# Research Papers — 2026-10 · bundle page 111 of 156 > Full-text markdown of papers #5501–5550 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/111 [← Previous bundle](https://aipod.dev/papers/archive/2026-10/page/110.md) [Next bundle →](https://aipod.dev/papers/archive/2026-10/page/112.md) ### Navier-Stokes lost in translation: Why Lean verification of AI autoformalisation does not guarantee correct natural language proofs Authors: Alexander Bastounis, Fabian Circelli, Anders C. Hansen Published: 2026-10-06 Categories: math.AP, cs.AI, math.LO arXiv: https://arxiv.org/abs/2610.08144v1 Autoformalisation is increasingly used to verify mathematical texts, including those generated by AI, as in OpenAI's announced proof of blow-up of solutions to the Navier-Stokes equations. In this process, an AI system translates the text from a natural language (NL) into a formal language such as Lean. Once this translation is done, the argument expressed in the formal language can easily be mechanically verified. The purpose of this article is to demonstrate why this process may offer no confidence in the original NL argument, owing to the various difficulties in performing the translation semantically faithfully. In particular, we highlight that the problem of resolving ambiguities in mathematical NL text, which is necessary in order to provide semantically faithful translation, is arbitrarily high up in the Solvability Complexity Index (SCI) hierarchy/arithmetical hierarchy (the SCI $= \infty$). Hence, informally, providing semantically faithful AI autoformalisation is harder than any computational problem including the Halting problem (which has SCI $= 1$). To demonstrate the effect of this result we provide several examples of AI mistranslations of NL statements and proofs into Lean in practice, resulting in mismatches between NL proofs and their Lean `verifications'. These include OpenAI's announced Navier-Stokes proof. In particular, we show that the formalised Lean proof does not correspond to the NL proof of blow-up of solutions to the Navier-Stokes equations. --- ### Partially Observable Zero-shot coordination by Predicting Intention of Partner Authors: Jinnyeong Yang, Yuhwan Jeong, Hoyong Kwon, Minseok Kim, Jihun Kim, Kuk-Jin Yoon Published: 2026-10-06 Categories: cs.AI, cs.MA arXiv: https://arxiv.org/abs/2610.08142v1 Zero-shot coordination in embodied settings requires acting while the partner is intermittently out of view, leaving existing methods with ambiguous partner representations and uncertainty over hidden partner states. We propose Predicting Intention of Partner (PIP) to jointly address these challenges. PIP uses a Joint-view VAE to distill richer training-time evidence from the union of both agents' local observations into a partner representation available from local observations alone. Partner-state Belief networks further infer the partner's hidden location and behavioral tendencies from the ego agent's interaction history. We evaluate PIP in Burrito-PO, Overcooked-PO, and a Melting Pot substrate, together with a human evaluation in Burrito-PO. PIP attains the highest mean performance among the compared methods across all three benchmarks. Human evaluation and diagnostic analyses further support coordination with unseen partners and the contributions of both components under partner occlusion. --- ### Test-Time Agent Evolution for Long-Horizon Legal Reasoning Authors: Haotian Chen, Shuaicheng Niu, Haocong Rao, Kaisong Song, Jun Lin, Lizhen Cui, Zhiqi Shen, Yonghui Xu Published: 2026-10-06 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.08138v1 Legal intelligence aims to support reliable decision-making across long-horizon legal processes involving evolving case states and multiple roles. However, real-world legal deployment exhibits substantial case heterogeneity in facts, evidence, and procedural contexts, exposing the limitations of static agent strategies. Moreover, legal reasoning is inherently interdependent across roles and procedural stages, making global reliability fundamentally different from isolated role competence. To address these challenges, we study training-free test-time agent adaptation, where agents continuously exploit deployment-time signals from preceding cases and ongoing interactions without updating model parameters. We propose \method, which introduces \emph{Test-Time Memory Evolution} to retrieve reusable experience from previous cases, adapt it to the current factual and procedural context, and consolidate accumulated experience for subsequent decision-making. Further, \emph{Rubric-Aligned Collaboration} verifies and revises role-specific actions according to behavioral and procedural requirements, enabling coordinated decision-making across roles and stages. Extensive experiments on J1-EVAL and LegalWorld across five backbone models demonstrate consistent improvements over representative reasoning and agent baselines with reasonable interaction and computational costs. Ablation and case studies further show that the two components provide complementary benefits in experience adaptation and cross-role coordination, improving the reliability and efficiency of long-horizon legal reasoning. --- ### Rethinking Visual Provenance: Detection and Watermarking Across Direct Visual Generation and LLM-Driven Code Rendering Authors: Zheng Gao, Xiaoyu Li, Zhicheng Bao, Jiaojiao Jiang Published: 2026-10-06 Categories: cs.CR, cs.CV arXiv: https://arxiv.org/abs/2610.08137v2 AI systems create images and videos with image/video generation models or by writing code and graphics descriptions that are then rendered. These routes can produce similar visible artifacts but expose different representations, intervention points, and provenance evidence. We develop a production-centered framework that compares detection and watermarking across both routes. An explicit verification specification distinguishes passive inference, message recovery, and authenticated provenance. We organize image, video, source-code, and rendering-aware watermarks by production stage. We examine the different requirements of generated images and video, plots and SVG, programmable video, and agent-composed workflows. Documented Claude, OpenAI, and rendering-tool interfaces connect the framework to concrete systems. We pose ten scoped research questions on identifiability, observability, fair comparison across stages, recoverable payload, reconstruction, synchronization, composition, hybrid local contribution, and private production-event authentication. The result is a conceptual research agenda grounded in published methods, inspected interfaces, and elementary boundary examples. It reports no experiments and claims no new theorems; its appendix results are elementary calculations, and documentation and source inspection establish interfaces, not empirical robustness. --- ### VLA-ACL: Action-Consistent Visual Token Pruning for Efficient Vision-Language-Action Models Authors: Owen Du, Yang Yue, Jie Zhang, Jiaqi Pi, Chi Bene Chen, Gao Huang Published: 2026-10-06 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2610.08133v2 Vision-Language-Action (VLA) models achieve strong robotic manipulation performance but incur high computational costs from processing long token sequences at every control step, limiting real-time deployment. Visual token pruning offers a direct solution, as visual patches dominate the input sequence and contain considerable redundancy. Existing approaches, however, either rely on indirect training-free heuristics, such as attention scores and motion thresholds, or require costly fine-tuning of the base VLA model. We introduce VLA-ACL (Action Consistency Learning), which learns a lightweight visual token pruning policy through action-level supervision while keeping the base VLA model entirely frozen. The training objective encourages actions produced from pruned visual contexts to remain consistent with the full-context teacher, with ground-truth actions as auxiliary supervision. This directly ties token selection to its effect on the downstream control output. Experiments on LIBERO and real-world manipulation tasks show that VLA-ACL prunes up to 87.5% of visual tokens while retaining competitive performance, reduces computation by up to 75%, and achieves a 1.5x inference speedup. These results establish a stronger performance-efficiency trade-off than existing frozen-VLA pruning methods and demonstrate the value of action-level supervision for visual token selection. Code is available at https://github.com/du-owen/VLA-ACL. --- ### Beyond Marginal Monitoring: Distributed Joint-Distribution Testing for Data Concept Drift in Large Scale E-Commerce Operations Authors: Cagdas Pullu, Mahmut Emir Arslan, Bugra Balkac, Aylin Ondersev Balta, Cihangir Celal Palaci, Fikri Cem Yilmaz, Altan Cakir Published: 2026-10-06 Categories: cs.DC, cs.LG, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2610.08132v1 Concept drift threatens production machine learning, yet the empirical behavior of multivariate two-sample drift detectors at scale remains under-characterized. Existing benchmarks rarely address the hundreds of millions of rows and high-cardinality features typical of industrial-operational datasets. We evaluate five multi-column two-sample tests (marginal, projection-based, and kernel embedding methods) across three complementary environments: the Harvard Dataverse, a validated Failing Loudly reproduction (mean absolute error between 0.030 and 0.053), and a novel synthetic-injection benchmark on the 137.5-million-row Trendyol collection-ranking feature table. Testing four drift types across two severity-scope regimes, we demonstrate that distributed Maximum Mean Discrepancy with Random Fourier Features on Apache Spark scales robustly. Averaged over the four drift types in the strong regime and under a calibrated threshold, it achieves a Pearson correlation of r = 0.940 with expected drift magnitude, an 80.4% true positive rate, and a 3.2% false positive rate. Conversely, the per-dimension Kolmogorov-Smirnov test failed due to statistic saturation from ID-like columns under asymmetric sampling, establishing a critical constraint for large-scale sampling design. At weak configurations (realized-flip fractions of at most 0.57%), detectors struggled to reliably discriminate, highlighting the need for future intensity-grid power analyses to distinguish fundamental sensitivity bounds from scalable threshold shifts. --- ### Mu-DisCoCat: A Variational Pipeline for Compositional Generalization on Quantum Processors Authors: Mina Abbaszadeh, Matilda Karabina Moore, Raem Haq, Martha Lewis, Mehrnoosh Sadrzadeh Published: 2026-10-06 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2610.08131v1 Achieving compositional concept generalization (CoCoGen), the ability to understand novel situations by recombining learned primitives, remains a fundamental challenge in artificial intelligence. Compositional semantic models such as Compositional Distributional Semantics (DisCoCat) offer solutions by generalising vectors to tensors, but suffer from scaling bottlenecks