Zero-Shot
-
LEMON: Equivariant Learning over Attention Maps for Zero-Shot Misbehavior Monitoring on New LLMs
2026-10-15 22:53:45Guy Bar-Shalom, Mor Geva, Fabrizio Frasca, Haggai MaronAbstract
Large language models exhibit misbehaviors that hinder deployment in high-stakes settings. One approach to monitor such failures is to use an external LLM as a monitor, but this is prohibitively expensive. A well-known and efficient class of monitors can detect such failures from internal representations (such as activation probes), but are tied to the model they were trained on, requiring costly re-annotation as new LLMs appear. A natural way to avoid this repeated annotation cost is to develop internal-representation based monitors that transfer zero-shot across LLMs. However, such transfer is challenging because internal representations are not naturally aligned across models; in particular, attention maps, known to contain signals of misbehavior, may differ in head function, ordering, and model depth. In this work, we characterize those cross-model attention mismatches as algebraic symmetries and use geometric deep learning principles to build LEMON (Language-model Equivariant MOnitoring Network) an online graph neural network-based monitor, whose predictions are invariant under these symmetries. LEMON processes attention maps token-by-token, as they are produced and outputs a score after every token (or every window of $W$ tokens) at negligible cost. Across 16 LLMs and six datasets, in the zero-shot cross-model setting, LEMON outperforms all baselines on both hallucination and jailbreak monitoring tasks, improving over the strongest by up to 14 ROC-AUC points.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22723
PDF
-
Repairing LLM-Generated Specifications with Deterministic Diagnosis
2026-10-15 19:12:54Giovanni Vaccarino, Emanuele La Malfa, Dehai Min, Hanrong Zhang, Pier Luca Lanzi, Lu Cheng, Michael J. Wooldridge, Philip S. Yu, Sara BernardiniAbstract
Large language models (LLMs) are increasingly used as formalizers: given a reasoning problem expressed in natural language, they produce a formal specification that a symbolic solver can then solve, with guarantees relative to that specification. Many specifications are runnable yet semantically wrong, encoding incorrect constraints, domains, objectives, predicates, or goals, and solver-error feedback dries up once execution succeeds. Purely LLM-based diagnosis, lacking a formal ground truth, often fails to spot these errors, and its ungrounded critiques cause regressions in 11.8% of initially correct specifications in our experiments. We propose DRAGO, a repair framework that replaces this ungrounded critique with deterministic, formalism-specific diagnosis, and retains a rewrite only when the diagnosed error weight strictly decreases under a reference-free acceptance predicate. Across five benchmarks spanning logical reasoning (SMT-LIB2), constraint programming (MiniZinc), and planning (PDDL), DRAGO cuts regressions by nearly an order of magnitude, to 1.3%, while improving answer accuracy by 12.9 percentage points over zero-shot formalization and 8.7 percentage points over LLM-diagnostic repair, averaged across benchmarks. We also uncover two rendering defects in Planetarium that cause 50% (3,086 instances) of its floor-tile split to be inconsistent with the reference formalizations, and release a corrected split.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22568
PDF
-
alkhder at ImageEval 2026 Shared Tasks: Constrained Selection and Rotation-Averaged Logit Scoring for Multimodal Hallucination Detection
2026-10-15 18:08:46Maher Pasa, Maria Hanini, Amro Najjar, Hasan AlkhderAbstract
We describe our submission to Subtask~1b of ImageEval 2026. The task consists of selecting the single visually valid statement from three candidates paired with an image, and is in both English and Modern Standard Arabic (MSA). Our system uses a pre-trained 7B vision--language model in a zero-shot setting, without task-specific training or fine-tuning. We formulate the task as a constrained selection problem and choose the answer using next-token log-probabilities. To reduce answer-position bias, we apply a Latin-square rotation to the candidate statements. Our system achieves CI scores of 0.0510 on the English test set and 0.0960 on the MSA test set, reducing CI relative to the official 3B baselines reported on devtest by 83.7\% and 80.4\%, respectively.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22515
PDF
-
LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation
2026-10-15 17:59:59Zhixu Li, Keqian Tang, Litian Gong, Jingyu Yao, Wenqian Zhang, Tianze Xu, Zehao Wang, Yuping Wang, Junge Zhang, Jiachen LiAbstract
