# Research Papers — 2026-09 · bundle page 260 of 266 > Full-text markdown of papers #12951–13000 of 13298 published in 2026-09, newest first. AI-readable mirrors of the full 13298-paper 2026-09 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-09/page/260 [← Previous bundle](https://aipod.dev/papers/archive/2026-09/page/259.md) [Next bundle →](https://aipod.dev/papers/archive/2026-09/page/261.md) ### When Tokenization is Secretly Output Supervision Authors: Tanja Baeumel, Josef van Genabith, Simon Ostermann Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01386v1 Tokenization in language models is treated by default as an input preprocessing decision. We argue that this framing is incomplete: in autoregressive models, tokenizer granularity determines what the model must resolve in a single forward pass, and therefore the supervision signal it receives. This affects both the difficulty of the learning problem and the representations that emerge inside the model. We test this in a controlled experiment on numeric reasoning with a novel decoupling of input and output tokenization. As the output supervision view predicts, differences in task performance, training dynamics, and model internals are induced by output tokenization and largely invariant to input tokenization. This may matter in practice, because models with different tokenization strategies differ not only in input representation but in the task they were trained on. Comparisons between models may thus partly reflect task definition rather than ability. A survey of 120 recent *CL papers on numeric reasoning confirms that this is rarely acknowledged: only about 10% report the numeric tokenization of the models they evaluate, while 69% compare across tokenization, and thus supervision, regimes without reporting it. While prior work documents that tokenization consistently affects model performance, there is no principled account of why. We argue that framing tokenization as output supervision provides that account. --- ### InSight: A Benchmark for Agentic Claim Verification in Interactive Visualizations Authors: Maeve Hutchinson, Syed Mahbubul Huq, Mohammad Albinhassan, Radu Jianu, Aidan Slingsby, Pranava Madhyastha Published: 2026-09-01 Categories: cs.CL, cs.CV, cs.HC arXiv: https://arxiv.org/abs/2609.01383v1 Vision Language Models have demonstrated remarkable proficiency in interpreting static visual artifacts, but modern data analysis is inherently dynamic, requiring the active interrogation of interactive environments. Existing benchmarks are predominantly constrained to static imagery and one-shot question answering and fail to capture the epistemic demands of this domain, where evidence is frequently occluded, distributed across linked views, or conditionally revealed through user agency. In this paper, we introduce InSight, a benchmark for agentic claim verification over interactive visualizations. The dataset consists of 21,349 claims derived from human-authored analytical narratives and grounded in fully interactive web-based environments. Agents must navigate these environments to determine whether a natural language claim is supported, refuted or not verifiable given the available evidence. Unlike traditional evaluations, InSight treats interaction traces as intrinsic proxies for reasoning, enabling a rigorous audit of how models seek and synthesize visual evidence. We evaluate state-of-the-art models, revealing that interactive verification remains a non-trivial challenge. We release InSight at https://github.com/maevehutch/insight. --- ### Polish ModernBERT: The Long and Short of Polish Language Understanding Authors: Michał Perełkiewicz, Sławomir Dadas, Rafał Poświata, Małgorzata Grębowiec Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01379v1 Encoder-only Transformers remain effective for discriminative and representation-learning tasks, yet Polish encoders still largely rely on BERT/RoBERTa-style architectures. We introduce \textbf{Polish ModernBERT}, a family of four Polish encoders available at Base and Large scales, each with 512-token and 8K context variants. We adapt the ModernBERT pretraining recipe through staged selection experiments and release a long-context benchmark covering legal topic classification, ideological decision-direction prediction, factual-consistency assessment over literary plot summaries, and human-rights violation assessment. Across 30 tasks, Polish ModernBERT achieves the best overall performance among the evaluated Polish encoders, reaching 83.99 and 85.11 for the Base-8K and Large-8K models, respectively. On long-context tasks, the 8K variants improve over matched Polish RoBERTa-8K baselines from 67.47 to 77.15 and from 75.88 to 78.49 at the Base and Large scales, respectively. The Base-8K model achieves this gain with 22\% fewer parameters (149M vs.\ 190M). Efficiency measurements in representative inference setups show lower peak memory usage and latency than matched Polish RoBERTa baselines in both 512-token and 8K settings. Polish ModernBERT-8K-Base additionally achieves the best result on a Polish retrieval benchmark among the evaluated encoders below 300M parameters. --- ### IntroConformal: Conformal Factuality Guarantees for Large Vision-Language Models via Introspective Signals Authors: Md. Atabuzzaman, Christian Alexander, Chris Thomas Published: 2026-09-01 Categories: cs.CV, cs.CL arXiv: https://arxiv.org/abs/2609.01375v1 Large Vision-Language Models (LVLMs) have achieved strong multimodal performance, yet ensuring the factual correctness of generated content remains challenging. Existing methods that provide statistical guarantees on factuality typically rely on external verifiers or generation-time confidence signals, which introduce auxiliary dependencies or often fail for confident but incorrect outputs. We argue that reliable factuality control can instead be achieved through introspective signals derived from the model itself. We introduce IntroConformal, a training-free Conformal Risk Control (CRC) framework that provides finite-sample, distribution-free factuality guarantees. We first instantiate it with layer-wise semantic stability, a conformity score derived from hidden-state representations, and then propose verification probability, a stronger score capturing the model's self-administered judgment on claim factuality. Across multiple LVLM architectures, IntroConformal satisfies the conformal risk guarantee while substantially reducing abstention and achieving competitive or superior claim-level discrimination relative to external verifier-based baselines. --- ### Behaviorally Effective LoRA Writes Are Sparse and Structured Authors: Haruto Sato, Yuki Tanaka, Ren Nakamura, Aoi Kobayashi, Mei Ito Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01374v1 Low-rank adaptation fixes the rank of the update, but it does not identify which parts of a trained write actually carry behavior. We study that question directly and show that behaviorally effective LoRA writes are sparse, structured, and far more concentrated than the raw low-rank parameterization suggests. We use Learned-Basis LoRA, a learned-basis continuation recipe, to expose that structure. The recipe warms up an unconstrained adapter, converts its learned write columns into a module-wise orthonormal basis, freezes that basis, and continues training inside the constrained parameterization. Across 14 exact switches from unconstrained to constrained form, held-out accuracy is unchanged at the conversion step and reconstructed write matrices differ by at most 0.25% relative Frobenius error. Same-state continuation then shows that the same trained checkpoint develops differently under different write subspaces, establishing write geometry as a causal state variable. A no-retraining projection test shows that useful write signal stays inside the learned write space and largely disappears from random or frozen-activation PCA controls. The concentration pattern is strong at both local and global scales. Across GSM8K, MathQA, and AQuA, per-module top-k continuation reaches its optimum at k in {2, 4} in all twelve seed-level cases we test. A stricter global ranking test shows that learned top-16 and top-32 subsets outperform matched random subsets, especially on GSM8K/Qwen and MathQA/Qwen. Single-direction ablations further reveal a sparse set of late q_proj, o_proj, and down_proj components with outsized behavioral impact. --- ### Diffusion Based Unpaired Data Learning for Inverse Problems Authors: Chenglong Bao, Yiming Dang, Chenguang Duan, Yuling Jiao, Defeng Sun Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01370v1 Data is important in many deep learning-based inverse problem solvers. However, obtaining sufficient paired data in many scenarios remains highly challenging, while unpaired data is cheap. To maximize data utilization, this paper proposes LUD-DIF, a diffusion-based approach for solving inverse problems with unpaired data. Starting from the evidence lower bound (ELBO) of the joint distribution, we decouple it into two independent diffusion processes under the weak-coupling assumption. The method provides theoretical support from a variational inference perspective, derives the loss function, quantitatively analyzes the error bound introduced by the assumption, and offers a theorem-motivated heuristic for hyperparameter selection. Experimental results demonstrate that LUD-DIF achieves outstanding performance on multiple image inverse problems, validating its effectiveness and generalization capability in unpaired inverse problem settings. --- ### How Correct Is Your Answer? A Semantic Correctness Framework for Open QA Evaluation Authors: Elitsa Yotkova, Violeta Kastreva, Petar Velkov, Hristo Boyanov, Dimitar Dimitrov, Ivan Koychev, Preslav Nakov Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01369v1 Reliable evaluation of open-ended question answering remains a bottleneck for measuring answer correctness of modern LLMs. Unlike multiple-choice tasks, free-form answers may be correct in many surface forms and may fail in qualitatively different ways, including incompleteness, contradiction, overgeneration, and endorsement of false premises. Existing judgment-based and similarity-based metrics often collapse these distinctions. We address this gap with three reusable contributions. First, we introduce a semantic correctness taxonomy that assigns open-ended answers to eight ordered classes, separating verbose-but-correct answers