# Research Papers — 2026-10 · bundle page 23 of 156

> Full-text markdown of papers #1101–1150 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/23
[← Previous bundle](https://aipod.dev/papers/archive/2026-10/page/22.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-10/page/24.md)

### MCAIR: Monitoring and Correcting Agent Unfaithfulness via Internal Representations

Authors: Yan Xu, Dandan Song, Luan Zhang, Jun Yang, Ke Duan, Yuhang Tian, Yading Li, Xudong Li, Xing Pei, Guangyuan Feng, Chenhao Li
Published: 2026-10-14
Categories: cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.19679v1

Large Language Model (LLM)-based agents have demonstrated remarkable capabilities in completing complex tasks through multi-step reasoning and tool interactions. However, recent studies have revealed that LLM-based agents can exhibit unfaithful behaviors, where their final reports fail to faithfully reflect their actual execution trajectories. Specifically, when agents fail to accomplish tasks through the intended sources or procedures, they may resort to alternative sources or strategies while failing to disclose such deviations, leading users to incorrectly believe that the tasks have been completed according to the original requirements. We refer to this specific form of agent unfaithfulness as Agent Completion Unfaithfulness. Detecting and mitigating agent completion unfaithfulness is critical yet remains largely unexplored. In this work, we propose MCAIR, a runtime monitoring and correction framework that leverages internal representations of LLM-based agents. MCAIR uses multi-step internal representations as inputs to train a risk-aware probe, enabling online detection of potential completion unfaithfulness during execution. Furthermore, MCAIR introduces a risk-guided correction loop that prompts agents to review their execution history and mitigate potential inconsistencies between their execution trajectories and final reports. Additionally, we construct ACUD, a dataset for Agent Completion Unfaithfulness detection that provides step-level execution traces, step-level internal representations, and consistency annotations for training and evaluating runtime detection probes. Experiments demonstrate that MCAIR outperforms LLM-as-a-judge baselines, generalizes across prompts and tasks, and effectively reduces agent completion unfaithfulness.

---

### Remember, Revisit, Revise: Scaffold Memory for Long-Horizon 3D Geometry

Authors: Soumick Sarker, Mengmi Zhang
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19674v1

Geometry foundation models estimate camera poses and 3D scene structure from video sequences, but maintaining consistent reconstructions over long horizons remains challenging. Feed-forward models process long sequences as independent, fixed-length frame windows and align their predictions, causing pose and reconstruction errors to accumulate across windows. Streaming models retain information from past frames to inform future predictions, but cannot revise earlier estimates when later observations reveal errors. To address this, we introduce Scaffold MemOry for Revision (SMORE), a framework equipping frozen geometry foundation models with a revisable external memory for long-horizon inference. Inspired by hippocampal associative memory in biological brains, which links spatial context with sensory experience, SMORE associates where an observer is with what it observes, enabling past observations to be stored, retrieved, verified against new evidence, and revised when inconsistencies are detected. We evaluate SMORE with eight competitive geometry models on five datasets spanning camera pose estimation, multi-view depth estimation, 3D point-cloud reconstruction, simultaneous localization and mapping (SLAM), and novel view synthesis. Across all evaluated models, SMORE consistently improves long-sequence pose estimation, reduces multi-view depth error by up to 58%, reduces Chamfer distance for 3D point-cloud reconstruction by up to 43%, and achieves average trajectory error as low as 2.9 cm in long-horizon SLAM. By retaining corrected poses and observations, SMORE further improves the fidelity and spatial consistency of large-scale 3D scene reconstruction. These results establish revisable external memory as a flexible interface for extending pretrained geometry models beyond fixed-context prediction toward iterative, error-correcting inference over long visual sequences.

---

### Learning When to Trust Medical Knowledge: Regression-Safe Routing for Clinical Reasoning

Authors: Hui Zhang, Yujia Zeng, Yuhao Xin
Published: 2026-10-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.19663v1

Clinical reasoning LLMs need external medical knowledge, but retrieved evidence and reasoning chains can be outdated, poisoned, or hallucinated. Harmful knowledge can cause regression: changing a correct answer into an incorrect one. We construct a medical knowledge-poisoning benchmark covering retrieved evidence and reasoning chains, with seven attack families filtered by a three-judge "subtle-but-wrong" vote and paired clean exams that separate poisoning effects from question difficulty. No fixed knowledge-use strategy is robust across these settings: strong RAG methods trade robustness to poisoned evidence against accuracy when evidence is useful. We formulate knowledge use as per-instance routing and introduce Med-TrustRouter, a lightweight controller that keeps the solver frozen. From leakage-free runtime signals, it estimates each action's expected correctness gain and uncertainty, preserving the base answer unless another action is confidently better. Across seven medical reasoning settings, Med-TrustRouter achieves 78.9 macro and 69.2 worst-case accuracy, matching the strongest competitor on macro accuracy while substantially improving the worst-case floor. Its gains transfer across solvers and evidence judges, unlike a hand-coded router using the same signals. Similar performance across linear, tree, and neural action-value heads suggests that the benefit comes from routing rather than estimator complexity. These results support learning when to trust external knowledge for robust clinical reasoning.

---

### Measuring and Reducing Unusable Replies of LLM Agents

Authors: Yilin Jiang, Fei Tan, Xuanyu Yin, Guanchu Wang, Xia Hu, Yutao Yue
Published: 2026-10-14
Categories: cs.CL, cs.AI, cs.HC
arXiv: https://arxiv.org/abs/2610.19659v1

People delegate multi-step work to LLM agents and read only the reply that the agent hands back. Current evaluation scores task completion and the faithfulness of reports, and it leaves open whether the requester can understand and use the reply. We call this property reply usability and measure it with a reader model that takes the requester's position and tests whether the reply can be interpreted from what the requester has seen, is in the requester's language and answers the request. Its agreement with the majority of five annotators lies within the range of the individual annotators. Across eight models from seven families, 54% of 480 replies fail, and the best model still fails on 30%. Failed replies mostly point to material the requester has not seen or use labels that the agent coined during its work. Stating explicit requirements for the reply reduces the failure rate from 0.52 to 0.43 together with errors and omissions, and a secondary model that rewrites the reply reduces it to 0.36, which human annotators confirm in a blind comparison. Reply usability can thus be scored alongside task success and improved with one appended message or one call to a secondary model.

---

### HACHIMI-II: From Role-Play to Learner Behavior Modeling for Faithful Student Simulation

Authors: Yilin Jiang, Fei Tan, Xuanyu Yin, Jing Leng, Aimin Zhou
Published: 2026-10-14
Categories: cs.CL, cs.AI, cs.CY, cs.LG
arXiv: https://arxiv.org/abs/2610.19654v1

Simulated students let researchers test tutoring systems and teaching strategies before real learners are available, but only if the simulation reproduces where a particular learner succeeds and fails. The common approach asks a large language model to role-play a student, but a role does not model what this learner knows or will do next, so the model's own competence fills the gap. Simulated students perform at a level set by the model, differ little from one another, and seldom make the real learners' errors. We model the learner's behavior and recast student simulation as third-person prediction of the learner's next action. HACHIMI-II builds a student model from the learner's history, infers the learner's state for the current item or tutor turn, and compiles it into a behavior harness that constrains the next action. On three item-response datasets, the Wasserstein-1 distance between simulated and real student ability is 2.4--8.4$\times$ smaller than that of the strongest baseline. On two dialogue datasets, HACHIMI-II reproduces real student errors with the same knowledge type, locus, and mechanism 1.8--5.1$\times$ more often than the strongest baseline, and ablations trace the gains mainly to the explicit learner state. Predicting the learner makes capable models usable as simulated students.

---

### On Global Joint Lower Bounds in Zeroth-Order Nonconvex Optimization

Authors: Yuxuan Xia, Xiao-Tong Yuan
Published: 2026-10-14
Categories: cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2610.19652v1

Recent work has sought to transfer first-order nonconvex optimization guarantees to zeroth-order methods through gradient estimators, typically incurring a factor linear in $d$. \citet{kornowski2024} obtained an $\mathcal{O}(dδ^{-1}ε^{-3})$ oracle complexity for nonsmooth nonconvex optimization and asked whether its dependence on all three parameters is jointly optimal. Although the dimension dependence and the first-order dependence on $(δ,ε)$ are separately known to be unavoidable, a matching zeroth-order lower bound has remained open. We answer this question in the unrestricted full-space setting. For smooth stochastic nonconvex optimization, we establish a scalar-oracle lower bound of $Ω\!\left(dL_0^2\min\{ΔH,L_0^2\}/ε^4\right)$ assuming a smooth, globally lower-bounded population objective and a second-moment bound on sample Lipschitz constants. A smooth reduction then yields a lower bound for Goldstein stationarity, proving the joint minimax optimality of the dependence on $(d,δ,ε)$ under explicit high-dimensional and small-accuracy conditions. Crucially, our results allow algorithms to query arbitrary points in $\mathbb{R}^d$, adaptively select sample-function labels, reuse the same sampled function at any number of locations, and output points without spatial restrictions. Thus, unlike lower bounds based on bounded query regions or fresh evaluation noise, our guarantees apply to a fully adaptive, reusable-sample scalar oracle. We also establish a separate noiseless exact-value lower bound against randomized algorithms, and hence also against deterministic methods. Our proof combines a globally defined sequential nonconvex construction with an exact Gaussian scalar channel, multistage random-field estimates, and an antithetic finite-sample realization that preserves information hardness under adaptive reuse.