when learning the tensors. Mapping DisCoCat onto Variational Quantum Circuits (VQCs) resolves this limitation for text, yet the methodology has not been expanded to multimodal situations such as the ones involved in CoCoGen. This paper introduces Mu-DisCoCat: a multimodal variational quantum learning framework for DisCoCat that achieves CoCoGen. The framework first learns stable object representations from single-object image-text pairs, then fixes these and uses them to learn the relations between them in multi-object situations. In classical simulations, the model used Uhlmann state fidelity to compute the overlap between the multimodal circuit representations and achieved higher relational OOD accuracy than the evaluated CLIP baseline. Its deployment was evaluated using the destructive SWAP test across noisy quantum emulators, including a range of IBM fake backends, IQM FakeAphrodite, and the IBM Marrakesh quantum processor. Despite real-world device noise, the hardware-executed models maintained a strong positive correlation with simulated fidelities, reliably distinguishing unseen similar and dissimilar pairs. Our work establishes a framework for executing CoCoGen on VQCs, demonstrating a viable use case for near-term quantum hardware. --- ### Do LLMs Act on What They Know? From Partner Representations to Cooperative Actions Authors: Yuhwan Jeong, Jinnyeong Yang, Kuk-Jin Yoon Published: 2026-10-06 Categories: cs.LG arXiv: https://arxiv.org/abs/2610.08129v1 Cooperation with unfamiliar partners requires adapting to communication conventions that are not known in advance. We study this problem in a controlled Hanabi-derived environment with scripted hint generation, LLM-controlled receiving decisions, and frozen model weights. Across eight LLMs, linear probes recover intent conventions substantially more accurately than target conventions, yet receiving choices do not consistently agree with the sender's convention. We compare probe-predicted and ground-truth conventions presented either as general rules or as externally computed action recommendations. Rule statements yield modest and model-dependent changes in cooperation, whereas action translation produces larger gains on average. In a Qwen3-8B case study, matched-state statement reversals reveal much greater sensitivity to action recommendations than to rule statements. Activation transfers from oracle-action and non-oracle hint-restatement donors improve intent accuracy on both action classes, but the tested alternatives do not reliably reproduce these benefits. Together, these results distinguish convention decodability, sensitivity to convention information, and cooperative performance, and highlight limitations in turning available partner information into receiving decisions. --- ### Supermarket Product Detection and Recognition: Utilizing Deep Learning with Rectified Imagery Authors: Mayank Sah, Jimson Mathew Published: 2026-10-06 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2610.08126v1 Product Identification has sprung up to become one of the most challenging problems in the automation of the retail industry. With the new industry 5.0 standards, automated inventory management, and catalog creation tasks are vitally important. Object identification models have emerged as a viable answer with their unprecedented identification and localization accuracy. However, the close-knit rack design of supermarkets generates the problem of angle variation in capturing images. The angle-variant densely packed images(a single image contains many objects) become overwhelming for these models alone. In this paper, we try to supplement object detection models with traditional Hough transform (HT) and homogeneous estimation concepts. We study the effect of rectified images using homography estimation and hough transform and their limitations on the problem of grocery identification. We make a case for creating a new dataset to test the effects of such rectification and produce analytical results on different scenarios of angle variation and object densities per image. Extensive experiments on different object detection models suggest that image rectification of angled images improves the detection accuracy of grocery products in images. The results also highlight the limitation of rectification on the angle of image capture and the object density of the image. --- ### Conversation Is a Two-Body Problem: Dyadic Evaluation of Full-Duplex Dialogue Models Authors: Sungnyun Kim, Sungwoo Cho, Jihwan Oh, Se-Young Yun Published: 2026-10-06 Categories: eess.AS, cs.CL arXiv: https://arxiv.org/abs/2610.08125v1 Full-duplex spoken dialogue models listen and speak at the same time, enabling voice agents to have natural, low-latency interactions that turn-based systems cannot offer. However, they are commonly evaluated against single-sided interlocutors: pre-recorded audio that cannot react, or an automated examiner that reacts in real time but only administers a fixed sequence of tests and is never graded. These single-sided frameworks evaluate only half of a two-body problem, where turn-taking, overlap, and interruption are joint products of two coupled speakers. We propose DyaFDB, a framework that evaluates full-duplex models in a dyadic setup: two models converse directly under assigned roles with cooperative or conflicting goals, and both sides are scored offline with an external judge. DyaFDB probes how the two models behave toward each other, such as how they take turns or carry an assigned role under different interests. We instantiate four tasks as 140 scenarios and record 7,560 conversations, covering six self- and cross-play pairings. Throughout the experiments, we observe that how a model behaves continually reshapes its partner. We thus demonstrate that each model must be both the examiner and examinee of the other, and no single fixed interlocutor can play both parts. We will release the scenarios, role prompts, and recording protocols between two full-duplex models, without any pre-recorded audio. --- ### Beyond Waypoint Regression: Query-Based Cost Learning over Reachable Ego Futures for End-to-End Driving Authors: Ahmed Abouelazm, Rupert Polley, Qingyuan Zhang, Yin Wu, Philip Schörner, Carl Esselborn, J. Marius Zöllner Published: 2026-10-06 Categories: cs.RO, cs.AI, cs.LG, eess.SY arXiv: https://arxiv.org/abs/2610.08123v1 End-to-end planners based on waypoint regression achieve strong open-loop accuracy, but they primarily learn to mimic expert geometry and remain difficult to adapt to deployment-time safety constraints. We propose a query-based cost-learning framework that estimates bounded costs for dynamically reachable ego trajectory queries, rather than dense BEV cells or a small regressed trajectory set. Compact joint scene tokens capture coherent multimodal agent futures, while contingency-aware cost aggregation and cost-guided intra-cluster MPPI mixing convert the learned cost topology into feasible ego plans. On nuScenes, our method improves over prior cost-estimation planners such as ST-P3 and NMP, outperforms most regression baselines in collision rate, while remaining competitive in L2, and retaining an interpretable cost interface. On real-world driving logs, the proposed planner reduces collision rates compared with SparseDrive and Alpamayo without fine-tuning, while maintaining a diverse set of candidate trajectories. --- ### Attenuated in-context identification in time-series foundation models: diagnosis under counterfactual inputs and repair by synthetic forced-system fine-tuning Authors: Hong-In Won Published: 2026-10-06 Categories: cs.LG, eess.SY arXiv: https://arxiv.org/abs/2610.08118v1 Covariate-aware time-series foundation models (TSFMs) promise training-free what-if answers for instrumented plants: the change in output that a different future input would cause. We test this on forced engineering systems with exact counterfactuals, comparing Chronos-2, TimesFM-2.5 and TabPFN-TS with classical system identification fitted to the same context. Through their default covariate interfaces, TimesFM-2.5 and TabPFN-TS are memoryless: the predicted effect of an input change is a same-time function of that change ($R^2 = 1.000$ for TimesFM-2.5). Chronos-2 identifies dynamics in context but attenuates them. Its predicted effect is 0.33-0.80 of the true effect, its recovered impulse response has the wrong shape, and its error on a one-degree-of-freedom oscillator levels off at 0.57 with 8192 context samples, where ARX fitted to 256 samples reaches 0.02. Context dither at inference lowers the what-if error on all six synthetic classes without training. A 26-minute fine-tune on synthetic forced systems restores the response magnitude (sensitivity 0.83-0.96) and outperforms structure-agnostic identification on Wiener-Hammerstein and a held-out friction class. A specialised in-context identifier trained on the same data comes close, so the forced-system data carry most of the gain. On three of four measured plants classical identification remains clearly better, and the fine-tuned model loses part of its univariate forecasting skill. Paired counterfactual inputs, together with shuffled future inputs on measured records, test two properties: whether the covariate interface can represent dynamics and whether the pretraining prior covers the plant's time scale. Only the counterfactual pairs expose the attenuation. --- ### Energy-Aware Path Following: Comparative Analysis of Reinforcement Learning and NMPC for Electric Vehicles Authors: Mohamed Sabaa, Mostafa Emam Published: 2026-10-06 Categories: cs.RO, cs.LG, eess.SY arXiv: https://arxiv.org/abs/2610.08112v1 Path-following control strategies typically follow the bi-objective optimization dilemma: minimizing deviations from a reference path while maintaining smooth speed profiles. The latter objective is especially relevant for Electric Vehicles (EVs), since their limited driving range can be extended by recovering energy through regenerative braking, a feature that has not yet been sufficiently studied in the literature. In this work, we perform a comparative analysis of four controllers under one common Frenet frame-based kinematic vehicle model, utilizing a validated energy model (VT-CPEM) with explicit regenerative braking. Herein, we implement the following controllers: Nonlinear Model Predictive Control (NMPC), Proximal Policy Optimization (PPO), gain-scheduled Ackermann state-feedback baseline (PID-SF), and a Stanley geometric baseline. To satisfy real-time requirements, we implement the NMPC using JIT-compiled CasADi. Moreover, we train the PPO using traditional straight and S-curve tracks, after which we successfully transfer the unmodified policy to unseen tracks, including: an ISO 3888-1 lane-change, a chicane, randomly-generated parameterized-splines, and a $\pm3^\circ$ graded road. In addition, the policy transfers to a dynamic single-track vehicle model with linear tires, zero-shot with an acceptable initial performance, which was optimized after brief fine-tuning. Thereby, we demonstrate that our PPO is readily transferable to more