Multi-turn Text-to-SQL is central to enterprise analytics yet remains predominantly evaluated in single-turn settings. We introduce EnterpriseMem-Bench, a multi-turn Text-to-SQL benchmark of 300 sessions and 1,400 turns built programmatically from three enterprise domains (BIRD financial, SEC EDGAR, Northwind), with deterministic ground truth and per-turn memory-critical annotation. We evaluate five frontier models -- GPT-5 mini, GPT-5.2, Claude Sonnet 4.5, Sonnet 4.6, and Opus 4.6 -- across five memory conditions enabling a three-way ablation isolating working-memory window size, episodic retrieval, and semantic augmentation as independent effects. All Claude models are evaluated with extended thinking enabled to maintain parity with GPT reasoning models. We introduce the Memory Benefit Score (MBS) as a per-turn diagnostic metric. Four findings emerge: (1) stateless multi-turn Text-to-SQL collapses to zero execution accuracy by Turn 3 across all five models, even under reasoning; (2) memory-architecture complexity does not monotonically improve accuracy -- working memory dominates, and additional components produce model- and dataset-dependent effects from +14 to -16 percentage points; (3) Claude Sonnet 4.6 underperforms Sonnet 4.5 by 17-33pp on SEC EDGAR across conditions, a generational regression persisting under reasoning; (4) under reasoning, Claude error distributions become mono-modal -- every non-correct turn is a wrong-result error. We release the benchmark, agent, and evaluation code.

Object re-identification (ReID) in egocentric kitchen videos is challenging due to rapid viewpoint changes, frequent occlusions, cluttered scenes, and large intra-class appearance variations. Objects may leave and re-enter the field of view, and the large diversity of instances with limited annotations makes supervised ReID difficult to scale, motivating zero-shot approaches. We study zero-shot object ReID on the EPIC-Kitchens benchmark, where the goal is to match active food and kitchen-tool instances across frames using only pre-trained visual features. We first evaluate five state-of-the-art feature extractors, including Vision-Language Models (VLMs) - CLIP, DINOv2, DreamSim, I-JEPA, and SAM3 - and show that zero-shot methods fail, with the best baseline achieving only 45.3% mAP. We then propose an Enhanced SAM3 ReID Pipeline, a zero-shot multi-stage method built around SAM3 segmentation as the core component. Stage 1 uses SAM3 to suppress background clutter. Stage 2 fuses embeddings from SAM3, DINOv2, and CLIP into a single L2-normalized descriptor. Stage 3 augments cosine similarity with mask-shape IoU for geometric consistency, and Stage 4 applies k-reciprocal re-ranking. The full pipeline improves performance by 7.5% mAP to 52.8%.

Deploying high-performing 3D medical image segmenters (e.g., nnU-Net) is often limited by memory footprint and inference latency. Compression is therefore necessary, but compact 3D encoders tend to lose fine structural cues (small lesions and sharp boundaries) as downsampling repeats across multi-resolution stages. We propose Detail Consistent Distillation (DCD), a stage-wise distillation framework that preserves structural detail across scales by aligning teacher-student features in a wavelet-decomposed representation. At each encoder stage, DCD distills directional detail components in the wavelet domain while leaving the coarse approximation comparatively unconstrained, avoiding over-regularization of global semantics. DCD is used only during training and introduces no inference-time overhead. Experiments on the BraTS 2024 and ISLES 2022 benchmarks demonstrate that our approach achieves superior performance in MRI segmentation using 3D multi-modal data. Code and implementation details for DCD are publicly available at https://github.com/ClinicaAlpha/DCD-3D-MedSeg.

We present a multi-modal classification framework that fuses satellite and street-level imagery through a Perceiver IO architecture operating on spatial patch tokens from a shared DINOv2 backbone. The design naturally handles a variable number of street-level views per building without padding or fixed-size pooling, and jointly predicts multi-label roof element and roof material classes. We construct a large-scale dataset of 32,135 buildings (61,672 segments) spanning ten countries, pairing satellite images with up to eight street-level views per segment and evaluating four masking strategies for isolating the target building. We propose an RGB-M masking strategy that appends the building footprint mask as a fourth input channel, providing a soft spatial prior that outperforms hard cropping across both modalities. The Perceiver IO fusion model improves over all other fusion strategies and yields substantial per-class gains for attributes visible from street level (e.g., +11.3 AP for slate, +1.3 AP for dormers), though the satellite-only baseline retains a slight advantage in macro-averaged mAP for classes that are predominantly visible from above. These results establish a scalable, flexible architecture for multi-modal building inspection that can accommodate heterogeneous inputs and multiple output tasks.

