Physical AI Brief

With the rapid advancement of image editing models and their widespread application across various domains, there is an increasingly urgent need to deploy these model capabilities directly into real-world scenarios. However, existing benchmarks remain confined to simple single-image tasks, suffering from limited coverage dimensions and an inability to effectively differentiate performance among diverse models. Consequently, they fail to reliably evaluate model performance in complex multi-image editing, highly demanding reasoning instructions, and practical deployment settings. To address these limitations, we propose CPI-Bench, a Comprehensive, Practical andIntelligent benchmark for real-world image editing. CPI-Bench comprises three core subsets: CPI-General-Bench, which comprehensively covers diverse editing tasks and pioneers the inclusion of multi-image editing evaluation; CPI-Practical-Bench, which focuses on high-frequency real-user application scenarios; and CPI-Intelligent-Bench, which is dedicated to evaluating capabilities in highly demanding reasoning-based editing. Evaluation results of mainstream image editing models based on CPI-Bench demonstrate that CPI-Bench enhances performance differentiation among models. It provides a comprehensive and reliable quantification of gaps in general editing capabilities, practical deployment efficacy, and advanced reasoning-based editing, offering invaluable guidance for the future optimization of image editing models. Crucially, our ranking analysis reveals that CPI-Bench achieves the highest alignment with the Arena Image Edit Leaderboard, indicating it faithfully captures the preferences and perceptual judgments of human evaluators, serving as a robust proxy for real-world user experience.

Interactive game world models typically autoregress visual observations directly in pixel or latent space, forcing structured properties such as pose, geometry, and occlusion to be implicitly maintained by the same generative sequence. Over long horizons, errors in these latent world properties accumulate, making consistency and controllability fragile. We explicitly model the evolving world state, delegate exact geometric computation to a fixed, zero-parameter renderer, and leave the neural model to synthesize appearance. We instantiate this idea as Marionette, a world model for interactive games with articulated characters. First, a two-stage autoregressive dynamics model predicts an explicit and interpretable 276-dimensional 3D world state comprising multi-entity articulated skeletons, metric root trajectories, and rotations. Second, a zero-parameter graphics bridge converts the predicted state into pose-control videos, computing world-space geometry and occlusion in closed form. Third, a control-conditioned video-diffusion observation model synthesizes photorealistic RGB observations from the resulting structured controls. Our experiments establish two properties of Marionette. First, the predicted world state is directly controllable. Forcing a mismatched action stream changes root-aligned joint error by 31% across 48 held-out segments. Second, long-horizon behaviour is determined in the state, and can be repaired there. Left free, the two generated characters drift to 21.2 m apart (recorded sessions stay near 5 m) and a third of frames show ground penetration. Two rules imposed on the explicit state, a terrain collider and a separation cap, cut penetration by 66% and keep the pair engaged, with no change to the observation model. Routing appearance through the predicted state costs no fidelity we can detect, at an FVD of 831 against 799 for recorded pose.

This study investigates the methodological and theoretical properties of session handover in applications that use large language models. A task may continue in a new session when the context reaches the model's input limit, when the application restarts, or when another agent is asked to finish the task. The application must then decide which information from the earlier session to pass on. We formulate handover as the transfer of a task-relative in-context learning (ICL) state and distinguish exact recovery of earlier material from preservation of the target distribution. Under an exogeneity condition, predictive equivalence characterizes the coarsest deterministic sufficient handover and gives a fixed-length bit requirement. The analysis isolates the effects of the memory constraint, the writer, and the continuation procedure, and quantifies the cost of writing before the realized downstream query is known. We propose a three-part record that stores decisions and constraints exactly, uses task-justified statistics for repeated evidence, and retains original observations whose effect is not preserved by those statistics. Gaussian linear regression gives an exact finite-dimensional handover and finite-bit perturbation bounds, while nonparametric regression gives upper and lower bounds that relate memory to squared prediction error. These results provide a theory and method for deciding what a handover must retain and how its memory requirement depends on the continuation task.

As AI systems make more morally loaded decisions across society, one response has been moral preference elicitation. In this approach, researchers poll participants on hypothetical dilemmas and use the aggregated votes to train a policy that an AI model then applies at scale. Before any vote is cast, developers make three key choices in the moral AI elicitation pipeline: feature scoping, voter sampling, and question framing. In other words, they decide which features go to a vote, which voters to include, and how to present the question. These choices are often opaque, undocumented, and treated as technical details rather than normative ones. We examine each of these choices within a common empirical study and show that each can shape the preferences produced by moral AI elicitation. Across two phases (N = 809) in three deployment contexts (i.e., AI kidney allocation, AI agents simulating absent workers, and generative AI depictions of the deceased), we examine the three main stages of the moral AI elicitation pipeline. First, morally relevant features shift across contexts. This suggests that feature schemas should not be assumed to transfer across deployment domains. Second, preferences differ by political ideology for roughly one-third of features, with some differences reversing direction. The ideological composition of the voter pool can therefore affect the resulting aggregated preference profile. Third, the wording of the elicitation question can narrow or widen ideological gaps by up to a full scale point. The framing conditions also change how moral foundations are associated with participants' judgments. Taken together, these findings suggest that voting-based alignment cannot deliver fair or transparent AI by aggregation alone; at minimum, each stage of the moral AI elicitation pipeline should be audited and disclosed.