comprehensive vehicle models. We conclude with a performance analysis of developed controllers and discuss ideas for future work. --- ### Beyond Training from Scratch: Foundation Models for Data-Efficient and Generalizable Cardiac MRI Reconstruction Authors: Anam Hashmi, Mayug Maniparambil, Julia Dietlmeier, Kathleen M. Curran, Noel E. O'Connor Published: 2026-10-06 Categories: cs.CV arXiv: https://arxiv.org/abs/2610.08109v1 Cardiac magnetic resonance imaging reconstruction aims to recover high-quality images from undersampled acquisitions, enabling faster scans while preserving diagnostic fidelity. Recent reconstruction methods are typically trained from scratch and often require large amounts of task-specific data, limiting their robustness under data scarcity and distribution shifts. In this work, we investigate whether pretrained vision foundation models can serve as effective priors for accelerated cardiac MRI reconstruction. We propose a reconstruction framework that integrates frozen and parameter-efficiently adapted visual encoders, including CLIP, BiomedCLIP, and DINOv2, within a transformer-based reconstruction architecture. Extensive experiments on the CMRxRecon2023 and CMRxRecon2024 benchmarks demonstrate that pretrained representations consistently outperform a transformer trained from scratch across multiple acceleration factors. We further evaluate performance under limited supervision and cross-dataset transfer, showing that foundation models provide superior data efficiency and generalization. While frozen representations are particularly effective in extreme low-data regimes, Low-Rank Adaptation (LoRA) yields additional gains when moderate amounts of training data are available. Among the evaluated backbones, DINOv2 achieves the strongest overall performance. These findings highlight the potential of vision foundation models as robust and transferable priors for cardiac MRI reconstruction. --- ### Enhancing Diffusion Language Models with Autoregressive Post-Training Weights Authors: Yiming Qin, Ke Wang, Amel Abdelraheem, Adam Hazimeh, Pascal Frossard Published: 2026-10-06 Categories: cs.LG arXiv: https://arxiv.org/abs/2610.08108v1 Diffusion language models (dLLMs) have emerged as a promising alternative to autoregressive (AR) language models, offering flexible token-update orders and parallel decoding. Recent dLLMs are often initialized from pretrained AR models before diffusion conversion in order to inherit their learned representations. After the conversion, however, they typically ignore the extensive post-training ecosystem of their AR ancestors. In this work, we show that these existing AR post-training weight updates can instead be effectively recycled to enhance diffusion models. Despite the changes by AR-to-diffusion conversion, directly adding an AR post-training weight update to a diffusion base model remains effective, bringing its performance close to that achieved by direct diffusion post-training. Notably, AR and diffusion post-training updates are nearly orthogonal in weight space, yet induce substantially more aligned representation changes in the diffusion model. Their distinct updates are also complementary: composing their weights can retain gains from both regimes and further improve the post-trained diffusion model. Based on these findings, we propose A2D, a simple training-free framework for enhancing diffusion models with existing AR post-training resources. A2D can transfer capabilities from AR post-trained models to diffusion base models, and further improve already post-trained diffusion models by composing AR and diffusion post-training updates. Across various dLLMs, including Dream, DreamReasoner, DiffuCoder, Dream-Coder, Nemotron-Labs-Diffusion, and DiffusionGemma, A2D reliably improves instruction following, mathematical reasoning, and coding with both supervised fine-tuning and reinforcement learning updates, without additional training, or inference-time computation. --- ### Exploiting Acoustic and Content-Oriented Speaker Verification Attacks Against Multilingual Voice Anonymization Authors: Ridwan Arefeen, Ze Li, Rong Tong, Ming Li, Xiaoxiao Miao Published: 2026-10-06 Categories: cs.SD, cs.AI arXiv: https://arxiv.org/abs/2610.08107v1 Attacker ASV systems for voice anonymization have been studied primarily in English, leaving their behavior in multilingual settings largely unexplored. Conventional ASV has shown that both acoustic and contextual information are important for multilingual speaker verification. Inspired by this, we investigate whether the same holds for attacker ASV on anonymized speech. We evaluate both acoustic- and content-oriented attackers on multilingual anonymized speech and construct a multilingual voice-converted dataset to improve cross-lingual generalization. Our results show that attacker effectiveness depends on the linguistic utility of the anonymized speech. Overall, acoustic-oriented attackers achieve better performance. However, when linguistic information is well preserved, the performance gap between content- and acoustic-oriented attackers narrows compared with conditions involving stronger speech distortion. The multilingual voice-converted dataset further improves performance and partially reduces the cross-lingual gap. These findings highlight the need for more comprehensive attacker modeling and evaluation protocols that consider both privacy and utility, rather than relying on a attacker strategy\footnote{Full code and pretrained models and MultiVC Dataset link are available at: https://github.com/monkeyDarefeen/DAST --- ### ChartBmkAgent: Harness-Governed Multi-Agent Construction of Chart QA Benchmarks from Sparse Error-Taxonomy Specifications Authors: Langxi Huang, Pingping Zhang, Lanyun Zhu, Chunyang Jiang, Jiawei Shao, Haocheng Yuan, Peilin Chen Published: 2026-10-06 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.08106v1 Multimodal large language models (MLLMs) advance rapidly, while conventional benchmark development lags behind, delaying investigation of newly observed capability gaps. Such investigation requires an expressive task format and an on-demand construction process: information-rich charts make chart question answering (Chart QA) suitable for probing coupled perception and reasoning. Automated Chart QA construction is intended to shorten the benchmark-development cycle by turning identified gaps into targeted samples on demand. Current methods, however, commonly separate target guidance from scratch generation: target-guided systems often require prepared data, charts, or templates, while scratch-generation systems primarily ensure artifact validity, without explicitly controlling whether newly synthesized requirements and content remain aligned with an externally specified diagnostic target. We introduce ChartBmkAgent, which turns an identified capability gap into targeted diagnostic evidence by constructing complete Chart QA samples from sparse error-taxonomy specifications. Throughout construction, a central harness governs specialized agents, requires stage-specific evidence of alignment with the original error category, and records the basis for each acceptance decision. On 300 taxonomy-wide samples, MLLM accuracies ranged from 32.7% to 84.3% with distinct category profiles, showing that generated samples reveal capability differences. Across three source-model comparisons, targeted follow-ups scored 50.0% versus 82.2% on matched controls ($p=8.96\times10^{-6}$); all six cross-model comparisons had the same direction, demonstrating targeted validation and diagnostic-data generation. Multiple evaluator models assessed whether each sample tested its specified error category; 86.4% met this criterion, providing empirical evidence of target preservation. --- ### DSV-Mem: Evaluating Multimodal Memory in Professional Workflows for MLLM Agents Authors: Jike Zhong, Ritwick Chaudhry, Xuanbai Chen, Tianchen Zhao, Linghan Xu, Yifan Xing, Nishant Sankaran Published: 2026-10-06 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.08102v1 Conversational MLLM agents are increasingly expected to assist in professional workflows, from AI research and engineering design to product management and business operations. Yet this capability remains underexplored: existing benchmarks largely focus on informal, everyday interactions and personal-life scenarios featuring photographic natural images, isolated static artifacts, and recall-oriented questions. In contrast, professional scenarios often involve structured, information-heavy artifacts that undergo frequent revisions and authority updates, and compositional queries requiring reconciliation of many artifact versions while tracking state precisely. To address these challenges, we introduce DSV-Mem, a benchmark for evaluating Dense Stateful Visual Memory. DSV-Mem comprises expert-reviewed scenarios and 1,000 questions across five user-oriented categories (Current State, Past State, Derived State, Change History, and Conflict/Refusal). A Hartley-inspired criterion favors questions with broader visual-evidence inspection demands. We also introduce a generation harness that produces evaluation suites by decoupling state-transition synthesis from conversation filling. Evaluation over 27 configurations spanning frontier and open-weight models and memory management methods reveals that the strongest baseline scores below 45% on DSV-Mem. Analysis surfaces findings: 1) multimodality and information density both contribute to difficulty, but state evolution, particularly the number of governing updates, is the dominant tested factor. Raw conversation/haystack length, OCR, and arithmetic are not the primary bottlenecks; 2) models often fail to verify user premises against prior state updates before answering; 3) increased reasoning effort and memory management methods yield limited gains, whereas state-aware designs prove more effective. The benchmark and code will be publicly released. --- ### Beyond Corrected Memory: Execution Consistency in Multi-Agent Systems Authors: Zhe Yu, Zixuan Wang, Peidong Wang, Hehai Lin, Ruochen Zhao, Chengwei Qin Published: 2026-10-06 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.08101v1 Shared memory coordinates agents' actions, but correct records do not establish that those actions satisfy task requirements. Memory governance and failure diagnosis regulate or inspect recorded information; they do not by themselves establish whether it is sufficient to judge task duties. We define execution consistency through duties governing state use, information handoffs, and final-state agreement, with explicit evidence conditions for judging fulfillment. Our core claim is that identical retained records can correspond to compliant and violating executions under the same task rule. Controlled removal of evidence such as receipt, action dependence, or response validity leaves 82.4% of opposite-label pairs indistinguishable; restoration separates 97.9% of the merged pairs. Natural-log annotations identify the defined violations in actual executions. However, existing logs do not always explicitly represent