authors: Zhaoqian Xue, Shaoze Zhou, Xuexin Li, Junting Mu, Haozheng Liu, Zhongyin Wang, Yichun He, Renkai Ma, Hao Zhou, Sarah Bratt, Siyuan Ma, Chen Chen et al. Published: 2026-10-15 Categories: cs.CL, cs.AI, cs.IR arXiv: https://arxiv.org/abs/2610.21042v1 Systematic literature reviews follow the PRISMA process of screening scholarly works and extracting evidence from included studies. Deep research tools based on large language models (LLMs) can search and summarize the literature but do not record which studies are excluded or where extracted values come from. We develop a conversational multi-agent system called ReviewPilot that automates the PRISMA workflow from paper screening to information extraction. Researchers can tailor criteria and schemas through dialogue, and a project memory keeps their configurations and revisions. Each screening decision comes with its reason and each extracted value with its source. On epidemiological expert reviews, ReviewPilot outperforms a single-prompt LLM and AgentSLR across three LLMs, with screening macro F1 up to 0.776 and extraction F1 up to 0.788. Case studies in biomedicine and human-computer interaction show close agreement with human annotators, suggesting that ReviewPilot can transfer beyond epidemiology. Code is available at https://github.com/ZhaoqianXue/ReviewPilot. --- ### Recursive Reward Improvement via Adversarial Meta-Prompt Evolution Authors: Wei Dai, Qianchu Liu, Guanghui Qin, Xing Han, Tristan Naumann, Paul Pu Liang, Sheng Zhang Published: 2026-10-15 Categories: cs.AI, cs.MA arXiv: https://arxiv.org/abs/2610.21040v1 Recursive Self-Improvement (RSI) has gained wide attention for its ability to work with limited data. In a proposer-solver loop, a proposer model writes training tasks and a solver model learns to solve them through reinforcement learning. Such loops have succeeded in mathematics and coding, where every answer can be checked automatically. Open-ended expert tasks, like clinical report generations, have no answer key, which makes simple question-answer generation not viable. Here the proposer must also write a rubric that lists the content an answer should include and the errors it should avoid. However, the unique challenge behind model-written rubric is the comprehensiveness of it. For naively generated rubric, solvers tend to exploit it, naming the expected topics without the substance behind them, which corrupts the learning process. To solve this problem, we introduce RRI, Recursive Reward Improvement, which uses a reward hacking adversary to supervise the reward. Before the solver trains on a generated task, an adversary writes an answer that games its rubric while being as useless as possible, the same model writes an honest answer without seeing the rubric, and up to two repair steps revise the rubric to make the honest answer win. The exploits found also rewrite how future tasks and rubrics are generated. On HealthBench Professional, RRI improves the adjusted score over the Generation Only self-improvement baseline by 10.9 points at 9B and 12.4 at 27B, and by 10.3 points without any external model. In particular, the gains are largest on difficult and writing tasks. We show that the 27B model trained on the HealthBench Professional description improves by 6.6 points on PRBench Hard and 8.0 on ProfBench without seeing a task from either. We further analyze the exploitable points of model-generated question-rubric pairs to facilitate future research on open-ended self-play. --- ### INCONVENIENT: A Benchmark to Assess LLM Generalization Capabilities from Convenience Samples Authors: Sara Rezaeimanesh, Chad Klochko, Kundan Thind, Mohammad Ghassemi Published: 2026-10-15 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2610.21038v1 Machine learning models for tabular prediction typically assume that training data are representative of the population on which they will be deployed. In practice, however, data are often collected based on convenience, availability, or cost, resulting in systematic sampling biases that can distort feature--target relationships and undermine real-world generalization. Existing tabular distribution-shift benchmarks primarily construct out-of-distribution (OOD) settings by partitioning observations across naturally occurring domains. While such partitions may induce changes in P(Y | X), the resulting shifts are neither explicitly controlled nor isolated, and the evaluation distributions themselves are often not representative of the broader target population. We introduce INCONVENIENT, a benchmark designed to study robustness to convenience-sample bias. Rather than relying on naturally occurring domain partitions, INCONVENIENT directly perturbs feature--target relationships in the training distribution while preserving the original population distribution as the deployment environment. This setting more closely reflects real-world scenarios in which models are developed from unrepresentative samples and subsequently deployed on a broader population. The distinction is particularly relevant for large language models (LLMs), whose pretraining may encode population-level knowledge that extends beyond the relationships observed in a biased training dataset. Unlike existing distribution-shift benchmarks, INCONVENIENT provides explicit control over both the direction and severity of relationship shifts, enabling a systematic assessment of robustness to convenience sampling. The benchmark comprises 200 tabular classification and regression datasets spanning diverse application domains and generates around 15,500 shifted training distributions which lead to substantial downstream degradation. --- ### Efficient Agnostic Learning of Univariate Densities Under Local Differential Privacy Authors: Mohammad Afzali, Hassan Ashtiani, Sivan Sabato Published: 2026-10-15 Categories: stat.ML, cs.CR, cs.DS, cs.IT, cs.LG arXiv: https://arxiv.org/abs/2610.21033v1 We propose a general framework for non-interactive locally differentially private (LDP) learning of structured univariate distributions that can be approximated by piece-wise polynomials. This gives the first sample-efficient LDP learners for several distribution classes, including mixtures of log-concave distributions, such as Gaussian mixture models (GMMs). These learners also provide agnostic guarantees and are computationally efficient. We also prove a lower bound for learning GMMs under LDP, improving on the standard pure-DP packing lower bounds of Bun et al. (BKSW19). To obtain our general learning framework, we develop an efficient non-interactive LDP algorithm for synthetic data generation over large domains, which may be of independent interest. We then use it together with a deterministic adaptation of the non-private piecewise polynomial learner of Acharya et al. (ADLS17). --- ### Learning When to Think: Utility-Guided Adaptive Reasoning for Efficient Text Retrieval Authors: Yingtao Luo, Kangyi Zhao, Shiyi Du, Yubo Li Published: 2026-10-15 Categories: cs.IR, cs.CL arXiv: https://arxiv.org/abs/2610.21030v1 Reasoning-conditioned embeddings can improve retrieval by progressively refining semantic constraints. However, additional reasoning may introduce over-generalized or semantically noisy abstractions that weaken retrieval specificity. Existing methods largely supervise reasoning through final retrieval outcomes, providing limited guidance over the refinement process itself. We propose AURA, a framework that directly supervises intermediate reasoning states according to how well they separate relevant documents from hard distractors. AURA further trains a lightweight utility estimator over prefix-conditioned embeddings to adapt reasoning depth at inference time. On the reasoning-intensive BRIGHT benchmark, AURA-4B achieves 34.7 average nDCG@10 while using 2.8 reasoning steps and 96 generated tokens per query. Under matched deployment conditions, adaptive stopping reduces latency from 4.6 to 0.76 seconds relative to full-trajectory reasoning. Results on MS MARCO further suggest that this specialization need not