Vision-language models (VLMs) enable embodied agents to reason and act from visual observations and language instructions. Reinforcement learning (RL) post-training enhances these capabilities using task feedback, but current on-policy RL runtimes execute rollout, reference scoring, and actor training in strict serial phases. While effective for text-only RL, this phase-granular execution is wasteful for VLMs, where processing dense video inputs and prompt prefixes occupies a large fraction of each phase. Because prefix processing is independent of the generated response, it can be run alongside rollout decoding, which leaves GPU compute capacity underutilized, without breaking synchronous on-policy semantics. We present Rollplex, a runtime that decomposes the reference and training phase and moves the prefix computation into the rollout decode window. Realizing this schedule requires more than concurrent kernel launches: naive colocation of Qwen2.5-VL-32\,B requires roughly 165\,GiB per GPU, while rollout and training prefer different tensor-parallel (TP) degrees and weight layouts. Rollplex addresses these constraints with two mechanisms. Phase-aware memory management controls HBM residency according to producer--consumer lifetimes. Parallelism-aware weight sharing uses the same physical storage for layout-compatible tensors across distinct TP degrees and reconstructs only incompatible tensors, avoiding a complete second actor copy. On 32 H800 GPUs, Rollplex achieves $1.23\times$--$1.30\times$ speedup over serial colocation and $1.57\times$--$2.24\times$ over disaggregation under the same GPU budget, while preserving the synchronous RL update.

Cross-Tabular Data Generation (CTDG) seeks to learn a generative model from multiple heterogeneous tables and produce new synthetic tabular datasets. However, existing synthetic tabular data generation methods are largely restricted to single-input-table scenarios and struggle to effectively handle multiple heterogeneous tables with diverse feature sets. To address this limitation, we propose a two-stage framework for cross-tabular data generation. In the first stage, each heterogeneous raw table is transformed into a standardized statistical table with the same set of columns across all tables. Each statistical table captures the marginal distributions of the original columns and the pairwise correlations among them. In the second stage, a diffusion transformer model is trained to capture structural patterns across these homogeneous statistical tables and to generate synthetic statistical tables. Synthetic raw tables are subsequently reconstructed from the generated statistical tables via multivariate Gaussian sampling followed by an inverse probability integral transform. This two-stage CTDG framework enables the learning of a unified generative model from multiple heterogeneous tables and supports the generation of an unlimited number of realistic synthetic heterogeneous tables. Experimental results demonstrate high fidelity in the learned statistical representations and a favorable fidelity-diversity trade-off in the generated synthetic data, validating the effectiveness of the proposed approach.

We present a Test-time World-model Inference (Twin) system, in which a frontier coding agent writes an executable world model for completing continual learning tasks, such as ARC-AGI-3 games. Traditional approaches hand-engineer such models, one custom design per task. Each game hides its rules and goal, and our system constructs them from simulation and interaction alone. Its inductive prior over grid games is strong enough to recover the true transitions of the game and the goal on nearly all levels. Replay validation happens in a twin world model. The harness enforces that an action is not made until the program reproduces every previous observed game transition. Each mismatch between a world model prediction and the actual action result becomes a counterexample that is used to repair the world model. Twin clears 179 out of 183 levels (97.8%), and does so more efficiently than humans in 158 out of 179 levels (88.3%). The system infers the goal before any reward on 156 of the levels it clears (87.2%), and in the remaining levels automatically discovers the goal by search. The benchmark scores completion and action efficiency, between 0 and 100, against humans playing each game for the first time. Played directly, the base model scores only 7.8%; an off-the-shelf harness increases it to 61.1%, whereas our twin world model increases the same base model to 93.3%, clearing 23 out of 25 games. Building a usable world model is simpler than anticipated, whereas the harder problem is inferring the right goal.

As heterogeneous robotic systems deploy across diverse urban zones, maintaining safety amid complex human-robot interactions remains a critical challenge. We present a unified framework that bridges systematic hazard analysis and runtime enforcement using hazard-informed safety envelopes. Rather than treating safety as a static constraint isolated within individual software modules, we introduce a cross-layer safety transformation process spanning symbolic, spatial, and dynamic world models. We show how this representation naturally interfaces with physical AI runtime harnesses to guarantee safe urban mobility.

An autonomous robot efficiently exploring an unknown environment, such as looking for water sources on Mars, faces two simultaneous demands: building an accurate information map while quickly finding the regions of greatest value, and paying for every meter of travel and the cost of every measurement it takes. Classical information-seeking and reward-seeking criteria address only one of these objectives at a time. Here, we propose Expected Free Energy (EFE), the principled action-selection objective from active inference, as a unifying criterion for budgeted robotic informative path planning. Maintaining a Gaussian-process belief over the information field, our agent plans continuous trajectories that minimize expected free energy under hard path-length constraints. The results from multiple realizations show that EFE-based planning yields accurate posterior maps and locates the highest-value regions simultaneously, outperforming information-theoretic baselines under the same settings. In robotic exploration, these unified, easy-to-tune principled information-gathering strategies facilitate autonomous deployment while enforcing efficiency and resource constraints.