Personalization is essential for general-purpose robots to transition into household environments and achieve true practicality. While general robotic manipulation has been extensively studied, personalized robotic manipulation (PeRM) lacks a controlled benchmark for separating task completion from preference satisfaction. We introduce LIBERO-PeRM, a benchmark for preference learning in embodied agents. LIBERO-PeRM studies five questions: 1) how policies follow preferences in zero-shot settings; 2) how they learn single preferences from demonstrations; 3) how efficiently they acquire preferences as demonstrations increase; 4) how learned preferences generalize across layouts and tasks; and 5) how policies compose preferences and behave under conflicts. We propose a controlled formulation and taxonomy of observable preferences in simulation, covering motion, skill, and user-profile choices. This formulation is restricted to language-expressible preferences that can be instantiated in LIBERO without changing the task goal, evaluated automatically, and reused across tasks. Building on LIBERO, our procedural pipeline generates paired demonstrations with and without preferences. We create five task suites (312 tasks) with high-quality demonstrations, rich metadata, and preference-satisfaction metrics. Experiments show that current VLA policies struggle with zero-shot personalization, while SFT makes preferences learnable. Mixed training achieves better no-preference retention than sequential adaptation and improves performance with contextual prompts compared with preference-only training, but can weaken motion-preference learning; preference generalization, decomposition, and conflict handling remain difficult. The code, datasets, and demos can be found on the project website: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22372
PDF
-
GeoState: Learning View-Decoupled Geometry States for 3D World Modeling
2026-10-15 17:56:18Hongchang Liu, Jian Liu, Yujuan Ding, Liang Peng, Xun Jiang, Fei Ma, Xing Xu, Yi BinAbstract
Recent 3D world models increasingly use features from geometry foundation models (GFMs) as predictive world states, benefiting from their rich geometric priors. However, the view dependence of these features entangles scene dynamics with camera-induced variation, making the resulting state evolution harder to predict. We propose GeoState, a view-decoupled geometry representation for predictive 3D world modeling that separates scene-state evolution from camera-dependent observation. Specifically, a scene tokenizer aggregates multi-view GFM features into a view-decoupled scene state, which a geometry decoder queries using camera rays. An autoregressive world model then predicts future scene states, while camera motion is modeled separately. The predicted states can therefore be decoded under either predicted or externally specified cameras, while retaining a single context-derived scale throughout the rollout without per-frame scale alignment. Experiments show that GeoState outperforms prior 3D world models on Waymo, supports geometrically consistent cross-view querying, and generalizes zero-shot to KITTI and nuScenes. Project page: this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2610.22343
PDF
-
Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries
2026-10-15 17:35:17Anikait Singh, Teresa Zhang, Yoonho Lee, Roshen Sanjay Nair, Aviral Kumar, Chelsea FinnAbstract
Reasoning over long document contexts remains a challenge for large language models (LLMs). Standard retrieval methods are inherently reactive, bottlenecking agents with complex, cross-document reasoning at inference time, while precomputed structures like knowledge graphs often fall short because they rely on rigid heuristics. Instead, can we allow agents to perform anticipatory directed curation: learning to proactively reorganize unstructured corpora prior to inference to efficiently serve diverse future queries. We build on a key asymmetry in modern agentic harnesses (e.g., Claude Code): while LLM agents are adept at navigating existing filesystems to solve an immediate task, they are rarely trained to construct reusable workspaces for future use. To instantiate this capability, we introduce Librarian Agents, a framework in which a builder agent transforms raw documents into a hierarchical filesystem of synthetic files and reference links, which forms a workspace tailored for a downstream answering agent. We formulate workspace construction as a Markov decision process and optimize the builder policy via reinforcement learning to directly maximize downstream answering accuracy. Empirically, Librarian Agents outperforms the strongest retrieval and precomputation baselines by 33 percentage points on BrowseComp-Plus, amortizes upfront construction costs across repeated queries, and generalizes zero-shot to unseen domains on BRIGHT with a 31-point average accuracy improvement.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22273
PDF
-
Change the Number, Change the Answer: Assessing Rating Scale Sensitivity in LLM-Based Text Scoring
2026-10-15 16:42:15Wonjeong Jo, Gyuwan Kim, Ming Shan HeeAbstract