---

### SPECTRA-VLA: Base-Fine Hyperspectral Token Transfer at Frozen Vision-Language-Action Interfaces

Authors: Chih-Chung Hsu
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19651v1

Hyperspectral imaging distinguishes RGB-similar objects, but a pretrained vision-language-action model needs a state its frozen decoder can use. SPECTRA-VLA provides a native-width spectral upgrade without retraining the visual encoder or action decoder. A learned continuous residual enriches each target token through its own normalized false-color reference; projection, norm control and validation-calibrated injection regulate the change. Across 86 held-out acquisitions, category balanced accuracy rises from 0.812 to 0.883. With identical residuals and scale, own-token addition improves reuse of an external category-derived probe on frozen decoder logits: 0.701 versus 0.590 with pooled replacement, at similar image-disjoint recognition. Five grouped tests covering the same 86 acquisitions recover within-category spectral shape with $R^2=0.802/0.337$ from continuous and posterior-code states. Category-matched spectral swaps further show that own-token composition approximately doubles the target-specific recovery increment after frozen decoding. Spectral enrichment and receiving-token preservation thus work together at the pretrained interface.

---

### AWARE-RG: Abstraction Aware Explanation of Resiliency Graphs for Cyber Physical Systems

Authors: Shadaab Kawnain Bashir, Rakesh Podder, Philip D. Hopkins, Sarath Sreedharan, Indrakshi Ray, Indrajit Ray
Published: 2026-10-14
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2610.19650v1

Understanding the security posture of cyber physical systems (CPS) requires reasoning across both information technology (IT) and operational technology (OT). However, a security analyst often holds an incomplete, role-specific view of the underlying system which has most likely resulted from their knowledge limited by a need-to-know framework. Such an incomplete view can potentially lead them to incorrectly assess if a cyber attack produces an untoward physical consequence resulting in making uninformed decisions about system resilience. In this work, we present AWARE-RG (Abstraction Aware Explanation of Resiliency Graphs), a semi-automated, human-in-the-loop framework that provides an analyst with a role-bounded, need-to-know explanations of resiliency paths in a CPS. The AWARE-RG abstraction structure is derived using Formal Concept Analysis over MITRE ATT&CK for ICS tactics, ordered by the ICS Cyber Kill Chain. For valid attack-fault queries, plan repair and backward means-end analysis identify the missing dependencies, explain why the physical hazard is reachable, and suggest defensive strategies. For invalid attack-fault queries, AWARE-RG localizes the first abstraction level at which the path fails, and reports confirmed blocking conditions as hardening pointers.

---

### Mapping Arctic Polynyas from Sea Ice Concentration Data Using Hierarchical Partitioning and Differentiated Region Growing: Sensitivity to Threshold and Product Choices

Authors: Tianyu Zhang, Haiyi Ren, Zhilun Zhang, Yang Hong, Chunlei Mi, Xiao Cheng, Fengming Hui
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19647v1

Accurately delineating Arctic coastal polynyas from sea ice concentration data remains challenging because of complex coastal geometry and spatially heterogeneous ice conditions. We develop a daily sea ice concentration (SIC) mapping framework based on hierarchical partitioning, connected-component analysis, and differentiated region growing for polynya and open-water regions. Applied to September-May ice seasons from 2002/03 to 2024/25, the framework achieves a mean overall accuracy of 96.74% and a mean kappa coefficient of 0.93 in the 63-scene visual assessment. Sensitivity experiments reveal a consistent pattern: the absolute magnitude of SIC-defined polynya extent is strongly configuration-dependent, whereas major temporal features are comparatively more stable. Relative to a 75% upper SIC threshold, seasonal extent integrals differ by 13.48%, 32.47%, and 65.04% at thresholds of 80%, 85%, and 90%, respectively, while the broad interannual structure remains recognizable. The response is also strongly regional and seasonal, with the largest threshold sensitivity in the Central Arctic Ocean, very weak sensitivity in the Gulf of St. Lawrence, and greater pan-Arctic curve separation from January onward. At an 80% threshold, product-comparison MAPE is 68.19% between the 6.25 and 25 km products and 11.47% between the 12.5 and 25 km products under their respective comparison periods and reference definitions. Regional product disagreement is likewise heterogeneous. Together, these results show that reproducible long-term polynya monitoring requires a fixed and explicitly documented combination of SIC product, threshold, and spatial classification rules, while broad temporal patterns can remain informative even when absolute extent differs substantially.

---

### Selective Reasoning for Typed Decision Models: When to Start, Read, and Stop

Authors: Bin Zheng, Zheng Chen
Published: 2026-10-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.19644v1

Typed decision models need an interface that can use additional reasoning while retaining candidate outputs. We train a shared candidate-answer reader on a frozen Qwen3.5-4B backbone to unify direct decisions with on-demand, interruptible reasoning. The reader returns distributions from empty and partial reasoning prefixes. A benefit gate starts reasoning, a natural-state trigger requests explicit reads, and an entropy rule accepts acquired answers and cancels generation. Measurements distinguish answer-query work, generated output, and pipeline time. Across 768 questions from ProofWriter, LogiQA 2.0, and ARC-Challenge, accuracy rises from 70.96% for direct reading to 84.38%. Relative to 512-token terminal reading, accuracy improves by 7.29 percentage points with 2.25% more mean output work. Relative to 1,024-token terminal reading, pipeline time falls by 46.30% with an accuracy difference of -1.17 points. Exact-prefix controls isolate the saving from interruption, while read-scheduling controls distinguish reduced observation work from shorter response time.

---

### Model What Matters: Decision-Guided Selective World Modeling for Language Agents

Authors: Maolin Ran, Minda Chen, Di Huang, Wei Xiong, Fei Huang, Weiwen Liu
Published: 2026-10-14
Categories: cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2610.19642v1

Reinforcement learning provides a direct mechanism for optimizing task objectives in large language model-driven agents, but task rewards are typically sparse trajectory-level signals that only evaluate task success, without directly supervising action-conditioned environment transitions, leaving the model with limited explicit understanding of action consequences and environment dynamics. Recent work introduces auxiliary supervision through policy and world modeling co-training, yet typically treats complete environmental observations uniformly as world modeling targets, overlooking the varying decision relevance of observation content. We propose DGS-WM (Decision-Guided Selective World Modeling), a decision-guided policy-world modeling co-training approach for language agents. DGS-WM employs a Decision-Guided Observation Selection component that filters observation units with decision value as world modeling targets based on their support for subsequent policy responses, and a Confidence-Adaptive Weighting component that adaptively adjusts supervision strength according to the model's prediction confidence on selected targets. Together, these components form a decision-aware auxiliary world modeling loss, jointly optimized with the RL loss on the same policy model. Across multiple benchmarks covering single-hop/multi-hop question answering and interactive web tasks, DGS-WM achieves consistent improvements under different model scales and RL algorithms, demonstrating more stable and effective performance compared to full-observation world modeling methods.

---

### Synthetic reaction data from transformation rules for graph-to-sequence pretraining in biodegradation pathway prediction

Authors: Irsyaad Rijwan, Liam Brydon, Aidan Jared, Jörg Simon Wicker
Published: 2026-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.19640v1

The scarcity of observed biodegradation reactions constrains the training of product and pathway predictors. We present enviGraph, a framework that generates and curates synthetic reaction data, uses these data to pretrain a graph-to-sequence predictor, and fine-tunes the predictor on observed environmental reactions for product and recursive pathway prediction. Applying inherited environmental transformation rules to structures from PubChem, NORMAN SusDat and EPA DSSTox, followed by curation, produces 776,851 canonical parent-product pairs for pretraining. We compare this dataset with three United States Patent and Trademark Office (USPTO) reaction datasets: USPTO-50K, USPTO-MIT and USPTO-Stereo. We evaluate enviGraph and the sequence-based enviFormer using ten folds split by pathway on the EAWAG Biocatalysis/Biodegradation Database (BBD), regulatory pesticide biodegradation data (Soil) and activated-sludge biodegradation data (Sludge). For both predictors, synthetic pretraining gives the highest BBD mean precision-recall area under the curve (PR-AUC) among the evaluated pretraining datasets for single-generation and multi-generation prediction. enviGraph achieves BBD single-generation PR-AUC of $0.202 \pm 0.050$ with synthetic pretraining, compared with $0.169 \pm 0.065$ for the retrospectively selected strongest USPTO condition, USPTO-Stereo (mean $\pm$ sample standard deviation across folds). Its exact BBD top-1 and top-5 accuracies with synthetic pretraining are 0.355 and 0.594. The strongest USPTO conditions give higher PR-AUC means than synthetic pretraining in most Soil and Sludge comparisons. enviGraph has higher Soil multi-generation PR-AUC means than enviFormer under each of the four main pretraining datasets, and with synthetic pretraining it also leads enviFormer on BBD single-generation PR-AUC and exact top-1 and top-5 accuracy.