Frontier multimodal large language models (MLLMs) have been reported to achieve over 90% accuracy on fine-grained perception benchmarks. However, such scores do not necessarily imply faithful use of visual evidence. Prior studies have identified three shortcuts that inflate benchmark performance. First, linguistic priors and lexical cues in questions often enable models to infer plausible answers without seeing the image. Second, coarse global semantics from the visual encoder can bypass fine-grained local details. Third, in some ``think-with-images'' benchmarks, corrupting the intermediate images returned by visual tools barely affects the final answer. These findings suggest that higher input resolution or larger question pools alone do not elicit genuine active visual search. To address this, we introduce VisualNeedle, a challenging, information-dense, and fine-grained benchmark for scenes where critical evidence is spatially constrained to minute regions and not discernible at a glance. We further propose a counterfactual crop-black setting, which replaces crops returned by tools with black images of the same size, to test whether tool-enabled performance truly relies on intermediate visual evidence. We evaluate 9 promninent MLLMs across three settings: no-tool, standard tool-enabled, and crop-black. No-tool accuracy stays below 20\%, and the best tool-enabled model reaches only 56.01\%, still trailing the 63.00% human majority-vote accuracy. These results reveal persistent limitations in fine-grained visual search, while the crop-black ablation confirms that success on VisualNeedle hinges on genuine intermediate visual evidence.

Hepatocellular carcinoma (HCC) is biologically heterogeneous, shaped by the interplay between hepatic functional reserve and tumor-related oncologic factors; thus, similar survival outcomes may reflect fundamentally different underlying biological processes. Prognostic modeling in HCC is informed by rich multimodal information from multiparametric MRI and radiology reports from routine clinical practice. Existing prognostic vision-language models (VLMs) learn a single entangled latent representation that blends hepatic and tumor-related factors, limiting both accuracy and biological interpretability. We present BioFact-MoE, a biologically factorized Mixture of Experts (MoE) framework that explicitly decomposes liver and tumor factors via biologically supervised experts within a residual MoE survival architecture. On a HCC cohort of N=588 patients (pretrained on 4,582 3D MRI image-report pairs), BioFact-MoE consistently improves survival prediction over all baselines across time horizons, achieving 12-, 18-, and 24-month AUCs of 75.33%, 75.85%, and 73.96%. Beyond scalar risk prediction, gated expert weights enable phenotype-aware risk stratification. Pathway-informed gating uncovers clinically meaningful treatment-associated survival heterogeneity. In held-out validation, hepatic and tumor embeddings show selective associations with liver function and tumor burden markers, respectively (p<0.05), without supervision. The code is available at https://github.com/jy-639/BioFact-MoE.

We study adversarial online learning with hidden-convex losses, i.e., nonconvex losses that become convex after a nonlinear reparameterization. Ghai, Lu and Hazan (2022) proved that, under geometric and smoothness assumptions, online gradient descent (OGD) on such nonconvex losses approximately simulates online mirror descent (OMD) on the underlying convex losses with a suitable regularizer, yielding $\mathcal{O}(T^{2/3})$ regret. They left open whether the optimal $Θ(\sqrt{T})$ regret from online convex optimization can be recovered in this hidden-convex setting. We answer this question affirmatively. More specifically, via a sharper discrete-time algorithmic equivalence argument, we prove that OGD achieves $\mathcal{O}(\sqrt{T})$ regret under the same assumptions, matching the optimal worst-case rate for adversarial online convex optimization. We also address another open question of Ghai, Lu and Hazan (2022) by clarifying the geometry required for this algorithmic equivalence. We replace the diagonal-Jacobian sufficient condition with a necessary-and-sufficient Hessian compatibility condition, thereby expanding the class of admissible reparameterizations. We complement our tight regret bound with a lower bound showing that the Hessian compatibility assumption is essential for OGD; when it fails, we construct a smooth reparameterization and an adversarial sequence of hidden-convex losses for which OGD suffers $Ω(T)$ regret. Finally, we extend our analysis to one-point bandit feedback and prove a $\mathcal{O}(T^{3/4})$ expected regret bound for bandit OGD with spherical smoothing, matching its classical rate on convex losses.