the execution relationships needed for these judgments. To assess the definition's practical value, we use CAVERT, a framework for consistency diagnosis and recovery, to extract supported relationships from logs and apply these criteria. It consistently outperforms contract-prompted LLM and rule-based baselines in diagnosis across all 12 benchmark-executor settings. Under the same gate and executor limits, it also outperforms rule-guided recovery in all four evaluated environments. These findings identify execution evidence that agent-memory and execution interfaces should preserve for reliable judgment. --- ### Surviving the Router: Optimizing Skill Injections for Retrieval and Execution Authors: Haneen Najjar, Luca Scionis, Haritz Puerto, Sahar Abdelnabi Published: 2026-10-06 Categories: cs.CR, cs.LG arXiv: https://arxiv.org/abs/2610.08098v2 AI agents increasingly rely on modular third-party "skills" that are dynamically selected by skill routers to execute complex tasks. While recent studies highlight the threat of prompt injections embedded in these skills, existing evaluations often assume settings where the malicious skill is already selected for execution. We show that this assumption can substantially overestimate attack success. In realistic multi-skill environments, injected skills must first compete for retrieval, reducing the effective attack success rate (ASR) of existing injections by 87-97%. To address this limitation, we introduce CORSA (Cluster Optimization for Router-Aware Skill Attacks), a router-aware attack that optimizes skill injections for both retrieval and execution across clusters of related tasks. We evaluate skill injection attacks under router-managed multi-skill settings by extending the benchmark introduced by SkillRouter with eight malicious payload categories. CORSA uses successive optimization stages to first improve retrieval and then optimize end-to-end attack success, while we evaluate user utility and injection naturalism separately. Our experiments show that CORSA substantially improves both retrieval and end-to-end attack success over existing skill injections while preserving user utility, and that the resulting attacks transfer across different router architectures and LLM backbones. --- ### When Tools Lie: Reliability of Mathematical Agents Under Corrupted Tool Feedback Authors: Kavienan Jegatheesan, Gayathri Lihinikaduarachchi Published: 2026-10-06 Categories: cs.CR, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2610.08097v1 Mathematical problem solving often requires deterministic computational steps that agents delegate to tools and implicitly trust. Yet tools can fail silently, returning plausible but incorrect results. How well can agents detect and correct corrupted tool call outputs? We study this through a controlled corruption framework where a hidden interceptor replaces tool call results with plausible incorrect information on targeted problems. We evaluate agents across 31 problems under four verification designs including no verification (baseline), mandatory same-context reflection, optional fresh-context verification, and optional structural verification. Without verification, corruption causes dramatic accuracy loss, from 100% down to 72.4%. Mandatory reflection fully recovers this performance to 100%. Optional verification improves accuracy only when models actively invoke it. Our results show that checking frequency is strongly associated with robustness differences, while unequal invocation prevents a controlled comparison of verifier quality. A supporting recovery experiment shows that full problem restart succeeds in 100% of cases after explicit detection. These findings demonstrate that verifier availability and verification policy are separate components of mathematical-agent reliability. Mandatory policies enforce verification while optional policies depend on the model's own choice to invoke it. --- ### Natural Language Questions as an Interface for Knowledge Graphs: QRAKEN Graph Distillation and Semantic Self-Healing Authors: Remo Grillo, Lukas Klic, Giovanni Colavizza Published: 2026-10-06 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2610.08095v1 Natural-language access to RDF knowledge graphs is a core Semantic Web ambition. Large language models (LLMs) have advanced Text-to-SPARQL, yet on unfamiliar graphs they often generate valid queries that misrepresent the populated data model. QRAKEN is a training-free, ontology-agnostic neurosymbolic pipeline grounding generation in empirical graph evidence rather than schema expectations. An offline distiller produces TTQL, a compact description of populated multi-hop patterns, conditional frequencies and path-conditioned literal examples, plus a class-property co-occurrence matrix. Online, TTQL guides the LLM, while deterministic syntax, vocabulary and data-model checks provide diagnostics for iterative refinement. On CK25 (First International Text2SPARQL Challenge), under matched-condition recomputation on a QLever snapshot, QRAKEN achieves strict F1 of 0.643 $\pm$ 0.026 with GPT-4.1 mini and 0.652 $\pm$ 0.012 with GPT-5.4: relative gains of 30% and 32% over the strongest recomputed participant, outperforming systems using the same base model family. Ablations identify TTQL patterns as the dominant driver (+0.31 strict F1 over a shape-only baseline); the refinement loop provides a cheap safety net, rejecting triple patterns unsupported by the co-occurrence matrix. Compared with auto-derived SHACL, TTQL yields 64% higher strict F1, supporting the value of empirical patterns beyond schema exposure. With two local 35B 4-bit open-weight models at zero marginal cost, the same pipeline matches the strongest recomputed participant, and TTQL advantages over shape-only and SHACL baselines persist. Results on a single, relatively small benchmark provide an initial empirical signal; monolithic TTQL injection on very open cross-domain graphs remains the main limitation. --- ### SAGE: Semantic Anchor-Guided Evolution for Grounded Medical QA Data Synthesis Authors: Chuan Li, Chengyu Wang, Cen Chen, Ye Lyu, Mingyuan Fan, Ming Gao Published: 2026-10-06 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2610.08093v1 Developing reliable models for clinical tasks, such as Medical Question Answering (QA), is severely constrained by the limited availability of high-quality, expert-annotated training data. This challenge is exacerbated by stringent privacy requirements and the impracticality of utilizing large open-source corpora or proprietary cloud APIs within resource-limited clinical settings. To address these obstacles, we introduce SAGE (\textit{Semantic Anchor-Guided Evolution}), a novel data synthesis framework that enables small, locally deployed models to generate high-quality medical training data. SAGE leverages lightweight, publicly available taxonomies such as MeSH as semantic anchors, imposing a structured prior to effectively guide and ground the data generation process. At its core, SAGE iteratively interleaves atomic (individual concept-based) and associative (relation-based) synthesis, bootstrapping training data from minimal seeds. This approach eliminates the need for large collections of medical documents or reliance on external APIs, providing a practical solution for on-premises data creation. Extensive experiments across multiple medical question-answering benchmarks demonstrate that models fine-tuned with SAGE-synthesized data consistently outperform those trained using self-derived or conventional document-based paradigms, highlighting tangible improvements in data efficiency and resource utilization for medical LLM development. Code is available at https://github.com/DIaacKr/SAGE. --- ### Explainable Rule Mining of IPv6 Extension-Header Presence Patterns from Paired-Vantage Captures Authors: Priyanka Sinha, Nikolaos Kekatos, Stylianos Basagiannis, Antonio Anastasio Bruto da Costa, Alexios Lekidis, Pabitra Mitra, Tom Nianios, Elpiniki Papageorgiou Published: 2026-10-06 Categories: cs.CR, cs.LG, cs.NI arXiv: https://arxiv.org/abs/2610.08090v1 IPv6 extension headers (EHs), such as fragmentation, segment routing, and in-situ telemetry, are operationally important yetwidely dropped in transit, and characterising their behaviour from packet captures is a recurring measurement problem. We ask whetheran explainable miner can recover human-readable rules of EH behaviour, and we contribute two reusable tools: a negative-control protocol that diagnoses whether a mined "temporal" network rule reflects genuine cross-packet dynamics or mere within-packetco-occurrence, and a sender-conditioned, per-family EH-retention measurement. Applying an interpretable temporal-logic rule miner to the JAMES paired-vantage dataset, we recover a portable Fragment-EH rule that the protocol reveals to be a within-packet,near-definitional co-occurrence rather than a temporal pattern, so the temporal-logic machinery does no work for this dominant rule;the retention measurement independently recovers the expected within-window ordering of EH observability. Our main result istherefore an honest, controlled negative finding, corroborated by executed decision-tree and large-language-model baselines: on theevaluated JAMES traces network-temporal structure does not carry the dominant Fragment-EH signal, and we supply the controls thatestablish when it would, validated on a synthetic positive control containing a genuine cross-packet dependency. --- ### When Plans Change Answers: Formalizing Cost-Accuracy Optimization for Semantic Queries Authors: Kyoungmin Kim Published: 2026-10-06 Categories: cs.DB, cs.AI arXiv: https://arxiv.org/abs/2610.08089v2 In semantic query engines, predicates are evaluated by machine-learned models, and the choice of a query plan affects not only the cost of a query but also its result. Existing systems either apply a fixed threshold to each semantic operator or tune accuracy per operator, without accounting for how errors propagate through joins. We give a formal problem definition for cost-accuracy optimization of such queries. Our starting point is the calibrated confidence that decision models such as Jev attach to each decision. It yields an expected error for every decision; weighting these errors by each decision's contribution to the output (in the simplest case, its fan-out) gives the expected output quality of a plan without any labeled data, and the same computation in reverse turns an output-level accuracy target into a price on each base or intermediate tuple. Building on this, we define an oracle semantics for relational algebra with semantic operators, physical plans as pairs of a logical plan and a decision policy, declarative output-level targets, and a hierarchy of plan equivalence. We show that accuracy is plan-invariant under pointwise-deterministic policies, and that selection pushdown is not quality-sound when escalation bands are calibrated on the plan's own candidates. Expected quality can be computed in polynomial time under bag semantics; under set semantics it follows the dichotomy of tuple-independent