Large language models (LLMs) are increasingly used in computational social science to assign scalar scores to social and psychological constructs. However, studies use different rating scales, and whether their scores remain comparable across scale designs is understudied. To address this gap, we define and systematically investigate scale sensitivity, the variation in LLM-generated scores caused by rating scale design. We distinguish numerical sensitivity, changes in normalized scores, from semantic sensitivity, changes in score interpretation. To assess both, our experimental design crosses 2 formats (continuous vs. ordinal), 4 ranges (3, 5, 11, and 21 points), and 2 signs (signed vs. unsigned) across 3 tasks (political leaning, sentiment, and toxicity), 2 prompting strategies (zero-shot and rubric-based), and 5 LLMs. We find that changing one design element shifts normalized scores by about 0.10 on average and changes the category of roughly 20% of texts. Sensitivity is greatest for toxicity, the only unipolar construct examined. Also, rubric-based prompting does not reduce sensitivity compared with zero-shot prompting. We provide practical guidance for designing more reliable LLM-based text scoring and highlight the need for robustness checks across alternative scales.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22187
PDF
-
Whole-Body Dexterity
2026-10-15 16:40:48Conor Mc Gartoll, An Dang, Pranay Thangeda, Arjun Gupta, Carolina Higuera, Nima Fazeli, Mustafa Mukadam, Manikantan NambiAbstract
Current humanoid dexterous manipulation remains largely confined to end-effectors and simple grasps, falling short of humans' ability to engage the entire body by bracing and sliding bulky objects against the arms and torso. We introduce Whole-Body Dexterity (WBD), a general framework that uses the robot's body and appendages as contact surfaces to grasp and manipulate objects. Our sim-to-real pipeline first generates a dataset of feasible whole-body grasps, then trains reinforcement-learning experts on a general any-to-any repositioning objective by sampling initial and goal grasps from the dataset, and distills these experts into a single policy that uses proprioception and object geometry. We show that the distilled policy transfers zero-shot to a real Unitree G1, where it repositions a basketball and a cardboard box to regions around the torso and completes a long-horizon task that stows a box under one arm, carries it from a shelf to a table, and places it. The controller operates interactively, steered by coarse motion references from a human teleoperator. Across real-world whole-body manipulation tasks, the distilled controller completes 90% of trials, compared with 18% for direct retargeting and 4% for a generalist whole-body controller.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22184
PDF
-
EMO: A Closed-Loop Test-Time Computation Framework via Alternating Context Optimization and Logits Steering
2026-10-15 16:23:55Wei Xia, Yehui Tang, Zhi-Hong DengAbstract
Test-time context optimization adapts frozen language models through textual feedback, but ordinary conditional generation provides no separate control over how strongly to reinforce the distributional changes induced by a context update. We introduce EMO, a training-free framework inspired by expectation--maximization that alternates question-specific context estimation (E-step) and distribution refinement (M-step). The E-step revises a system context using textual gradients from an external judge without access to gold answers. The M-step contrasts next-token distributions under the revised and preceding contexts at a shared prefix to construct a confidence-constrained logit correction. Feedback on the steered response drives the next context update, coupling adaptation across rounds with intervention within each generation. Theoretically, we connect parameter and context optimization through a shared target-KL objective and additive logit representations, and derive the steering vector from context-induced logit displacements. Under context-improvement and faithful-steering assumptions, we prove a one-round decrease in target KL divergence for sufficiently small steering strength. Under greedy decoding across eight models and six benchmarks, EMO improves the zero-shot baseline in 39 of 40 aggregated settings, averaging $+9.10$ percentage points. With the same judge, it adds $+1.53$ points on average over context-only optimization (TextGrad), with consistent MMLU-Pro gains across models. At matched target-model generation budgets, accuracy advantages over best-of-$N$ are strongest on Llama, while oracle coverage is higher in most comparisons despite fewer candidates. The coverage--selection gap motivates allocating test-time computation jointly to candidate generation and verification.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22160
PDF
-
PROBE: Learning to Audit Policy Compliance in Tool-Using LLM Agents
2026-10-15 16:01:38Kshitij Mishra, Abhijith Sharma, Nils Lukas, Salem LahlouAbstract