A frozen language model on reasoning tasks has two coupled weaknesses: it under-uses evidence its own residual stream already encodes, and it fails to detect when the input is insufficient to answer, so it confabulates. This paper consolidates two research lines that address these on the same residual stream: a conditional steering probe writes the stream at mid-stack layers and recovers reasoning accuracy from a frozen backbone, and a zero-shot sufficiency direction reads the stream and abstains when information is insufficient. Deployed in one forward pass they interfere: the steering write shifts the state the direction reads, costing up to 8 AUROC points of cross-domain transfer on small models; a separate clean pass doubles inference cost. We keep the direction fixed and train a small network to reconstruct the pre-steering residual from the steered one -- mean-squared error on (steered, clean) pairs, no sufficiency labels -- and read the direction on the reconstruction. The resulting system, YOPO (You Only Pass Once), answers, steers, and abstains in one forward pass of a frozen Qwen2.5 backbone (1.5B/3B/7B). End to end, three-way accuracy more than doubles the frozen baseline (0.375->0.798 on 1.5B alphaNLI) and one pass beats the two-pass reference at every scale (0.798/0.830/0.893 vs 0.753/0.790/0.863) and on ten backbones across six model families. We chart the capacity-transfer frontier quantifying the principle that abstention should not be trained in; a source-side audit catches our own alphaNLI construction leaking a surface artifact, so architectural claims are anchored on native-label replications (SQuAD2, RepLiQA, MuSiQue); and on the standard four-domain suite we contribute, to our knowledge, the first answer-or-abstain benchmark, where our gate tops every in-domain dataset and the label-free direction is the only gate family to survive domain transfer.

This paper presents THRIVE (Therapeutic Humanoid Robot In Virtual Environment), an at-home rehabilitation platform that integrates a suite of virtual-reality upper-body rehabilitation games, a real-time camera-based motion-tracking system, and a socially interactive robot therapist. The system is designed for therapy and intervention in children with upper-limb motor impairments, which can be improved through consistent, task-specific practice. THRIVE features a set of newly designed, engaging games that target functional reaching, grasping, and object-manipulation movements through customizable popping, hitting, catching, and grabbing tasks, while the camera-based tracking system captures the child's kinematic performance during play. A robot therapist - deployable either as a physical robotic coach or as a remote-presence virtual agent - delivers adaptive, dynamic feedback to motivate the child and guide their movements toward therapeutic goals. THRIVE decouples the therapeutic games from the robot embodiment, extending the platform to support various embodiments and different robots within one modular system. This robot-agnostic design makes THRIVE affordable, scalable, and readily adaptable for sustained use in the home, offering a practical pathway to more consistent and engaging upper-limb therapy for children with motor function impairments.

Effective Hamiltonians connect microscopic interactions to measurable collective behavior in quantum materials, but determining their parameters directly from experiment remains a challenging inverse problem. We introduce a supervised machine-learning framework that infers Hamiltonian parameters from nonlinear terahertz two-dimensional coherent spectra. A calibrated forward model generates spectra from candidate Hamiltonians, a common preprocessing pipeline maps simulated and experimental spectra into the same representation, and a neural network learns the inverse map from spectral fingerprints to microscopic parameters. We demonstrate the approach for rare-earth orthoferrites using a two-sublattice Landau--Lifshitz--Gilbert spin model with exchange, Dzyaloshinskii--Moriya interaction, anisotropies, and damping. Synthetic benchmarks show that nonlinear spectra encode parameters beyond those fixed by the linear response, with inference accuracy tracking the physical spectral sensitivity and robustness against noise improved by using multiple inter-pulse delays. Applied to experimental THz-2DCS data from Sm$_{0.4}$Er$_{0.6}$FeO$_3$, the inferred parameters yield physically reasonable forward simulations, while remaining discrepancies identify limitations of the reduced model. These results establish THz-2DCS as a data-rich platform for effective-Hamiltonian inference and model refinement, enabling experimentally driven identification of microscopic interactions while providing a foundation for understanding, predicting, and ultimately controlling the emergent properties of quantum materials.

The majority of work on summarization evaluation focuses on general summary quality (e.g., ROUGE, BERTScore) or specific desired properties (e.g., readability, factuality). However, these metrics fail to measure the utility of a summary to an individual user. For example, a biomedical researcher learning about the latest vaccine research will have different informational needs from a family doctor. Query-focused summarization captures part of this need, but in practice, users rarely state everything relevant in a query: a single short query is likely inadequate to distinguish the needs of a researcher from those of a physician. By contrast, a reader's background or persona (their role and expertise) is comparatively stable across queries and recovers much of this missing context, which makes it a practical signal for assessing whether a summary satisfies that reader's needs. In this work, we assess how sensitive popular summarization metrics are to both informational and persona differences, and find that many popular metrics, including strong LLM-as-judge metrics, fail basic perturbation tests of informational content. We additionally conduct an expert human evaluation, measuring summary preferences based on information satisfaction given a specific person's background and use case. We find that both traditional and LLM-based metrics are insufficient measures of information satisfaction and agree poorly with human judgment.

Spreadsheets are widely used to organize, analyze, and manipulate semi-structured data, yet automated spreadsheet reasoning remains challenging for large language models (LLMs). Real-world workbooks often contain implicit cross-table associations, fine-grained column dependencies, and complex spatial layouts. Existing methods typically flatten these multidimensional structures into sequential strings, losing important intra-sheet boundaries and inter-sheet semantics. Consequently, LLMs cannot exploit the global spatial context that human experts naturally use when inspecting spreadsheets. We propose SheetCompass, a graph-guided and memory-driven agentic framework for spreadsheet reasoning and automation. SheetCompass explicitly models structural relationships within and across worksheets while maintaining task-relevant information in memory, enabling agents to reason more effectively over complex workbooks.