from those contaminated by hallucinated content. Second, we release CAP-Correctness, an 8.8k-example benchmark spanning widely used QA datasets, and CAP-Statements, an 11k-example dataset for converting question-answer pairs into declarative statements for natural language inference (NLI) training and statement-based evaluation. Third, we introduce CAP (Context-Aware Precision), a reference-based metric that scores question-conditioned statements using bidirectional NLI. Under a monotonicity protocol testing whether metrics respect the taxonomy's intended ordering, CAP outperforms established baselines. --- ### Accurate Reconstruction of Gas Turbine Blade Geometry Using 3D/2D Rigid Registration and CT View Optimization Authors: Hristo Valtchanov, Nicolas Piché, Vladimir Brailovski, Justin Byers, Catherine Désrosiers, François Guibault Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01368v1 Non-destructive X-ray and computed tomography (CT) testing are essential for ensuring the dimensional accuracy of manufactured components with complex internal structures, such as the cooling channels in gas turbine blades, which directly affect thermal performance and service life. This study presents a multipart 3D-2D rigid registration approach for aligning CAD models with X-ray projections as an alternative to CT reconstruction for part inspection and measurement. A greedy registration algorithm sequentially aligns the blade's exterior before registering its internal components by maximizing the mutual information between simulated and acquired X-ray images. This stepwise approach reduces problem complexity and improves alignment accuracy. View angles are optimized using a greedy method that iteratively selects angles to minimize dimensional measurement errors. The results indicate that a small number of oblique views provides the best accuracy, although a broad range of angles yields acceptable results. The method achieves subpixel registration accuracy, with errors below one-fifth of the magnified detector-pixel pitch. Image noise and defects reduce registration precision, but direct registration in projection space mitigates these effects compared with CT reconstruction. Appropriate view selection can therefore preserve acceptable subpixel accuracy in the presence of image noise and defects. --- ### Investigating Linear Probe Robustness to Linguistic Register, Medical Specialty, and Corpus Shifts in Medical QA Authors: Nishant Mishra, Ameen Abu-Hanna, Iacer Calixto Published: 2026-09-01 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.01361v1 Linear classifiers trained on hidden states of a large language model (LLM), linear probes, can flag factual errors from a single forward pass. Geometrically, that implies that true and false statements separate along a stable direction in hidden state space, i.e., the truth direction. Prior work disagrees on whether this generalises across input shifts, but the disagreement is hard to interpret because cross-dataset probe transfer experiments confound several kinds of input change at once. We isolate three such variables in medical question-answering (QA): writing style (register), domain (medical specialty), and corpus (dataset). We build a benchmark using 500 MedQA entries, each rewritten into four styles (textbook, patient, clinical note, colloquial), annotated with clinical specialty, and grouped with two other exam corpora, MedMCQA and MMLU-medical, for cross-dataset evaluation. Probing four open-weight LLMs (2--8B), we find that the truth direction is largely robust to writing style (mean $Δ_\text{register} \approx 0.10$ AUROC on held-out facts) and to medical specialty ($Δ_\text{specialty} \approx 0.03$), but degrades unevenly across corpora: by $0.12$ AUROC on MMLU-medical and by $0.21$ on MedMCQA, roughly twice the register gap. The register result replicates with a second generator and carries over to human-written patient questions. The truth direction is therefore largely stable within the medical domain but breaks under some corpus shifts, and question format does not explain the break, which suggests that the signal a linear probe recovers is partly bound to dataset structure rather than to medical knowledge alone. --- ### EDGE: Error Dependency Graph-Guided Multi-Error Attribution in Multi-Agent LLM Systems Authors: Jun Hou, Priya Pitre, Yi Fang, Xuan Wang Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01360v1 Large language model (LLM) agent failures often contain multiple related errors rather than a single mistake. Existing attribution methods usually identify a responsible agent, step, or root cause, but do not explicitly model dependency between errors. We introduce EDGE, an Error Dependency Graph-guided multi-Error attribution framework. EDGE constructs an error dependency graph from observed error events and validates a reliable causal subset through counterfactual rollout. The inference graph guides a two-stage LLM-as-judge detector for error attribution, and the intervention-validated subgraph provides a more reliable basis for explanation and repair analysis. Experiments on TRAIL and MAST show that EDGE improves category-level multi-error attribution across most evaluated models and settings. Experiments with adapted Who&When-style prompts show that the graph helps across prompting strategies. These results suggest that dependency structure is a useful diagnostic prior for agent failures beyond isolated root-cause prediction. --- ### PopPert: Population-level Joint-Distribution Modeling for Single-Cell Perturbation Prediction Authors: Handong Wang, Jiaxin Qi, Haochen Feng, Baisheng Lai Published: 2026-09-01 Categories: q-bio.GN, cs.AI arXiv: https://arxiv.org/abs/2609.01357v1 Predicting transcriptional responses to specific perturbations is critical for understanding cellular regulatory mechanisms and accelerating drug discovery. Single-cell RNA sequencing destroys each measured cell, yielding only unpaired populations of control and perturbed cells. However, existing methods typically model perturbation prediction at the single-cell level and assume cell-to-cell correspondence, which conflicts with the unpaired nature of the observed data. To address this challenge, we propose PopPert, a framework that explicitly parameterizes population-level joint gene expression distributions for collective transcriptional state modeling. Given a control population distribution and a perturbation condition, PopPert predicts perturbation-induced changes in distribution parameters, eliminating the need for cell-level correspondence and reducing sensitivity to single-cell noise. To effectively capture gene co-expression patterns, PopPert leverages a low-rank Gaussian Copula to model cross-gene statistical dependencies and construct the joint gene expression distribution, additionally allowing sampling of synthetic perturbed single-cell profiles. Across multiple single-cell benchmarks spanning both genetic and chemical perturbations, PopPert achieves superior overall performance in differential expression recovery, perturbation effect estimation, and population-level distribution matching. These results establish population-level joint distribution learning as an effective paradigm for predicting transcriptional responses from unpaired single-cell populations. Code for PopPert is publicly available at https://github.com/whd1125/PopPert. --- ### Separating Syntax from Language: A Mechanistic Account of Translation in Multilingual LLMs Authors: Mikhail Sonkin, Tanja Baeumel, Daniil Gurgurov, Josef van Genabith, Simon Ostermann Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01356v1 Multilingual large language models (mLLMs) achieve strong performance in machine translation, yet our understanding of the mechanisms by which they transform representations from one language to another remains incomplete. Prior work suggests that translation decomposes into separable processes within an mLLM, where conceptual content is first represented independently, followed by a production into language-specific form. In this work, we show that translation is even more modular than previously assumed and that the output language production in translation processes is actually further separable into a syntax and a surface language process. We construct controlled multilingual datasets that isolate cross-linguistic differences in word-order and use causal interventions and probing to track how representations are transformed during translation. We find that models first construct target-side word-order before realizing the target language surface form. We identify individual attention heads that are selectively sensitive to syntactic transformations while remaining largely invariant to language identity. These results establish the commitment to a syntactic structure as an independent stage in translation, extending prior decompositions and showing how translation is implemented by functionally different components within mLLMs. --- ### Exact Risk-Complexity Laws for Projective Boundaries in Scenario Optimization and Distribution-Free Certification Authors: Giuseppe C. Calafiore Published: 2026-09-01 Categories: eess.SY, cs.LG, math.OC arXiv: https://arxiv.org/abs/2609.01355v1 Scenario optimization, conformal prediction, and related distribution-free certification methods use finite samples to construct decisions or prediction sets with violation-risk guarantees for fresh observations. In several classical settings, the conditional violation risk follows an exact beta law, whose tail has a beta-binomial representation and whose parameter is a support, calibration, or compression dimension. This paper identifies the deterministic boundary mechanism behind these formulas and derives the corresponding law when the observed boundary size is random. A decision rule is represented by an acceptance set for future observations, together with a boundary map selecting the sample points responsible for that set. The resulting pair is called a {\em proper projective boundary scheme} when held-out samples are accepted precisely if the full-sample boundary is retained, and accepted non-boundary samples can be deleted without changing that boundary. For every such scheme, the conditional law of the violation risk given the observed boundary size is determined by the boundary's cross-sample complexity profile. A stable profile yields the usual beta law, whereas a varying profile produces an exact profile