probabilistic databases when every relation carries a semantic predicate. Choosing which tuples to drop is NP-hard, while the optimization problem decomposes into per-tuple decisions through two Lagrange multipliers, and, with what we call confidence-centric skipping, tuples that can no longer affect the target are skipped without being scored. Simulations on a synthetic workload illustrate these effects; an evaluation on real engines is left for future work. --- ### Multi-Dataset Diagnostic Utility of Clinical Visual Concepts in AI Systems for Dermatology Authors: Linda Wermelinger, Simone Lionetti, Fabian Gröger, Nipun Ranasekara, Philippe Gottfrois, Ludovic Amruthalingam, Labelling Consortium, Marc Pouly, Alexander A. Navarini Published: 2026-10-06 Categories: cs.CV arXiv: https://arxiv.org/abs/2610.08086v1 The clinical integration of AI systems in digital dermatology relies heavily on human trust. Clinically interpretable visual concepts can act as intermediate representations enhancing trust and reliability. However, research in this domain is currently limited by scattered, heterogeneous dataset annotations. In this work, we introduce SkinLex, a harmonized dataset of 48 clinical morphological attributes across four public datasets (SkinCon, DermaCon-IN, MM-Skin, and PASSION) for a total of 20,411 records. Supervised nine-partition classification of skin conditions shows that limiting features to specific visual groups, like shapes or colors alone, reduces diagnostic accuracy. Bootstrapped backward elimination reveals that the set of 48 visual concepts has some degree of redundancy for algorithmic nine-partition diagnosis on the examined dataset. This demonstrates that coarse diagnosis on the selected dataset requires a relatively small but varied combination of clinical concepts, and motivates further research to improve concept taxonomy. Results can be translated into clinical benefits by reducing inputs for concept-based models, improving efficiency for annotation and modeling, and further enhancing interpretability. Code and prompt templates are available at https://github.com/Digital-Dermatology/SkinLex. --- ### DirectSpeech2LLM: A Simple End-to-End Framework to Mitigate Prompt Overfitting in Speech-LLMs Authors: Hemant Yadav, Sunayana Sitaram, Roger Zimmermann, Rajiv Ratn Shah Published: 2026-10-06 Categories: cs.CL arXiv: https://arxiv.org/abs/2610.08085v1 Speech-LLMs often exhibit prompt overfitting, where models solely trained on automatic speech recognition (ASR) instruction fail to generalize to new instructions such as speech translation and continue to behave primarily as ASR system. We propose DirectSpeech2LLM, a simple end-to-end framework that preserves the instruction-following ability of the LLM on unseen tasks when conditioned on speech. It computes distance-based CTC loss over the frozen LLM embedding matrix and uses greedy CTC labels to derive geometrically and temporally aligned speech embeddings respectively as an input to the LLM. Trained solely on 960 hours of LibriSpeech ASR data, DirectSpeech2LLM outperforms the cascaded system on ASR (seen task) and generalizes zero-shot to speech translation and emotion recognition (two unseen tasks), closely matching the cascaded system upper bound on these two new instructions despite seeing neither during training. We also find that geometric alignment strength plays a smaller role than previously assumed, as our modified CTC loss is shown to provide sufficient implicit geometric grounding without requiring an explicit regression loss. Results are consistent across two LLM families and scale with both more training data and model capacity. --- ### POLAR: Ontology-Guided Risk Prevention for Tool-Calling LLM Agents Authors: Yunju Kang, Seonghyeon Cho, Irene Li, Yeo-Chan Yoon, Chanjun Park Published: 2026-10-06 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2610.08082v1 LLM tool-use agents operate in dynamic environments where many actions carry operational risk. However, most safety mechanisms react only after errors manifest. Existing pre-emptive approaches either fine-tune the agent on chain-of-thought deliberation or compile natural-language guardrails into runtime checks, but they do so without exposing a structural, auditable verdict. We propose POLAR, a guardrail framework for small tool-calling agents that assesses reversibility through a structured two-layer ontology. POLAR assigns each action a graded reversibility score by deriving a candidate inverse sequence; calls failing a threshold are pruned before execution. Evaluated on $τ^2$-bench across six agent models, POLAR improves mean task reward by 0.11 to 0.18 points on airline for four of six agents, but only eight of eighteen model--domain cells improve overall; retail and stronger agents often regress. POLAR provides an auditable structural check and characterizes its task-utility trade-offs. Reward is not a direct measure of prevented harm. --- ### ProximalFM: Amortized Proximal Causal Inference under Hidden Confounding Authors: Christophe Muller, Ayub Kharel, Alex Luedtke, Chan Park, Eric Tchetgen Tchetgen, Juan L. Gamella, Rahul G. Krishnan, Ricardo Silva, Jakob Zeitler Published: 2026-10-06 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2610.08078v2 Standard causal identification methods often assume no unmeasured confounding and can fail when relevant confounders are unobserved. Proximal causal inference instead uses proxy variables to identify effects under hidden confounding. However, nonparametric proximal estimation can be challenging in practice: recovering causal estimands such as the conditional average treatment effect (CATE) requires solving an ill-posed integral equation that is data-hungry, hyperparameter-sensitive, and optimization-unstable. Bayesian inference for such models provides a desirable alternative, mitigating these difficulties by regularizing through the prior. However, computing a posterior is itself challenging, as a typical likelihood function will include latent variables. Following the recent success of tabular foundation models in backdoor, instrumental variable, and frontdoor settings, we propose that prior-data fitted networks (PFNs) are uniquely suited to resolve this bottleneck. Indeed, by training on synthetic data sampled from compliant structural causal models with access to oracle counterfactuals, we simplify the task substantially, amortizing the implied Bayesian operator inversion into a single transformer forward pass. Compared to prior literature that focuses primarily on point estimation, our model, ProximalFM, explicitly targets the Bayesian posterior distribution of the CATE. One unique aspect of this problem is that we need to provide Monte Carlo estimates of the oracle CATEs, leading to a novel variation of PFNs that accounts for the added stochastic error. Across a diverse suite of proximal regimes, ProximalFM achieves consistently strong CATE-estimation performance without dataset-specific tuning, with its largest advantage when latent confounding is substantial and the proxies are weakly informative; it also provides fast inference through a single amortized forward pass. --- ### Self-Retrospection Distillation: Turning Post-hoc Experiences into Prior Foresight Authors: Haoxiang Zhang, Qinglin Chen, Hiroaki Hayashi, Zhuofeng Li, Siming Zhang, Jiaxin Zhang, Jixuan Chen, Fang Wu, Pan Lu, Silvio Savarese, Julian McAuley, Chien-Sheng Wu Published: 2026-10-06 Categories: cs.AI, cs.CL, cs.IR, cs.LG arXiv: https://arxiv.org/abs/2610.08077v3 Reinforcement learning with verifiable rewards (RLVR) turns agent experience into learning signals primarily through scalar outcome rewards after interaction. For group-relative objectives, however, this signal vanishes when all rollouts receive the same reward, even though their trajectories may reveal useful information about what the task requires and how the agent fails. We ask a complementary question: can hindsight teach an agent what it could have anticipated before acting? We introduce prospective learning, which uses post-hoc experience to supervise foresight predictions from the pre-interaction view, and instantiate it with Self-Retrospection Distillation (SRD). Intuitively, a completed trajectory reveals knowledge that would have been useful and pitfalls that should be avoided; SRD distills this privileged hindsight into trajectory-blind foresight of the same policy. Foresight serves only as a training target and need not be explicitly generated at inference time. Across 10 tool-integrated reasoning and long-horizon agentic tasks, SRD complements RLVR and self-distillation baselines with gains of up to 24.2 pp. Its advantage is especially pronounced when reward contrast is scarce: when 37--98% of rollout groups are reward-uniform across model scales, yet SRD can still exploit learning signal from sampled trajectories. In the 2B setting, where 98% of groups are all-failure, the RLVR training ends up at 0.0% success, while adding SRD reaches 60.6% under the same rollout budget. Our results suggest that post-hoc agent experience is useful not only for evaluating or improving behavior, but also for shaping predictive representations before available interaction. Training code can be found at https://github.com/SalesforceAIResearch/SRD. --- ### SpeedrunBench: Challenging LLM Agents with Video Game Speedrunning Authors: Yoshinari Fujinuma, Keisuke Kamahori, Ryuto Koike, Abdelrahman Madkour, Varun Prashant Gangal, Monty Bichouna, Martyna Markiewicz, Shivani Jain, Duncan Curtis, Rebecca Qian, Anand Kannappan Published: 2026-10-06 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.08076v1 Frontier LLM agents have been shown to be capable of solving increasingly complex tasks for which humans have measurable solutions. This begs the pertinent question of whether LLM agents can go beyond what humans have already solved. The ability to develop sophisticated strategies to tackle consequential problems becomes paramount as well-trodden, human-developed solutions become insufficient for problems for which we lack context or enough training data. We study agents' capability of such strategy formation through the communal practice of video game speedrunning. In speedrunning, practitioners compete to find the fastest way to complete a video game under certain conditions, and in so doing uncovering interesting unorthodox play styles that require a thorough understanding and mastery of the underlying game mechanics. We introduce SPEEDRUNBENCH, a benchmark that evaluates frontier LLM agents across 9 different games. To perform well in this benchmark, agents must repeatedly improve their strategy, reflect on their performance, exploit their gained knowledge, and reason across a long-horizon of actions to improve on an increasingly difficult problem: being faster than themselves and everyone else. Our experiments show that while frontier agents approach human world records in simple platformer games, they remain behind human performance on longer, more complex games under practical budgets. These results suggest that SPEEDRUNBENCH is a useful testbed