In the current artificial intelligence-driven innovation era, the pace of knowledge growth is accelerating, and is hard to keep up with. While generative models are increasingly used to synthesize content, they often lack in information grounding. To address these peculiarities of our time, we propose Wyvern, a multi-agent framework for the automated generation of grounded, multimodal technical reports. Wyvern allows for the generation of multimodal outputs, integrating images, tables, and text with supporting references in a unified report. Additionally, a particular focus is placed on the grounding of the content, with the implementation of a claims auto-revision stage. We conduct a human evaluation study to assess the quality of our proposed framework. The results show that the figures' informativeness is perceived as superior to that of a recent baseline in 87% of cases. Furthermore, Wyvern's reports are rated as more useful than those produced by three alternative methods in 63% to 100% of instances. We also carry out automatic evaluations showing that Wyvern gains up to 2.3$\times$ in citation recall and 1.6$\times$ in citation precision with respect to the baselines.

Self-evolving agents improve future behavior from interaction experience, yet existing evaluations typically optimize under fixed execution conditions and do not test recovery after those conditions change. To address this gap, we introduce PACE-Bench (Physics Adaptation via Code Evolution), a simulator-grounded benchmark of 144 source-to-target adaptation pairs across six physics domains. Each pair links a source environment to a mutated target environment with the same goal and interface. A code-driven design that succeeds in the source fails in the target, where agents must iteratively adapt it into a working target design using diagnostic sandbox feedback within a limited attempt budget. We compare ten self-evolving methods from four paradigms. The benchmark remains far from saturated: Reflexion + Qwen3-14B succeeds on only 35.9\% of full-benchmark pairs, while GPT-5.5 solves 66.7\% of the Statics subset under the full budget. Together, these results show that simulator-grounded reflection is more reliable than unverified self-revision, while memory anchors agents to early designs and broad tree search explores without converging. Even revealing exact physical changes does not raise the performance ceiling, pointing to mechanism redesign rather than parameter inference as the central bottleneck. Data and code are available at https://github.com/thunlp/PACE-Bench.

Reinforcement learning (RL) post-training provides a direct way to align diffusion models with human preferences and task-specific rewards. However, current RL algorithms for diffusion models remain fragmented: reverse-trajectory methods rely on discretized likelihood ratios, whereas forward-matching methods train on reward-labeled noising versions of the rollout samples. This paper shows that these seemingly different losses arise from a single path-space principle. Starting from the regularized diffusion-RL objective, we use importance sampling between sampling SDEs to obtain an explicit policy-gradient estimator on trajectory space. The estimator contains the stochastic Itô integral underlying Flow-GRPO-type updates; we derive an equivalent variance-reduced value-gradient form that recovers the forward-matching structure of AWM and DiffusionNFT. This identifies the empirical gap between these method families as a variance-reduction effect rather than a difference in RL principle. The derivation yields a unified design space organized by value-gradient estimation, weight functions, and sampling choices. Within this space, we propose a multi-sample KDE value-gradient estimator that reuses rollout groups, together with scale-bounded weight families that retain stable existing recipes while excluding singular ones. Experiments on SD3.5-M and Qwen-Image models validate the variance-reduction explanation and show that the resulting recipe improves over prior diffusion-RL baselines.

LLM evaluations often use fixed sampling budgets, testing every item the same number of times even after estimates are precise. We introduce optstop, a precision-based adaptive stopping framework that treats evaluation as a sequential measurement problem: keep sampling where uncertainty remains high, and stop where estimates are precise or stable enough. The framework builds on hierarchical Bayesian inference, supports binary, ordinal, and continuous outcomes, and keeps every benchmark item eligible for sampling, without requiring a calibrated item bank. It runs live or retrospectively, and includes a safeguard that samples more cautiously as measured performance approaches zero, where rare successes matter most. In an illustrative 200-item, 10-epoch evaluation, it removes 57%-97% of planned trials across nine validation settings, with overall conclusions equivalent to the full run. These results show that LLM evaluation compute can be allocated by uncertainty rather than by fixed repetition counts, with the magnitude of savings depending on evaluation design.

Power Sampling sharpens a language model's distribution over complete generation trajectories, offering a verifier-free way to improve reasoning at inference time. It also has the potential to serve as a general-purpose front end for a broad range of downstream sampling methods. However, we uncover a striking paradox: Power Sampling can drive more probability mass toward correct trajectories while degrading the downstream inference it is intended to enhance. Using self-consistency as a representative case, we observe accuracy drops of up to 18.5 percentage points across models and reasoning benchmarks. We trace this paradox to two mismatches. Dose mismatch arises because a fixed exponent induces drastically different amounts of distributional change across problems. Coverage mismatch arises because global sharpening concentrates mass on a narrow set of dominant paths: high pass@k, often interpreted as evidence of preserved diversity, can therefore coexist with the loss of broad reasoning-path support required for downstream aggregation, search, and selection. Guided by this diagnosis, we replace uniform trajectory exponentiation with a deformation-controlled, support-preserving Power target that calibrates sharpening across problems while limiting the suppression of moderate-probability paths. In a same-budget instantiation with weighted self-consistency, the repaired sampler reverses the losses caused by global Power and outperforms standard multi-sample inference across reasoning benchmarks.