correction. The framework covers scalar order-statistic calibration, support-reconstructive scenario programs, cascaded support-removal certificates, coordinatewise envelopes, and Pareto-frontier calibration with vector scores. It also yields conditional probabilistic certificates and a no-go result explaining why observed complexity alone is insufficient. --- ### Where the Verifier Fails: A Category-Level Audit of Reward Signals in RLVR Authors: Esther Xin Published: 2026-09-01 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.01354v1 Reinforcement learning with verifiable rewards (RLVR) and standard benchmark evaluation both rely on an automatic verifier that turns a free text answer into a binary reward. Prior work reports that one evaluation harness accepts only about 94% of its own ground truth answers, blaming LaTeX parsing. That is an aggregate: it does not say which answer forms consume the error budget. We supply the decomposition. We apply metamorphic testing to the verifier rather than the model, generating certified equivalent answer variants, that is, rewrites that preserve mathematical meaning by construction, so that any rejection is a provable false negative needing no human adjudication. We then measure rejection per answer category across four widely used verifiers over 307,420 verdicts. We find three things. (1) Self validation ranges from 53.8% to 95.2% on identical inputs, a spread of 41.3 points. The published figure describes one implementation, not the task; two configurations of the same library disagree on 49.9% of pairs. (2) The residual is not spread across parsing categories but concentrated in whitespace and punctuation, which account for 93.0% of in contract failures for the default LaTeX configuration. A trailing period or newline dominates the budget. (3) Separating rejection from execution failure shows that verifiers with similar aggregate error fail for opposite reasons, and that a reference numeric cascade accepts off by one wrong answers as a step function of magnitude, from 0% below 10^4 to 100% at or above, because its relative tolerance is scale invariant. --- ### SymFold: Synergizing Evolutionary and Structural Priors for Accurate Protein Inverse Folding Authors: Handong Wang, Jiaxin Qi, Baisheng Lai, Jianqiang Huang Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01353v1 Protein inverse folding aims to recover amino acid sequences for a given 3D protein structure, underpinning broad applications such as enzyme engineering and drug discovery.Current methods often follow a serial pipeline, in which a structure encoder predicts a coarse sequence, which is then refined by protein language models (PLMs). However, because PLMs only perform post-hoc sequence edits, the refinement is bounded by the quality of upstream predictions.Thanks to recent multimodal protein language models (MPLMs), we could directly encode structure to generate sequences with pretrained structural knowledge, but we observe that they are not effective for inverse folding. Therefore, we introduce a symmetric dual-path architecture that both leverages PLMs for pretrained sequence evolution knowledge and MPLMs for pretrained structural knowledge to iteratively guide protein sequence generation.Through extensive experiments across standard protein inverse folding benchmarks, our method achieves state-of-the-art performance, surpassing prior approaches, and ablation studies validate the rationale of our symmetric design, revealing a promising direction for the community. --- ### CHARM: Character Hallucination for Multicultural Role Play Benchmark Authors: Sunkyung Han, Nahyeon Park, Gaeun Seo, Seunghyun Yoon, JinYeong Bak Published: 2026-09-01 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.01352v1 Role-playing large language models (LLMs) are expected to adopt a character's style while also respecting that character's knowledge boundaries. Prior evaluations detect character hallucination but rarely distinguish whether errors arise from failure to recognize a boundary or from failure to comply despite recognition. We introduce CHARM, a multicultural benchmark of 40 real and fictional characters drawn from five cultural-linguistic regions, and validated by native reviewers. It probes two boundary types, Temporal (historical vs. modern) and Cross-Universe (entities outside a character's narrative or historical universe), using abstention-enabled multiple-choice questions. We propose a two-stage evaluation that separates Boundary-Awareness (explicit recognition that a query is out of scope) from Boundary-Compliance (abstention when answering concrete questions). Evaluations across six LLMs show that hallucination is driven predominantly by compliance failures. Models frequently acknowledge that a query lies outside the character's knowledge yet still provide factual, out-of-character answers. By re-posing the same questions to the target character, we confirm that a large fraction of these cases are verified parametric overrides; the model stores the relevant fact but fails to suppress it. We also observe systematic cultural variation in these failures, consistent with imbalances in how characters from different regions are represented in model knowledge. --- ### Scalable Rao-Blackwellized Online Planning for High-Dimensional POMDPs Authors: Jiho Lee, Nisar Ahmed, Kyle Hollins Wray, Zachary Sunberg Published: 2026-09-01 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2609.01351v1 Online planning under uncertainty remains a fundamental challenge for robotic systems operating in partially observable environments with high-dimensional state spaces. While sampling-based POMDP solvers enable approximate decision-making in large or continuous domains, their performance degrades as belief dimensionality increases due to the high variance inherent in Monte Carlo-based estimation. In this work, we extend the Rao-Blackwellized online POMDP (RB-POMDP) framework to improve its generalizability in high-dimensional settings through hybrid continuous-discrete belief representations. By analytically propagating uncertainty associated with marginalized state components during tree-based planning, the proposed approach reduces sampling-induced variance in value estimation. We demonstrate the effectiveness of this framework in a robotic search-and-rescue task by integrating it with FastSLAM 2.0. Experimental results show that the proposed planner achieves higher cumulative rewards using significantly fewer particles and planning simulations than purely sampling-based methods under equivalent computational budgets. These results suggest that structured high-dimensional robotic problems admitting tractable sufficient statistics can be effectively leveraged within the RB-POMDP framework for computationally feasible online decision-making. --- ### Cheap Verifiers, Large Blind Spots: Measuring the Reliability Cost of Cost-Saving Cascades Authors: Dushyant Rajput, Nirdesh Chauhan, Siddharth Kosaraju Published: 2026-09-01 Categories: cs.AI, cs.CR, cs.LG arXiv: https://arxiv.org/abs/2609.01345v2 Inference cascades cut cost by answering most queries with a cheap model and escalating a hard tail to a frontier model that acts as verifier. A natural extension closes the loop: fine-tune the cheap student on the verifier's rejections so the escalation rate, and cost, fall each round. We measure this loop on real LLMs and report four findings. First, the verifier's blind spot, the fraction of the student's wrong answers it accepts, is large and moves adversarially: it grows with student capability ($β$ from 0.12 to 0.55 as the student scales 0.5B to 32B) and shrinks with verifier capability, so it is worst in the cheap-student, cheap-verifier regime cascades exist to create. Second, buying it away returns the saving: a frontier verifier drives $β$ to about 0.05 but then escalates on 46% of hard-MATH queries against a 39% true error rate, paying the frontier price on nearly half of all traffic. Third, naive corrective fine-tuning on the verifier-rejected tail does not improve the small student but degrades and ultimately collapses it, across every teacher we tried (cross-family and same-family), so at this scale the self-improving loop is self-defeating. Fourth, through all of this the cascade's own dashboard, every metric computed through the verifier, reads a flat 3% error while true delivered error swings up to 32%: the system is blind to its own degradation by construction. We then give the theory that explains the blindness, a two-population conservation law, $ε_\infty \lesssim q_0 β_0$, under which every in-loop metric improves while true quality does not, and a synthetic study that validates the mechanism. The practical conclusion: the reliability of a self-improving cascade cannot be read from any metric computed through its own verifier. --- ### Cheap Verifiers, Large Blind Spots: Measuring the Reliability Cost of Cost-Saving Cascades Authors: Dushyant Rajput Published: 2026-09-01 Categories: cs.AI, cs.CR, cs.LG arXiv: https://arxiv.org/abs/2609.01345v1 Inference cascades cut cost by answering most queries with a cheap model and escalating a hard tail to a frontier model that acts as verifier. A natural extension closes the loop: fine-tune the cheap student on the verifier's rejections so the escalation rate, and cost, fall each round. We measure this loop on real LLMs and report four findings. First, the verifier's blind spot, the fraction of the student's wrong answers it accepts, is large and moves adversarially: it grows with student capability ($β$ from 0.12 to 0.55 as the student scales 0.5B to 32B) and shrinks with verifier capability, so it is worst in the cheap-student, cheap-verifier regime cascades exist to create. Second, buying it away returns the saving: a frontier verifier drives $β$ to about 0.05 but then escalates on 46% of hard-MATH queries against a 39% true error rate, paying the frontier price on nearly half of all traffic. Third, naive corrective fine-tuning on the verifier-rejected tail does not improve the small student but degrades and ultimately collapses it, across every teacher we tried (cross-family and same-family), so at this scale the self-improving loop is self-defeating. Fourth, through all of this the cascade's own dashboard, every metric computed through the verifier, reads a flat 3% error while true delivered error swings up to 32%: the system is blind to its own degradation by construction. We then give