for studying agents' strategy formation capabilities as well as being a saturation-resistant evaluation measure, as there is almost always a faster completion time waiting to be discovered. --- ### Optimization Encoders: Rethinking Second-Order Meta-Learning for Neural Fields Authors: Rudolf L. M. van Herten, Soufiane Ben Haddou, Rachit Saluja, Johannes C. Paetzold Published: 2026-10-06 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2610.08075v1 Conditional neural fields represent signals continuously, but their effectiveness depends on how the conditional latent representations are inferred from observed data. In meta-learning, this encoding occurs through gradient updates induced by the decoder, tying representation learning directly to decoder design. We formalize this connection by interpreting latent optimization as an optimization encoder, unifying the roles of second-order differentiation, latent parameterization, and task supervision. This concept enables second-order meta-learning for end-to-end training of the encoding procedure alongside the decoder, and clarifies which learning pathway first-order approximations discard. Guided by this view, we introduce Attentive Latent Fields (MetaLF), an equivariant transformer-based neural field that contextualizes a latent pointcloud through self-attention. These interactions shape both field predictions and the updates that construct their representation, allowing local observations to inform coherent non-local structure. Disentangling the inner encoding objective from outer task supervision unifies reconstruction, classification, and segmentation within an end-to-end meta-learning framework, using reconstruction-only latent adaptation at test time. Controlled experiments on polynomial fields link latent coordination to lower effective rank and stronger alignment with the underlying function space. Across image and 3D shape reconstruction, MetaLF improves fidelity within three to five gradient updates, while supporting semantic prediction across images, shapes, and volumes. Together, these findings position the optimization encoder perspective as a unified basis for designing neural fields around how representations are constructed, coordinated, and used. --- ### Two Halves are More than One: Phase-wise Velocity Distillation for Fast and High-Quality Image Generation Authors: Zhen Guo, Rongyuan Wu, Qiaosi Yi, Chenxi Xie, Xinyu Wei, Lei Zhang Published: 2026-10-06 Categories: cs.CV arXiv: https://arxiv.org/abs/2610.08070v1 Recent diffusion-based image generation backbones have grown substantially in scale, making the network inference cost increase rapidly. While diffusion distillation techniques can reduce the number of inference steps, high-quality image generation within a single full-backbone-forward compute budget remains challenging. Existing one-step methods typically allocate this budget to a single evaluation of a monolithic student. However, approximating the heterogeneous coarse-to-fine transport with a single monolithic mapping is difficult and often leads to over-smoothed outputs. To address this issue, we propose Phase-wise Velocity Distillation (PVD), which partitions the generation timeline into a coarse and a fine phase, and models the transition within each phase via the average velocity. A dedicated half-sized expert is assigned to each phase, decoupling structural composition from detail refinement while keeping the cumulative computation equivalent to one full-backbone forward pass. We show that the use of two half-sized phase-specific experts outperforms a single full-size monolithic student. On class-conditional image generation, PVD achieves an FID of 1.48 on ImageNet 256 x 256. On more complex text-to-image (T2I) tasks, PVD-distilled models (Stable Diffusion 3.5-Medium, FLUX.1-dev, Qwen-Image) produce results competitive with their multi-step teachers, significantly outperforming prior distillation methods. Moreover, across the evaluated T2I backbones, PVD reduces active parameters by 49.10-50.89% and peak VRAM by 45.76-48.36% compared to the corresponding teachers. Source code and distilled models are available at https://github.com/PolyU-VCLab/PVD. --- ### Detecting a Shift Is Not Enough: Exact Minimax Limits of Linear Representation Repair Authors: Anuar Aimoldin, Yankai Chen, Ayana Mussabayeva, Xue Liu Published: 2026-10-06 Categories: cs.LG, math.ST, stat.ML arXiv: https://arxiv.org/abs/2610.08069v1 A mean shift between two data sources can be easy to detect but hard to remove without substantially changing their representations. We cast its removal as a statistical decision problem: from noisy differences between paired calibration measurements in $\mathbb{R}^d$, learn one linear map, applied to both sources under a hard distortion budget, that leaves as little of the shift as possible on fresh data. We derive the exact finite-sample minimax risk over all such maps, $(d-k) \mathbb{E}(1/(d+2J))$ with $J\sim\mathrm{Pois}(κ/2)$, where the budget allows deleting $k$ directions and $κ$ is the calibration signal-to-noise ratio. Projecting out the mean calibration difference attains it without knowing $κ$ or the noise scale. This exposes a detection-repair gap: detecting the shift needs only $κ\gg\sqrt d$, whereas removing a fixed fraction of it at constant distortion needs $κ\asymp d$, as for estimating its direction. Standard linear concept erasers (MP, SAL, LEACE) remove the same calibration difference, so the formula gives, before fitting, exactly how much shift they leave on fresh data and how much calibration a target requires. The limit is robust: pairing keeps it exact for non-Gaussian shared content, the projection keeps its guarantee under anisotropic noise, and selective abstention cannot close the gap. On paired clinical and wearable sleep EEG, where differences between participants act as calibration noise, the formula predicts the device shift left in new participants, and more recordings per person soon stop helping. Together, these results tell whether a correction that falls short needs a better method, more recordings, or more participants. --- ### PhysTacGen: Physics-Aware Visual-Tactile Sensor Image Generation Authors: Guo Tang, Yongtao Wang Published: 2026-10-06 Categories: cs.CV arXiv: https://arxiv.org/abs/2610.08068v1 Realistic physical interaction is a cornerstone of embodied intelligence, yet collecting paired visual--tactile data remains costly. Visual-to-tactile synthesis offers a promising approach to augmenting such data, but learning this mapping is complicated by the gap between visual appearance and contact-related material properties, as well as spatial misalignment in paired observations. To address these challenges, we present \textbf{PhysTacGen}, a visual-to-optical-tactile image generation framework that integrates material-aware descriptions with geometric conditioning. First, we introduce Group Tactile Policy Optimization (GTPO), a reinforcement learning strategy that refines a vision--language model to generate structured material descriptions using task-specific rewards. Second, we combine DINOv2-based pair curation with monocular relative-depth estimation to select training pairs and provide geometric priors. Finally, an SDXL ControlNet synthesizes optical tactile images conditioned on RGB, relative depth, and GTPO-generated text. Experiments on curated SSVTP data demonstrate improved structural similarity over the compared baselines, while a blinded user study shows a preference for GTPO-generated descriptions. Generated tactile inputs also improve performance on an attribute-derived force-coefficient prediction proxy. Together, these results demonstrate the effectiveness of PhysTacGen for optical tactile image synthesis and its utility in the evaluated downstream task.The code will be available at https://github.com/VDIGPKU/PhysTacGen. --- ### HINTT Submission to the 2nd MLC-SLM Challenge: Comparing Cascaded and Unified Approaches to Diarization and ASR Authors: Takanori Ashihara, Kohei Matsuura, Masato Mimura Published: 2026-10-06 Categories: eess.AS, cs.CL arXiv: https://arxiv.org/abs/2610.08063v1 This paper presents the HINTT system submitted to the 2nd Challenge and Workshop on Multilingual Conversational Speech Language Model (MLC-SLM). We address multilingual speaker-attributed ASR, where systems must determine who spoke when and what was spoken. We investigate two modeling strategies for this problem: a cascaded pipeline that combines speaker diarization with speech-LLM-based ASR, and a unified speech LLM that directly generates speaker labels, timestamps, and transcriptions. Our final submission is based on the cascaded pipeline, consisting of a fine-tuned DiariZen diarization model, a fine-tuned Qwen3-ASR model, and LLM-based generative error correction. For comparison, we also fine-tune VibeVoice-ASR as a unified model using the same official training data. All task-specific fine-tuning and model selection are performed using only the official MLC-SLM data, without external data or pseudo-labels. Experimental results demonstrate that the cascaded system remains more reliable under the MLC-SLM Task 1 conditions, while unified speech LLMs offer a promising direction for future speaker-attributed ASR. --- ### Language Carries the Expert's Impression: Instrument-Anchored LLM Judges Transfer Counseling-Quality Assessment and Beat In-Domain Training Authors: Tobias Hallmen, Elisabeth André Published: 2026-10-06 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2610.08055v1 Automatic assessment of communication quality in dyadic counseling conversations is bottlenecked by data: expert-rated corpora are small and expensive to grow. We study cross-domain transfer of expert overall-impression prediction across three German corpora of simulated counseling (two general-practice medical, one school-related parent-teacher; $n=195$ expert-rated sessions, one corpus after scale equating). Training on the other domains beats training in-domain: leave-one-domain-out transfer reaches nested Spearman $ρ= 0.54$ against $\le 0.48$ within the target domain, a paired session-level gap of $+0.15$ that holds at $+0.12$ when the training-set sizes are matched, so it is not simply data volume. The decisive features are session-level construct scores from small open-weight LLMs reading the two-speaker transcript, with the constructs largely derived from the experts' rating instruments: the instrument-derived battery lifts a single judge from $0.32$ to $0.41$ over generic dialogue qualities, judges from three model families ensemble to $0.51$ language-only, and a nonverbal-dyadic block adds $+0.03$ more, not separable from noise at this sample size. We also price the recording setup: one corpus lost its per-speaker audio, 16% of its diarised segments carry the wrong speaker, and repair is worth $+0.07$ there. At practically attainable corpus sizes, the expert's overall impression is carried by what is said, and by other communication programs' data more than by one's own. --- ### A Riemannian Geometry for Low-rank Adaptation Authors: Shoichiro Takeda, Shin'ya