We present a method for jointly predicting instance-level roof segment masks together with three continuous geometric attributes -- building height, roof slope, and roof azimuth -- from a single aerial orthophoto. Our approach extends Mask R-CNN with a dedicated attribute regression branch and introduces two key innovations: a conditional azimuth loss that suppresses supervision for flat roof segments where azimuth labels are inherently noisy, and a log-normalized height representation that addresses the heavily skewed distribution of building heights. We train and evaluate on a large-scale dataset of Dutch aerial images paired with automatically derived ground truth from 3DBAG, a nationwide LiDAR-based 3D building dataset. Using a DINOv3 ConvNeXt-Base backbone, our method achieves a mean absolute error of approximately 4 degrees for roof slope, 7 degrees for azimuth, and 1 meter for building height, with an instance segmentation AP$_{50}$ of 0.566. The predicted per-segment masks and attributes are sufficient to reconstruct simplified 3D building models (LoD2) from a single overhead image, requiring expensive 3D reference data only for training.

Large Language Models (LLMs) often exhibit homogenized cultural perspectives. While the World Values Survey (WVS) provides a gold standard for mapping human values, traditional direct prompting of LLMs on WVS often fails to access the model's latent cultural depth, leading to safety-aligned refusals or neutral responses. Here, we propose a generalizable framework for cultural evaluation and intervention that transitions from abstract queries to scenario-based behavioral probing. By extracting implicit token probabilities across 300 situational dilemmas, we bypass surface-level alignment to map the latent coordinates of LLMs cultural value. We further introduce activation steering to shift these internal alignments during the forward pass without retraining. Across multiple LLMs, we find substantial variation in adaptability and uncover a consistent phenomenon of latent entanglement, where interventions along one cultural dimension induce shifts along another. These results suggest that cultural values are encoded as coupled structures, limiting precise alignment. This work establishes a computationally efficient framework for cultural steering, highlighting the structural complexities when navigating global value with LLMs.

In many reasoning tasks, large language models (LLMs) rely on structured external knowledge, such as graphs and tables, which is typically linearized into sequential token representations. However, even when sufficient knowledge is available, LLMs can still produce hallucinated outputs, and the underlying mechanisms behind such failures remain poorly understood. We investigate these mechanisms and find that hallucinations arise from systematic internal dynamics rather than random noise. First, attention disproportionately concentrates toward shortcut-like structural cues rather than distributing across the full context. Second, feed-forward representations fail to ground the provided knowledge, causing the model to revert to parametric memory. Moreover, our results indicate that hallucination is consistently associated with failures in semantic grounding within feed-forward layers, while attention allocation exhibits greater task-dependent variability. Finally, we show that these mechanistic patterns generalize beyond single-hop graphs to multi-hop and tabular settings, enabling effective hallucination detection across structured knowledge formats.

In-context learning has recently been linked to implicit gradient descent in linear self-attention models, suggesting that context can induce a forward-pass update. Retrieval-augmented generation (RAG) also relies on context, but retrieved documents are usually treated as static evidence rather than signals for adaptation. We study RAG as an in-context optimization process. First, we show that one linear self-attention layer can implement one gradient-descent step on a unified linearized RAG objective covering both projection-based and dot-product retrieval interfaces. This gives an exact regime where retrieval-augmented prediction and in-context optimization coincide. We use this result not as a literal model of LLM computation, but as a guide for adapting the interaction between queries and retrieved evidence. We then test the boundary of this correspondence: it remains stable under controlled linear extensions, but becomes feature-distribution dependent under nonlinear architectures. Finally, we turn this view into a lightweight method for frozen RAG LLMs. The method keeps the retriever and backbone fixed, and predicts a context-conditioned update to a generator-side evidence-use interface. Across seven QA benchmarks, two retrievers, and two frozen LLM backbones, this forward-only update improves a shared-interface baseline, transfers to held-out tasks, and approaches test-time gradient adaptation at much lower per-query cost.