substantially degrade conventional in-domain retrieval. --- ### SCARCE: Emergent Institutions in Long-Horizon Multi-Agent Coding Systems Authors: Abhinav Kumar, Rediet Abebe, Eugene Bagdasarian, Sahar Abdelnabi Published: 2026-10-15 Categories: cs.MA, cs.AI arXiv: https://arxiv.org/abs/2610.21029v1 As autonomous agents become widely deployed, coordinating access to shared computing resources and services will become increasingly important. Recent incidents have already revealed unplanned coordination between agents, highlighting the safety implications of these interactions. Yet prior studies of agent negotiation often use stylized games and simulated bargaining, leaving open how these behaviors develop during sustained, long-horizon tasks. We propose SCARCE (Shared Compute ARena for Coding agEnts), a framework in which agents solve sequences of coding tasks under individual deadlines while negotiating access to a shared execution resource. This negotiation determines when and for how long they can inspect programs, compile code, and run tests. We find that agents spontaneously develop informal institutions: persistent rules and commitments grounded in reputation, reciprocity, and precedent. These arrangements, which we call institutional behaviors, carry across tasks and influence resource allocation, with agents sometimes foregoing immediate advantages to honor earlier agreements. We develop a 21-category taxonomy of negotiation strategies and emergent norms and use it to analyze eight models and audit more than 1,500 trajectories across different environmental settings. Among the models tested, newer frontier models exhibit more institutional behaviors than older or lightweight models. Alongside cooperation, we observe exclusionary coalitions and rare divide-and-conquer strategies. SCARCE provides a controlled setting for proactively and systematically auditing how these relationships emerge, shape access to resources, and affect the safety and coordination of multi-agent systems. --- ### On the Last-Iterate Convergence of Clipped Gradient Methods Authors: Zijian Liu Published: 2026-10-14 Categories: math.OC, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2610.21026v1 In stochastic (strongly) convex optimization, establishing last-iterate convergence guarantees for first-order methods has recently garnered increasing attention, as this not only deepens the theoretical understanding of these algorithms but also aligns more closely with practice. Notably, people have demonstrated that one of the most famous, simple, and popular methods, Stochastic Gradient Descent ($\mathtt{SGD}$), provably achieves last-iterate convergence. However, the existing results may have limited applicability, since they typically rely on strong conditions on the gradient noise (e.g., an exponentially decaying tail) rather than the more realistic heavy-tailed noise. Unfortunately, when faced with heavy-tailed noise, $\mathtt{SGD}$ is known to exhibit undesirable behavior or even fail to converge. To address the challenge of heavy-tailed noise, people have proposed $\mathtt{Clipped}\text{-}\mathtt{SGD}$, an algorithm that combines $\mathtt{SGD}$ with a simple mechanism, gradient clipping. Although $\mathtt{Clipped}\text{-}\mathtt{SGD}$ performs well in practice, its last-iterate convergence property remains highly underexplored. In this work, we prove the first optimal last-iterate rates in high probability for $\mathtt{Clipped}\text{-}\mathtt{SGD}$ under heavy-tailed noise, thereby closing a gap in the literature. --- ### Learning to Read Out: Unembedding Dynamics in Language Model Pretraining Authors: Matteo He, William F. Shen, Alex Iacob, Andrej Jovanovic, Xinchi Qiu, Nicholas D. Lane Published: 2026-10-14 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2610.21025v1 When does a language model acquire a capability? Developmental studies usually answer this by tracking when models start producing the correct token. But that prediction passes through the unembedding matrix, or readout, which is itself still being learned during pretraining. We show that a model's hidden states can encode task information before its own readout expresses it, so timelines read from outputs conflate what the model encodes with what its readout expresses. To track the readout directly, we introduce parameter-trajectory crosscoding, which fits one sparse dictionary to every token's readout row at every checkpoint. Unlike crosscoders on activations, it needs only checkpoint weights. We use it to locate a narrow early window, 0.4-1.1% of the way through Pythia's pretraining, in which the readout reorganizes and WordNet word categories become decodable from its rows. By then, linear probes on the hidden states already decode task information that the model's own outputs do not yet show. To test whether such gaps depend on how fast the readout learns, we pretrain small models from scratch that differ only in the readout's learning rate. We find that raising it makes the readout develop earlier and the gap close sooner, with little effect on final loss. Together, these results imply that early in training, where many developmental transitions are dated, a behavioral date can reflect when the readout learns to express task information and not when the model learns it. --- ### Acceleration and its Limits in Gaussian Variational Inference Authors: Jaesung R. Park, Uijeong Jang, Yifan Chen Published: 2026-10-14 Categories: stat.ML, cs.LG, math.OC arXiv: https://arxiv.org/abs/2610.21020v1 We study whether the classical momentum-based acceleration that improves the condition-number dependence of gradient descent from $κ$ to $\sqrtκ$ for strongly convex, smooth objectives extends to Gaussian variational inference (GVI) for strongly log-concave, log-smooth target distributions. With exact Gaussian expectations, a convex composite formulation of GVI yields the classical accelerated $\sqrtκ$ convergence rate. However, when only pointwise derivatives are available, we construct a family of strongly log-concave targets for which any randomized algorithm requires $κ^{1-o(1)}$ queries to achieve a fixed objective accuracy, even when the initial objective gap is bounded independently of $κ$. Thus curvature bounds alone cannot guarantee uniform accelerated total-query complexity. We then ask which additional structure permits acceleration and give positive examples: under a Hessian-Lipschitz assumption, variance-reduced stochastic methods yield accelerated guarantees when the Hessian variation is sufficiently small. Finally, experiments on real datasets show substantial empirical speedups from acceleration across a broad range of methods. --- ### Failure Is the Mother of Success: Recovering Supervision with Grounding Hindsight Authors: Sheldon Yu, Tong Yu, Nikki Lijing Kuang, Zhendong Chu, Zhenwei Tang, Sungchul Kim, Lina Yao, Julian McAuley, Junda Wu Published: 2026-10-14 Categories: cs.AI, stat.ML arXiv: https://arxiv.org/abs/2610.21018v1 When every rollout for a problem fails, GRPO receives neither a successful exemplar nor a nonzero group-relative advantage, so it learns nothing from exactly the problems it most needs to learn. The failed reasoning can nevertheless suggest what to try next. We present Replay-Selected Hindsight Distillation (RSHD), which turns failed attempts into an additional, tested source of supervision during GRPO training. The policy writes a corrective note about each failed attempt without seeing a reference answer or solution, then re-solves the problem with the note in context. A note is admitted only if its replays succeed more often than the original group did, after screening for notes that state the answer, and an admitted replay is distilled into the plain-prompt policy. Because admission is decided by fresh successes rather than by the failed rollout itself, the procedure can recover supervision from groups in which every attempt failed, and it applies equally to failures in mixed groups. Only the policy is trained, the only judge is the outcome verifier, and deployment requires no notes