We present a survey of the past and future of AI Scientists: machines capable of automating science. AI Scientists can originate hypotheses, deduce their consequences, design and execute experiments, interpret their results, and revise their beliefs. Such systems are integrated scientific agents, connected to the literature, formal knowledge, mathematical models, simulations, data-analysis systems and physical laboratories. Adam was the first machine to make novel scientific discoveries through cycles of hypothesis formation and physical experimentation. Eve established the architecture of the modern self-driving laboratory. Foundation models, autonomous agents and laboratory robotics now make it possible to build systems far more general than either Adam or Eve. The central problem is no longer whether individual components of science can be automated. They can. The problem is integration. AI Scientists must combine neural learning with logic, probability, mathematics, causal reasoning, simulation, experimental design, robotics and formal scientific records. AI Scientists have the potential to transform science: to make science faster, cheaper, more systematic and more reproducible. AI Scientists could investigate systems too complicated for unaided human science, and enable thousands of AI scientists to work together on single problems. The Nobel Turing Challenge sets the goal of developing by 2050 AI systems capable of automating Nobel-quality discoveries. Progress is ahead of schedule. When we succeed it will create a new form of science and transform the world.

Textile-integrated capacitive sensors offer flexible and conformable tactile sensing for wearable electronics and human-robot interaction; however, the influence of yarn-level architecture on capacitive transduction characteristics remains insufficiently quantified. This work presents a textile capacitive sensing platform based on silver-coated yarns coated with polydimethylsiloxane and assembled into one-, two-, and four-layer twisted configurations. The influence of effective electrode overlap area and inter-fiber separation on the capacitive response is systematically investigated, enabling architecture-dependent tuning of pressure and proximity sensing characteristics. Pressure was calculated using the localized single-fiber contact area, corresponding to stresses of 0.4-3.9 MPa. Increasing the layer number improved mechanical strength and sensing performance: elongation at break increased from 37.5% to 62.5% and 85.0%, while the maximum load increased from 23.3 to 42.7 and 89.7 N. Sensitivity increased with layer number and frequency, reaching 0.1331 MPa$^{-1}$ for the four-layer sensor at 100 kHz. The four-layer configuration also exhibited low hysteresis, minimal thermal drift from 25 to 90 $^\circ$C, and stable operation over 15,000 cycles. Proximity detection ranges of 60, 50, and 40 mm were obtained for the one-, two-, and four-layer sensors, respectively, revealing an architecture-dependent sensitivity-range trade-off. A 4$\times$4 textile sensing array enabled spatial contact mapping, while robotic-arm integration demonstrated real-time touch and proximity detection with an end-to-end robotic system latency (from detection to robot reaction) of 403 ms. The results establish yarn architecture as a tunable design parameter governing the measurement characteristics of textile-integrated capacitive sensing systems.

Zahavy [2026] argues that Large Language Models, despite their capabilities in induction and deduction, cannot perform the abductive "Jump" that produced Einstein's equivalence principle, and attributes this limitation to the absence of embodied simulation. Zheng-Xin [2026] and Farmer [2026] question whether embodiment is necessary for abduction, pointing to alternative routes to General Relativity and forms of abduction that require no sensorimotor grounding. Max Planck resolved the blackbody radiation problem in 1900. Planck's move to E = hν required no embodied simulation. It was motivated by a mathematical consequence of classical theory, an infinite predicted energy for a finite measured quantity, that could not be physically accepted. We show that neither induction nor deduction could have produced the postulate and argue that its adoption required a coupling between epistemic error and physical cost. We formalize this distinction through thermodynamic coupling and show that fixed-weight transformer inference lacks such coupling, regardless of model scale. This is consistent with empirical results showing that output entropy remains nearly unchanged across tasks with sharply increasing causal difficulty, even as accuracy falls from 100% to 17%. We therefore argue that the missing ingredient in machine abduction may lie deeper than embodiment: a system must have some physical mechanism through which epistemic error becomes costly enough to force revision.

Perception is a fundamental component of autonomous driving systems. While LiDAR-based methods have achieved remarkable progress in object detection, their reliability can degrade under adverse weather conditions. Radar point clouds provide a robust alternative due to their resilience to bad weather and low-illumination scenarios. However, radar point clouds are typically sparse, unordered, and less informative than LiDAR data, making it challenging to directly apply existing LiDAR-based perception methods. To address these challenges, we propose IRGNN, an Invariant Radar Graph Neural Network for radar point cloud object detection. IRGNN first reconstructs radar point clouds into graph representations using translation- and rotation-invariant feature designs, enabling robust modeling of sparse radar measurements. It then employs an improved message passing neural network (MPNN) with residual connections and a virtual node layer to enhance local feature propagation and global context modeling. Finally, task-specific heads are applied to the learned graph representations for object classification and bounding box prediction. Experimental results on the RadarScenes dataset show that IRGNN outperforms existing radar-based object detection methods and achieves competitive performance. In addition, IRGNN significantly reduces computational cost and memory usage during inference, demonstrating its effectiveness and practical potential for efficient radar-based perception in autonomous driving.