Standard transformer attention computes pairwise token similarity but treats all tokens as equally salient and all positions as equally local, regardless of the informational structure of the input. We identify two complementary inductive biases that standard attention lacks: energy salience (which tokens concentrate informational energy, learned end-to-end without explicit frequency decomposition) and scale-selective locality (how far positional influence extends at each frequency, implemented via Morlet wavelet encoding). We address both with two simple components. Energy-Gated Attention (EGA) gates value aggregation by a learned energy estimate of key token embeddings, computed via a single linear projection; it selects what to attend to. Morlet Positional Encoding (MoPE) replaces fixed sinusoidal encodings with learned Gaussian-windowed wavelets that adapt the joint position-frequency localization to the corpus; it specifies where attention operates at each scale. On TinyShakespeare, EGA alone achieves +0.092 validation loss improvement over standard attention (+0.103 over Phase 1-3 baseline); MoPE alone is -0.032 (below baseline as a standalone encoding); but their combination achieves +0.119 -- more than the sum of parts. This superadditivity, observed across two independent training runs, is the central empirical finding: salience and locality are complementary inductive biases, each addressing a gap the other cannot fill alone. Ablations confirm that structured spectral priors (Morlet wavelet gates, scale-initialized heads, fixed sinusoidal PE) consistently underperform their unconstrained learned counterparts, while complementary learned components interact superadditively. All experiments are at small scale (<=6M parameters, character-level benchmarks, single seed); larger-scale multi-seed validation is the most important direction for future work.

Different visual patterns appear with different frequencies in the world: e.g., beach balls appear on sand more often than they do on a road. These statistics are reflected in vision datasets, and as a result trained models more easily recognize objects in common scenarios. However, recognizing a beach ball on a road may arguably be even more important than recognizing it on sand. We study how to mitigate this discrepancy. Since collecting uncommon images in the real world may be difficult, we explore whether generating images with less frequent contexts can serve as effective training augmentation. A key challenge is guiding generations to remain close to the original dataset distribution while creating diverse images with uncommon contexts. We introduce Decoupling Contextual Patterns with Generations (DecoupleGen), a method that personalizes text-to-image diffusion models to facilitate coherent synthesis of images with rare contexts while preserving original visual details. The generated images contain semantically meaningful content and remain visually aligned with the original datasets. We further apply verification constraints to ensure relevance of the augmented data. We evaluate our approach on object classification and recognition tasks on complex scene datasets. Our experiments demonstrate consistent improvements over previous approaches, and our analyses identify factors underlying these improvements.

Retrieval is increasingly moving from one-shot matching toward interactive reasoning, where language agents iteratively inspect evidence, reformulate queries, and search again. Training such agents raises a credit-assignment challenge: executable actions such as queries or summaries can be directly evaluated by the retriever, while latent reasoning steps are not directly observable and only affect future executable actions. This asymmetry makes outcome-level reward assignment unreliable, as the same final reward may credit reasoning steps that did not actually shape retrieval success. We propose RICE-PO, a critic-free policy optimization framework that converts retrieval interactions into localized learning signals. RICE-PO selects high-uncertainty executable actions as anchors, evaluates local counterfactual branches using retrieval metrics, and propagates credit to latent reasoning steps only when reasoning-to-action influence is strong and future residual effects are stable. On BRIGHT and BEIR, RICE-PO consistently outperforms prompt-based agents and group-based RL baselines under the same retriever setting. These results show that the structure of agent-environment interaction itself can provide useful supervision for training reasoning-based retrieval agents.

In-context learning (ICL) is often motivated by the intuition that demonstrations help because they provide correct input-output examples. However, we reveal a counterintuitive phenomenon: correctness does not guarantee exemplar utility, and some correct demonstrations can even reduce ICL accuracy. To study this correctness-utility gap, we introduce task-preserving perturbations, where only the exemplar input is changed, while the example remains a correct instance of the same task. Concretely, each perturbed exemplar is assigned the target induced by the task mapping. This framework covers both label-updating perturbations, where task-relevant semantics change and targets are recomputed, and stricter target-preserving perturbations, where the original target remains valid. We formalize the resulting failure mode as contextual evidence shift: task-preserving perturbations can change the effective mixture of evidence used by the model for contextual inference, thereby separating exemplar correctness from exemplar utility. Across sentiment classification, logical reasoning, and math word problems, we find that task-preserving perturbed demonstrations can substantially degrade ICL performance, especially for smaller models, harder tasks, and higher perturbation ratios. Our results show that robust ICL requires evaluating not only whether demonstrations are correct, but also how they influence contextual inference. Code is available at https://github.com/Chenghao-Qiu/Task-Preserving-ICL.