the theory that explains the blindness, a two-population conservation law, $ε_\infty \lesssim q_0 β_0$, under which every in-loop metric improves while true quality does not, and a synthetic study that validates the mechanism. The practical conclusion: the reliability of a self-improving cascade cannot be read from any metric computed through its own verifier. --- ### ExBind: A Controlled Diagnostic Benchmark for Visual-to-Executable Correspondence Authors: Ziqian Wang, Yuxiao Cheng, Tingxiong Xiao, Jinli Suo Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01344v1 Multimodal coding and editing systems must map a visible or semantic referent to the exact executable object that can be edited. A wrong reference may select a valid but incorrect DOM node, SVG element, graph endpoint, hierarchy member, or table cell, while final execution success alone does not reveal the source of the failure. ExBind isolates this visual-to-executable correspondence layer as a controlled diagnostic benchmark between semantic localization and action execution. It samples representation-independent latent binding instances and compiles them into SVG, DOM, canvas, tree, graph, and table cases with deterministic mappings to executable references. Models output only a strict reference; the evaluator maps predictions back to latent structure and scores structural constraints without requiring reasoning traces. The release contains a 250-case broad suite, a disjoint 240-case targeted suite, and 50 paired latent groups. Qwen2.5-VL-3B achieves 98.4% candidate validity but 76.4% exact accuracy, while Qwen3-VL-4B achieves 100.0% validity and 98.8% exact accuracy. In the targeted table suite, all Qwen2.5-VL-3B residual errors are valid correct-row/wrong-column selections. Candidate-order perturbations change case-level outcomes while preserving this error pattern. ExBind is designed for controlled diagnosis rather than population-scale ranking or end-to-end editing evaluation. Code and benchmark records are available at https://github.com/Daerwang2020/Exbind and https://huggingface.co/datasets/Ziqianwwww/ExBind. --- ### SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers Authors: Shaowen Wang, Ge Zhang, Kairong Luo, Yuhao Wu, Shaofan Liu, Jiaheng Liu, Wenhao Huang, Shen Yan, Jian Li Published: 2026-09-01 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.01343v3 Looped Transformers increase effective depth by iterating a shared block of layers, but most evaluations compare at fixed model size, conflating architectural advantage with extra FLOPs. We study looping on Mixture-of-Experts Transformers while closely matching per-token FLOPs, total non-embedding parameters, and KV cache. Through a series of ablations, we arrive at a recipe we call SMELT (Sparse MoE Transformer, middle layers Loop Twice), which loops the middle half of layers twice while matching the unlooped Baseline on all three budgets. We scale SMELT across four sizes up to 54B non-embedding parameters and fit a separate Chinchilla-style scaling law for each architecture. SMELT's loss drops faster with compute, saving 6.8--18.0\% of training FLOPs on the compute-optimal frontier. The advantage transfers to downstream benchmarks beyond what validation loss predicts, is largest on Code, and grows with sample length and the number of in-context examples. Mechanistic analysis shows that the second visit reduces the attention sink and redirects mass toward content-relevant tokens, an inductive bias that may underlie the observed performance gains. These results show that looping can improve Transformers even under budget matching, offering a practical recipe that turns depth reuse into measurable gains. --- ### SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers Authors: Shaowen Wang, Ge Zhang, Kairong Luo, Yuhao Wu, Shaofan Liu, Jiaheng Liu, Wenhao Huang, Shen Yan, Jian Li Published: 2026-09-01 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.01343v1 Looped Transformers increase effective depth by iterating a shared block of layers, but most evaluations compare at fixed model size, conflating architectural advantage with extra FLOPs. We study looping on Mixture-of-Experts Transformers while closely matching per-token FLOPs, total non-embedding parameters, and KV cache. Through a series of ablations, we arrive at a recipe we call SMELT (Sparse MoE Transformer, middle layers Loop Twice), which loops the middle half of layers twice while matching the unlooped Baseline on all three budgets. We scale SMELT across four sizes up to 54B non-embedding parameters and fit a separate Chinchilla-style scaling law for each architecture. SMELT's loss drops faster with compute, saving 6.8--18.0\% of training FLOPs on the compute-optimal frontier. The advantage transfers to downstream benchmarks beyond what validation loss predicts, is largest on Code, and grows with sample length and the number of in-context examples. Mechanistic analysis shows that the second visit reduces the attention sink and redirects mass toward content-relevant tokens, an inductive bias that may underlie the observed performance gains. These results show that looping can improve Transformers even under budget matching, offering a practical recipe that turns depth reuse into measurable gains. --- ### Probing Factual Knowledge Transfer with Training Data Interventions Authors: Romina Oji, Marc Braun, Marcel Bollmann, Marco Kuhlmann, Jenny Kunz Published: 2026-09-01 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.01341v1 Do multilingual language models transfer factual knowledge across languages during continued pretraining, or do they mostly recall facts learned directly from the target-language data? To answer this question more reliably, we propose an intervention-based framework: starting from an English-pretrained model, we continue pretraining on Persian data from which specific facts have been systematically removed at varying levels of granularity. We construct SIFT, a resource of 500 triples across 20 relations, stratified by the cultural origin of each fact's subject into general (globally prominent) and Persian-related entities, designed for both systematic fact removal from training data and evaluation, with natively written Persian cloze templates. Our results show that fact transfer is very limited: under the strictest removal condition, a large majority of English-acquired facts fail to transfer into Persian. We further show that sentence-level co-occurrence removal is insufficient to eliminate fact signal, and that easier (randomly selected) negative candidate sets substantially inflate apparent transfer by rewarding shallow associative heuristics, while performance on a harder candidate set that allows for less reliance on heuristics is much lower. Finally, we show that source-language entity frequency has a large influence, with Persian-related facts, which are orders of magnitude rarer in the English corpus, hardly transferring. --- ### mzCache: On-Device LLM Memory Management under Multitasking Authors: Hongseung Yu, Minsung Kim, Jongseok Park, Kyunghan Lee Published: 2026-09-01 Categories: cs.OS, cs.DC, cs.LG arXiv: https://arxiv.org/abs/2609.01338v1 On-device mobile Large Language Model (LLM) inference is gaining significant attention. However, mobile devices operate in highly dynamic multitasking environments where users frequently switch between applications. This creates memory pressure, forcing LLM memory (model weights and KV cache) to be evicted by the operating system. When a new inference request arrives, the inference system must restore the evicted memory through slow storage reads or recompute the entire KV cache, severely degrading responsiveness. To address this, we present mzCache, an on-device LLM inference system with specialized memory management for multitasking environments. Under unpredictable memory pressure, mzCache elastically evicts LLM memory and leverages the unified memory of mobile SoCs to enable zero-wait inference on the GPU with concurrent CPU-side restoration. mzCache realizes this through restoration-oriented memory management: LLM memory is partitioned into fine-grained shared buffers to enable partial eviction and restoration with concurrent cross-processor access, while hybrid swap and backward-out eviction policies ensure low-latency restoration from any eviction state. Implemented on llama.cpp and deployed as an Android application, mzCache achieves 2.1-5.5$\times$ reduction in Time-to-First-Token compared to storage-backed partial offload and demonstrates its effectiveness in real multitasking scenarios. --- ### LEAP: Likelihood Elicitation and Aggregation for LLM-based Probabilistic Forecasting Authors: Yufei Chen, Yiran Zhao, Xiaogang Xu, Qipeng Xie, Jiafei Wu, Zhe Liu Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01337v1 LLM-based forecasting systems have improved on real-world tasks such as financial markets and sports outcomes, largely through stronger search and tool use. Many systems still ask an LLM to read all collected evidence together and produce the final forecast. We call this design Monolithic Prediction. It can obscure how individual evidence items affect the result and collapse uncertainty across competing outcomes. We propose LEAP (Likelihood Elicitation and Aggregation for Probabilistic forecasting), which reorganizes how collected evidence is used in the prediction stage. LEAP examines each evidence item separately and elicits likelihood parameters that describe its implications for the target. An explicit prior and a deterministic probabilistic model then combine these likelihoods into a posterior distribution. This procedure supports continuous, single-choice, and multi-choice forecasts while preserving reproducible evidence contributions. We build a benchmark covering forecasting, information-seeking, and browsing tasks, and evaluate LEAP on our own agent loop and several agent CLI frameworks. Given the same evidence, LEAP improves most prediction and calibration metrics across models and remains stronger under controlled comparisons of prior access, inference budget, and aggregation. --- ### Bandits in Prod: Hyperparameter Optimization at Inference Time Authors: Louis Abraham, Tuan-Anh Nguyen, Nicolas Devatine Published: 2026-09-01 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.01335v2 Many production systems can assess a configuration only by using it on live requests and observing noisy feedback. Modern agentic systems are