Recent video generators can fabricate realistic depictions of wars, disasters, public emergencies, and other real-world crises, creating substantial risks of misinformation. Existing benchmarks, however, provide limited evidence on detector and generator behavior in such settings, including how detectability varies with generation conditions, how people perceive generated videos, and whether detectors remain reliable during social dissemination. To address this gap, we introduce RA-Bench, a benchmark for AI-generated video detection that uses Real videos as Anchors. RA-Bench contains 17,886 videos, comprising 1,830 real-video anchors across 10 social-risk categories and 16,056 generated clips from four open-source and five closed-source generators. Based on RA-Bench, we organize our evaluation along three dimensions. We first assess detector generalization across seven traditional detectors, ten zero-shot multimodal models under three review settings, and two MLLMs specifically fine-tuned on AI-generated video detection. Across these methods, none of the three detector families generalizes consistently across RA-Bench instances. We then examine how detectability varies with generation quality, conditioning information, and sampling seeds. These analyses show that generation properties affect detector families differently, while source-level detection patterns remain stable across seeds. Finally, we study human authenticity judgments and detector reliability during social dissemination. We find that videos that mislead people are also difficult for current detectors, and that social dissemination makes detection harder. Together, these findings show that current methods struggle to detect realistic AI-generated videos, highlighting the need for detectors robust to evolving video generators.

This paper studies policy learning for distributed task allocation in open multi-agent systems, where agents may join and leave in a time-varying fashion, with submodular stage team utilities. At each time, the active agents select actions from local categorical policies such that the feasible joint agent-action pairs form a partition matroid. Standard continuous relaxations of submodular set functions are based on independent Bernoulli sampling, making them inconsistent with agents' policies.To solve this mismatch, we propose the \emph{partition multilinear extension} (PME), a policy-based relaxation whose continuous support matches feasible actions under categorical policies.We prove that the marginal gains of the stage utility provide an unbiased estimator of the gradient of the PME and that maximizing the PME over action distributions is equivalent to maximizing the stage utilities over agent actions, which are critical to devise principled policy gradient.Building on this, we design \emph{SubMAPL}, a centralized-training decentralized-execution KL-mirror policy-learning method that uses local marginal gains as stochastic PME gradients during training. KL-mirror updates preserve categorical feasibility without Euclidean projection.In the case where agents run tabular-softmax policies, we introduce open policy migration and an open-system KL tracking variation to handle agent arrivals and departures. Using dynamic regret analysis, we establish a lower bound on the cumulative utility which accounts for the openness of the environment and for the gap between optimal stage-wise and global utilities. Simulations on multi-agent coverage demonstrate that SubMAPL outperforms policy-gradient and online-learning baselines.

Existing palm presentation attack detection (PAD) datasets are often limited by static imagery, restricted acquisition conditions, or insufficient multimodal video data, hindering systematic evaluation across environments, modalities, and attack types. We present GBU-Palm, a large-scale multimodal video dataset and benchmark containing 21,326 videos from 105 subjects and 210 palms across six acquisition environments, including bona fide, Print, and Replay presentations, with 6,310 synchronized RGB-NIR samples. We construct leakage-controlled protocols that separate palm identity and attack lineage and benchmark four representative video architectures under environment-matched and held-out-environment settings. Results reveal substantial architecture-dependent degradation under environmental shift and show that RGB-NIR fusion does not consistently outperform RGB-only input. We further analyze model behavior through true accept (TA), true reject (TR), false accept (FA), and false reject (FR) decomposition, spectral masking, temporal-order intervention, and frozen-backbone NIR probing, revealing distinct failure patterns and evidence utilization across architectures. GBU-Palm provides a unified and challenging benchmark for developing and evaluating robust multimodal palm PAD methods under cross-environment conditions.

Mixture-of-Experts (MoE) models have been widely adopted in real-time interactive applications such as coding assistants, real-time audio-video interaction systems. To meet the extremely low response latency requirements of these scenarios, practitioners commonly employ small-batch decoding, under which MoE inference becomes memory-bound and is severely bottlenecked by expert weight loading. However, this bottleneck has received limited attention, and existing solutions such as post-training weight compression or fine-grained expert design during pre-training either degrade model accuracy or introduce additional computation and communication overhead. To tackle this issue, we propose DeaMoE, a decoding-efficient MoE architecture, in which the experts are grouped into several departments, and the experts belonging to the same department share most parameters since they come from the same professional field, and additionally each expert contains a few private parameters to reflect its uniqueness. Moreover, we design customized two-stage routing strategy for DeaMoE to avoid redundant loading, under which DeaMoE greatly improves the efficiency during LLM decoding. Compared with vanilla MoE, DeaMoE reduces per-step loaded weights by up to 50.9% and achieves up to 1.33 end-to-end TPOT speedup for the pre-trained 7B model on A40, and up to 2.00x and 1.97x peak speedup for DeepSeek-V3 on A40 and H100 in microbenchmarks.

Many real-world tasks require LLM agents to interact with their environments over long execution horizons. Errors that occur early in execution may propagate through both the agent context and environment state, and their effects may be difficult to reverse through subsequent actions. Existing methods mainly seek to reduce such errors through plan refinement and safety checks but provide little support after errors occur. To enable recovery during long-horizon execution, we present AgentRewind, a runtime recovery framework that records aligned checkpoints of the agent context and controlled environment, allowing agents to return to an earlier state and resume execution with information from previous attempts. We also construct MettleBench, a benchmark for evaluating task completion and partial progress on long-horizon engineering assignments containing a series of related requirements. Experiments across tasks, multiple models, execution strategies, and agent harnesses show that AgentRewind improves task success rate and average checklist progress over the compared baselines.