Modern agentic systems extend language models with external tools, enabling them to complete complex tasks through multi-turn interactions. As these agents are deployed in consequential settings, they are often constrained by policies governing pricing, privacy, confirmation, safety, and other requirements. However, existing evaluation methods leave a critical gap: capability benchmarks measure what agents can accomplish, while content-safety red-teaming measures harmful generations, but neither tests whether agents comply with predefined policies during otherwise valid tool use. Such policy violations are difficult to audit because their effects may emerge only after several turns, span multiple tool calls, or surface in downstream systems. We propose PROBE (Policy Rule Observation via Behavioral Exploration), an algorithm for training LLM-based auditors that interact with deployed agents as realistic users and probe them for policy violations through multi-turn, tool-mediated dialogue. We formalize agent auditing as a two-player Markov game and propose a composite reward that balances violation discovery, behavioral plausibility, conversation progress, and policy diversity. To study training objectives and reward design, we compare GRPO, DPO, and GFlowNet-based optimization in a controlled setting. Across extensive experiments, auditors trained with PROBE uncover substantially broader and higher-yield policy violations than same-architecture zero-shot adversaries, even when the auditor is significantly smaller than the target agent. Our results show that multi-turn behavioral auditing is a distinct and necessary axis of agent evaluation, complementing capability and safety benchmarks. PROBE offers a scalable path toward automated policy auditing for real-world agentic systems and suggests a co-evolutionary loop in which auditors expose failures and agents improve from auditor feedback.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22124
PDF
-
SCCL-DT: Semantic-Conflict-Aware Contrastive Learning with Dual-Rule Teacher Distillation for Zero-Shot Chest X-Ray Diagnosis
2026-10-15 15:44:40Ting Liu, Haoran Lai, HaobinChen, Ming Liu, S Kevin ZhouAbstract
Contrastive vision-language pre-training supports zero-shot chest X-ray diagnosis, but standard objectives treat every off-diagonal image-report pair in a batch as a negative. In radiology, recurring findings across patients mean that instance mismatch need not imply semantic incompatibility, creating false-negative supervision that we term a semantic conflict. A training-trace audit shows that pair-observations selected by the online mask used during training are enriched for report-derived finding agreement, although this proxy analysis does not establish clinical equivalence. Fixed thresholds can become miscalibrated as similarity distributions evolve, while soft reweighting leaves residual repulsive gradients. We therefore propose Semantic-Conflict-Aware Contrastive Learning (SCCL), which tracks modality-specific similarity distributions with exponential-moving-average (EMA) statistics, derives adaptive thresholds, and hard-masks flagged candidates before softmax normalization. Inclusive (OR) and consensus (AND) rules define two SCCL teachers, whose bidirectional cross-modal rankings are consolidated through standard distillation into a single deployable student, SCCL-DT. Across five public chest X-ray benchmarks, SCCL-DT achieves a mean zero-shot AUC of $0.850$, compared with $0.837$ for the strongest evaluated baseline in our comparison. Code will be made publicly available soon.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22094
PDF
-
Learnable Splines for Unstructured Neural Operators
2026-10-15 15:15:31Vignesh Gopakumar, Zhuoyuan Wang, Ander Gray, Valentin Duruisseaux, Josh Williams, William Michael Edward Ellis, Nuno Nobre, Matthew Field, James Panayis, Alexander I. Blair, Andrew Davis, Stanislas Pamela, Robert Akers, Marc Peter Deisenroth, Anima AnandkumarAbstract
Neural operators approximate solution operators of partial differential equations at a fraction of the cost of numerical solvers. Efficient grid-based architectures such as the Fourier neural operator (FNO) assume uniform rectangular grids, whereas engineering simulations use large, irregular, and locally refined meshes. We introduce L-SplineFNO, a memory-efficient neural operator that couples an FNO backbone with a sparse, learnable tensor-product B-spline transfer between unstructured meshes and a latent Cartesian grid. A local normalised scatter avoids the geometry-dependent solve of analytic spline transformation while preserving compact support, partition of unity, and spline smoothness by construction. The neural operator, conditioned on global design and operating parameters, learns the mapping of geometry to the field solution. Across six three-dimensional engineering benchmarks, L-SplineFNO achieves strong predictive accuracy while requiring substantially less memory and compute than baseline operators, scaling to 26.6 million nodes in training and 46 million in inference on a single A100 GPU. It also supports zero-shot super-resolution and efficient closed-form spatial derivatives. Uniquely, the analytic spline decoder yields rigorous prediction bounds over continuous space and uncertain parameters, via Bernstein convex hulls and Taylor models, demonstrated on a 29.2-million-node MAST-U tokamak model.