or extra inference. Across model families and sizes and two training domains, RSHD consistently improves over GRPO on in-domain benchmarks, and the gains carry over to out-of-domain evaluation sets. --- ### CausalEvo: Causal Skill Self-Evolution for Health LLM Agents Authors: Yuchen Zhang Published: 2026-10-14 Categories: cs.LG arXiv: https://arxiv.org/abs/2610.21016v1 Large language model (LLM) agents for health can accumulate reusable textual skills from task feedback, such as calculation criteria, measurement time points, and formula units. Deciding which skills to retain is consequential: useful rules improve later answers, while faulty generalizations propagate errors. We propose CausalEvo, a method for causal skill self-evolution that estimates the relative effects of deliberately changed skill sets while holding the model weights, feedback cases, and rule count fixed. At each evolution step, a frozen LLM proposes candidate rules, which are pooled with the incumbent skills so that the current skill set remains a feasible choice. Paired skill replacements against shared background rules are scored on the same feedback cases; least squares converts the score differences into relative skill values, and the top-ranked rules are retained, favoring incumbents on ties. A decision-loss bound separates candidate support, approximation, and estimation errors. On six MedCalc-Bench-Verified tasks with 120 held-out cases, CausalEvo achieves 60.00% accuracy, compared with 55.83% for SkillOpt and 51.67% without skills. Across the recorded training configurations, which use different optimization budgets, it consumes 55.6% fewer training tokens than SkillOpt. Gains concentrate in particular calculations alongside regressions on others, and paired accuracy intervals include zero. CausalEvo turns task feedback into controlled decisions about skill retention and replacement. Project repository: https://github.com/txbrook/CausalEvo --- ### Audio-Visual Dog Identification and Tracking in Real-World Videos Authors: Aryahvishwa Babu, Kingsley Jin, Lillian Wang, Tuan M. Dang, Theron S. Wang, Jeorge Joshi, Annie Cao, Kenny Q. Zhu Published: 2026-10-14 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2610.21011v1 Identifying individual dogs from barks in real-world videos remains challenging and largely unexplored. Solving this problem would enable long-term behavioral analysis, veterinary monitoring, and animal welfare monitoring by attributing vocalizations and behaviors to the correct individual. Real-world videos are challenging because multiple dogs may bark simultaneously, become occluded, or move in and out of view. Prior studies show that canine vocalizations and visual appearance contain discriminative identity information, but these modalities have largely been studied independently. We present a multimodal dog-fingerprinting framework combining visual identification, multiple-object tracking, and canine vocal fingerprinting to associate bark events with individual dogs while maintaining persistent profiles. The framework synchronizes bark events with tracked dogs and updates profiles using visual appearance and vocal characteristics, enabling identification despite occlusion, overlapping vocalizations, and off-screen barking. We also introduce a benchmark of 46 real-world videos containing 213 manually annotated bark events. Our framework correctly localizes the active visible dog in 61.83% of visible events at IoU $\geq 0.30$ and achieves a 91.33% F1 score for visible versus off-screen barking detection. --- ### Deep Reinforcement Learning for HPC Job Scheduling: A Statistical Evaluation Authors: Justin M. Cheney, Clement N. Nyirenda Published: 2026-10-14 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.21008v1 Deep Reinforcement Learning (DRL) offers learned job selection for High-Performance Computing (HPC), but performance against heuristics alone does not establish value beyond action masking. Six DRL treatments are compared: Proximal Policy Optimization (PPO), Advantage Actor-Critic (A2C), and Deep Q-Network (DQN), each with and without masking. Three heuristics and a seed-paired masked uniform random control provide baselines across ten seeds on two real Slurm (Simple Linux Utility for Resource Management) workload traces, one homogeneous and one heterogeneous, under a common non-parametric framework. All masked treatments complete evaluation, whereas unmasked treatments are truncated. On the homogeneous trace, no masked DRL treatment significantly improves average waiting time or average and maximum bounded slowdown over the control. On the heterogeneous trace, MaskablePPO improves all three metrics relative to the control. Its average waiting time is equivalent to that of the three heuristics within a 10% margin, and its maximum bounded slowdown is 22% lower than that of Last-Come, First-Served (LCFS). MaskablePPO has the leading average waiting-time rank among complete treatments on both traces, but Nemenyi comparisons detect no significant difference from MaskableA2C. Among masked variants, MaskablePPO has the lowest recorded mean training time, while MaskableDQN has the lowest mean decision latency. These findings distinguish completion from baseline-specific scheduling gains and support including a masked random control when assessing the value of learned selection. --- ### Mind the Tail: Data Selection Through Sparse Concepts Authors: Ziheng Zhang, Neha Hulkund, Colin Lee, Zheda Mai, Hui Shen, Ying Ding, Tanya Berger-Wolf, Sara Beery, Wei-Lun Chao, Jianyang Gu Published: 2026-10-14 Categories: cs.CV arXiv: https://arxiv.org/abs/2610.21007v1 Data selection reallocates training support across visual concepts, including objects, attributes, color, texture, and backgrounds. Different allocation strategies may produce varied downstream model behavior, especially for tail concepts supported by only a few samples. Discarding them can hurt test-time robustness, while overemphasizing samples lying on category boundaries may introduce unstable training signals. This work explicitly investigates this relationship across fine-grained, spurious-correlation, and general image classification, as well as multimodal pretraining. Specifically, we use sparse autoencoders (SAEs) to extract concept activations from individual images. During selection, we accumulate the dataset-level concept support for each added sample. Applying diminishing returns to this distribution down-weights well-represented concepts and shifts the focus to the distribution tail. We then incorporate local support to find samples that reliably strengthen under-supported concepts. Empirical results show consistent improvements over strong data selection baselines across all benchmarks. We also reveal the connection between concepts our method enhances and the induced behavioral changes, including improved robustness against out-of-distribution and spurious contexts. This work establishes concept support, beyond individual sample importance, as a useful principle for data-efficient and robust learning. --- ### Can LLMs Anticipate Their Cruxes? Authors: David Reber, James Reber, Ari Holtzman, Victor Veitch Published: 2026-10-14 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2610.21006v1 To use language models to help us make decisions we would like to understand the rationale behind the language model's outputs. In principle, the language model provides its rationale as part of its text output. However, the sheer volume of text makes this hard to use in practice. In this paper, we are interested in extracting the cruxes that lead a model to recommend one action over another. We take a crux to be a sub-question that a model's rollouts disagree on, where changing its value may have altered the conclusion. A crux is therefore a property of the distribution of a model's rollouts, not of any one rollout. We consider three strategies of increasing computational cost. First, we just ask a model to identify, a priori, what the cruxes of a decision will be. Second, we have the model actually go through