---

### J-tau: A Fully Localized Japanese Benchmark for Tool-Agent-User Interaction

Authors: Chihiro Yano, Jun Hirako, Hiroaki Yamagiwa, Ryota Hirobuchi
Published: 2026-10-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.19638v1

We present J-tau, the first publicly available Japanese benchmark for evaluating tool-using LLM agents in interactive customer-support scenarios. While tau-bench is widely used to evaluate LLM agents, it is English-centric and cannot evaluate agents in other languages. Existing multilingual extensions offer no public Japanese version and rely primarily on machine translation, leaving locale-specific entities such as addresses and currencies inconsistent with the target region. To enable reliable evaluation in Japanese, we carefully localize tau-bench through machine translation, entity adaptation, and human review. During this process, we also uncover and fix two flaws in the original tau-bench: invalid tasks and a score aggregation error that inflates certain models' results. Experiments show that J-tau yields fewer evaluation errors than a naively machine-translated counterpart and that our fixes resolve both flaws.

---

### Sparse Autoencoders Are a Shared Coordinate System for Symbolic Matching

Authors: Xiaocong Yang, Xuanyu Meng, Yuji Zhang, Jiawei Han, ChengXiang Zhai
Published: 2026-10-14
Categories: cs.LG, cs.AI, cs.CL, cs.SC
arXiv: https://arxiv.org/abs/2610.19636v1

Large language models (LLMs) hold broad knowledge in one set of weights, but lack interpretability and trustworthy reasoning, especially in high-stakes use cases. A neuro-symbolic model, with its interpretable architecture and System 2 reasoning capacity, is a promising alternative if built at the scale of LLMs; however, its symbolic backbone is usually constructed with human effort, so existing neuro-symbolic models are bottlenecked by the scope of the symbol component and usually domain-specific. A data-driven method could instead build a larger symbolic system by aligning and merging existing ones. We target its first, foundational step: matching units across these heterogeneous systems. Our method, SAE-IDF, encodes a symbol in its native context into a sparse code with a sparse autoencoder (SAE), re-weights each dimension by its rarity and matches across systems. On 12 tasks across 6 kinds of symbolic system, SAE-IDF outperforms the same LLM's dense baseline on all tasks with Gemma-2-9B and all but one with Llama-3.1-8B, by up to 45 points, consistently across SAE layers and configurations. SAE-IDF is also interpretable: every dimension of the sparse code has a human-readable meaning, so each match decomposes into named concepts (features). Our work paves the way to a universal symbolic backbone.

---

### Eye, Robot: Learning Expressive Gaze Dynamics via a VLM-Guided Transformer-VAE

Authors: Yuhang Hu, Junlin Lyu, Yuan Fang, Xiaoying Shen, Lechen Zhang, Jiong Lin, Judah Goldfeder, Steven Tian, Hod Lipson
Published: 2026-10-14
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2610.19635v1

Gaze coordinates attention in human interaction through coupled eye, head, and eyelid motion. A humanoid robot must choose where to look while generating a suitable movement toward that target; fixed tracking rules alone do not model the temporal structure of human gaze. We separate these tasks with an asynchronous vision-language model that selects a scene referent, a reactive perception module that maintains fixation, and a Realistic Gaze Transformer (RGT) that generates target-conditioned motion. RGT is a Transformer-VAE trained on markerless human recordings retargeted to the robot's 11-DoF joint space: self-supervised pretraining encodes recent full-head history, and autoregressive fine-tuning exposes the model to its own predictions during rollout. At inference, a four-dimensional binocular-eye trajectory conditions coordinated neck and eye motion; an auxiliary blink predictor triggers eyelid templates during playback. On a two-second benchmark in which all controllers receive the same ground-truth target, RGT-Base obtains the lowest reported overall seven-joint MSE, and physical demonstrations show the integrated targeting and control pipeline. The evidence establishes robot-space kinematic fidelity and real-time RGT inference; perceived naturalness and open-vocabulary target-selection accuracy remain questions for controlled evaluation.

---

### Risk-Controlled DFT Selection for Machine-Learned Interatomic Potentials

Authors: Kun Ryu, Yeo Jin Jung, Shyue Ping Ong, Claire Donnat, Ying Shirley Meng
Published: 2026-10-14
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2610.19630v1

Automated materials discovery requires evaluating large pools of candidate configurations, with density functional theory (DFT) as a major throughput bottleneck. Pretrained machine-learned interatomic potentials (MLIPs) approximate DFT cheaply, but their errors vary by orders of magnitude across configurations, making it difficult to decide when to trust the MLIP predictions and when to fall back to DFT. We frame this decision as a risk-controlled selection problem and solve it within a conformal risk control (CRC) framework. Each configuration is mapped from atomic numbers and Cartesian coordinates to a fixed geometric descriptor, which a lightweight predictor uses to estimate an MLIP-DFT disagreement score in energy, average force, and maximum force. CRC then calibrates this score into a selection threshold for DFT labeling such that the MLIP error on the unselected configurations, which keep their MLIP labels, stays below a user-chosen fraction $α$ of the expected error. Under an identical guarantee across eight dataset-MLIP pairs, our predictor routes 7% and 14% fewer configurations to DFT at $α=0.3$ than committee baselines. Because committee disagreement overlooks errors the MLIPs share, our gains are largest when the DFT labels and MLIP training data differ in DFT functional and basis set, and narrow when the two agree. Without model access, the structure-only predictor also ranks errors as well as the target MLIP's latent representation and outperforms widely used descriptors such as Smooth Overlap of Atomic Positions (SOAP) and Atom-Centered Symmetry Functions (ACSF).

---

### Self-Conditioning Enables Deep Sequential Reasoning in Flow Matching

Authors: Zeyang Zhang, Linxiang Gao, Meiqi Gu, Xingyan Chen, Minrui Luo, Kaifeng Lyu, Tianxing He
Published: 2026-10-14
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2610.19628v1

Recent work has explored visual reasoning through image generation, where models solve visual-reasoning problems by directly generating solution images. Intuitively, iterative samplers such as flow matching are well suited to this task, since each sampling step may resemble one reasoning step. However, in this work, from a circuit complexity perspective, we show that this intuition is misleading. We prove that as long as the model outputs satisfy certain mild regularity conditions, the trajectory of the vanilla flow-matching sampler can be simulated by a constant number of sequential rounds of parallel model evaluations. Consequently, additional sampling steps cannot increase the expressive power of a constant-depth backbone model under the standard complexity-theoretic model of transformers. We then reveal that self-conditioning, which conditions the model on the previous step's estimate of the final image, plays a previously unrecognized role in removing this bottleneck. Controlled experiments on diverse visual-reasoning tasks show that flow matching with self-conditioning achieves higher accuracy and greater reasoning depth than vanilla flow matching, and its performance improves with more sampling steps. Theoretically, we prove that with self-conditioning, a constant-depth backbone can achieve expressive power equivalent to that of space-bounded Turing machines. Our study suggests a design principle for future visual reasoning research: carry more information between sampling steps, so that additional steps translate into deeper reasoning.

---

### DAGFormer: Global Topology Prediction for Transformer Cross-Layer Connections

Authors: Xiaocong Yang, Yuren Hao, ChengXiang Zhai
Published: 2026-10-14
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.19625v1

We propose DAGFormer, a Transformer language model whose cross-layer connections are planned by a jointly trained topology predictor. Beyond the residual connections in a standard Transformer, DAGFormer adds direct connections from earlier layers to later ones, and the predictor controls the strength of every connection based on input tokens. Unlike existing cross-layer routing architectures, which set each layer's connections separately, DAGFormer plans the routing of the whole model at once. On matched OLMo-2 pairs from 75M to 1B parameters, DAGFormer lowers WikiText-2 perplexity by 17% to 25% over the vanilla Transformer, and a 1B DAGFormer trained on 5B tokens outperforms the vanilla model trained on 10B. Against four routing baselines with the same data and budget, it reaches the lowest perplexity at 75M, 300M and 1B. The gain carries over to 13 commonsense, science, reading and math evaluation tasks (2.0 to 2.5 points higher mean accuracy at 300M and 1B) and to instruction finetuning (63% to 70% AlpacaEval win rate against the vanilla Transformer, leading every routing baseline), and under structured pruning DAGFormer degrades the least in five of six settings.