Vision-Language-Action (VLA) models have recently achieved promising performance in robotic manipulation. However, existing benchmarks mainly evaluate generalization on static manipulation tasks and largely overlook dynamic interaction scenarios. To address this gap, we present ReflexBench, a benchmark for reaction-critical manipulation. ReflexBench contains six dynamic tasks and introduces an evaluation framework that decouples simulator stepping from robot control while supporting configurable latency under synchronous and asynchronous inference. Building upon ReflexBench, we propose ReflexVLA, an efficient VLA model designed for reaction-critical manipulation without large-scale robot-data pretraining. ReflexVLA enhances temporal reasoning through latent future prediction and multi-frame temporal fusion within the vision backbone, while reducing deployment latency through batched visual encoding and CUDA Graph replay. Experiments show that ReflexVLA consistently improves dynamic manipulation performance while maintaining competitive accuracy on standard static manipulation benchmarks, and real-world experiments further demonstrate its effectiveness under practical deployment conditions. Project website: https://reflexvla.github.io

Sports have witnessed growing global enthusiasm in recent years, serving as a vital force for physical health, cultural exchange, social connection, and economic growth. The rapid advancement of large models, particularly (multimodal) large language models (M)LLMs, has demonstrated transformative potential to reshape sports understanding, analysis, and interaction across diverse domains. This paper presents a comprehensive survey of large models in sports, including (i) an overview of tasks and applications across different participant groups; (ii) a detailed analysis of sports-related datasets and benchmarks; and (iii) a critical discussion of current challenges and future directions. Our goal is to establish a foundation for advancing research and practical development of large-model-driven sports intelligence. An open-source GitHub repository is maintained at: https://github.com/Road2Redemption/Awesome_Large_Models_In_Sports1.

Multi-agent reasoning systems often use agreement, confidence, or automated scores to decide which messages should shape a final answer. Such filtering assumes that a message likely to be correct is also worth keeping. Yet a wrong answer can contain a useful decomposition, constraint, or scientific principle. We test this distinction with Diverse Hypothesis Deliberation (DHD), a controlled measurement protocol that caches five independently generated messages and replays the same downstream solver, called the integrator, with each message available or hidden. The replay comparison measures a message's trajectory value: whether making the message available helps or harms subsequent reasoning. Across five mathematics and science benchmarks and two openly available model families, gpt-oss-120b and gemma-4-31B-it, wrong-helpful messages appear in every benchmark-model combination. Among wrong-answer messages that change final correctness, more than four in ten changes are helpful in each model. Controlled repeats show that the number of repeatable message effects is unlikely to arise from replay variation alone (p=0.0002). A focused intervention on repeatable wrong-helpful messages finds that the complete message works best, while retaining its reasoning preserves more success than retaining only its answer; the source of the complete-message advantage remains open. Within the same problem, repeated trajectory-value evidence also identifies a better keep-or-remove choice than answer correctness alone. Answer correctness is therefore informative but does not determine trajectory value. DHD measures this missing property and produces reusable labels for learning when agents should listen.

Spatial transcriptomics (ST) enables the simultaneous profiling of gene expression and tissue morphology, creating an opportunity to learn multimodal representations capturing shared morpho-transcriptomic structure. However, standard multimodal models often compress modalities into a common latent space without explicitly separating shared and modality-specific sources of variation, which may limit downstream utility. We investigate whether explicit disentanglement of shared and private latent components improves multimodal representation learning for paired Hematoxylin \& Eosin (H\&E) and ST data. We compare VAE-based and contrastive approaches, each in standard and disentangled variants, across two cancer cohorts under matched experimental conditions. Representations are evaluated using cross-modal reconstruction, downstream probing and cross-modal probe transfer. The experiments suggest two main trends. First, contrastive objectives yield higher downstream probing performance than VAE-based models. Second, disentangled variants improve the selected reconstruction and probing metrics, although the gains depend on the model family, task, direction, and disentanglement strength. Overall, our results suggest that explicitly factorizing shared and modality-specific information can improve multimodal representation learning for spatial transcriptomics and provides a useful evaluation framework for future foundation models.

Enabling LLM agents to sustain productive, stable, and goal-aligned research over extended horizons is a central challenge for autonomous machine learning and scientific discovery, as progress hinges on continuously managing evolving state, exploration decisions, and computational resources. Pioneering autoresearch agents, despite great success, still lack mechanisms for continuity, recovery from dead ends, and value-driven compute allocation, which inherently undermines overall search efficiency, wastes computational resources, and lowers the chance of ultimate success. To bridge this gap, we introduce ScienceFlow, an end-to-end autoresearch agent framework that organizes long-horizon research work into research segments grounded in executable workspaces. It represents research progress as recoverable executable states, enabling efficient exploration, revision, and execution. Transitions between research segments are governed by Executable-State Transition through Re-Anchoring (ESTRA), which selects either the live state or an archived state as the next anchor and determines whether to continue or redirect the research trajectory. An evidence-aware execution controller allocates resources to physical jobs based on resource availability, remaining budget, and validated progress. We evaluate ScienceFlow on tasks spanning machine learning, scientific modeling, and mathematical optimization. Results on diverse long-horizon benchmarks demonstrate its ability to sustain effective research processes, highlighted by a SOTA 70.22 percent Any-Medal score on the full MLE-bench within a 24-hour budget, outperforming prior reported results by 4.92 percentage points. The efficacy of ScienceFlow further demonstrates that efficient state management, adaptive exploration, and objective-aligned execution are critical for scaling autonomous research beyond short-horizon interactions.