the process of generating a response and then extract the cruxes that arose in its reasoning. Finally, we generate many responses and extract the cruxes by comparing the reasoning across the rollouts. We make this last procedure computationally tractable with cheap calls to a reranker, an estimated 86 times less compute than a judge. We then conduct an extensive empirical evaluation to assess whether the computationally inexpensive options are sufficient. They are not. Across domains and models, we find that a model's a priori assessment of the cruxes has little relation to the factors that are actually identified as important a posteriori. Using a single rollout helps, but is still inadequate for identifying the full set of cruxes. Accordingly, for high-stakes decisions (and to understand the space of language model decision making) the many rollout procedure is unavoidable. --- ### Imperfect Is Perfect: Learned Column Prefilling for Branch-Price-and-Cut Authors: Haoran Liu, Yu Yang Published: 2026-10-14 Categories: cs.LG arXiv: https://arxiv.org/abs/2610.21001v1 Column generation folklore warns that initial columns hurt: a warm start may bias the dual and slow the root of branch-price-and-cut. We show the opposite once the columns are imperfect in the right way. Preloading the support of the optimal root LP, the best columns one could know in advance, fails to finish on half of our hardest instances; a pool of ordinary routes that need not belong to any optimal solution, held outside the master and priced at the current smoothed dual, speeds the root up. Because the columns need not be optimal, producing them becomes a learning task: a neural edge scorer guides a constructive generator that builds the pool in seconds. A second network predicts the initial center of Wentges smoothing from the instance file, and with it the direction of the smoothed pricing queries; the solver's own updates run unchanged. With prefill and center in place, exact pricing dominates root pricing time, so we rebuild bucket-graph labeling around its dominance test for the GPU and certify on the CPU. Inside RouteOpt, the state-of-the-art exact solver for capacitated vehicle routing, the pipeline reaches the same certified root bound in 13.2% of vanilla RouteOpt's root pricing time over 95 instances with 300 to 1000 customers, prefill alone in 26.1%, and 17.5 times faster in the best case. In an exact branch-price-and-cut solver for unrelated parallel machine scheduling, prefill with smoothing and the learned center certifies all 25 benchmark instances, 12 times faster than the solver's own column generation, which certifies 20 of them. --- ### EigDiff: Eigensample Diffusion - or how to significantly reduce training dataset size with marginal performance variation? Authors: Andrei Jarca, Vlad Hondru, Florinel-Alin Croitoru, Radu Tudor Ionescu Published: 2026-10-14 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2610.20999v1 Diffusion models are notoriously expensive to train, requiring large-scale datasets to learn how to gradually reverse the noise from images over multiple small denoising steps. Current approaches focus on distilling the number of diffusion steps, neglecting potential redundancies of the training data. In this paper, we explore the potential of harnessing eigensamples to train diffusion models on significantly smaller datasets, while preserving model performance. To mitigate the sensitivity to affine transformations of eigenimages, we harness the latent space of a pre-trained CLIP model. We next compute latent eigenvectors by performing principal component analysis (PCA) in CLIP space. Further, we employ eigenvectors as reference points to guide the selection of a representative training set comprising comprising both exemplars (images whose embeddings are closest to eigenvectors) and diverse images. Our comprehensive experiments with different diffusion models applied on multiple datasets indicate that our approach (EigDiff) can reduce training data by up to 14x and accelerate subset selection by up to 1607x, with marginal performance differences. --- ### Halve-Then-Greedy: Adaptive Sampling for Posterior Success with Near-Optimal Instance Dependence Authors: Yinan Li, Kwang-Sung Jun, Kapilan Balagopalan, Tuan Nguyen, Yao Zhao, Abbas Zaidi, Houssam Nassif Published: 2026-10-14 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2610.20997v1 We study fixed-budget adaptive experimentation aimed at maximizing the probability of obtaining sufficient posterior evidence that a recommended design's mean performance exceeds a target. Unlike best-arm identification, this objective does not require finding the best design. For Gaussian bandits with a Gaussian prior, we propose \emph{Halve-Then-Greedy} (HTG), which combines a large initial balanced allocation, successive halving, and posterior-greedy sampling. We prove instance-dependent, finite-budget exponential bounds on posterior-failure probability that capture both separation from unfavorable arms and the presence of multiple favorable arms. On one-positive-arm families, we establish an asymptotic sample-complexity lower bound that HTG matches up to a logarithmic factor. We further analyze uniform sampling and sequential halving, revealing regimes where they require more samples than HTG by linear and logarithmic factors, respectively. Simulations illustrate HTG's sample-efficiency advantages and their dependence on instance structure. --- ### One Bias Score, Two Factors: How Mitigations Affect Option Bias in Multiple-Choice Evaluation Authors: Hyunjong Ok, Minjae Park, Jaeho Lee Published: 2026-10-14 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2610.20993v1 Reordering multiple-choice options can change a model's answer, an effect studied as selection bias (a preference for certain option letters) and as order sensitivity (changes in answers across option orderings). We connect these perspectives through the pooled selection-rate standard deviation (pooled SD), which measures how unevenly option letters are selected across questions and orderings. When every option of each question appears equally often at each position, the pooled SD is exactly the product of two label-free factors: the flip rate and the positional skew. The flip rate is the fraction of questions whose chosen option changes across orderings; the positional skew is the standard deviation of option-letter selection rates among those questions. Across 41 checkpoints and 12 benchmarks, the flip rate tracks accuracy and output uncertainty, while the positional skew tracks the estimated option-symbol prior. Prior correction for selection bias mainly reduces the positional skew. Aggregating each ordering with its reversal, a mitigation for order sensitivity, mainly reduces the flip rate. Both reduce the range of accuracy across correct-answer letters by similar amounts, but aggregation leaves fewer model-rank reversals. Applying prior correction before aggregation reduces both factors and further narrows this accuracy range. However, aggregation lowers the flip rate partly by settling questions on the same wrong option. We therefore recommend reporting both factors to distinguish mitigation effects, alongside accuracy to distinguish stability from correctness. --- ### DeSpatial: The devil of Agentic Spatial Reasoning Lies in 3D Itself Authors: Yifei Liu, Xiaolong Zhou, Ziyang Gong, Yifan Zhou, Yuanyuan Gao, Yuning Gong, Yuchen Yang, Hongjie Zhang, Xue Yang, Xiao Sun, Zhihang Zhong Published: 2026-10-14 Categories: cs.CV arXiv: https://arxiv.org/abs/2610.20990v1 Spatial agents have access to reconstruction and segmentation tools, yet inconsistent predictions and fragmented object identities can prevent their outputs from becoming reliable 3D scene information. A simple oracle experiment exposes this perception bottleneck: replacing tool-predicted perception with ground-truth 3D bounding boxes raises an agent to nearly full score across five object-centric VSI-Bench tasks, while retaining the same reasoning tools. Motivated by this finding, we believe