a prominent example, with inference-time choices such as model selection, retrieval depth, prompting strategy, and decoding temperature, yet often with no representative validation data. We formalize this setting as Online Hyperparameter Optimization (OHPO) and cast it as an infinitely many-armed bandit over mixed and conditional search spaces. We introduce IMABO, a general framework that combines any bandit policy for choosing among already sampled configurations with any oracle for proposing new ones. We instantiate it with IMOSS, a restart-free anytime policy whose active set grows as $t^β$, and prove an expected cumulative quantile-regret bound of $O(p_ρ^{-1/β} + T^{(1+β)/2})$, where $β\in(0,1)$ controls active-set growth and $p_ρ$ lower-bounds the probability that a proposed configuration falls in the top-$ρ$ fraction of the search space. We combine IMOSS with three practical oracles: a Tree-structured Parzen Estimator, an incumbent-mutation oracle driven by a per-coordinate bandit, and a pretrained tabular foundation model, all three improving over the uniform random oracle baseline. IMABO outperforms all baselines in terms of regret across diverse OHPO settings, from tuning classical machine-learning models to configuring LLM-based agents. Our implementation is available at https://github.com/Tiime-Software/IMABO. --- ### VerTox: Verifiable Reward-Guided Corpus Poisoning Against Neural Ranking Models Authors: Zhiqi Huang, Vivek Datla, Zhichao Xu, Puxuan Yu, Vivek Srikumar, Alfy Samuel Published: 2026-09-01 Categories: cs.CL, cs.IR arXiv: https://arxiv.org/abs/2609.01325v1 Neural ranking models have become core components of modern information retrieval systems and important building blocks of AI systems such as retrieval-augmented generation (RAG) pipelines. However, their robustness remains insufficiently understood in the presence of large language models (LLMs), which can generate fluent and deceptive content at scale. This work investigates the vulnerability of neural ranking models to corpus poisoning attacks, in which an adversary injects a small number of maliciously crafted documents into the corpus to distort ranking behavior. We propose VerTox, the first framework to formulate corpus poisoning as a verifiable reward-guided reinforcement learning (RLVR) problem. By explicitly coupling ranking distortion with factual corruption through specialized reward shaping, we fine-tune compact LLMs into adversarial generators. Experiments demonstrate that our method achieves near-perfect attack success rates, producing adversarial documents that frequently rank higher than target documents across major neural ranking architectures, as well as a proprietary commercial embedding model. The generated adversarial documents are fluent and exhibit low perplexity, making them difficult to detect. Furthermore, by explicitly encouraging factual corruption, our adversarial documents significantly degrade the performance of a downstream RAG application. --- ### Exploring Sparse Autoencoders in Text-Based Causal Confounding Adjustment Authors: Mian Zhong, Katherine A. Keith, Anjalie Field Published: 2026-09-01 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.01322v1 In many settings, studying causal questions based on text data requires adjusting for confounding information within texts. Yet there is a tradeoff in constructing text representations for adjustment: they must be sufficiently large and/or dense to preserve the confounding variables necessary for unbiased effect estimation, but sufficiently small and/or sparse to satisfy finite-sample overlap and yield low-variance estimates. To address this tradeoff, we turn to sparse autoencoders (SAEs), and propose a novel causal adjustment pipeline that iteratively selects a minimal set of SAE features via conditional independence tests. We find that SAE representations achieve better adjustments (lower bias and and higher coverage) than alternative representations in standard semi-synthetic evaluations with binary confounders, and their interpretability offers opportunities for falsification. We also introduce a more realistic semi-synthetic evaluation that uses multi-label data as the unobserved confounders and find off-the-shelf adjustment methods require increased investigation for these more complex settings. Code: https://github.com/mianzg/sae-text-confounder --- ### Automated Event Log Generation from Unstructured Text Using Finetuned LLMs Authors: Maximilian Seeth, Gabriel Marques Tavares, Daniel Schuster Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01320v1 Process mining (PM) provides a powerful framework for discovering and optimizing operational processes from event data. However, the efficacy of PM techniques is strictly predicated on the availability of structured event logs. Thus far, event logs have often been laboriously created by domain and process mining experts. This costly effort causes large portions of organizational knowledge, including incident tickets, manuals, and textual reports, to remain underutilized. We address this bottleneck by investigating the efficacy of Large Language Models (LLMs) as automated data translators. We propose a scalable framework that leverages LLMs as data translators to bridge the gap between unstructured textual resources and structured event data. We finetune LLMs on a newly created text-to-log dataset, demonstrating that the resulting models can extract high-fidelity event logs from unstructured resources. Our results show that this finetuning approach outperforms few-shot or zero-shot prompting by a large amount, highlighting finetuning as a necessary pre-condition for generating reliable event data. We conclude that our method provides a promising pipeline for making previously unused data available to process mining ecosystems, effectively expanding the possibilities of using PM to further investigate organizational workflows. --- ### Matched Queries for Curvature and Density at Branching Junctions Authors: Ziqi Zhao, Qingjian Ni Published: 2026-09-01 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2609.01319v1 At a junction, a score field can reveal weighted tangent rays, yet these first-order quantities do not determine how individual branches bend or how their densities change away from the center. Recovering this missing information is necessary for describing local continuation beyond a single point, but finite observations must separate branchwise second-order effects while allowing error in the estimated center. We address this inverse problem using matched score queries at noise scales $σ$ and $λσ$. For a finite union of $C^{2,α}$ half-branches in $\mathbb{R}^D$, the normalized score has the expansion $F_σ=F_0+σG+O(σ^{1+α})$. Matched subtraction cancels the tangent contribution and exposes $G$, which depends linearly on branchwise curvature and log-density slope. Given tangent directions and weights on distinct rays, $G$ uniquely identifies all $sD$ branch parameters, and $sD$ scalar component observations are necessary. An $O(σ^2)$ center error introduces $D$ translation modes, leading to $(s+1)D$ observations under full-rank calibration, except for a translation-invariant full line. We also establish a perturbation bound and a conditional kernel-density-estimation rate. Experiments reproduce the predicted population and $N^{-1/5}$ trends and remain full rank up to $D=20$ with 16 supplied branches. In end-to-end tests for $D=3$--$5$, a known-count first-order frontend yields full rank in all 135 population systems and a median relative jet error of 0.132. With strong first-order error, matched responses reduce median parameter error by a factor of 49.4 relative to naive tangent subtraction. --- ### Reliability Challenges in Diffusion Vision-Language Models Authors: Md. Atabuzzaman, Chris Thomas Published: 2026-09-01 Categories: cs.CV, cs.CL arXiv: https://arxiv.org/abs/2609.01318v1 Diffusion-based Large Vision-Language Models (dLVLMs) have recently emerged as a compelling alternative to autoregressive (AR) LVLMs, offering advantages in parallel decoding, bidirectional context, and controllable generation. Despite rapid progress, their reliability properties remain largely uncharacterized. We present the first systematic reliability evaluation of hallucination and bias in dLVLMs, benchmarking six diffusion models against competitive AR baselines across four dimensions. Our key findings are: (1) dLVLMs reverse the yes-bias of AR models in binary visual queries; (2) they achieve competitive hallucination rates yet exhibit degraded linguistic quality; (3) they collapse to near-zero accuracy on underrepresented racial groups with opposite-polarity gender bias; and (4) they exhibit accuracy collapse in multiple-choice settings when the correct option is shorter than its distractors, associated with a length prior that emerges at the first denoising step. Tokens committed at late denoising steps with low confidence further correlate with hallucinated content, pointing to a mechanistic signal unique to diffusion generation. These patterns vary across model families, suggesting reliability is shaped by the generative paradigm together with training data. --- ### MIDR: Enrichment-Augmented Indexing for Multimodal Document Retrieval Authors: Debanjan Mahata, Atharva Tendle, Daniel Preotiuc-Pietro, Yong Zhuang, Ozan Irsoy Published: 2026-09-01 Categories: cs.IR, cs.AI, cs.CL, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.01316v1 Retrieval over visually rich documents has a representation problem: important content often lives in tables, charts, figures, and layout relations that plain OCR linearizes, corrupts, or omits. ColPali-family visual retrievers address this with patch-level multi-vector indexes and late-interaction scoring, keeping image-derived retrieval on the query-time serving path. We introduce MIDR (Multimodal Indexing for Document Retrieval), a training-free framework for enrichment-augmented indexing that shifts multimodal reasoning to index time. During ingestion, a multimodal LLM converts rendered pages into verified textual fields that are indexed with BM25F and optionally fused with dense retrieval, enabling text-centric serving over multimodally grounded evidence. On ViDoRe V3, MIDR Hybrid achieves 0.6219 average nDCG across five English domains, a 23.0% relative gain over BM25, remaining competitive with ColQwen2.5. On two French-document domains, enrichment bridges English queries and