Industrial automation is at a pivotal moment, as Physical AI is driving a transition from rigid, hand-engineered automation systems toward more flexible and adaptive systems. This shift has created a growing demand for large-scale, real-world robot demonstration data, making teleoperation an increasingly important mechanism for data collection. However, high-quality teleoperated demonstrations remain difficult to obtain in practice, as novice operators often produce episodes that are task-successful but suboptimal for downstream use due to inefficient motion, repeated corrections, or operation near robot joint limits. We present a Data Quality Assessment and Feedback (DQAF) framework that closes the loop in teleoperation by providing immediate post-episode feedback grounded in semantic task progress and robot telemetry. The framework extracts quality relevant signals such as sub-task progress, motion smoothness, stalls, kinematic limits and converts them into structured quality assessments and actionable natural-language feedback. Unlike binary success or failure feedback, the proposed system explains why an episode is suboptimal and highlights specific behaviors to correct in the next trial. We evaluate the framework through a diagnostic validation study and a pilot user study. In the validation study, the system is compared with a human reviewer during dataset curation, producing rejection reasons and actionable feedback for improvement. In the pilot study with three novice operators across two manipulation tasks, the operator who received the systems immediate, automated post-episode feedback improved faster than those who did not, producing higher-quality demonstrations sooner.

Mobile robots can fail before they collide: a velocity that is safe now may commit the robot to a passage that moving obstacles will soon close. We study this predictive near-miss commitment problem and propose Risk-Sensitive Conjectural Scenario Planning (RCSP), a planning layer that evaluates candidate commands against plausible short-horizon obstacle futures. RCSP maintains a lightweight belief over local motion conjectures, samples future interactions, penalizes high-risk tails, and executes through a local safety check. In controlled MuJoCo bottleneck tasks, the RCSP planner reaches the goal without collisions and yields higher secondary safety and path-quality point estimates than a non-adaptive predictor, with additional latency. In ROS2/Gazebo, adding the local safety layer to a standard Nav2 stack reduces dynamic near-miss failures. On official DynaBARN/Jackal transfer, tuned DWA and TEB remain stronger on strict benchmark success, revealing the boundary of the approach. These simulation results position RCSP as a predictive-risk module that complements existing navigation stacks in dynamic bottleneck regimes.

Objective: To describe the design and early clinical evaluation of The Daily Dose (TDD), an LLM-driven, automated clinical summarization and clinical-trial identification system integrated into routine radiation oncology practice. Design: Mixed-methods evaluation using a cross-sectional, anonymous clinician survey administered after 1 month of system deployment. Exposure: Daily automated delivery of physician-specific email summaries generated using RadOnc-GPT, including patient schedules, concise EHR-derived clinical-status summaries, and automated identification of potentially relevant clinical trials for new or consult visits. Main Outcomes and Measures: Primary outcomes included self-reported usability, satisfaction, perceived usefulness, perceived impact on workflow, time savings, and intention for continued use. Internal consistency reliability was assessed using Cronbach's $α$. Results: Among 55 respondents, 52 (94.5\%) worked in radiation oncology, and 38 (69.1\%) were attending physicians. Most participants (83.6\%) reported using TDD daily or several times per week. Mean (SD) scores were 3.89 (1.04) for usability and satisfaction, 3.43 (1.24) for perceived usefulness, and 3.80 (1.17) for impact and future use (5-point Likert scale). Overall satisfaction was positively associated with perceived time savings ($p < .001$). Participants reported variable time savings, with 27\% estimating $\geq 10$ minutes saved per day. The questionnaire demonstrated excellent internal consistency (overall Cronbach's $α$ = 0.97).