the devil of spatial reasoning lies in 3D itself, and introduce DeSpatial(Devil of Spatial), which explicitly decomposes spatial understanding into perception and reasoning. DeSpatial has a shared 3D perception tool to reconcile independent perception outputs into a unified, metric, gravity-aligned 3D scene. This perception tool has two core components: a) a joint segmentation--depth co-refinement module that consolidates masks of the same object and removes projection artifacts caused by inconsistent segmentation and depth within a single frame; b) a cross-view instance fusion module that consolidate voxel-overlapping instances but not allowed to merge same-frame instances. This fusion module additionally fuses spatially disjoint fragments caused by insufficient multi-view observations, by checking their 3D structure and source mask appearance. After receiving the complete and unified 3D scene information, the agent chooses reasoning or computational tools to solve the spatial problem. DeSpatial achieves leading performance on VSI-Bench, MindCube-Tiny, and ViewSpatial, with additional long-video results on VSI-SUPER, revealing the bottleneck of spatial understanding lies in a precise 3D perception itself. --- ### Do As I Do, Not As I Say: On-Policy Distillation Inherits the Teacher in Vision-Language Models Authors: Frank F. Yang, Kefei Yan, Yiren Wang, Yiyang Wang Published: 2026-10-14 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2610.20989v1 Improving the reasoning capabilities of vision-language models (VLMs) often relies on post-training with supervision from stronger models. Two common strategies are supervised fine-tuning (SFT), which trains on teacher-generated demonstrations, and on-policy distillation (OPD), which provides teacher feedback on trajectories generated by the student itself. While OPD can be more data-efficient by supervising student-generated trajectories, it is unclear how the teacher's supervision at each state affects the student's learned behavior. To study this question, we conduct a controlled comparison of OPD and filtered SFT across data budgets and tasks where the teacher exhibits different levels of reliability, and analyze how these differences shape student accuracy and completion behavior. Across the two tasks, we find that OPD is effective only when the teacher remains reliable on student-generated trajectories. On ChartQA, where the teacher completes 99% of its solutions, OPD trained on only 10% of the data comes within 1.4 accuracy points of full-data SFT. In contrast, on Geometry3K, where the teacher fails to complete 22% of its solutions, OPD underperforms SFT by 7.9 points, with most of the gap explained by incomplete responses. More strikingly, starting OPD from an SFT student reduces its completion rate from 76% to 62%, effectively erasing the completion behavior learned through SFT. Filtering teacher demonstrations for completeness alone recovers much of SFT's advantage. These results show that on-policy supervision can transfer not only teacher knowledge but also persistent failure modes on student-visited states. Code is available at https://github.com/ffy208/opd-inherits-teacher-vlm. --- ### The Blume-Capel Model as a $φ^6$ Field Theory: A Monte Carlo and Flow Model Study Authors: Ejaaz Merali, Ryan Larson, Ehsan Khatami, Richard T. Scalettar Published: 2026-10-14 Categories: cond-mat.stat-mech, cond-mat.dis-nn, cs.LG, hep-lat arXiv: https://arxiv.org/abs/2610.20982v1 We report Monte Carlo simulations of a continuous spin variant of the Blume-Capel model. We compute the phase diagrams in the crystal-field-temperature plane, as a function of the softness of the spins. We show that normalizing flows, a class of generative models capable of direct sampling without autocorrelation, are effective samplers across the phase diagram, including in the first-order transition region and in the vicinity of the tricritical point. Finite-size scaling analyses of flow samples locate the first- and second-order transitions as well as the tricritical point with accuracy comparable to canonical and extended-ensemble Monte Carlo. --- ### How to Train Your Little Dragon: Teaching a Small Language Model to Construct and Refine GPU Optimization Scripts Authors: Annabelle Sujun Tang, Wei Tang, Songyao Jiang, Tianhui Li, Shengzhe Xu, Jiin Woo, Kaan Ozkara, Youngsuk Park, Zhen Jia, Hadi Esmaeilzadeh Published: 2026-10-14 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2610.20980v1 Compiler transformations interact, making GPU kernel optimization a search for combinations that preserve the computation while improving performance. We introduce HOW TO TRAIN YOUR LITTLE DRAGON, a training method that teaches a small language model both to construct complete optimization scripts and to improve existing ones. Instead of ranking candidates or generating kernel code, our model generates optimization scripts. We devise a compact, lossless representation whose deterministic decoding reconstructs the exact compiler trace, preserving its transformation parameters and dependencies. We use a compiler framework TVM MetaSchedule to build training data on 54,116 distinct workloads, and construct 108,594 examples in two training sets: 60,236 teach the model to generate an optimized script from an unoptimized kernel, while 48,358 teach it to produce a better script given the kernel and a somewhat optimized script. Supervised fine-tuning teaches the model to construct and improve optimization scripts from these search-derived examples. We then add reinforcement learning with a shaped reward that combines correctness with measured performance, awarding a latency-based bonus only to numerically correct kernels. We apply this training method to a model with 8 billion parameters. After supervised fine-tuning and reinforcement learning, the model achieves per-attempt correctness of 91.27% for construction and 92.01% for refinement, while reinforcement learning produces faster kernels than supervised training on 88 of 102 construction tasks and 145 of 170 refinement prompts. With only 8B parameters, our LITTLE DRAGON achieves higher per-attempt correctness and better kernel performance than six frontier LLMs: GPT-6 Astra, Opus 5, Kimi-K3, and GLM-5.3 among them, each with hundreds of billions to trillions of parameters, for which task-specific training is unavailable or expensive. --- ### Two-Stage Score Estimation for Group-Specific Diffusion Models Authors: Huitao Yang, Oscar Hernan Madrid Padilla Published: 2026-10-14 Categories: stat.ML, cs.LG, math.ST arXiv: https://arxiv.org/abs/2610.20979v1 Generative modeling across related distributions raises a question: how can shared structure support accurate generation for a target distribution with limited observations? We develop a statistical framework for diffusion transfer that analyzes a shared pooled score together with a target-specific offset. Under Besov regularity of the pooled density and a stable hierarchical composition model for the offset, we derive convergence rates that separate the costs of learning these components and identify when a simpler offset yields a faster risk-bound rate than separate fitting. Under suitable boundary and regularity assumptions, we prove that the offset remains bounded and jointly regular as diffusion time approaches zero. These results allow the lower time cutoff to vanish polynomially and provide total-variation guarantees for diffusion models as density estimators. Synthetic simulations and image benchmarks provide a proof of concept for the framework. --- ### PiCo: Physics-Informed Concept Probing for Geo-Foundation Models Authors: Yihang She, Andrew Blake, Jianbo Qi, Srinivasan Keshav Published: 2026-10-14 Categories: cs.CV arXiv: https://arxiv.org/abs/2610.20973v1 Geo-foundation models (GFMs) are typically evaluated by downstream task predictive performance, revealing little about which physical features they are sensitive to or use in prediction. This limits developers' ability to test hypotheses about model behaviour and users' ability to select models