French page text, lifting BM25 from 0.1532 to 0.5448 nDCG and outperforming ColQwen2.5. Across all seven domains, MIDR leads ColQwen2.5 on four while using approximately 9x smaller index memory and approximately 2x lower query latency. These results establish index-time multimodal reasoning as a compelling accuracy-deployment alternative to serving-time visual late interaction. --- ### A Composable Evaluation System for Reproducible Omni-Modal Foundation Model Evaluation Authors: Hodong Lee, Sanghee Park, Dohoon Ryu, Jungwhan Kim, Junyeob Kim, Soyoon Kim, Geewook Kim Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01315v2 Building an omni-modal foundation model means evaluating it across text, image, video, and audio. Excellent evaluation toolkits exist for each modality, but their inference engines, prompt conventions, and metric implementations are mutually incompatible, so practitioners end up maintaining separate environments for every toolchain and still struggle to compare results across them. OmniEvaluator grew out of this need in our own model development: rather than reimplementing benchmarks, it connects existing inference engines and curated evaluation libraries at a higher level, exposing four inference backends, four evaluation frameworks, and over a thousand benchmarks through a single interface. Every run is recorded as an artifact capturing the full configuration for exact reproduction, and results flow into a shared dashboard for cross-model comparison. A federated mode shares GPU inference servers across concurrent evaluations, and a built-in verifier, small enough to run on CPU, keeps its score stable across engines and prompts where rule-based scoring fluctuates under configuration mismatch, matching cost-efficient commercial LLM judges without their recurring API cost. The system, demo video, and dashboard are publicly available. (https://github.com/naver-ai/omni-evaluator) --- ### One-Layer Transformer Provably Learns Multiclass One-Nearest Neighbor in Context Authors: Skanda Athreya, Yutong Wang Published: 2026-09-01 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2609.01311v1 We extend recent work establishing an equivalence between one-layer transformers and nearest-neighbor classifiers in the binary setting to the multiclass case. By leveraging the simplex encoding, we show that one-layer transformers with an argmax classification head behave identically to a one-nearest-neighbor classifier in the multiclass setting. This closes a gap left by prior work, whose multiclass result relied on a non-standard rounding-based approach rather than the typical argmax head used in practice. --- ### GazeRefine: Expert Gaze as a Test-Time Prompt for Training-Free Medical Image Segmentation Authors: Mohammed Oussama Benyahia, Marouane Tliba, Mohamed Amine Kerkouri, Taifour Yousra, Bin Wang, Max Bengtsson, Gorkem Durak, Elif Keles, Zuheng Ming, Marek Penhaker, Azeddine Beghdadi, Ulas Bagci et al. Published: 2026-09-01 Categories: eess.IV, cs.AI, cs.CV, cs.HC, cs.LG arXiv: https://arxiv.org/abs/2609.01310v1 Medical image segmentation remains difficult to scale because high-performing methods typically rely on dense expert annotations and task-specific training. We introduce GazeRefine, a training-free framework that uses gaze as an inference-time prompt for zero-shot medical image segmentation. Sparse, duration-weighted fixations are converted into foreground and background priors that initialize semantic prototypes in frozen DINOv3 feature space. These prototypes are iteratively refined through foreground-background discrimination, feature-space affinity propagation, and anchoring to the initial gaze guidance, allowing segmentation to extend beyond directly fixated regions while limiting semantic drift. GazeRefine requires no segmentation masks, fine-tuning, adapters, prompt encoders, or gradient updates. We evaluate the method on gaze-annotated polyp segmentation and prostate MRI segmentation. The results show strong performance on colonoscopy images and competitive performance on prostate MRI, supporting gaze-guided prototype refinement as a promising approach for segmentation-label-efficient, human-in-the-loop medical image segmentation. Our tools and code can be found in the following repository: https://github.com/MohammedOussamaBEN/GazeRefine.git --- ### CMRVision: A Foundation Model for Cardiac MR Image Analysis Authors: Athira J. Jacob, Puneet Sharma, Daniel Rueckert Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01308v1 Cardiac magnetic resonance (CMR) imaging provides complementary information on cardiac anatomy, function, and tissue characterization across multiple sequences and views. In this work, we investigate foundation model pretraining for 2D CMR and introduce CMRVision, a CMR-specific foundation model trained using DINOv3-style self-supervised learning on a multi-center, multi-sequence cohort of 36 million CMR images. We systematically evaluate architectural and training design choices for domain-specific pretraining. CMRVision is evaluated on two downstream tasks: multi-task segmentation across cine, late gadolinium enhancement (LGE), and mapping sequences, and cine view classification. Our experiments show that CMR-specific pretraining, smaller patch sizes, and patch-level objectives consistently improve downstream performance. Across a multi-task segmentation benchmark, CMRVision achieved the strongest overall performance, outperforming prior natural-image (NI), medical-image, supervised, and CMR foundation model baselines. Improvements were modest but consistent across structures and sequences, with Dice scores ranging from 0.940-0.967 for LV and 0.855-0.905 for myocardium, and reaching 0.929 for RV, 0.920 for LA, and 0.931 for RA. The largest gains were observed for myocardium segmentation in LGE and mapping images. In a zero-shot segmentation task on unseen LGE long-axis views, the model achieved an average Dice score of 0.692, demonstrating cross-view generalization. For cine view classification, CMRVision achieved the highest average accuracy (0.906), compared to prior methods reported in the literature. These results highlight the potential of CMRVision to support robust and generalizable cardiac MRI analysis across multiple sequences and views. --- ### MeshSplatBench: A Unified Benchmark for Triangle- and Mesh-Based Neural Rendering Authors: Kaixuan Zhang, Minxian Li, Mingwu Ren, Xiatian Zhu Published: 2026-09-01 Categories: cs.GR, cs.CV arXiv: https://arxiv.org/abs/2609.01306v2 Triangle- and mesh-based neural rendering aims to bridge neural scene representations and existing graphics engines (\textit{e.g.}, Unity and Blender) by leveraging triangle primitives compatible with standard rasterization hardware. However, existing methods are developed and evaluated under inconsistent settings, with limited comparison and little investigation into practical graphics engine deployment. This gap significantly hinders the understanding of their real-world usability. To address this issue, we introduce MeshSplatBench, the first benchmark for systematic evaluation of triangle- and mesh-based neural rendering from native rendering to graphics engine deployment. We propose a hierarchical deployment protocol with two options: (1) Standard deployment, using a conventional opaque mesh pipeline with vertex colors and hardware Z-buffering; and (2) Dedicated deployment, incorporating method-specific engine implementations to preserve appearance and compositing properties (e.g., alpha blending). For mesh splatting, we further introduce a structural audit to evaluate the topological and geometric integrity of exported surfaces for downstream graphics applications. Extensive evaluations reveal three key findings: (1) graphics engine deployment introduces noticeable quality degradation across methods, while mesh splatting approaches achieve relatively better robustness under standard deployment; (2) dedicated deployment can preserve most rendering fidelity at the cost of approximately 6-30$\times$ slowdown; and (3) explicit connectivity and shared vertex indexing in current mesh splatting methods remain insufficient to guarantee manifoldness or global connectivity. Our benchmark demonstrates that rasterizability alone does not imply graphics readiness and highlights the importance of evaluating practical engine compatibility. The benchmark and source code will be publicly released. --- ### Explore Before Committing: Hypothesis-Guided Search for Deep Research Agents Authors: Ruochen Zhou, Zhengyu Chen, Luan Zhang, Siyang Gao, Yee Whye Teh, Shiqi Chen Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01294v1 Deep-research agents answer complex questions by interacting with search and browsing tools, yet they often search along a single evolving trajectory. Our trajectory-level analysis reveals a common failure mode in which the agent may encounter an early search state with several plausible directions, but follow one direction before collecting enough comparative evidence. Once this happens, subsequent tool calls tend to reinforce the same path, increasing the chance of failure when the initial direction is misleading. We further find that successful trajectories reduce this risk through two behaviors: grounding vague exploration in concrete candidates and shifting directions when the current path is weak or incomplete. Based on these findings, we propose HypoSearch, which generates lightweight hypotheses as soft search hints, explores them through bounded independent branches, and compares branch-level evidence before commitment. Across four deep-research benchmarks and three backbone models, HypoSearch consistently outperforms single-trajectory search and standard parallel baselines, improving Qwen3.5-122B from 46.7 to 60.0 on BC-small while using fewer tool calls than five independent trajectories. A pilot supervised fine-tuning study further shows that these behavioral signals can curate compact training trajectories and reduce degradation from unfiltered data. --- ### Relational Task Generation Language: A Declarative Specification Framework for Relational Deep Learning Authors: Oleksii Kolesnichenko, Jakub Peleška, Gustav Šír Published: 2026-09-01 Categories: cs.PL, cs.DB, cs.LG arXiv: https://arxiv.org/abs/2609.01292v1 Relational Deep Learning (RDL) has become a powerful paradigm for learning from multi-tabular data. However, manually defining RDL prediction tasks is a laborious process that