Mechanistic interpretability has identified small sets of attention heads that implement specific behaviours in transformer language models, but recovering these circuits typically requires a bespoke analytical pipeline for each new task. We recast circuit discovery as a reinforcement-learning problem. An agent operates over the 144 attention heads of GPT-2 small as a discrete action space; each action triggers a zero-ablation and a contrastive reward that subtracts the ablation's damage to general next-token prediction from its damage to the target task. A single PPO policy, trained on two tasks (induction and IOI) in a vectorised multi-task environment, attains the per-episode oracle on both training tasks and on a held-out third task (docstring completion). Its preferred heads coincide with the canonical heads of established literature on precisely the axes those papers identify as causally non-redundant under single-head ablation; the categories they identify as redundant are correctly de-prioritised by the agent. On the held-out task, best-of-five planning recovers 96\% of the oracle ceiling with no task signal supplied at evaluation. These results indicate that reinforcement learning over causal interventions is a viable, transferable substrate for identifying the single-head bottlenecks of mechanistic circuits, complementary to existing path-patching approaches.

Physics-informed machine learning (PIML) integrates mechanistic knowledge, typically in the form of partial differential equations (PDE), into data-driven models. Despite strong empirical performance, its statistical generalisation properties remain poorly understood, particularly in the regression setting with unbounded losses. Existing analyses rely on approximation or stability arguments and do not fully capture how physical structure influences generalisation from finite data. In this work, we develop a PAC-Bayesian framework for PIML that provides high-probability generalisation guarantees in the presence of unbounded losses. We adopt a multi-task perspective that jointly treats data fidelity, PDE residuals, initial and boundary conditions, avoiding the looseness induced by standard union-bound approaches. Our analysis leverages the structure of physics-informed objectives to derive novel bounds where the complexity scales with input-gradient norms of the losses, revealing a direct link between physical regularity and generalisation. We instantiate this framework under Sobolev and Poincaré-type assumptions, yielding two classes of bounds that trade off statistical complexity and smoothness in different regimes. Building on these results, we propose a self-bounding-aware learning algorithm that directly optimises tractable surrogates of the derived bounds, along with a practical procedure to estimate the associated constants in realistic settings. Empirical evaluations on standard PDE benchmarks demonstrate that our bounds are non-vacuous, significantly tighter than union-bound baselines, and can be effectively minimised during training. Overall, our results provide a principled statistical foundation for the generalisation of physics-informed models.

Autonomous research agents produce competitive solutions and professional-looking manuscripts, yet their outputs contain verifiability failures undetectable by surface-level evaluation: fabricated citations, unreproducible scores, and method descriptions that diverge from the implementation. We address this through three contributions. First, Chain-of-Evidence (CoE), a verifiability framework requiring every claim to be traceable to its evidence source. Second, ScientistOne, an end-to-end autonomous research system that maintains evidence chains by construction throughout literature review, solution discovery, and paper writing. Third, CoE Audit, a post-hoc audit whose four integrity checks -- score verification, specification violation, reference verification, and method-code alignment -- apply uniformly to all systems. Across 75 papers spanning five systems and five frontier research tasks, every baseline exhibits at least one systematic failure mode: hallucinated reference rates reach 21%, score verification passes in as few as 42% of papers, and method-code alignment ranges from 20% to 80%. ScientistOne achieves zero hallucinated references (0/337), perfect score verification (12/12), and the highest method-code alignment (14/15), while matching or exceeding human expert performance on all five tasks. ScientistOne further generalizes to six additional tasks spanning medical imaging, fine-grained recognition, 3D perception, and language modeling, achieving state-of-the-art on Parameter Golf and gold medals on MLE-Bench tasks where baselines fail entirely.

Scalar post-training quantizers discard pairwise coordinate structure within weight rows. We introduce QAM-W (Quadrature Amplitude Modulation for Weights), a codec that recovers this structure: each row is L2-normalized, block-Hadamard rotated, paired into 2D coordinates, and quantized against a single Lloyd-Max codebook trained on the unit circular Gaussian, with activation-aware per-channel scaling. In a cross-model study spanning five LLMs from four families (1.1B--13B parameters) and eight quantized configurations, the activation-aware variant at $\approx 5.5$ bpw stays within $\pm 0.4\%$ of BF16 WikiText-2 perplexity on every model, matching the SmoothQuant W8A8 quality envelope at $32\%$ fewer weight bits. Joint 2D coding outperforms polar (amplitude $\times$ phase) coding by 2--15~pp $Δ$PPL at equal bitrate, and paired KL against BF16 tracks $Δ$PPL\% at Spearman $ρ= 0.99$ across 37 (method, model) rows, consistent with a monotone composite bound from codec distortion to KL divergence. A 3.5~bpw variant is competitive on quantization-tolerant architectures. At strict 4~bpw, the rotated-codebook frontier method QTIP outperforms QAM-W; the contribution is the quality-preserving 5--6~bpw band.