---

### Adaptive Observation Control for Active Visual Recognition

Authors: Dikai Li, Weijia Wang, Yuan-Gen Wang
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19624v1

Visual recognition models usually process a fixed view under a fixed computation budget, although images often benefit from different scales and amounts of spatial evidence. Active vision can adapt by observing a coarse scene and then selecting informative fixations, but the public AdaptiveNN-DeiT-S pipeline still fixes both field of view and observation count. We extend this fixed-step recognizer with a controllable observation interface that adapts both the field of view and the number of fixations to the input. Specifically, the model predicts a per-fixation scale that sets its field of view, refines each selected crop with optional scale-gated foveation and a residual high-resolution detail-token branch, and uses label-grounded stop heads for input-dependent early termination. Experiments across fine-grained benchmarks, full ImageNet-1K, and ImageNet-A/R-derived classification splits show that the interface remains competitive with fixed-step references while exposing a controllable observation budget. The same checkpoint enables threshold-controlled accuracy-compute trade-offs without per-threshold retraining. On sequential ImageNet-1K, the learned stop rule outperforms fixed-step truncation of the same checkpoint, sustaining 81.9% top-1 using 2.45 of the four focus steps and exceeding three-step truncation at lower cost. On each benchmark, a variant combining scale control and detail-token fusion achieves the highest accuracy among the tested component-added non-exit variants. Our code is available at https://github.com/GZHU-DVL/LI-Dikai.

---

### MagicGarment: Benchmarking Single-Image Standalone 3D Garment Reconstruction

Authors: Jia Chen, Ruyang Zhao, Jun Bai, Chao Huang, Saishang Zhong, Xinrong Hu
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19623v1

Single-image 3D reconstruction has advanced rapidly, yet standalone garments remain challenging due to thin-shell geometry, open boundaries, and layered structures. Existing datasets and evaluation protocols provide limited support for systematic evaluation in this setting. We introduce MagicGarment, a benchmark for single-image standalone 3D garment reconstruction with 1,518 professionally constructed assets across 17 composite garment types, aligned multimodal annotations, and a unified evaluation protocol. We further establish a garment-aware reference baseline by adapting a pretrained image-to-3D backbone at two levels: Grounded Semantic Refinement (GSR) provides garment-focused visual conditioning, while Flow-Time-Route LoRA (FTR-LoRA) adapts the frozen 3D generative backbone using flow-timestep, token-routing, and expert-load information. Experiments demonstrate MagicGarment's benchmark utility and the effectiveness of garment-aware adaptation. Project page: https://1019538157.github.io/MagicGarment/.

---

### Are AI Decision Models Biased? A Case Study of Jev

Authors: Ilaria Imperatrice, Taisia Kameneva, Adam Eric Greenberg, Alessandro Romano
Published: 2026-10-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.19619v1

Do AI models designed for fast, structured decisions exhibit the cognitive and social biases observed in human judgment? We evaluate Jev, a model described as a System One decision-maker, using 41,500 responses across cognitive reflection tasks, five classical cognitive-bias paradigms, and a pricing experiment testing ethnic and gender bias. At the level of selected answers, the results are largely encouraging. We confirm previous findings that Jev performs extremely well on cognitive reflection problems. We also observe that it generally avoids the characteristic choice patterns associated with several well-documented human biases, and exhibits only small differences among demographic groups in the tested pricing scenario. However, this apparent robustness conceals important vulnerabilities. Its judgments are sensitive to irrelevant numerical information and the ordering of response options. Moreover, several cognitive-bias manipulations substantially affect returned probabilities even when selected answers remain unchanged. Most strikingly, Jev answers a well-known probability problem correctly in every response but violates the same underlying probability rule in every response to an adapted version. We investigate this discrepancy using 15,000 additional responses across 30 conjunction tasks. Jev violates the rule in 54\% of responses, with accuracy varying dramatically across formulations and answer orders. These findings suggest that evaluating decision models solely on familiar benchmarks and selected answers may substantially overstate their reliability in practical applications. They also identify vulnerabilities for developers to address and practical considerations for users integrating such models into automated decision-making systems.

---

### Instruction-Conditioned Motion Control for Robotic Arms: Language-to-Code Decomposition and Reinforcement Learning

Authors: Fan Huang, Dezhong Tong, Changyu Chen, Haewoon Kwak, Jisun An
Published: 2026-10-14
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2610.19617v1

Instruction-Conditioned Motion Control (ICMC) separates parameter specification, instruction generation, and code synthesis for language-driven robot control, making execution outcomes traceable to instructions and programs. We compare one-shot generation with ICMC under o1 and o4-mini on cube pick-and-place and letter-reshaping tasks for a UR5e arm in MuJoCo. We then train an open-weight instruction policy by GRPO through a frozen code generator on cube placement onto two- and five-cube letter layouts, averaging the rewards of five programs per instruction. The logged positive-reward rate rose from 31.6% to 35.3% (253/800 to 282/800) between the first and last five training steps. A log audit identifies positive binary labels associated with geometric-failure messages: at least 19 in that run, 18 in the dense-reward run, and 5 in the run with one program per instruction. In one case, all five rewards were 1.0 while the log reports a 0.9515 m miss against a 0.05 m tolerance; the verdict came from the program's own output stream. We distinguish logged reward rates from independently verified task success and report descriptive results for each configuration's single training trajectory. We introduce an independent evaluation protocol that checks task completion against fixed specifications, separating robot control from success assessment to support reproducible evaluation.

---

### Divergent Paths: Learning an Outcome-Predictive Representation of LLM Agent Trajectories

Authors: Alexander Gill, Guram Gogia, Francisco Guzmán
Published: 2026-10-14
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.19614v1

In this work, we reframe agentic trajectories as paths through a shared embedding space of actions and environmental observations. We show that successful and failed rollouts of the same task separate in this space, with simple per-task discriminators predicting held-out rollout outcomes at up to 0.95 AUROC across three agent benchmarks and three policy models. To exploit this signal across tasks, we introduce Precog-RM, a transformer-based model that meta-learns a representation in which outcome prediction can be performed in a few-shot setting. We show that Precog-RM generalizes to unseen tasks and models, and gives confident predictions of trajectory success long before termination. Deployed as an early stopping controller for Group Relative Policy Optimization (GRPO) on Tau2-Bench, Precog-RM cuts training FLOPs by up to 47.3\% while maintaining performance comparable to baseline GRPO.

---

### Are We Really Measuring Forgetting? Pitfalls in Evaluating Machine Unlearning for Multimodal LLMs

Authors: Oscar Chew, Po-Yi Lu, Yi-Jie Cheng, Kunal Jain, Hsuan-Tien Lin, Zhengzhong Tu, Kuan-Hao Huang
Published: 2026-10-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.19613v1

As machine unlearning for multimodal large language models (MLLMs) rapidly advances, rigorous evaluation of its effectiveness becomes increasingly important. However, existing benchmarks largely inherit evaluation protocols from text-only unlearning and overlook the unique challenges of unlearning in multimodal settings. In this work, we identify three major pitfalls that may lead to misleading conclusions: the lack of heldout evaluation to distinguish genuine concept-wise forgetting from instance-wise overfitting, the reliance on synthetic, fine-tuned knowledge rather than real-world knowledge acquired during pre-training, and unfair method comparisons caused by applying the same training settings to different unlearning methods. To address these challenges, we introduce GEMMU (Generalizable Evaluation for Multimodal Machine Unlearning), a benchmark built on real-world knowledge and a principled evaluation protocol. Re-evaluating representative unlearning methods under GEMMU reveals substantial generalization gaps and challenges conclusions drawn from previous benchmarks. Our findings highlight the importance of rigorous evaluation for advancing multimodal machine unlearning and provide guidelines for future benchmark design.