Abstract (translated)
URL
https://arxiv.org/abs/2610.22446
PDF
-
Flow-CLIP: Image-Conditioned Residual Scoring for CompositionalImage-Text Matching
2026-10-15 14:12:30Minghong Wei, Enguang Zhu, Pu Cao, Lu Yang, Qing SongAbstract
Vision--language pre-training has enabled substantial progress in image--text retrieval and zero-shot recognition by learning transferable cross-modal representations. However, sentence-level alignment remains insufficient for compositional matching, where captions can describe the same objects and attributes while differing in how these elements are bound. To address this limitation, we propose Flow-CLIP, which refines global image-text alignment through image-conditioned changes in token--patch evidence. Global Compatibility Calibration (GCC) learns a sentence-level similarity metric, while Image-Conditioned Evidence Refinement (ICER) incorporates visual patch context into token queries and uses the resulting change in local matching evidence to adjust the global score. We further introduce Cross-Image Margin Ranking (CIMR), which compares the same positive--negative caption pair across matched and mismatched images, encouraging a larger ranking margin when the image supports the correct description. Experiments on compositional matching and attribute--object binding benchmarks show that Flow-CLIP improves binding accuracy over pretrained CLIP and learned global alignment through local evidence refinement, while cross-image ranking supervision mitigates reliance on language priors. Code will be released at link.
Abstract (translated)
URL
https://arxiv.org/abs/2610.21937
PDF
-
RoleQ: Role-Aligned Scales for Converting Language Models to Ternary
2026-10-15 12:56:57Johannes Knittel, Alec Helbling, Hanspeter PfisterAbstract
Weight quantization below four bits shrinks memory and bandwidth at a perplexity that looks acceptable, but often at the cost of factual knowledge and task accuracy. Ternary weights have mostly been obtained by training from scratch or long fine-tuning. We introduce RoleQ, a quantization framework that places scales by the functional roles of Transformer units rather than by the layout of the weight matrix, as conventional formats do. Vectors that read from the residual stream take one scale per output unit; those that write to it, one per input unit. The rule is structural, needs no search, calibration or stored state, and holds on integer and ternary grids. In a rotated basis it matches a per-layer axis search at half the solver work, and at ternary it cuts perplexity by up to half against layout-based placement in two families at every size from 0.6B to 14B, where per-row scales leave MMLU at chance. Our lightweight post-training pipeline converts an existing checkpoint to 1.59 bits per weight in a few GPU-hours, a fraction of the cost of learned ternarization, and leads every post-training ternary method at matched or lower memory-resident bits on perplexity, zero-shot accuracy and MMLU, most widely on knowledge. A sparse second ternary plane placed by a per-unit Fisher saliency keeps the matmul multiplication-free and recovers what ternary loses first: at 8B, a fraction of a bit returns about half the MMLU that conversion costs, well ahead of random or Hessian-based placement at equal bits.
Abstract (translated)
URL
https://arxiv.org/abs/2610.21808
PDF
-
PW-Pruning: Phase-Aware Expert Selection for Speech-Centric Omni-Modal Models
2026-10-15 12:36:40Junjie Cao, Fengping Tian, Wenjie Zhou, Xuanfan Ni, Chenyang Lyu, Siyou Liu, Longyue Wang, Weihua LuoAbstract
Sparse routing reduces per-token computation but leaves the full expert bank resident in memory. We study expert pruning for speech-centric omni-modal models, where pooling calibration statistics can obscure the different expert demands of acoustic-input processing and text-answer generation. We propose PW-Pruning, a phase-aware extension of DO-ACP that constructs separate importance-weighted kernels for acoustic-prefill and teacher-forced answer tokens, then combines phase-weighted log-determinant marginal gains to select one static expert bank. On Qwen3-Omni, retaining 32 of 128 experts per layer improves MMAU-mini and MMSU accuracy over matched pooled selection by 3.20 and 4.38 percentage points, respectively, after supervised fine-tuning, averaged over three runs. Single-phase and shuffled-label controls on this public backbone suggest that semantic phase information contributes to these gains. Experiments on a non-public Qwen3.5-Omni backbone and English-only zero-shot evaluation of Nemotron provide complementary evidence. Across the tested Qwen budgets, AU gains are larger under stronger pruning, without inference-time mask switching.