frequently results in data leakage. To address this issue, we introduce Relational Task Generation Language (RTGL) - an open-source declarative language that streamlines RDL task formulation by abstracting away low-level SQL details. We showcase RTGL by reconstructing existing RDL benchmark tasks and uncovering their inconsistencies stemming from manually crafted SQL definitions of RDL prediction targets, thereby underscoring the value of a dedicated declarative language. In addition, we demonstrate the practical utility of RTGL by designing various new tasks with diverse forms and target types. Our experiments confirm the robustness and usability of RTGL, as well as its seamless integration with the existing RDL frameworks, making it widely accessible to the community. --- ### Agentic Multimodal Models for Environmental Hyperspectral Unmixing Authors: Michał Cholewa, Luca Ciampi, Nicola Messina, Przemysław Głomb, Giuseppe Amato Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01289v1 Hyperspectral unmixing is a key task in remote sensing that aims to decompose mixed pixels in hyperspectral images into their constituent material signatures, or endmembers, and their fractional abundances. Conventional modular approaches estimate the scene composition through successive model-order estimation, endmember extraction, and abundance estimation stages, whose errors can lead to redundant or ambiguous candidate components and ultimately affect the recovered decomposition. We introduce an algorithm-agnostic, large vision-language model (LVLM)-driven agentic framework that refines the outputs of such pipelines rather than replacing their underlying numerical algorithms. Starting from an initial decomposition, the agent iteratively gathers complementary spectral and spatial evidence through dedicated tools, including spectral-library retrieval and abundance-map visualization, and modifies the active endmember set through merge and discard operations followed by abundance re-estimation. We apply the same refinement procedure to several modular pipelines combining different model-order, extraction, and abundance-estimation methods, and evaluate it on HYDICE Urban, Jasper Ridge, and Stonewall Playa. Experiments show that the proposed agent consistently improves endmember cardinality and generally improves the recovered spectral signatures and abundance maps across heterogeneous modular pipelines, while remaining competitive with integrated end-to-end unmixing methods, including CNN-AE, uDAS, and R-CoNMF. These results highlight the potential of tool-using LVLM agents to combine spectral and spatial evidence for algorithm-agnostic refinement of physically grounded hyperspectral unmixing decompositions. Code is publicly available at https://anonymous.4open.science/r/agentic-hu. --- ### Analog-DB: An Agent-First Analog Integrated Circuit Database, From Blocks to Systems Authors: Danial Noori Zadeh, Mohamed B. Elamien Published: 2026-09-01 Categories: cs.AI, cs.AR, eess.SP arXiv: https://arxiv.org/abs/2609.01286v1 Sharing analog integrated circuit designs remains difficult: foundry non-disclosure agreements restrict the process details a design depends on, and the testbenches behind published results are rarely released. We present analog-db, an open-source, versioned database built on a shareable design representation. A domain-specific language captures each design as a process-neutral topology, reusable testbenches, and a machine-readable datasheet under one schema, so a design is shared in full and re-simulates on the process kits it is bound to. A parameterization scheme exposes functional sub-blocks and device sizes as named parameters that carry their matching constraints, making circuits composable and retargetable; a schema-governed contract and queryable catalog let AI design agents discover and reuse them directly. Across the regulator corpus, all 23 circuit-kit bindings on three open kits meet their own recorded specification bands (typical corner, matched devices, no layout) and 10 of 23 meet a common class band. Seventeen of the 23 imported sizings failed their testbenches and closed under a gm/ID sizing loop driven by the annotated sub-block roles, typically within one to three iterations. In a supervised case study, a coding agent working from the released artifacts sized the op-amp cores of a chopper instrumentation amplifier on an open 130nm kit, locating four hand-entry defects and a missing common-mode feedback loop that the sizing-only baseline did not repair. The database holds 68 circuits across sixteen classes, verifiable at schematic level under a tiered harness and tracked on a power/performance scoreboard, released at https://github.com/MacAnalog/spicexplorer-release. --- ### HiLRP: Toward One Trustworthy Explanation for Vision Transformer: Conservation-Valid Attribution via Attention Primitives Authors: Sathiyamohan Nishankar, Pubudu Sanjeewani, Asanka Perera, Selvarajah Thuseethan Published: 2026-09-01 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2609.01282v2 Vision Transformer (ViT) design has become increasingly diverse, with backbones combining convolutional stems, windowed, linear, or multi-axis attention, patch merging, and spatial reduction in various configurations. This diversity poses challenges for existing attribution methods, whose assumptions often do not hold across ViT variants: Grad-CAM requires a terminal spatial feature map, attention rollout assumes global softmax attention, and layer-wise relevance propagation (LRP) requires module-specific rules. To the best of our knowledge, no existing method provides a unified attribution framework across this architectural space. We show that this architectural diversity can be captured by a simpler underlying structure. The attention and resolution-reduction operators in current ViTs can be decomposed into four operation types: linear maps, bilinear mixing, normalization or gating, and reindexing. Each operation admits a relevance rule that satisfies conservation. Based on these rules, HiLRP supports new backbones by construction rather than by architecture-specific derivation, and its attribution maps decompose the prediction rather than relying on heuristic assumptions. We prove conservation and conditional equivariance and verify both to machine precision. Across 14 attribution methods and 10 architectures, we find that no prior method remains reliable across ViT families, while Faithfulness Correlation becomes uninformative for backbones robust to spatial masking. HiLRP alone preserves conservation across windowed, spatial-reduction, multi-axis, and linear-attention models, where naive extensions can produce zero or inflated relevance. It also localizes attribution failures in class activation mapping, achieving 0.97 Pointing compared with 0.55 for competing methods on EfficientViT. --- ### EmbodiedSkills: A Unified Framework for Orchestrating, Training, and Deploying VLA Agents Authors: Wei Wang, Wenqiao Zhang, Yutong Lin, Yuqian Yuan, Tianwei Lin, Jinhao Mao, Zhenxuan Fan, Mingjian Gao, Yang Dai, Wentong Li, Zheqi Lv, Zheng Dong et al. Published: 2026-09-01 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2609.01281v1 Vision-language-action (VLA) models map visual observations and language instructions directly to robot actions, but long-horizon tasks require more than action prediction. An agent must coordinate perception, planning, execution, progress verification, and recovery as the physical state evolves. An action prediction or a model-generated skill decision does not, by itself, guarantee that the proposed operation is valid in the current state or that its outcome will be verified. We propose EmbodiedSkills, a unified framework that treats each skill decision as an execution proposal: the runtime checks its prerequisites before execution and verifies the outcome afterward. A shared executable-skill interface connects high-level skill selection, bounded low-level VLA execution, and post-action verification within a single agent loop. Because this interface remains fixed, low-level VLA policies can be replaced or adapted without changing the agent loop. The interface also records planning, execution, verification, and recovery events as structured trajectories, which provide supervision for individual components and can support optional online adaptation when interactive feedback is available. We instantiate EmbodiedSkills with Qwen3-VL and OpenPI/pi0.5 on RoboTwin 2.0 and LIBERO. Task-adapted low-level VLA policies achieve an average success rate of 86.20% across 50 RoboTwin 2.0 tasks and 97.40% across the four LIBERO suites. These results establish the execution performance of the task-adapted low-level VLA policies used in EmbodiedSkills. On four memory-dependent RMBench tasks, the same task-adapted execution approach achieves 12.5% average success. The framework provides a trainable and inspectable agent layer for turning these policies into closed-loop embodied systems. --- ### Some Emotions Run Deeper: Layer-wise Probing and Causal Intervention in Large Language Models Authors: Tian Fang, Gaël Guibon, Davide Buscaldi Published: 2026-09-01 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2609.01279v1 Emotion is expressed in text along a wide spectrum, from surface lexical cues to inferences entangled with content. Most layer-wise analyses of emotion in LLMs use a single corpus, leaving open whether the depth at which emotion becomes accessible is a property of the model or also of the text source. We investigate this across three datasets spanning different degrees of explicitness and contextualization in emotion expression (Twitter posts, Reddit comments, and autobiographical narratives) and eight 1B--9B open-weight LLMs from the Llama, Qwen, and Granite families. We combine layer-wise probing with offline feature scaling and online forward interventions, transfer analyses, and an early-exit classifier. We find that (i) the best probing layer shifts systematically across corpora, from input-adjacent layers to over half model depth, and this ordering persists after matching label-by-length-bin distributions; (ii) across the evaluated settings, forward-pass interventions on probe-selected bands reduce test accuracy by 5--6 points more than same-width random bands ($q < 0.01$); (iii) selected bands transfer across datasets and emotion categories, suggesting partially shared affective information rather than strictly per-emotion