for new tasks. We propose Physics-Informed Concept Probing (PiCo), a post-hoc framework for explaining GFM predictions. PiCo identifies concept vectors in a GFM's embedding space by generating synthetic multispectral satellite observations from a 3D radiative transfer model coupled with a game engine. These vectors reveal physical sensitivities that inform domain-based hypotheses about downstream performance. Projecting embeddings onto these concept vectors produces low-dimensional representations for quantifying concept importance in making predictions. We evaluate PiCo on three GFMs with different input modalities using seven physical concepts covering forest leaf optics and canopy structure. In ground-cover estimation, RGB restriction reduces dry-matter sensitivity and disproportionately degrades predictions for dry vegetation and bare ground, supporting a hypothesis derived from sensitivity analysis. For hierarchical forest type and tree species classification, concept representations retain 70.0-93.8% of full-embedding accuracy. Models capture similar physical contrasts between classes, such as structural differences between broadleaf forest and cleared ground and leaf-water-content differences between oak and beech, yet differ in the strength of these contrasts and the concepts driving classification. PiCo thus provides a physical basis for hypothesis testing and model assessment, with potential applicability beyond Earth observation wherever forward models enable controlled simulation and pretrained embeddings are accessible for probing. --- ### SOMA: Stack-of-Objects Manipulation using a Quadruped Robot Authors: Stergios E. Bachoumas, Panagiotis Artemiadis Published: 2026-10-14 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2610.20968v1 Executing heavy-duty physical manipulation in unpredictable environments remains a major challenge for quadrupedal robots lacking dedicated manipulators. This paper presents SOMA, a hierarchical reinforcement learning framework for whole-body non-prehensile manipulation of multi-object stacks using a quadruped robot. Unlike traditional methods restricted to two-dimensional planar pushing, SOMA integrates explicit robot base-height control, unlocking full three-dimensional pushing strategies. This capability enables the quadruped to manipulate complex payload stacks as a unified system while preventing relative sliding or tipping of upper layers. The architecture factorizes control into a high-level manipulation policy using a Simplified Structured State Space Sequence (S5) encoder to reason over long-horizon temporal history and predict bounded target tracking commands via Beta distributions, and a low-level locomotion policy that dynamically adapts leg stance configurations. In extensive simulation benchmarks, SOMA achieves a $90.3\%$ success rate for single-object reposing and up to $90.1\%$ for dynamic two-box stacks under stringent error margins ($0.05\text{ m}$ position error, $1^\circ$ orientation error). Finally, we validate zero-shot sim-to-real transfer on a Unitree Go1 quadruped robot across diverse payload masses and geometries, demonstrating active base-height adjustments to maintain traction and balance during heavy pushing tasks. --- ### MIDI-REPA: Improving Text-to-Music Generation with Symbolic Representation Alignment Authors: Song-Ze Yu, Fang-Duo Tsai, Yi-Hsuan Yang Published: 2026-10-14 Categories: cs.SD, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2610.20967v1 State-of-the-art text-to-music models increasingly rely on scaling model capacity, data, and compute, while still learning musical structure directly from dense audio. Yet melody, rhythm, harmony, and structure can be represented compactly in symbolic form. We introduce MIDI-REPA, a representation-alignment framework that derives MIDI supervision from automatic music transcription (AMT) to align intermediate representations of a flow-matching DiT. Building on the 122M-parameter FluxAudio-S baseline from the 2026 ICME Academic Text-To-Music Challenge, we investigate what, where, and when to align. Aligning a MuScriptor-derived piano roll with an early DiT layer throughout the full flow trajectory performs the best. Without additional training data or techniques, MIDI-REPA remains competitive on the ATTM objective metrics while remaining trainable on a single RTX 3090 in about 31 hours. Our results highlight symbolic supervision as a simple and inexpensive learning signal for improving text-to-music generation. We release our transcription pipeline, JamendoMIDI, training code, and audio samples at https://midi-repa.vaclis.net. --- ### MSP: Modality Self-Play for Training Multimodal LLMs Without Paired Cross-Modal Data Authors: Robert Moseley, Md Kaykobad Reza, Ameya Patil, Edward Ayrapetian, Salman Asif Published: 2026-10-14 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2610.20965v1 Large multimodal models have achieved remarkable progress by scaling model capacity and training on massive paired multimodal data. However, this paradigm introduces a fundamental bottleneck: aligned multimodal datasets are expensive to curate, difficult to scale across diverse modality combinations, and increasingly constrain further progress. We introduce Modality Self-Play (MSP), a training framework for multimodal foundation models based on proxy-mediated modality composition. Instead of relying on paired cross-modal data, MSP uses anchor-supporting proxy textual descriptions of unseen modalities as semantic bridges during training. Given an observed modality (e.g., image or audio), the model generates proxy descriptions for complementary unseen modalities, which are then encoded and aligned with the observed modality through learned bridges, enabling cross-modal interaction without real paired non-text supervision. Our central hypothesis is that proxy text, combined with contrastively trained pretrained encoders, provides sufficient semantic structure for multimodal composition to emerge. We show that MSP enables zero-shot modality composition: models trained only with a real anchor modality and accompanying proxy descriptions can integrate and reason over multiple modalities at inference time despite never observing real paired non-text data during training. Empirically, MSP achieves state-of-the-art or competitive performance on multiple multimodal benchmarks. Ablation studies further demonstrate that learned bridges improve alignment, while anchor-supporting proxy descriptions enable effective cross-modal composition. MSP removes the need for real paired non-text (e.g., audio-vision) supervision, while continuing to rely on modality-text pairs, pretrained contrastive encoders, and modality-distribution bridge calibration. --- ### VELD-QA: A Large-Scale Dataset for Visual Evidence Grounding and Question Answering in Long Documents Authors: Zhivar Sourati, Marianne Menglin Liu, Rongguang Wang, Zheng Wang, Yazhe Hu, Lan Guo, Yuying Wang, Mengqing Guo, Bhavesh Dnyaneshwar Gore, Shubham Pathak, Ramapriya Ranganath, Sandeep Sharma et al. Published: 2026-10-14 Categories: cs.CV, cs.CL arXiv: https://arxiv.org/abs/2610.20964v1 Existing long-document QA benchmarks rarely combine large-scale human-authored questions with detailed visual evidence annotations, limiting diagnosis of evidence localization and interpretation failures. We present VELD-QA, a large-scale benchmark for Visual Evidence grounding and question answering in Long Documents. It contains 15,000 fully human-authored question-answer pairs across 5,859 real-world PDFs totaling 372,445 pages, balancing chart, table, and complex-layout tasks. Human annotations identify supporting pages and evidence elements and describe cited visual content, enabling separate retrieval and interpretation evaluation. We evaluate nine vision-language models under four evidence settings and seven retrieval-augmented and agentic systems. Our analysis identifies bottlenecks in evidence localization and interpretation. Retrieval-based methods substantially underperform the oracle setting with ground-truth