---

### What Breaks in Multi-Image Schema Application? Factorial Oracle Interventions Reveal an Instance-Information Bottleneck

Authors: Yiliu Xu, Xinjin Li, Yu Ma
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19611v1

Modern vision-language models can accept multi-image contexts, but aggregate accuracy rarely reveals where a multi-image reasoning pipeline fails. We study a controlled schema-application task in which a model must locate two target characters in a 16-image sequence, identify their visual attributes, consult an item-specific visual codebook, and compose two ordered nonce code names. We introduce a within-item 2-by-2 factorial Oracle design: Oracle-A supplies a verified textual schema, Oracle-B supplies a verified ordered target-appearance representation, and both interventions retain the original images and final question. Across 72 new items and 1,188 analyzed model responses, Oracle-B increased strict exact accuracy by 54.6 percentage points (95% confidence interval: 46.3 to 63.0) and outperformed Oracle-A by 60.6 points (Holm-adjusted p-value no greater than .0002). An auxiliary retrieval diagnostic reached only 49.1% ordered-appearance accuracy even after removing code-name generation. Providing both Oracle sources yielded 100% accuracy across all models. These results localize the dominant recoverable deficit to upstream ordered instance recovery rather than primarily to downstream schema use and composition.

---

### CORE: Contrast-and-Reflect for Self-Supervised Prompt Optimization in Anomaly Synthesis

Authors: Haoxiang Rao, Fang Zhao, Zhao Wang, Weijia Li, Chenyang Si, Caifeng Shan
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19610v1

Industrial anomaly detection (AD) is bottlenecked by the scarcity of anomalous samples, making high-fidelity anomaly synthesis crucial for improving downstream performance. However, existing methods either rely on scarce real anomaly references or auxiliary data, while lacking complex reasoning and iterative optimization capabilities for generating realistic defects in open-set scenarios. To address these challenges, we propose CORE, a cost-efficient and training-free prompt optimization framework based on Large Multimodal Models (LMMs), which unifies zero-shot and one-shot anomaly synthesis. CORE first introduces an Anomaly Context Grounding module to transform sparse user context into defect-aware generation context, which is subsequently instantiated as structured anomaly prompts that disentangle normal-region preservation from anomaly-specific variation, enabling controllable and faithful defect generation. Furthermore, CORE autonomously refines prompts through a self-supervised contrast-and-reflect mechanism. This design empowers general-purpose LMMs to synthesize realistic anomalies under both zero-shot and one-shot settings without task-specific training. Extensive experiments on MVTec-AD and VisA demonstrate that CORE synthesizes realistic and diverse anomalies, yielding superior improvements in downstream AD performance. Furthermore, its model-agnostic and plug-and-play paradigm enables effortless integration with diverse LMMs, facilitating adaptation to unseen open-set scenarios.

---

### Twenty Days Later: A Pre-Registered Test of History-Based Forecasts of Brand Mentions in Gemini Answers

Authors: Dmitrij Żatuchin
Published: 2026-10-14
Categories: cs.IR, cs.CL
arXiv: https://arxiv.org/abs/2610.19608v1

Counts of which brands a large language model names are useful only if they hold long enough to act on. We test this with a pre-registered replication. The same 162 requests (54 questions in English, Estonian and Finnish about three software categories, four of twelve candidate brands per question, three repeats) were sent to gemini-3.1-pro-preview in three waves within about ten hours on 22 and 23 September 2026, and again on 13 October 2026, 20.3 days after the third wave, with identical request bodies. The plan and its decision rule were frozen, hashed and registered on OSF before the first call. The outcome is a lexical mention label per answer and brand, 648 per wave. The primary forecast, each brand-and-question cell's mention rate over the nine earlier answers, scored Brier 0.0281 and log loss 0.126, against Brier 0.0259 for the six-answer forecast of the 8-hour wave and a registered ceiling of 0.0559. It beat brand history (0.104), the overall rate (0.234) and a constant 0.5 (0.250) on both scores and in all nine question families (exact sign-flip p = 0.0039), so durability holds under the registered rule. A forecast frozen from the first wave alone scored 0.0599, 0.0587 and 0.0604 at horizons of 1.5 hours, 9.6 hours and 20.7 days. Of the 202 cells unanimous in the 8-hour wave, 194 were unchanged at D1 and none reversed. The latest wave alone was nearly as accurate as the pooled history (Brier 0.0292; pooled lower in 5 of 9 families, p = 0.28). Every answer reported the requested model version, and none issued a search query although search was enabled. A blind human check of 96 labels agreed with the rule on 94; both disagreements were adjudicated as rater omissions, so after adjudication the rule matched on 96 of 96. The test covers one model, one interval and twelve known brands, and estimates no decay rate.

---

### A Graph-Based Analysis of Comparability, Scale, and Attribution in Private Transformer Inference

Authors: Souvik Das, Ariful Azad, Nitesh Saxena
Published: 2026-10-14
Categories: cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2610.19607v1

Rising privacy concerns around large language models have accelerated research in private Transformer inference (PTI). Synthesizing the empirical claims in this line of research, however, is hindered by multi-dimensional cryptographic progress and inconsistent evaluation contracts. In this work, we systematize the field by constructing an empirical graph of 62 PTI systems, mapping system attributes to node features and evaluation compatibility to edges. A three-tiered analysis of this graph illuminates the state of private inference. Node-level attribution shows that cryptographic progress drives performance gains across most systems, yet always co-occurs with other factors (model fidelity, hardware, cost accounting). Edge-level auditing shows that only 1 of 1891 candidate system pairs is directly comparable under shared assumptions. Finally, graph-level analysis shows that the scaling frontier of private inference has plateaued at 13B parameters and projects the cost of deploying current PTI techniques on production-scale Transformers. Overall, this graph-based framework equips the community to evaluate PTI claims transparently and channel technical speedups into deployment-scale private inference.

---

### PN-Mix: Pair-Norm Guided Mixed-Precision Key Cache Quantization for Efficient Long-Output LLM Inference

Authors: Gil Jae Song, Sungjoo Yoo
Published: 2026-10-14
Categories: cs.LG, cs.CL
arXiv: https://arxiv.org/abs/2610.19605v1

Long-output inference in large language models suffers from severe KV cache memory and bandwidth bottlenecks. While KV cache quantization alleviates this issue, low-bit quantization (e.g., 2-bit) severely degrades model accuracy. Recent mixed-precision key-cache quantization methods mitigate this degradation by allocating higher precision to important key channels, but estimate channel saliency using only key statistics. Through attention logit perturbation analysis at the RoPE channel-pair level, we show that quantization errors in channel pairs with large query--key pair-norm products contribute significantly to logit variation. Based on this insight, we propose PN-Mix, a mixed-precision key-cache quantization method for RoPE-based decoder-only LLMs, which allocates higher precision to salient channel pairs identified once during the prefill stage and reused throughout decoding. Across standard reasoning, code-generation, and AIME benchmarks, PN-Mix with 2.26-bit average precision achieves comparable or better average accuracy than prior mixed-precision methods using 2.32--2.39 bits. Furthermore, in our Pareto sweep, PN-Mix degrades gracefully down to 1.5 bits, whereas the compared baselines exhibit sharper performance drops below 2.25 bits.

---

### Beyond Static Positional Encoding: Bayesian Time-aware Attention for Non-Stationary Long-term Time Series Forecasting

Authors: Xinjie Lan, Yunbo Mei, Leheng Zhou, Hao Qi
Published: 2026-10-14
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.19603v1

Transformers have demonstrated strong performance in long-term time series forecasting (LTSF), yet modeling non-stationary temporal dependencies remains challenging. In this paper, we interpret self-attention as a Bayesian posterior over attended positions, combining a content-based prior induced by scaled dot-product similarity with a temporal-evidence likelihood induced by positional encoding. This formulation provides a probabilistic interpretation of how positional encoding shapes the temporal dependencies captured by self-attention. Building on this perspective, we derive a population-level evidence lower bound (ELBO) that characterizes the role of positional encoding in forecasting weakly stationary time series. Our analysis further highlights a limitation of static positional encoding: when temporal dependence structures vary across input windows, a fixed temporal bias must accommodate heterogeneous dynamics and can therefore be suboptimal for non-stationary forecasting. Motivated by this analysis, we propose Bayesian Time-aware Attention (BTA), which employs a mixture-of-experts (MoE) router to adaptively combine temporal kernels and generate input-conditioned, head-specific positional biases. Extensive experiments on standard LTSF benchmarks demonstrate that BTA consistently improves forecasting performance across multiple Transformer backbones and achieves competitive overall performance with modest computational overhead and stable training.