Yamaguchi, Satoshi Suzuki, Yasunori Akagi Published: 2026-10-06 Categories: cs.LG arXiv: https://arxiv.org/abs/2610.08049v1 Low-rank adaptation (LoRA) is widely used as a parameter-efficient fine-tuning technique for pre-trained deep neural networks, which approximates the weight update via full fine-tuning by a low-rank matrix $BA^\top$. This parameterization leads to the equivalence relation $(B, A) \sim (BG^{-1}, AG^\top)$ for any invertible matrix $G$ because $BA^\top = BG^{-1}(AG^\top)^\top$ and thus both pairs yield the same loss value. This relation induces a quotient manifold where matrices $(BG^{-1}, AG^\top)$ for all $G$ are identified, eliminating redundant directions along which the loss value remains unchanged. To respect the geometry of this manifold, the original search space is endowed with a Riemannian metric that is invariant under the equivalence relation. Such a metric induces preconditioning at each gradient step and ensures that each weight update via LoRA changes the loss value, leading to efficient optimization. In this paper, we propose a new Riemannian metric that is specifically tailored to LoRA to close the gap to full fine-tuning at the weight level. We theoretically show that LoRA with our preconditioning induced by this metric satisfies the following two properties at each iteration: (i) The weight update follows the direction closest to the gradient of full fine-tuning within the subspace of first-order weight changes allowed by the LoRA parameterization. (ii) The updated weight matrix is closer in Frobenius norm to that of full fine-tuning than the updated weight matrices of LoRA with conventional preconditioning and without preconditioning. These theoretical insights suggest that our preconditioning makes LoRA better approximate full fine-tuning, thereby leading to more efficient optimization. Experiments show the effectiveness and efficiency of our preconditioning for LoRA on fine-tuning tasks with language and vision domains. --- ### DAEDALUS: Bootstrapping Agent Memory from Self-Generated Tasks Authors: Antoine Edy, Max Conti, Victor Xing, Marc-Antoine Allard, Nawfal Benhamdane, Gautier Viaud Published: 2026-10-06 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2610.08048v1 LLM agents often lack the operational knowledge to act reliably in new environments, as they must discover specific tool behaviors or environment conventions on their own. Without memory of past attempts, they repeat the same mistakes across tasks, leading to more task failures and longer trajectories. To address this, agentic systems typically rely on human-written guidelines or on procedural memory built from training tasks and an oracle verifier, both of which require prior knowledge of the environment. We present DAEDALUS, a method for bootstrapping reusable agent memory from self-generated practice without existing tasks or oracle verifiers. DAEDALUS pairs two agents: an explorer that interacts with the environment to generate challenging yet solvable tasks, and a solver that attempts them. A heuristic is derived from each solver failure and accepted only after the solver repeatedly succeeds with that heuristic in context. These outcomes also provide feedback for the explorer to refine the difficulty of future tasks. Accepted heuristics are then consolidated into a memory bank for test-time use. Across AppWorld, $τ^2$-bench, and AutomationBench, DAEDALUS improves mean success rates by up to 15.9 points and pass^5 by up to 2.2x over a no-memory baseline, and is competitive with methods using training tasks, at a lower inference cost than most. We show that performance gains already emerge with a small exploration budget, and that its heuristics also benefit agents from other model families. Our ablations further reveal that solver traces provide the key information needed to derive effective heuristics, while factorizing early discoveries makes exploration more cost-efficient. Beyond memory construction, we find that the tasks generated by DAEDALUS can serve as a proxy for benchmark tasks when ranking models by performance. Code and artifacts: www.github.com/illuin-tech/daedalus. --- ### SepsisLens: Structure-Preserving Sequence Modelling for Decomposable Early Sepsis Warning Authors: Yikun Ou, Wei Li Published: 2026-10-06 Categories: cs.LG arXiv: https://arxiv.org/abs/2610.08046v1 Early sepsis warning from ICU records can be cast as a structure-preserving prediction problem. A model needs to detect deterioration from irregular measurements while keeping each alert connected to the physiological signals that support it. Many temporal models fuse clinical variables into a patient-level representation, supporting scalar risk prediction but weakening the structure needed for clinical decomposition. We present SepsisLens, which preserves variable-indexed temporal states until risk composition. Observation-aware representations encode each variable's dynamics and measurement history, while a shared temporal encoder models each trajectory without collapsing the variable axis. The StructuredRiskHead composes multi-horizon risk from explicit variable-level and organ-level components. We evaluate SepsisLens on three public ICU cohorts and one private-hospital cohort under a common pre-onset protocol. SepsisLens achieves strong discrimination on all four cohorts and lower alert burden at matched event recall on MIMIC-IV. Structural ablations support the design, while input-side masking shows that the ranked components reflect variables with greater influence on prediction. --- ### Are Language Models Script-Aware? Authors: David Kletz, Sandra Mitrović, Ljiljana Dolamić, Fabio Rinaldi Published: 2026-10-06 Categories: cs.CL arXiv: https://arxiv.org/abs/2610.08037v1 Language models frequently generate outputs in unintended languages or scripts, a phenomenon known as off-target generation. While existing research has focused on language selection, the dimension of script knowledge remains understudied: before any linguistic understanding can occur, users must recognize the graphic symbols in a model's response. We investigate whether Small and Large Language Models (SLMs and LLMs) possess script knowledge by testing them on multi-scriptic languages. Through two complementary experiments, we evaluate whether models (1) adapt their output script to match the input, and (2) follow explicit instructions to generate text in a specified script. The models we tested demonstrate substantial script knowledge: they all achieve a near-perfect Latin script fidelity (more than 98%) and follow script instructions with high frequency. Nevertheless, we notice differences between LLMs and SLMs, with higher scores for LLMs including for non-standard script combinations. --- ### Same Feedback, Different Answer: Measuring Run-to-Run Instability in Frontier-Model Customer Feedback Analysis Authors: Viraj Bagal, Raviraja Ganta, Prabhath Chellingi Published: 2026-10-06 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.08036v1 AI agents are increasingly being programmed to automate knowledge work over large collections of unstructured data. Such automation requires repeatability: when the underlying evidence is unchanged, the agent's categories, priorities, and counts should not shift materially between runs, even if each individual answer appears plausible. We introduce a repeat-run evaluation framework that aligns semantically equivalent categories and focuses on two operating metrics: theme churn, the normalized change in the returned category set, and volume disagreement, the change in counts for categories that persist. We evaluate three recurring customer-feedback tasks across eight frontier models, corpus sizes from 100 to 5,000 records, multiple prompts, and three execution designs: raw generation, taxonomy-free hierarchical decomposition, and a taxonomy-grounded agent (TGA) using persistent themes, subthemes, and record-level predictions. With Claude Opus 4.8 and the 1,000-record corpus fixed, TGA reduces theme churn by 86--88% relative to both raw generation and hierarchical decomposition, while matched-theme volumes have zero disagreement. The taxonomy-grounded agent is more stable than every raw model in the screen, remains more stable at each corpus size, and keeps this advantage when theme matching is made stricter or looser. Although evaluated on customer feedback, the framework targets repeated synthesis of unstructured corpora more broadly, including financial reports, legal documents, incident records, and scientific literature. Overall, these results show that taxonomy grounding produces more consistent and repeatable outputs for recurring knowledge work. --- ### Learning in Dreams, Winning in Reality: A Continuous Dyna Loop for a Ten-Hero MOBA Authors: Jordy Kieto Published: 2026-10-06 Categories: cs.AI arXiv: https://arxiv.org/abs/2610.08033v1 World models are usually judged from the inside: by prediction loss, by the return a policy earns in imagination, or by how convincing their frames look. We judge one from the outside. We learn a structured, multi-agent world model of a complete ten-hero MOBA (206 units, every hero acting every tick, games of up to 6,000 ticks), train a policy only inside it with 1,400-tick free-running imagined episodes, and measure that policy in the real game against the opponent the game ships with. The real game never provides a gradient; it provides the policy's own games as training data for the world model, and an online evaluation that selects and anchors the policy. Run as a continuous asynchronous Dyna loop, the policy wins 70.2% of real games as radiant (421 of 600; 95% CI 66.4-73.7) on seeds never used for any decision, up from 0% for dream training alone and 33.7% before the loop. It wins none as dire, and neither does the shipped opponent when it plays itself. Four findings explain the result. Model exploitation is invisible from inside the dream: every unanchored run collapsed within a few updates while no in-dream metric tracked the collapse. A world model that is accurate on its training corpus is badly wrong on the policy's own games, and Dyna repairs it there, which is worth +9.2 points of real win rate with the policy recipe held fixed. Finally, the policy inherits its world model's fidelity profile mechanic by mechanic: the model represents the macro game but not crowd control, cast timing or lethality, and the policy wins by map-wide pressure with almost no coordinated fighting. We release the world model, the dream-PPO harness, a world-model debugger, the evaluation protocol, and every policy and log. --- ### The Labeling Problem in Hallucination Detection Benchmarks: An Empirical Evaluation Authors: Jorma Valjakka, Juhani Kivimäki, Juha Mylläri, Jukka K. Nurminen Published: 2026-10-06 Categories: cs.CL arXiv: https://arxiv.org/abs/2610.08026v1 In recent years, several methods for detecting when large language models (LLMs) hallucinate have been developed. These methods are often benchmarked with open-domain question answering (QA) datasets containing questions and corresponding short reference answers. First, an