supporting pages; full-document access does not close the gap. Localization alone is insufficient: oracle accuracy peaks at 81.7%, highlighting persistent visual interpretation and reasoning challenges. Enriching inputs with our visual-element annotations narrows the performance gap by 17-24%, demonstrating value beyond evaluation and motivating data enrichment for multimodal document QA. Our code is available at https://github.com/oracle-samples/veld-qa. --- ### StepSafe: Learning Timely Intervention with Context-Aware Safety Reasoning Authors: Siyi Gu, Kunhan Gao, Chao Peter Yang, Sukanta Ganguly, Rex Ying, Arman Cohan Published: 2026-10-14 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.20960v1 Safety in agentic systems is inherently sequential: whether an action is safe depends on prior actions, tool outputs, information disclosures, and interactions among agents. Recent work has shifted agent safety guardrails from post-hoc trajectory-level assessment toward step-level action diagnosis and proactive risk prediction, yet existing training objectives provide limited supervision for how step evidence accumulates over time and when it becomes sufficient to justify intervention. We propose \methodlong~(ASPO): a training framework for step-level timely intervention. Pre-fix level stage alignment allows crediting each decision at its own step: it compares each response with the other responses sampled at the same prefix and rewards the action that fits the stage, continuing before the risk is visible, asking while clarification can still avert harm, and stopping before harm is irreversible. Adaptive feedback self-distillation further credits the reasoning behind each decision, with the model itself as a teacher that sees the step annotations on intervention steps and rules mined from its own false alarms on safe steps. Its training data carry a violation taxonomy and an intervention window at every step and are extended through self-evolving multi-agent red-teaming. Finally we present \textsc{StepSafe}, a step-level reasoning guardrail with timely intervention, that achieves the state-of-the-art peformance among open-source guardrails across five agent-safety benchmarks on safety-utility pareto curve with Youden's $J$ of 53.0 (4B) against 46.2. --- ### Same Context, Different Agents: A Behavioral Study of Semantic Layers, Orchestration and Compaction in Data Agents Authors: Zhivar Sourati, Marianne Menglin Liu, Zheng Wang, Rongguang Wang, Sai Ashish Somayajula, Tao Sheng, Kyu J. Han, Sujith Ravi, Dan Roth Published: 2026-10-14 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2610.20959v1 Answering questions over heterogeneous data means combining evidence from files, databases, documents and the web. Data agents automate this, but their success depends on context. Three context-engineering techniques (CETs) help by shaping it: semantic layers describe the data, orchestration splits work among agents, and compaction condenses interaction history. Existing studies validate each CET alone, leaving open how agent behavior shapes its effect. We present a comprehensive analysis of the three CETs, alone and combined, with seven large language models (LLMs) on 1,514 long-horizon tasks from five benchmarks, covering behavior, accuracy and cost. We find that data agents without CETs form three behavior groups: investigators, premature answerers that answer without inspecting data, and early stoppers that give up. Each CET delivers its known benefit, from more investigation to smaller prompts, but only the behavior group it matches turns that benefit into accuracy, such as premature answerers under orchestration. For investigators, the same CETs can hurt: the semantic layer stops them from looking, orchestration makes them overconfident, costing 5 percentage points, and compaction makes them repeat steps. Combining CETs yields smaller gains and losses than their sum. We derive guidance for choosing CETs by agent behavior, task demands and accuracy-cost trade-offs. --- ### Line-Augmented cuVSLAM: Robust Visual Odometry in Low-Texture Environments Authors: Dara Molloy, Terry Wu, Wonju Lee, Abhilash Nand Kumar, David Adley, Conor Ferguson, Hesam Rabeti Published: 2026-10-14 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2610.20958v1 Autonomous mobile robots in warehouses and production facilities operate in precisely the environments that defeat point-feature visual odometry: featureless aisles, uniform walls, repetitive racking and specular floors. Such environments are, however, rich in straight-edge structure: rack uprights, door frames, wall junctions and cabinet outlines present high-contrast linear edges stable across wide viewpoint changes. A line-feature extension to cuVSLAM, a CUDA-accelerated visual-SLAM system, is presented that recovers this signal. Lines are parameterised as Plücker coordinates for geometric transport and as the minimal 4-DOF orthonormal representation of Bartoli and Sturm for optimisation; a Cauchy-robust point-to-line residual with analytic Jacobians is folded directly into cuVSLAM's per-frame Gauss-Newton pose solve, and the same line state is jointly refined in the Schur-complement sliding-window bundle adjustment. An adaptive information weight causes lines to assume more of the estimation burden exactly as point features become scarce, and the extension adds residual blocks to existing solvers rather than introducing a second estimator, reducing exactly to the unmodified point pipeline when disabled. Evaluated in RGB-D mode under SE(3) Umeyama alignment with no scale correction, line features reduce mean absolute trajectory error by 38.5% over the 22 sequences of the TUM RGB-D Freiburg3 collection and by 30.1% over the 8 Replica room and office sequences, where median translation error falls by 38.9% and median rotation error by 66.4%, with zero tracking failures in any configuration tested. --- ### Cross-subject decoding of imagined speech from scalp EEG: a leakage-controlled evaluation on three public corpora Authors: Blessing Isoyiza Adeika, Onyema Osuagwu Published: 2026-10-14 Categories: eess.SP, cs.LG, q-bio.NC arXiv: https://arxiv.org/abs/2610.20954v1 Objective. A decoder of imagined speech that works for a new user without calibration would make a non-invasive speech interface practical. We tested whether imagined-speech content recorded with scalp electroencephalography (EEG) transfers across people, and examined why accuracy within a single person remains low. Approach. We analyzed three public corpora, ds003626, KaraOne and Track 3 of the 2020 International Brain-Computer Interface Competition, under a protocol that enforces split integrity by assertion and reports exact binomial thresholds, group tests and bootstrap confidence intervals. We trained decoders on each participant and tested them on every other, compared representational geometry across participants without a classifier, applied subject-level alignment, and tested five explanations for the within-subject ceiling. Main results. Across 482 ordered pairs of participants, no pair survived correction for multiple comparisons; mean pair-transfer accuracy lay within 0.006 of chance on every corpus, while every within-subject reference lay above it. Imagery against pre-stimulus rest generalized at about 0.61 after artifact and elapsed-time controls, but its decoder labelled the interval preceding imagery as imagery as often as imagery itself, so the contrast reflects a post-stimulus state, not imagery. Within subject, only the competition corpus decoded above threshold for every participant (0.3147; chance 0.200); syllable count explained little of this, but its frequency profile failed an artifact audit. Significance. In these corpora, imagined-speech content did not transfer between people through any representation we examined. Trial count per class was the only variable that ordered the corpora, which favors larger single-condition datasets over further model development. --- ### LEV: Thinking Fast and Deep with Looped Decision Models Authors: Dingjie Song, Zhiling