---

### PA1-OPD: One-Step Prefix-Augmented On-Policy Distillation for Language Model Pretraining

Authors: Zhenhe Wu, Jian Yang, Yaping Jin, Xinlai Kang, Jie Hu, Hanting Chen
Published: 2026-10-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.19601v1

Knowledge distillation (KD) improves language-model pretraining by matching teacher distributions at corpus prefixes, but provides no direct supervision at contexts induced by student predictions. We introduce PA1-OPD, which augments corpus-path distillation with teacher feedback after one student-sampled transition per selected prefix. This expands the supervised states while retaining next-token prediction and corpus-path KD. Data-anchored roots support parallel sampling, and differentiable tree attention shares prefix computation without changing inference. In two continued-pretraining settings, full coverage improves held-out cross-entropy over KD across C4, PG-19, and Pile at every evaluated checkpoint. In Stage2, selecting the 10% of roots with highest student surprisal adds 48.85% to KD's Macro cross-entropy gain over next-token training. Gains remain when branch supervision excludes successor states already on the corpus path, supporting the value of feedback beyond observed contexts. Additional depth or breadth improves aggregate results at unchanged per-term supervision weight, with benefits varying by corpus. These findings support local student-induced supervision as a complement to pretraining distillation.

---

### JevCoder: When Does Decision Delegation Pay Off in Code Repair?

Authors: Jiahao Zhang, Yifan Zhang, Yu Huang
Published: 2026-10-14
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2610.19596v1

A coding agent repeatedly makes decisions as it repairs code. These decisions shape repair trajectories and affect the token cost of successive model requests. We ask which repair decisions can be delegated to a low-cost probabilistic judge to reduce token use while maintaining repair quality. The judges answer questions with probabilities over predefined answers, without generating explanations. We evaluate judgment quality across repair stages and test how individual interventions change subsequent repair work. We compare these interventions with unguided repair, which uses no judge, and with the same workflow using the repair model itself as the judge. The benefit depends on the work an intervention changes, rather than the cost of the judgment alone. Based on these studies, we propose JevCoder, a workflow built around Jev, a low-cost probabilistic judge. It combines code-snippet recommendations with action-type guidance and uses judgments to determine when to submit. We evaluate three repair models, each on the same 400 tasks: 200 from SWE-bench Verified and 200 from SWE-Bench Pro Verified. Compared to unguided repair, JevCoder reduces mean repair-model token use by 21.6%-50.6%, with higher observed repair rates across all three models.

---

### InstructPro2: Natural Language and Biological Grounding Guided De Novo Protein Design

Authors: Charles Novak, Zhenqiao Song, Chuan Li, Jianwen Xie, Lei Li
Published: 2026-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.19592v1

The ability to generate proteins with desired functions is a central goal of computational biology. Although natural language can express diverse design objectives, existing protein generation methods often support only narrow tasks or make limited use of biological constraints. We develop InstructPro2 to translate natural-language design objectives into protein sequences while incorporating task-specific biological context. A text encoder captures the design instructions, while dedicated ligand and antigen encoders provide biological context to the sequence-generating decoder. For pretraining, we assemble InstructPro2Bench, comprising 205.7 million <function description, protein> pairs from UniProtKB. We subsequently adapt the model to ligand-binding protein and antibody design. On functional protein generation benchmarks, InstructPro2 outperforms Pinal, the strongest baseline, achieving a functional recovery F1 of 74.8 versus 62.2 and lowering global RMSD from 5.33A to 3.04A. For antibody design, the model obtains a DockQ score of 0.39, while ligand-binding protein design reaches a docking success rate of 72.6%. These findings highlight how integrating language-based instructions with biological conditioning can extend protein generation to distinct design settings.

---

### L3M: A Verifiable Harness and Runtime for Self-Extending and Self-Improving Agents

Authors: Carson Eisenach, Robert Joseph George, Vincent Quenneville-Belair, Kate Gwimm, Dean Foster
Published: 2026-10-14
Categories: cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.19591v1

The lasting gains in AI have come from methods that absorb more computation, and language models already scale computation within a call. Useful work, though, happens across calls: over hours or days, through tools and external systems, among many agents and the humans they work for, and increasingly through code the agents write themselves. Whether that work compounds is enabled by the harness around it, and the harness faces a constraint the model does not: of the three participants in such a computation, the human, the model and code, only the human's attention does not scale, so the computation can grow only if the human is not in every loop. L3M is a runtime built on that premise, written in Lean 4 so that one language implements the harness, states its invariants, and is the language agents extend it in. One computation spans humans, agents, models and code, and all of its state lives in Git: a rich state trajectory -- rather than a chat conversation -- is an object the runtime can resume, rewind, fork, compare and merge. Waiting, communication, delegation and undo are primitives, not patterns the model reasons through each time. When an agent works out a procedure, it can write it once as a sealed Lean program whose type is its permission request, and the harness offers it back as a tool -- machinery for the model to use. To take the human out of the loop, L3M uses reversibility to decide where that is safe: a tool's review tier follows from how far its effects can be undone, so agents act freely where the runtime can recover. Because the machinery can change, the properties that make this safe are stated as kernel-checked theorems about the code that runs. On Terminal-Bench, an untuned L3M reaches about two thirds of its (unsafe) shell-equipped pass rate using only safe (not needing human review) tools.

---

### JockeyBench: Evaluating LLM Agents on Horse-Racing Decision-Making

Authors: Zhongwei Xie, Xingsu Li, Jingqiao Yu, Zihao Wang, Hong ChiTat, Zhaowei Wang, Weizhi Du, Yuxuan Liu, Huihao Jing, Jiaxin Bai, Yangqiu Song
Published: 2026-10-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.19589v1

Repeated event forecasting requires extracting predictive signals from noisy evidence and evaluating their value across successive decisions. We introduce JockeyBench, a benchmark and evaluation protocol that studies this problem through historical Japanese horse racing. The protocol combines time-gated evidence, priced tools, paired frozen-program controls, and two workflows: executing an LLM-written predictor or acquiring evidence for a fresh per-race LLM forecast. Evaluation separates forecast quality, benchmark-specific net utility, and retrospective betting returns and risk under fixed decision rules. Across 28 model--scaffold configurations on 324 races, paired tests identify seven revision gains and two declines after multiple-testing correction. Yet every configuration has negative mean profit under capped-Kelly settlement. In a four-model comparison, the reusable-program workflow has higher mean betting profit and lower research fees, although both workflows lose on average. Further diagnostics reveal that mean probability bias reverses on selected bets and that profitable individual runs can depend on concentrated returns. These findings motivate evaluating predictive competence, adaptation, resource use, and decision value separately when assessing agents for repeated event forecasting.

---

### Projected Wasserstein-Fisher-Rao Flows for Hawkes Background-Intensity Estimation

Authors: Isaias Banales, Arturo Jaramillo
Published: 2026-10-14
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2610.19588v1

We consider the problem of estimating the background intensity of a Hawkes process when the triggering mechanism is known, treating the background as an unknown finite positive measure with a regular density. We propose a likelihood-based variational approach based on the Wasserstein-Fisher-Rao geometry. In this setting, the relevant first variations are naturally distribution-valued and therefore fall outside the classical gradient-flow formulation. We develop a distributional framework for the resulting dynamics and give them a finite-dimensional meaning through projections onto statistical manifolds. We identify the true background as the unique zero of a limiting generalized gradient and, under suitable stability, compactness, integrability and approximation conditions, establish a conditional consistency theorem for measurable estimators satisfying the empirical projected equations on events whose probabilities tend to one. Finally, we specialize the construction to Gaussian-mixture manifolds and obtain an explicit finite-dimensional scheme.

---

### Moral Combat: Exploiting Moral Foundations Theory to Jailbreak LLMs

Authors: Adel Khorramrouz, Yi-Chia Wang, Sharon Levy
Published: 2026-10-14
Categories: cs.CL, cs.CY
arXiv: https://arxiv.org/abs/2610.19587v1

Social psychology has long described how people can be persuaded to interpret a harmful act as moral. Recent research suggests that Large Language Models (LLMs) echo aspects of human psychology, from cognitive biases to social conformity. This raises a question with direct safety implications: can LLMs be moralized to bypass safety guardrails? We answer this question with a multi-agent framework grounded in Moral Foundations Theory (MFT). Given a harmful request that is initially refused, our agents diagnose which moral foundation (MF) the prompt is violating, identify a justifiable foundation under which the same request appears virtuous, rewrite the prompt accordingly, and reinforce it with generated precedents of comparable moral reframing. We evaluate our framework on harmful prompts from three well-known jailbreak datasets across five open-source and proprietary models. Our analysis reveals that models are highly susceptible to moralization, exhibiting high single-trial jailbreak rates with success increasing further under repeated attempts. We also find that safety benchmarks are imbalanced in the MFs they violate, concentrating on foundations toward which models already exhibit biases.