Abstract (translated)
URL
https://arxiv.org/abs/2610.26891
PDF
-
AVALON: Zero-Shot Indoor Navigation with Human-Seeded Topological Memory
2026-10-15 09:33:04Jing Xia, Changyan Li, Yuduo Wang, Guiyin Tian, Bo Wang, Yifan Liao, Mingjun Li, Shiyan HuAbstract
Indoor service robots often enter unmapped buildings whose layouts and task constraints are already familiar to users. We address this human-familiar, robot-unseen setting with AVALON, a topological navigation framework organized around human-seeded layered spatial memory, node-verified topological execution, and task-triggered, evidence-gated human--robot memory co-updating. The robot has no prior on-site traversal or site-specific training. Deployment requires only a floor plan, place clues, and an initial topology-odometry alignment. Visual depth point-goal navigation leverages verified node arrivals to perform translational re-anchoring, without requiring a robot-constructed global SLAM map. Human--robot memory co-updating fuses complementary sources of information. Human operators specify task constraints that are unobservable via vision and report offline environmental modifications, whereas the robot provides physical-state evidence through active in-situ validation. Task relevance governs whether human-reported environmental changes undergo robot validation during the active mission. Field-collected perceptual evidence gates memory updates, maintaining an explicit separation between human-specified permissions and robot-observed traversability. Our experiments on a Unitree Go2 show that AVALON with X-NavDP completes a 12.81 m Intersection-to-Dining mission in 46.4 s (5/5) with no mapping, 40.9% less robot time than DimOS-Go2 (78.5 s, 4/5), while policy-only X-NavDP times out in all of the 5 runs. These demonstrate the superiority of the proposed topological navigation framework.
Abstract (translated)
URL
https://arxiv.org/abs/2610.21586
PDF
-
MYOPIA: Understanding the Limits of Tree-Based Reasoning in Vision-Language Models
2026-10-15 09:25:10Islam Mesabah, Sary Elmansoury, Gerrit Gro{\ss}mann, Peter Neigel, Sebastian J. VollmerAbstract
Decision trees can make vision-language model (VLM) classification interpretable by replacing a single prediction with a sequence of root-to-leaf questions. However, the accuracy cost of this greedy, non-backtracking procedure remains unclear. We introduce \textbf{MYOPIA}, a benchmark comprising four LLM-generated, human-refined decision trees for GTSRB, MNIST, CIFAR-10, and MSCOCO, together with matched prompts, diagnostic protocols, and an evaluation harness. Across GPT-4o, Qwen-VL Max, and LLaMA-3.2 11B Vision, tree traversal underperforms zero-shot prompting in all 12 dataset--model pairs; paired bootstrap analyses consistently favor zero-shot. Our diagnostics characterize interacting failure modes associated with this gap. Models often recall tree-encoded class attributes more accurately than they ground the same attributes in images: the mean label--image accuracy gaps are 9.5, 13.8, and 12.6 points for GPT-4o, Qwen-VL Max, and LLaMA, respectively. Incorrect local decisions then irreversibly exclude the correct class, while answer extraction introduces additional model-dependent loss. A node-level audit of the four trees finds that split entropy is weakly but consistently associated with lower node accuracy, whereas depth, branching factor, and within-node grounding differences are less stable. \textbf{MYOPIA} therefore provides a framework for diagnosing hierarchical visual classification and motivates testing tree designs that prioritize perceptual answerability and recovery from early errors. Code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.21573
PDF
-
FlashAttention with Statistics: Low-Overhead Attention Features for Compact Hallucination Detectors
2026-10-15 08:23:42Andrei Volodichev, Maksim Savichev, Nikita Sviridov, Artem Yuliy, Dmitry Simakov, Alexey ZaytsevAbstract
Large language models can produce fluent answers that the supplied context does not support. A compact frozen proxy model can detect such answers by reading the context and another model's response, even from a proprietary generator. Attention-based detectors use summary statistics of the proxy's attention weights, but standard pipelines materialize the attention matrix, whose memory grows quadratically with sequence length. We compute these statistics without materializing attention, by accumulation in the FlashAttention pass, sparse lookups after it, or a second tiled pass over the scores, and classify them by this path into nearly free and costlier ones. On an H100, the statistics of five published detectors add 16% to the latency of a Qwen3-8B forward at 2K tokens, falling to 1.4% at 35K, and store 22.5 KiB per response token. On three benchmarks, probes trained on frozen Qwen3 proxies of 0.6B to 8B parameters prove strong detectors: a 4B proxy outperforms the strongest of nine zero-shot 4-9B LLM judges in F1 on two of the three, the proxies match or outperform specialized grounding checkers of up to 7B parameters, probing each proxy beats prompting it as a judge, and adding attention features significantly improves F1 in half of the 12 benchmark-proxy pairs, never significantly lowering it.