Educational virtual laboratories can make experimental training more scala-ble, adaptive, and accessible, especially when students have limited access to physical laboratory facilities. However, authoring new simulated laboratory procedures remains costly: educators must describe new equipment, define how instruments and materials interact, and specify valid procedural flows that can be executed or assessed inside the virtual environment. Large lan-guage models can assist in this authoring process by generating detailed ex-perimental procedures, but their output should not be treated as directly exe-cutable plans. They may omit necessary actions, arrange steps in the wrong order, or produce instructions that are logically incorrect or incompatible with the laboratory equipment. This paper presents a prototype framework for managing uncertainty in LLM-generated procedural knowledge for virtu-al laboratory planning. The framework aims to reduce procedural uncertainty by using structured domain representations and uncertain LLM-generated state-transition samples to extract candidate procedural rules, transform them into explicit and inspectable constraints, and use them to repair uncertain procedural steps. Although the motivating domain refers to educational vir-tual laboratories, the underlying problem is more general: managing uncer-tain procedural knowledge for action planning in structured interactive envi-ronments. We illustrate the approach in a virtual laboratory domain involving laboratory instruments, containers, tools, and material-transfer actions.

Machine unlearning aims to remove specific concepts from pretrained text-to-image diffusion models, yet several white- and black-box attacks have been introduced to make the model generate such unlearned concepts. These attacks, nevertheless, do not assume a realistic threat model, i.e. they either assume access to the model weights, or result in gibberish adversarial prompts that could be easily detected even through naive rule-based safeguarding. We aim to address this gap in this paper. We introduce BEAP, a black-box, embedding-aware adversarial prompting attack that leverages a large language model (LLM) to iteratively generate effective adversarial prompts and exploit such hidden vulnerabilities. BEAP performs an embedding-aware search in text space, combining multiple reward signals: unlearned concept presence, text-image alignment, and image quality, to refine generated prompts. Unlike previous attack methods, BEAP keeps its prompts undetectable to safety filters while producing high-quality images. Extensive experiments show that BEAP improves the Attack Success Rate (ASR) by more than 60% over prior methods, while requiring only an average of fifteen prompts per successful attack. Warning: This paper contains model outputs that may be offensive or upsetting in nature.

Small aerial robots are particularly well-suited for search and rescue in confined and hazardous environments due to their agility, low cost, and ability to traverse through cluttered spaces that are inaccessible to larger platforms. However, enabling autonomous navigation in complete darkness remains a significant challenge, because small aerial robots cannot easily accommodate perception systems that demand substantial payload, power, or computation. In this work, we present a lightweight perception approach that combines a monocular event camera, a coded aperture lens, and an infrared dot projector to enable navigation in such conditions. The projected pattern, when imaged through the coded aperture, produces depth dependent blur signatures that implicitly encode scene geometry. We train a convolutional neural network to decode these signatures into dense depth maps using only synthetic data generated from a simple planar wall setup. Despite this minimal training regime, the model generalizes zero-shot to complex real-world scenes. Our system operates in real time at 20 Hz on a NVIDIA Jetson Orin Nano, demonstrating suitability for resource-constrained platforms. We further analyze the impact of different coded aperture designs on depth estimation performance. Our approach gives high accuracy (l1 error 7.0cm) upto 2.5m range (2.80% error). These results highlight the potential of combining structured illumination, coded optics, and event-based sensing for enabling robust perception and navigation in complete darkness.

Current benchmarks for occupational AI agents are scoped primarily by economic values, telling a replacement story. We introduce JobBench, which evaluates AI agents on the workflows that experts identify as high-priority for delegation, empowering humans based on their needs instead of replacing them with GDP value. JobBench covers 130 agentic tasks across 35 occupations. Each task is packaged as a workspace of heterogeneous reference files, requiring the agent to reason through the cluttered information streams of real professional work. Outputs are graded by a fact-anchored chain of rubrics, averaging 35.6 binary criteria per task. We evaluate 36 models; the strongest, Claude Opus~4.7 under Claude Code, reaches only 45.9 %. We hope JobBench shifts the community's target labour-market effect from replacement to enhancement: building agents that do what humans actually want delegated, not only what is most economically valuable.