substrates; and (iv) probe-selected early-exit representations outperform full-depth exits by $6.9$ percentage points on average. --- ### TimeSteer: Inference-Time Speech Scheduling in Joint Audio-Visual Diffusion Models Authors: Chao Zhou, Yiling Chen, Qi Chu, Tao Gong, Nenghai Yu, Tianyi We Published: 2026-09-01 Categories: cs.CV, cs.AI, cs.MM arXiv: https://arxiv.org/abs/2609.01277v1 Although pretrained joint audio-visual diffusion models offer rich control over \emph{what} to generate, they provide no explicit control over \emph{when} an utterance should occur. To address this, we study \emph{inference-time speech scheduling}, a novel task that places coupled speech and visual articulation within user-specified begin--end intervals without finetuning the backbone model. We uncover two intrinsic properties of the denoising process that enable this task. First, a timing-sensitive text-to-audio cross-attention head exposes each utterance's model-implied source span along the latent timeline. Second, the predicted clean latent already organizes coupled speech and visual articulation, allowing their temporal placement to be edited without regenerating the content. Building on these discoveries, we propose \textbf{TimeSteer}, a training-free framework that localizes each utterance's source span through \textbf{Source Span Localization} and transfers the associated audio-visual latent content from the source interval to the specified target interval through \textbf{Region-Aware Latent Remapping}. We further introduce \textbf{SpeechShift}, the first benchmark for interval-level speech scheduling in joint audio-visual generation. Experiments across two representative backbones show that TimeSteer substantially improves interval controllability over training-free baselines while maintaining competitive overall generation quality. --- ### Seeing the World and the Self from Egocentric Video Authors: Kai Guan, Minchao Jiang, Ruichen WangLi, Wentao Zhu, Lei Zhang Published: 2026-09-01 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.01276v1 Complete 3D perception from egocentric video requires recovering the surrounding scene and the wearer's full-body motion in a shared metric frame. Existing methods typically address scene reconstruction and motion estimation separately: scene reconstruction methods ignore the wearer, whereas motion estimation methods lack explicit scene geometry and often depend on external trajectories. Joint recovery is challenging because the two tasks exhibit asymmetric visibility and require different prediction paradigms. The largely visible scene supports deterministic geometric regression, whereas the severely occluded body requires generative motion inference. We therefore propose RESELF (REconstructing the Scene and the sELF), a unified framework that couples deterministic metric geometry reconstruction with geometry-conditioned motion generation. RESELF adapts a geometry foundation model pre-trained on large-scale exocentric data to egocentric video using frame-wise scale and relative-pose consistency objectives. The resulting camera trajectory and latent geometric features condition a diffusion model that recovers the wearer's motion. A subsequent closed-loop kinematic feedback stage further refines the camera head while preserving the reconstructed scene geometry. To support training and evaluation, we curate EE4D-JSM from EgoExo4D by aligning egocentric video, sparse metric scene geometry, camera trajectories, and full-body motion annotations. Experiments show that RESELF outperforms state-of-the-art methods designed for the individual tasks across depth estimation, camera tracking, and full-body motion estimation. Code, models, and datasets will be available at https://ka1guan.github.io/RESELF/. --- ### The Constitutional Coverage Trilemma in AI Governance Authors: Natalija Mitic, Soona Sedahmed A. O., Mamadou Selly Ly, Moustapha Cisse Published: 2026-09-01 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2609.01275v1 Frontier AI systems function as \emph{constitutional institutions}: each deployed model encodes an implicit ranking among safety, helpfulness, honesty, autonomy, and equity. We ask whether the supply of frontier constitutional types covers human demand. Combining a paraphrase-controlled audit of the as-shipped default constitutions of $23$ frontier LLM archetypes with a pairwise-tradeoff study of $1{,}649$ US participants on the same instrument, we report three facts. \emph{Demand is broad}: it spans all five values, with the largest constituency under one-third. \emph{Supply is narrow and drifting}: the $23$-archetype hull occupies ${\sim}2\%$ of the demand hull under conservative noise-matched estimation ($0.10\%$ at full audit precision), no archetype puts helpfulness or autonomy first ($37\%$ of users are constitutionally homeless), and across six model families autonomy decreases in $5/6$, equity increases in $5/6$, and safety increases in $4/6$, with monotone within-family version trends (order-permutation $p = 0.013$) and the autonomy decline concentrated in scenarios where safety is not at stake. The drift's importance is directional: \emph{away} from a value already undercovered, mechanically worsening the welfare floor for the least-served users. \emph{The fix is sparse}: a $2$-vertex menu $\{e_{\mathrm{HON}}, e_{\mathrm{AUT}}\}$ beats the full $23$-archetype frontier by $47\%$ on mean regret (CI $(43\%, 52\%)$); three vertex additions cut mean/worst-group regret by up to $81\%$/$64\%$. We formalize these findings as a budgeted-pluralism trilemma, show the binding regime is empirically realized, and verify the conclusions are robust to distance-based welfare and to degraded routing. The instrument and audit harness are described in full in the appendices. --- ### From Base Rollouts to RL Reasoning: A Budgeted Search Perspective Authors: Wenhe Sun, Cunxiang Wang, Zijun Yao, Yixin Cao Published: 2026-09-01 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.01274v1 Reinforcement learning with verifiable rewards (RLVR) improves language-model reasoning, but how these gains relate to inference-time decoding and search remains unclear. Does RL create reasoning the base model lacks, or shift the rollout distribution toward trajectories it can already reach but rarely samples? We study this behaviorally with a Unified Decoding Framework (UDF), which expresses token-level sampling, beam-like search, tree search, and sequence-level resampling as executable policies over a shared budgeted operating space, scored post hoc with pass@$k$, self-consistency, best-of-$N$, and first-finish success. Using paired Base/RL checkpoints from SimpleRL-Zoo, we ask whether an RL default-policy curve can be approximated by a structured path of Base operating points. On Math500, AIME, GPQA, and IFEval, the pass@$k$ recovery path follows a Budgeted Operating-Point Transition Rule (BOPTR), $N_{\mathrm{Base}} \approx αN_{\mathrm{RL}}^β$, with benchmark-conditioned exponents. On Qwen2.5-7B, BOPTR gives the lowest transfer error among the non-oracle rules we test, 3.41 pp (95% CI (2.32, 5.53)); a three-seed replication gives 3.07 $\pm$ 0.39 pp. The rule extends to ten models across four families (3.28 to 4.87 pp on checkpoints added after fitting), to four benchmarks it was never fitted on (5.03 pp vs. 4.44 pp in fit), and holds without an RL checkpoint for the target model (4.19 pp) or without RL supervision of any kind (5.08 pp). These results support a qualified internalized-search reading: under the recipe we test, much of the measured RL gain corresponds to a change in sampling efficiency toward operating points the base model can already reach under search. We treat the scaling patterns as descriptive of this recipe and cohort, report where they break down, and use UDF and BOPTR as behavioral diagnostics rather than evidence of parameter-level equivalence. --- ### Position: Privacy Is a Claim, Not a Property of Synthetic Data Authors: Jiachen Zhao, Antonia Januszewicz, Taeho Jung Published: 2026-09-01 Categories: cs.LG, cs.CR arXiv: https://arxiv.org/abs/2609.01273v1 Synthetic data has become a common component of machine learning research. While widely adopted, its use in privacy-sensitive contexts has quietly shifted from a claim of residual inference risk under stated assumptions to an appearance-based property inferred from data generation itself. In this position paper, we argue that this shift reflects an implicit change in community standards for what counts as sufficient privacy evidence, rather than a misunderstanding of well-established privacy principles. Drawing on an empirical analysis of recent publications across major ML venues, we show that synthetic data is frequently used in privacy-sensitive settings without explicit articulation of threat models, inference risks, or falsifiable privacy claims. As a result, privacy assurance often remains implicit, difficult to verify, and unevenly distributed, with heightened exposure for rare and minority records. We argue for treating privacy as an explicit, evidence-based scientific claim and recommend that ML venues adopt norms requiring privacy-relevant assertions to be clearly scoped, testable, and contestable. --- ### Making Prospective Memory SLM-Shaped: Typed Intention Stores for Small-Model Agents Authors: Jinqing Zhao, Chengcan Wu Published: 2026-09-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2609.01272v1 Prospective memory means carrying out a deferred intention at the right future cue while other work continues. Benchmarks now isolate it as an agent skill, yet frontier LLMs still struggle: the best published PM-Bench scaffold reaches only 65.1% Set-F1. We argue that this loop is schema-constrained state tracking rather than open-ended reasoning, and that small models can execute it when the action space is typed. We propose the Prospective Intention Store (PIS) that puts lifecycle logic in code and scoped language work on the model. The scaffold is agentic and training-free: no selector fine-tuning and no trajectory distillation. On PM-Bench, DeepSeek-Chat with PIS reaches 82.9% Set-F1. On Gemma-E2B, Set-F1 is only 4.2% without a store and at most 6.6% under seven retrospective memories, while PIS reaches 66.2%. PIS further reaches 70.1% Set-F1, where retrospective memory methods stay at most 54.4%. PIS sets a new state of the art on this benchmark and enables small models to surpass the published large-model scaffold. ---