Abstract (translated)
URL
https://arxiv.org/abs/2610.21483
PDF
-
Cross-Prompt Inference Mitigates Rare-Class Suppression in Imbalanced VLM Finetuning
2026-10-15 08:13:26Emanuel Ben Baruch, Sivan Biham, Alon Shoshan, Yochai Zvik, Avi Ben-CohenAbstract
When a vision-language model is finetuned on class-imbalanced data, rare-class recall can fall well below the model's own zero-shot performance. In this regime, we uncover a surprising matched-prompt penalty whereby the standard choice of reusing the descriptive finetuning prompt at inference can yield lower rare-class recall than using a different descriptive prompt. We call this inference-time strategy cross-prompt inference. We evaluate diverse VLMs across six datasets and find that cross-prompt inference consistently recovers rare-class recall, with varying trade-offs in frequent-class recall. The effect reflects a train-eval prompt interaction rather than prompt quality alone, and does not arise with non-descriptive prompt changes. It is strongest when imbalance and visual confusability produce substantial rare-class suppression, and is weak or absent when such suppression does not occur. Using Tuning Contribution (TuCo), which decomposes a model's computation into pretrained and finetuning parts, we find that cross-prompt inference reduces reliance on finetuning-induced computation and shifts the model toward its pretrained component. Reducing that finetuning contribution also substantially reduces the cross-prompt advantage. These results show that rare-class knowledge suppressed by imbalanced finetuning can remain accessible under a different inference prompt.
Abstract (translated)
URL
https://arxiv.org/abs/2610.21470
PDF
-
Beyond Sparse Captions: Aligning Slide-Level Text and Patch-Level Vision in Pathology
2026-10-15 05:24:40Daniel Shao, Dongmin Bang, Luca L. Weishaupt, Sophia J. Wagner, Nic G. Reitsam, Ming Y. Lu, Long P. Le, Richard J. Chen, Faisal MahmoodAbstract
Contrastive vision-language (VL) pretraining in computational pathology has generally relied on curated image-caption datasets sourced from social media, educational videos, and research publications. These corpora are sparse relative to the diversity of tissue morphology and difficult to scale further, as pathologists do not routinely produce descriptive captions in clinical practice. We propose to instead leverage slide-level pathology reports which can be generated during routine diagnostic workflows in a scalable manner. A key challenge in this weakly-supervised setting is that pathology reports describe macroscopic, diagnostic findings while patch-level visual features describe microscopic findings. We introduce Locked Image Multiple Instance PreTraining (LIMIT), which addresses this gap by freezing a pretrained vision encoder and fine-tuning only the text encoder: report embeddings serve as cross-attention queries over bags of patch features, producing report-conditioned slide representations optimized via InfoNCE. Using LIMIT, we establish CONCH-Z, which aligns the text encoder of CONCH v1.5 to 335k clinical reports. Evaluated across 22 tasks spanning patch-level classification, tumor detection, and slide-level subtyping, CONCH-Z establishes state-of-the-art zero-shot performance over VL encoders trained on curated caption datasets, while simultaneously closing the gap with slide-level foundation models trained with substantially more parameters and compute. The code and model is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.17233
PDF