---

### Who's Behind the Text? Robust Model Authorship Attribution

Authors: Minseok Ko, Hyunseok Lee, Woomin Song, Beomjun Kim, Jinwoo Shin, Kunwoong Kim
Published: 2026-10-14
Categories: cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2610.19583v1

As large language models (LLMs) are increasingly used to generate text, their misuse has become a growing societal problem, with implications for academic misconduct, model distillation, and cyberattacks. For these reasons, Model Authorship Attribution (MAA), which identifies the source model of a given text, is important for promoting the responsible use of LLMs and improving AI safety. In practice, attribution systems must remain reliable beyond the distributions observed during training. They may encounter texts from unseen domains and writing settings, as well as generations from future versions of model families that were not available when the system was trained. However, existing MAA methods often exhibit substantial performance degradation under such distribution shifts. To address this challenge, we propose Authorship Tracing via Likelihood-based Attribution of Sources (ATLAS), which formulates MAA through source-conditioned text modeling rather than direct source prediction. ATLAS learns how text from each source family is generated and attributes a given text by comparing its likelihood under each learned source condition. We aggregate this likelihood evidence at the token level with clipping so that no single token dominates the decision. We further introduce MAABench, a broad MAA benchmark for evaluating practical OOD settings such as domain and version. Our experiments show that ATLAS consistently outperforms existing baselines across MAABench and OpenTuringBench. In particular, when both the generation domain and the model version change on MAABench, ATLAS achieves 74.27% accuracy, 9.00pp higher than the strongest baseline. Beyond MAA, ATLAS can also help identify distillation source models by measuring which source model a target LLM most closely resembles. Our extended experiments show that ATLAS recovers the known teacher of every distilled model we evaluate.

---

### XferAgent: Cross-Agent Experience Transfer for Self-Improving LLM Agents

Authors: Weihao Li, Jun Bai, Shijie Chen, Qincheng Lu, Xinyu Wang, Ziyang Song
Published: 2026-10-14
Categories: cs.MA, cs.AI
arXiv: https://arxiv.org/abs/2610.19581v1

Large language model (LLM) agents can reuse experience from prior task executions to improve performance on subsequent tasks. Although existing methods enable cross-agent experience sharing, differences in agents' execution histories and task requirements make shared experiences difficult to reuse effectively. We propose XferAgent, a framework for cross-agent experience transfer that organizes reusable experience in local and shared experience graphs. XferAgent reconciles compatible contributions from heterogeneous agents while avoiding unreliable merges. For each new task, it selects shared experiences and integrates them with the target agent's local experience to support task execution. Experiments across diverse task domains show improvements over representative memory-based methods and effective experience transfer between heterogeneous agent frameworks. Cumulative task success also increases over successive task passes as agents update and reuse their experiences.

---

### Bridging the Topological Divide: Dual-Branch Graph Distillation with Semantic Priors for Robust Cold-Start Recommendation

Authors: Yuanfeng Qiu
Published: 2026-10-14
Categories: cs.IR, cs.LG
arXiv: https://arxiv.org/abs/2610.19578v1

Graph Collaborative Filtering models have achieved remarkable success in recommendation systems. However, they inherently suffer from a catastrophic representation collapse in strict cold-start scenarios due to the topological isolation of unobserved items. While integrating multi-modal semantic features offers a potential remedy, directly mapping isolated semantic attributes into a globally topology-driven collaborative space frequently precipitates representation degradation and optimization conflicts. To bridge this fundamental structural divide, we propose DualGCN, an innovative Dual-Branch Graph Knowledge Distillation Framework. Operating under a decoupled two-stage training paradigm, DualGCN first constructs a robust collaborative teacher network where item representations are anchored by a topology-smoothed semantic prior, effectively reshaping the latent manifold. Subsequently, a semantic student branch employs a residual mapping mechanism and a magnitude-agnostic cosine distillation objective to achieve high-fidelity cross-modal alignment without disrupting established collaborative signals. Furthermore, we introduce a dynamic norm reconstruction mechanism to rigorously calibrate the physical scales of novel and warm items during global retrieval. Extensive experiments across three real-world benchmark datasets demonstrate that DualGCN significantly outperforms state-of-the-art baselines in strict cold-start scenarios while comprehensively preserving recommendation accuracy for highly active items.

---

### Learning Across Temporal Graph Representation Silos

Authors: Minwoo Yu, Young-guk Ha
Published: 2026-10-14
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.19576v1

Temporal graph data are recorded in different representations, but whether these differences also determine the scope of learned knowledge reuse remains unclear. This study investigates whether predictive structure learned from relational histories can be reused between continuous-time dynamic graphs (CTDGs) and temporal knowledge graphs (TKGs). Bidirectional reuse emerges while retaining distinct input and prediction interfaces and independent vocabularies, and sources with the same representation are not always more useful. Controlled reorganizations of source histories show that the organization contributing to reuse varies across source-target conditions. We show that learned aspects of Relational Dynamics, the predictive structure present in relational histories, can be reused across conventional representation boundaries. Independently pretrained TKG and CTDG learners can also be fused at the score level to improve prediction on a third dataset without entity or relation-semantic alignment. Two proofs of concept in recommendation and security detection illustrate that external temporal sources without direct alignment can also become useful pretraining resources.

---

### On Inherent Privacy of Posterior Sampling: A Unified Rényi-Divergence Framework

Authors: Ruyu Zhou, Fang Liu
Published: 2026-10-14
Categories: stat.ML, cs.LG, math.ST
arXiv: https://arxiv.org/abs/2610.19573v1

Differential Privacy (DP) provides a rigorous framework for quantifying privacy guarantees. While most methods achieve DP by injecting calibrated random noise, the intrinsic randomness of certain procedures such as sampling can yield DP guarantees ``for free''. In this work, we develop a unified Rényi divergence framework to characterize the inherent DP guarantees of sampling from a Bayesian posterior distribution. We show that the privacy loss incurred by One-Posterior-Sampling (OPS) is governed by posterior exponential moments of the single-record log-likelihood ratio, and establish explicit DP guarantees under uniformly bounded, sub-Gaussian, or sub-exponential tail regimes. We apply the framework to representative Bayesian models, including categorical likelihoods with arbitrary priors, Gaussian-Gaussian conjugate pair, generalized linear models from the exponential family and linear regression with Gaussian likelihood, both with Gaussian priors on the regression coefficients. Our results elucidate sample size, model structure, neighboring relations in DP (bounded or unbounded), and how prior if applicable, jointly determine the inherent DP guarantees of OPS. Furthermore, our results recover the existing DP guarantees for OPS as special cases -- often with tighter privacy loss bounds -- and substantially broaden the class of Bayesian models for which the inherent DP guarantees of OPS can be rigorously characterized.

---

### Divergence-Routed On-Policy Distillation for Language Models

Authors: Tiantian Peng, Guannan Lv, Hongyang Li, Qiuhe Hong, Hang Cheng, Hongjian Dou, Xiangyu Wu, Yuyang Liu
Published: 2026-10-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.19572v1

On-policy distillation (OPD) transfers knowledge from a teacher model to a student along student-generated trajectories, typically by minimizing the reverse Kullback-Leibler (RKL) divergence. However, RKL can provide weak corrective signals for teacher-supported tokens to which the student assigns very low probability. This raises two questions: which positions are most consequential for the behavioral gap between the student and teacher, and what additional supervision is needed at these positions? Through cross-sampling, we find that divergence-based routing more effectively pinpoints positions contributing to the teacher-student performance gap than entropy-based routing. Further analysis reveals teacher-mode under-coverage at high Jensen-Shannon (JS) divergence positions, where standard OPD increases the large teacher-student probability gaps. Motivated by these findings, we propose Divergence-Routed On-Policy Distillation (DrOPD). DrOPD retains RKL supervision at all positions and selectively adds forward Kullback-Leibler (FKL) divergence supervision at positions with high JS divergence. Experiments across six math benchmarks show that DrOPD improves over standard OPD and entropy-based routing. DrOPD also reduces severe under-coverage cases by 37.9 percent relative to OPD, supporting targeted FKL as an effective complement to OPD.

---

### LED-BENCH: A Benchmark for Large-format Engineering Drawing Understanding

Authors: Ke Li, Tianyi Wang, Enning You, Chaoqun Wan, Zhen Huang, Shuxian Liang, Xian-Sheng Hua
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19571v1

Real-world engineering drawings are often large-format and information-dense. Answering practical engineering questions therefore requires both fine-grained understanding of engineering information and complex reasoning over objects, relations, and evidence distributed across different regions or views. Existing benchmarks cover a range of engineering-drawing understanding tasks involving local regions, individual views, specific elements, or more constrained visual settings. However, systematic evaluation of understanding over realistic, large-format engineering drawings remains limited. To address this gap, we introduce LED-BENCH, a multidisciplinary benchmark designed to evaluate fine-grained understanding and complex reasoning over large-format engineering drawings, spanning the mechanical, architectural, and transportation domains. Extensive evaluation of 11 representative multimodal large language models, including 8 proprietary models and 3 open-source models, reveals substantial remaining difficulty: GPT-5.6 Sol achieves 52.65% average accuracy, performance varies across engineering domains, and all evaluated models perform worse on questions requiring broader evidence integration. These results highlight fine-grained understanding, evidence integration, and complex reasoning over large-format engineering drawings as persistent challenges for current multimodal models.