LLM is used to generate answers to questions within the QA dataset. Then, some automated labeling strategy is used to label these answers as hallucinated or not by comparing them with the reference answers in the dataset. This evaluation setting creates a methodological ambiguity between two criteria: reference faithfulness (whether the answer is fully supported by the reference) and factual correctness (whether the answer is free from contradictions and factually false specific claims). In practice, automated labelers may apply the former criterion even when the intended target is the latter. We study this potential criterion mismatch using 900 human-labeled question-answer pairs spanning three commonly used QA datasets and three generator models, with labels targeting answer-level factual correctness. We evaluate lexical similarity metrics, a reference-entailment NLI baseline, and seven LLM judges under controlled prompt variants as automated labelers. Our experiments reveal substantial disagreement both among automated labeling strategies and between these labels and human annotations. Many strategies also exhibit strong directional error biases, and for most judge-generator pairs, replacing a faithfulness-oriented prompt with a factual-correctness prompt improves agreement with human annotations and reduces false-positive dominance, indicating that automated hallucination labels depend strongly on how the target criterion is specified. Label-source choice should therefore be considered a fundamental part of benchmark design and made explicit, validated, and matched with the benchmark goal. --- ### Spectra: Exact Component Transport for Test-Time Prior Adaptation in Simulation-Based Inference Authors: Xin Zhao, Nico Scherf, Robert Trampel, Kerrin J. Pine, Nikolaus Weiskopf Published: 2026-10-06 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2610.08021v1 Simulation-based inference (SBI) has become a powerful approach to Bayesian inference in complex scientific models whose likelihoods are difficult or impossible to evaluate. Amortized SBI learns reusable inference models from simulated data, enabling rapid posterior inference for new observations, and modern generative models have made these models increasingly expressive. However, this reuse is limited to the prior distribution chosen during training, whereas scientific analyses often need revised priors as knowledge accumulates or alternative assumptions are tested. We introduce Spectra, a test-time adaptation method for diffusion-based SBI. Spectra uses an exact score-transport identity to obtain the adapted score from a frozen diffusion model in closed form for structured prior changes, without additional simulation or training. Across six SBI benchmarks, Spectra achieves accurate adaptation under strong prior shifts at low online sampling cost. This enables pretrained SBI models to incorporate updated prior information at test time. --- ### Learning consistent molecular mechanics force fields from first principles Authors: Berkay Günes, Leif Seute, Jigyasa Nigam, Frauke Gräter Published: 2026-10-06 Categories: physics.chem-ph, cs.LG, physics.comp-ph arXiv: https://arxiv.org/abs/2610.08020v1 Classical force fields (FFs) remain the workhorse for large-scale simulations even as machine-learned interatomic potentials (MLIPs) approach ab initio accuracy. They decompose total configuration energies into simple effective interactions whose parameters are traditionally assigned based on atom or bond types, enabling efficient simulations but also limiting their ability to adapt across configurations. Recent machine learning approaches have improved the accuracy and transferability of bonded parameters in these FFs by inferring them as functions of local atomic environments, but still rely on empirical nonbonded parameters for practical simulations. In this work, we introduce a unified approach, \texttt{grappa-fullFF}, which learns both bonded and nonbonded parameters \emph{consistently} and simultaneously from ab initio reference data. By incorporating physically inspired regularization via supervision of the electrostatic potential and an architecture that facilitates charge equilibration, our model recovers accurate electric response properties, achieves state-of-the-art accuracy on geometry optimization benchmarks, and reproduces the conformational sampling of both classical and existing machine-learned FFs, without relying on externally assigned nonbonded parameters. --- ### Structured but Silent: Probing Capability Requirements in LLM Hidden States Authors: Kyojun Choo, Minsoo Song, Yunju Kang, Chanjun Park Published: 2026-10-06 Categories: cs.CL arXiv: https://arxiv.org/abs/2610.08018v1 Reliable tool use requires more than triggering a mechanism or matching a query to an API description. Before selecting a specific tool, an agent must first infer the capability requirements implied by the user query. In this paper, we investigate whether these query-side capability requirements are linearly decodable from LLM hidden representations prior to generation, and how this hidden-state accessibility compares with explicit verbal classification. We introduce TACIT, a framework that decomposes external requirements along three fundamental axes: Source, Transformation, and World Effect, defining eight structurally distinct capability classes. Using 1,600 balanced training queries from benchmarks, synthetic examples, and new domain scenarios, we train linear probes on pre-generation hidden states from four open-weight LLM families. Our empirical results demonstrate that fine-grained capability structures are linearly decodable with high accuracy across all models. Crucially, however, we expose a representation-to-verbalization gap: these same models are significantly less reliable when asked to explicitly classify the same queries in natural language. This disconnect indicates that information about required external capabilities is linearly accessible in LLM hidden representations but not reliably expressed, a phenomenon we define as "structured but silent." --- ### FOSLS-deRhaNN: native de Rham neural classes for H(div) and H(curl) with applications to first-order system least-squares neural network methods for partial differential equations Authors: Shun Zhang Published: 2026-10-06 Categories: math.NA, cs.LG arXiv: https://arxiv.org/abs/2610.08016v1 We construct neural approximation classes native to the graph spaces H(div) and H(curl), in two and three dimensions and, for H(div), in any dimension. Every realization lies in the space for all parameter values, and with kinked potentials, such as ReLU networks, the admissible jumps appear at finite width. The classes are images of scalar and componentwise networks under fixed operators of the de Rham complex, and do not involve a mesh or finite element emulation. For H(div) in R^n two native classes are given on an equal footing, with a skew-symmetric potential $A$: $\mathrm{Div}\,A+R_nq+\mathbf{h}$, with the divergence $q$ as an explicit unknown, and $\mathrm{Div}\,A+\mathbf{z}$ with an $H^1$ field $\mathbf{z}$; for H(curl) the analogous classes are $\mathrm{grad}\,φ+Sr+\mathbf{h}$ in two dimensions and $\mathrm{grad}\,φ+\mathbf{z}$ in two and three dimensions. In all of them every interface jump of the field is carried by the potential term, $\mathrm{Div}\,A$ or $\mathrm{grad}\,φ$, while the remaining part has no interface jump (it is an $H^1$ field in the regular-decomposition classes); the classes with $\mathbf{z}$ are the componentwise approach enriched by this term. Known or learned interface geometry enters the potential through factors with trainable amplitudes, and the remaining part if the divergence jumps. The classes lead to the FOSLS-deRhaNN method, first-order system least squares with de Rham neural networks, whose loss is the least-squares functional posed in the natural spaces of the weak formulation; for elliptic equations this includes $H^{-1}$ right-hand sides and $H^{1/2}$ Dirichlet data. Elliptic equations with discontinuous coefficients and curl-curl problems are treated as instances, with the functional equivalent to the error; linear transport with discontinuous solutions and conservation laws with shocks use the same flux classes. --- ### Can phenotypic activity be predicted without experimental readouts? Authors: Télio Cropsal, Rocío Mercado Published: 2026-10-06 Categories: cs.LG arXiv: https://arxiv.org/abs/2610.07997v1 Molecular encoders contrastively pretrained on paired molecule-morphology data, such as CLOOME and CellCLIP, have been proposed as cheap surrogates for phenotypic prediction, avoiding the need to run a Cell Painting assay. We evaluate this idea for these molecular encoders under a protocol designed to control for two confounds that can inflate apparent performance: leakage across an encoder's own pretraining boundary, and the correlation between phenotypic activity and cytotoxicity. Testing six representations, including a non-pretrained MLP control matching CLOOME's input and layer count, on two distinct Cell Painting screens, we find that once these confounds are controlled for, the pretrained molecular encoders show no clear advantage over plain physicochemical descriptors, and that toxicity is generally easier to predict than phenotypic activity across representations. Our results suggest leakage-aware, confound-controlled evaluation should be standard practice before phenotype-pretrained encoders are trusted as surrogates for phenotypic drug discovery. --- ### TICDA: Tabular In-Context Data Attribution Authors: Yacine Benihaddadene, Milan Bhan, Eliot Dugelay, Mohammed Jawhar, Benjamin Wong, Nicolas Chesneau, Duong Nguyen Published: 2026-10-06 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2610.07996v1 Tabular foundation models (TFMs) achieve strong predictive performance by conditioning on labeled demonstrations provided in context, without any parameter update. Yet how individual demonstrations shape a given prediction remains poorly understood. This gap matters in practice: the context is often assembled from whatever labeled data is available, potentially leading to the inclusion of mislabeled, redundant, or low-quality examples that degrade performance. Standard data attribution methods do not transfer to the TFM setting: resampling-based approaches such as DemoShapley require a combinatorial number of forward passes, and gradient-based estimators such as influence functions require computing training point's effect on the model parameters, which in-context learning never updates. We introduce TICDA, a method that measures the influence of every demonstration in the context directly from linear surrogates trained on TFM latent embeddings, in a single forward pass and at negligible cost. We show that TICDA offers the best compromise against competitors across four tasks: detecting labeling errors, curating context to preserve predictive accuracy while lowering inference cost, producing attribution scores that transfer across TFMs, and supporting an acquisition strategy for efficient active learning. ---