Yan, Yuxuan Zhang, Lichao Sun Published: 2026-10-14 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2610.20951v1 Decision models answer typed questions about an input with one probability per option, computed in a single forward pass without generating text. Like the fast System 1 of dual-process theory, they answer at once, but their depth is fixed at training time, so they spend the same computation on easy and hard questions. Models that reason in generated text can think longer, but they pay for every extra step in decoded tokens. We present LEV, a decision model that thinks deeper by rerunning its layers in a loop instead of generating tokens. LEV builds on a language model pretrained as a loop, reads the options after every loop step and trains all steps jointly, so one checkpoint can answer at any depth. On three looped backbones, accuracy rises with depth on the same weights, mostly on hard questions, and a 2.6B LEV goes from 63.2% at depth one to 76.2% at depth four on JevBench. On-demand depth, set per question by a small stopper with a threshold fitted on unlabelled requests, saves 15% to 24% of the computation at a cost of under one accuracy point, level with or slightly above a fixed mix of depths. On a locked test set, the 1.4B and 3B models are 5.4 to 8.8 points more accurate than dense (non-looped) models and cascades of the same latency below 50 ms. --- ### Ranked Right, Sized Wrong: Ensemble Spread Fails Silently at Streamflow Extremes Authors: Justin Gallant, Tyler Wilson Published: 2026-10-14 Categories: stat.ML, cs.LG, physics.geo-ph arXiv: https://arxiv.org/abs/2610.20947v1 Deep learning models, particularly Long Short-Term Memory networks, are increasingly used for operational streamflow forecasting, with ensemble spread providing a signal of predictive uncertainty. As climate change increases the frequency of hydroclimatic extremes, these models will increasingly operate under flow conditions that are rare in their historical records. We therefore evaluate whether ensemble spread remains informative under increasing flows, examining both its magnitude relative to error and its ability to rank errors. Prediction error is approximately $2\times$ the corrected spread at ordinary flows and up to $4\times$ at the highest flows, with the discrepancy increasing with flow (log-log slope $1.30$, 95% CI: $(1.20,1.41)$). Under-dispersion is strongest in observed extreme flows beyond the training record, reaching an $18\times$ discrepancy. This is pervasive and not explainable by conditional mean bias. Area Under the Sparsification Error, a standard distribution-free metric of uncertainty ranking, describing how well a model knows when it is uncertain, indicates near-oracle ranking for ensemble spread ($0.071$), yet ordering by predicted discharge alone nearly matches it. Informative uncertainty ranking can coexist with severe under-dispersion. In deployment, error ranking cannot be verified, yet the spread is available, and it is relied on as the uncertainty signal under increasingly extreme conditions, while systematically understating risk, most severely for the extreme events that matter most. --- ### Screwtape: Adversarial Transcript Editing for Monitor Red-Teaming Authors: Gosha Kozhevnikov, Tyler Tracy Published: 2026-10-14 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2610.20941v1 We present Screwtape, a pipeline that converts real agent transcripts into attacks that evade action-only monitors. An LLM attacker edits a transcript so that the agent causes a specified harm, such as leaking an API key or spawning an unmonitored agent. The edited actions are never executed, so we use an LLM verifier to reject attacks that are inconsistent with the original transcript or below a severity bar. The attacker revises the edit until the monitor no longer flags it. We applied Screwtape to 94 Claude Code and Codex transcripts from Redwood Research staff, optimizing against our own monitors. Across 252 runs, it produced 35 attacks that passed the verifier and evaded these monitors. Six of them, confirmed valid by human review, also evaded a held-out monitor provided by a frontier developer. We therefore demonstrate that a red-teamer without access to a developer's infrastructure can find confirmed failures of the developer's monitor. We expect to find more attacks when the pipeline runs inside the developer's infrastructure, with frontier attacker models that lack refusal safeguards and direct optimization against the production monitor. Developers can use these attacks to fix monitor blind spots, and evaluators can use them as counterexamples to safety claims about the monitor. --- ### MAI-Gym: An Open All-in-One Platform for GUI Agent Evaluation and RL Training across Mobile and Computer Environments Authors: Panrong Tong, Quyu Kong, Pengxiang Li, Tianyu Xia, Hanzhang Zhou, Lei Wang, Chenglin Cai, Long Li, Long Chen, Jianan Zhang, Zhengwen Zeng, Xu Zhang et al. Published: 2026-10-14 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.20939v1 Online reinforcement learning (RL) with verifiable rewards has driven rapid progress in mathematics, coding, and tool use, but scaling this success to GUI agents remains challenging due to the high cost of interactive environments and the scarcity of realistic training tasks. We introduce MAI-Gym, an open all-in-one platform that unifies GUI agent evaluation, data collection, and online RL training across mobile, computer, and web environments. Through a shared rollout pipeline, MAI-Gym ensures consistent agent behavior across evaluation and training, supports ten established benchmarks, and scales to 20,000 parallel environment instances. Beyond scalable infrastructure, MAI-Gym enables automated synthesis of diverse GUI applications and challenging tasks, including resettable mocked apps and coherent cross-app worlds with execution-validated task verifiers. Using MAI-Gym, we train MAI-Gym-9B, a unified 9B agent for mobile and computer use, achieving state-of-the-art performance among open GUI agents under 10B parameters: 78.4% on MobileWorld and 76.2% on OSWorld-Verified. Importantly, RL in a mocked train-ticketing app transfers to its real-world counterpart, improving task success on a phys- ical phone from 28.8% to 48.1%. We release the framework, synthesis skills, mocked apps, 2,000 verified tasks, and the model checkpoint to support repro- ducible research and further advances in scalable GUI agent training. --- ### RAC: A Relational Agent Contract for Constraint-Governed Engineering Tasks Authors: Kaiyuan Zheng, Fengqing Jiang, Yichen Feng, Zubin Aaron Assadian, Yuetai Li, Bhaskar Ramasubramanian, Basel Alomair, Luyao Niu, Radha Poovendran Published: 2026-10-14 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.20935v1 Engineering agents must produce artifacts that satisfy exact, coupled constraints. Existing solutions often require agents to make semantic decisions and also derive consequences such as precise geometry or timing from those decisions, leading to failures in constraint-governed engineering tasks. In this paper, we introduce RAC, a harness-agnostic contract for the action and feedback layer of engineering agents. It specifies authoritative observations, semantic actions, tool-based realization, independent verification, and repair evidence. The agent decides which relations an artifact should satisfy; specialized tools derive their concrete consequences. On PCB placement, routing, and repair tasks, RAC achieves the highest clean-pass rate for all nine evaluated models, averaging 90.9%, compared with 74.0% for the strongest baseline, requirements-only delegation. In job-shop scheduling, direct relational actions achieve 96.3% quality pass versus 59.6% for code-mediated delivery with access to the same scheduling backend. Our evaluations on 141 embedded-code tasks across four microcontroller identify the action contract as a source of agent capability: preserving semantic control while delegating mechanically derivable outputs improves the reliability of constrained artifacts. ---