---

### Geometric Kernel Diversity: An Axiomatic Diversity Measure in Euclidean Space

Authors: Ditbul Ban, Younghun Roh, Julian Shun, Vihan Lakshman
Published: 2026-10-14
Categories: cs.CG, cs.CC, cs.LG
arXiv: https://arxiv.org/abs/2610.19570v1

Quantifying the diversity of a set of objects is a fundamental problem in machine learning and statistics. Mironov and Prokhorenkova (2025) recently showed that three axioms-monotonicity, uniqueness, and continuity-are crucial for a reliable diversity measure and that none of the standard metrics in the literature satisfy all three. They also constructed two measures that do satisfy the axioms, but both are NP-hard to compute, leaving open the question of whether efficient axiomatic diversity measures can exist. We study a Euclidean-restricted version of this question. We introduce Geometric Kernel Diversity (GKD), which associates a radial kernel solid with each point and measures the volume of their union. We show that canonical GKD satisfies all three Mironov-Prokhorenkova axioms on Euclidean-realizable inputs. For canonical GKD, we give exact polynomial-time algorithms in dimensions $d \le 3$. In arbitrary dimensions, we give a randomized algorithm that returns a $(1 \pm \varepsilon)$-approximation with probability at least $1-δ$ in time $O \left(\frac{nd}{\varepsilon^2}\log\frac{2}δ\right)$ for $n$ points. The axiomatic guarantee applies to exact canonical GKD; the randomized algorithm approximates this deterministic quantity. We further show that the two diversity measures of Mironov and Prokhorenkova (2025) remain NP-hard on Euclidean-realizable inputs and, unless $\mathrm{NP} = \mathrm{RP}$, admit no FPRAS even on this restricted domain.

---

### INTACT: Inducing Latent Reasoning on a Frozen Backbone

Authors: Kutay Tire, Atula Tejaswi, Haris Vikalo, Sujay Sanghavi
Published: 2026-10-14
Categories: cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2610.19568v1

Latent reasoning replaces natural-language chains of thought with compact continuous "soft-token" representations, reducing the number of intermediate tokens that must be decoded autoregressively. We use a small external encoder to induce latent reasoning in a frozen backbone model. The encoder iteratively maps the backbone's hidden states to new latent blocks, constructing a latent reasoning prefix, after which the frozen model writes its remaining reasoning and its answer as text. Our method, INTACT, trains this encoder through distillation to reproduce the effect of written reasoning steps while keeping the model's behavior close to that of the frozen backbone, and then reinforces it on answer correctness. Across four frozen backbones from 1B to 8B, INTACT improves both GSM8K and average out-of-domain accuracy, including GSM8K gains from 75.8% to 79.3% on Llama-3.2-3B and from 77.6% to 79.7% on Qwen2.5-3B. It also reduces end-to-end latency per problem by 19% on Llama-3.2-3B and outperforms the evaluated latent-reasoning baselines at the 3B and 8B scales. Beyond the mathematics setting, the same recipe trained on a multi-domain corpus improves the frozen backbone on twelve evaluation benchmarks spanning mathematics, science, and commonsense, without task-specific adaptation.

---

### Frame Evidence:A Benchmark for Reference Frames and Visual Evidence in Multi-Image Spatial Reasoning

Authors: Chunxi Luo, Qishuai Ding, Xingchen Peng, Hengwei Ye, Ke Zhang, Wenxin Lian, Tianyi Lu, Chunlong Yu, Zhenya Lin
Published: 2026-10-14
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2610.19564v2

A spatial answer depends on the reference frame specified in language and its supporting visual evidence. Item accuracy alone obscures whether correctness persists when either changes. We introduce Frame Evidence, a benchmark of 320 matched triplets and 960 Chinese multiple-choice questions from captured and rendered scenes. One family compares three grounded reference frames over fixed images. The other pairs image deletion annotated as insufficient with answer-preserving addition. Joint metrics require correct answers across all three frames and paired evidence variants. Across 20 configurations on 288 held-out questions, GPT three-shot reaches 50.00\% item accuracy, but the best three-frame group accuracy is 5/48 (10.42\%), reached by GPT two-shot. Few-shot prompting and task-specific adaptation improve item accuracy without consistently improving joint correctness. Input controls show where images support matched success, while repeated executions reveal answer instability beneath similar scores.

---

### SaTPO: Text-to-Image Safety Alignment via Transition Preference Optimization

Authors: Ziyi Fang, Jingzhi Li, Xiaojun Jia, Xingxing Wei, Wenqi Ren, Haijun Zhang
Published: 2026-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.19563v1

Safety alignment of text-to-image diffusion models aims to suppress harmful content without compromising useful generation capabilities. However, preferences over completed images compress multidimensional generation factors into a coarse signal, potentially driving changes beyond those required for safety. To address this, we propose Safety Transition Preference Optimization (SaTPO), which enforces step-wise alignment during denoising. SaTPO leverages the insight that an unsafe prompt and its safe rewrite naturally form a tightly coupled preference pair when conditioned on a shared parent state and noise realization. By anchoring optimization on identical parent states, SaTPO eliminates historical trajectory divergence, confining safety adaptation to an incremental alignment while preserving task-relevant semantics. Multi-scale drift constraints and benign replay further limit behavioral deviation and preserve ordinary generation. We additionally construct CoProPV, a refined dataset ensuring visually verified contrasts. Experiments on Stable Diffusion v1.5 and SDXL demonstrate improved safety-quality trade-offs. Compared with AlignGuard, PerspectiveVision unsafe-detection rates on Inappropriate Image Prompts decrease from 12.70% to 6.00% and from 15.04% to 9.30%, respectively, while FID improves from 32.28 to 25.95 and from 36.50 to 27.51. Warning: this paper includes potentially offensive content.

---

### REWIND: Uncertainty-Guided Trajectory Repair for Test-Time Reasoning

Authors: Yi Ding, Mengyu Wang, Min Zhou, Menglin Yang
Published: 2026-10-14
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2610.19561v1

An early mistake in the generated reasoning can constrain all subsequent reasoning. Repeated-generation methods, like MCMC-based approaches, attempt to improve output quality by sampling multiple generations, but their effectiveness can be limited by randomly selecting where to regenerate. On MATH500 with Qwen2.5-Math-7B, we find that incorrect responses have substantially higher mean token entropy than correct responses, with an average within-question relative increase of approximately 112%, suggesting that entropy can signal reasoning errors. Motivated by this finding, we introduce REWIND, a training-free controller that uses entropy dynamics to propose targeted suffix regenerations. Across three model families and five benchmarks, REWIND outperforms all baselines, with gains of 0.51-10.00 percentage points over the strongest baseline. These results demonstrate the effectiveness of targeted, uncertainty-guided trajectory repair across models and tasks. Our code is available at https://anonymous.4open.science/r/rewind-D310.

---

### $τ$-Emotion-Bench: Exposing the Interaction Fragility of LLM Agents through Real-World User Emotions

Authors: Rishitosh Kumar Singh, Amir Saeidi, Shamanthak Hegde, Shrinidhi Kumbhar, Chitta Baral
Published: 2026-10-14
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.19560v1

In long-horizon tool-calling tasks, LLM agents can fail not only because of tool-use errors, but also because of ineffective interaction with users. Yet existing benchmarks largely assume cooperative and behaviorally stable users, leaving the effect of emotional variation on agent reliability underexplored. We introduce $τ$-Emotion-Bench, a ten-domain benchmark for evaluating tool-calling agents under controlled emotion-conditioned user interactions while preserving the underlying task objective. To scale both benchmark construction and agent training, we introduce Tracer, a scalable agentic data-generation pipeline that constructs grounded, policy-valid tasks and collects trajectories through an interactive agent--user--environment harness. By allowing dissatisfied users to terminate inefficient interactions, the harness provides an interaction-aware signal that filters trajectories with excessive user dependence and favors effective tool use. Experiments with two user simulators show that frontier agents remain unreliable, with substantial gaps between average task success and consistent task completion. Tracer-generated supervision consistently improves Qwen3 models across scales, generalizes to unseen domains, and transfers to $τ$-Bench and $τ$-Trait. Behavioral analyses further show that higher task success is associated with a better balance between tool execution and necessary user follow-up. Overall, $τ$-Emotion-Bench and Tracer provide a scalable framework for evaluating agent reliability under diverse user behaviors and improving tool-calling agents through interaction-aware supervision.

---