Radars are an ideal complement to cameras: both are inexpensive, solid-state sensors, with cameras offering fine angular resolution, while radars provide metric depth and robustness under adverse weather. However, radar data is more difficult to interpret than camera images and varies significantly between sensors, necessitating increased reliance on simulation for prototyping sensors and processing pipelines. Recent work treating radar reconstruction as a novel view synthesis problem has shown great promise in reconstructing radar-relevant geometry and simulating low-level radar data. However, such methods are constrained by the low spatial resolution of the underlying radar. To address this, we propose a unified differentiable renderer, RadarSim, which leverages the high angular resolution of RGB cameras to generate Doppler radar range images from a camera-initialized neural field. Using a novel data set of calibrated radar camera recordings from a custom hand-held rig, we demonstrate that RadarSim produces sharper geometry and Doppler range frames than radar-only reconstructions.

Shampoo-based methods, such as KL-Shampoo and SOAP, have demonstrated strong performance in training neural networks and rely on QR decomposition. Because existing QR implementations require single-precision (FP32) arithmetic and remain computationally expensive, these methods become time- and memory-intensive when their preconditioning matrices are large. Moreover, using BFloat16 (BFP16) storage to reduce memory usage can degrade the performance of Shampoo-based methods. We propose a reparametrization of the preconditioner that supports BFP16 storage and forms a complete basis by combining updated basis vectors with unchanged ones. By updating only part of the basis through QR decomposition in a subspace, our approach reduces computational overhead while mitigating the performance degradation caused by BFP16 storage. Our approach applies broadly to Shampoo-based methods that employ QR decomposition, including KL-Shampoo, SOAP, and KL-SOAP. In particular, it improves the performance of SOAP and KL-SOAP under BFP16 storage, enabling KL-SOAP to match or exceed KL-Shampoo. Overall, our approach makes Shampoo-based methods more memory- and time-efficient.

Learned physics simulators are often evaluated by one-step or short-horizon prediction error, but these metrics can miss failures in temporal composition and long-horizon rollout. For autonomous, state-complete systems, exact solution maps satisfy a semigroup law: direct evolution over $s+t$ should agree with evolution over $s$ followed by $t$. We propose normalized semigroup error as a post hoc, model-agnostic diagnostic comparing these direct and composed learned predictions. On one-dimensional heat and Burgers dynamics with time-conditioned ConvNet and FNO baselines, semigroup error is positively associated with rollout degradation, with trajectory-level Spearman correlation $ρ= 0.635$ and $95%$ CI $[0.621, 0.649]$. Semigroup regularization has mixed effects, supporting semigroup consistency primarily as an evaluation diagnostic rather than a universally beneficial training objective.

Theory of Mind (ToM), the ability to infer others' knowledge, intentions, and emotions, is commonly evaluated in large language models (LLMs) using end-point question answering, where performance is judged solely by the final answer to a social reasoning query. This paradigm obscures whether the model actually constructs the underlying mental-state representations required for robust reasoning, particularly in scenarios involving divergent, evolving, or mistaken beliefs. In order to address this research gap, we introduce OmniToM, a benchmark that directly evaluates these representations by requiring explicit modeling of belief structures for all relevant actors within a narrative. These structures are composed of belief propositions: minimal statements of what an actor takes to be true about the world or another actor's mental state, allowing knowledge, intentions, emotions, and false beliefs to be analyzed in a common format. Models are evaluated in two stages: Stage 1: Belief Extraction, which extracts from the story the beliefs relevant to its social dynamics, and Stage 2: Belief Labeling, which assigns each belief a seven-dimensional schema label covering recursive order, truth status, knowledge access, explicitness, content type, mental source, and context. Built from 895 stories from the existing ToMBench story corpus and augmented with 22,343 labeled belief propositions, OmniToM uses a human-calibrated LLM-assisted annotation pipeline. Across diverse models in zero-shot evaluation, OmniToM reveals an actor-specific belief-tracking bottleneck: current LLMs struggle with the knowledge-access and representational decisions required to transform narrative facts into actors' beliefs and shared mental states.