Large Language Model (LLM)-based agents are increasingly used for complex tasks such as software testing and cybersecurity assessment. While these agents demonstrate impressive capabilities, their behavior is difficult to understand, explain, and analyze. Existing evaluations focus mainly on task success and execution traces, offering limited insight into the strategies employed by the agent. We present ATLAS (Automata Learning for Agent Trajectory Analysis and Strategy Discovery), an approach for recovering interpretable behavioral models from agent trajectories. ATLAS combines trace abstraction with automata learning to infer finite-state models that capture observed agent-environment interaction strategies. These models provide human-interpretable insights and support automated analyses of recurring behaviors, decision points, successful task-completion paths, and failure loops. As a proof of concept, we apply ATLAS to trajectories generated by an LLM-based penetration-testing agent. The resulting models expose high-level behavioral strategies for exploiting vulnerable machines that are difficult to identify from raw execution traces alone. We discuss how learned behavioral models can support explainability, model-guided exploration, auditing, and analysis of agentic systems. We further demonstrate symbolic model-based knowledge transfer from powerful frontier models to compact language models. In addition, we show how model transformations can derive concise explanations of agent behavior in a penetration-testing case study comprising 12 vulnerable machines. ATLAS highlights a new opportunity for model-driven engineering: transforming agent trajectories into explicit behavioral models that enable systematic understanding and analysis of otherwise opaque AI agents.

We present a training-free method for multi-modal trajectory prediction that achieves comparable accuracy to a 57M-parameter transformer while requiring no GPU and zero learned parameters. The method builds a transition table of historical state-to-next-position pairs and retrieves neighbors using a product kernel over spatial proximity, bearing, speed, and temporal context. Two inference modes operate over this shared representation: diversity-penalized sampling produces trajectories covering distinct plausible routes, while beam search finds the highest-likelihood path. On the TrAISformer benchmark (Danish Maritime AIS), our method achieves competitive accuracy at full data availability and dramatically outperforms the transformer in data-scarce regimes---remaining stable down to 10% of training data where TrAISformer degrades catastrophically. This enables deployment in new geographic regions from an order of magnitude less historical data, and with no GPU training.

We study proactive exploration in LLM agents, i.e., the ability to explore an environment to acquire information that improves future decision-making. In this regard, we first identify two fundamental bottlenecks that hinder this capability and then propose \ours, a novel method designed to instill and refine proactive exploration. Specifically, \ours\ consists of two components: (1) Exploratory Data Construction, which synthesizes exploration-rich trajectories to mitigate the hindsight bias of standard demonstrations; and (2) RL Optimization with Contrastive Signal Guidance, which leverages contrastive trajectory pairs to distinguish productive exploration from redundant wandering. Extensive experiments demonstrate the effectiveness of \ours\ and provide insights into the characteristics of proactive exploration. Our code is available at: https://github.com/GuanZhizhao/SAFARI.

Principle-based regulation, with evaluative standards such as "fair, clear, and not misleading" or "deliver good outcomes", cannot be reduced to binary predicates, and LLM-as-judge is increasingly used as the substitute. Our position is that any such judge must be evaluated on four axes: accuracy, paraphrase robustness, adversarial robustness, and calibration. We release Principle-Bench, 168 cryptoasset financial-promotion scenarios mapped to two UK FCA principles, with paraphrase, adversarial keyword-stuffing, and boundary perturbations authored under a pre-registered rubric; the first benchmark covering all four axes for principle-based regulation. We also introduce Ceca (Calibrated Exemplar-Cluster Assessment): a calibrated, auditable assessor that emits exact per-exemplar counterfactual attributions. Across keyword counting, three sentence-transformer embedders, an open-weight LLM-judge, and a calibrated cascade, no method dominates all four axes. A 120B LLM-judge, strongest on benign inputs, loses 47 accuracy points (0.74 to 0.27) on keyword-stuffed Consumer Duty inputs: "compliance theatre." A second judge from a different model family agrees only at Cohen's kappa = 0.16 on that split, localising the failure to the model rather than the corpus. Any deployment-grade LLM-judge for principle-based regulation must report per-principle adversarial deception and post-hoc calibration alongside aggregate accuracy.

The anchoring effect is a cognitive bias in which an initial reference value shifts a later judgment toward itself. This effect is well established in human judgment and decision-making, and recent work suggests that large language models (LLMs) exhibit similar behavior. However, existing work on anchoring in LLMs typically evaluates only a narrow set of anchor pathways and rarely distinguishes irrelevant from plausible anchors. We introduce AnchorBench, a benchmark for the anchoring effect in LLMs that evaluates multiple anchor pathways under an explicit anchor relevance axis. Across fourteen models, including ten open-weight models and four frontier API models, and a large set of controlled prompts, we find that (1) anchoring is strongly pathway-dependent, (2) plausible anchors usually induce larger shifts than irrelevant ones when introduced through stronger pathways, (3) anchor influence generally weakens as the anchor moves farther from the evidence-supported answer, most clearly on External and RAG, and (4) high task accuracy on the anchor-free control condition (Acc$_{10}$: answers within 10 points of gold) does not guarantee robustness: even frontier API models above 95% control accuracy remain susceptible to plausible anchors.

Reinforcement learning (RL) for terminal agents needs executable training environments with reliable rewards and useful difficulty. Fixed recipes such as few-shot, Self-Instruct, and Evol-Instruct apply the same prompting policy to every seed, even when the current policy would benefit from a harder, easier, or simply different task. We present Envs-FORGE, a prompting policy that converts verifier rewards into per-seed environment-synthesis actions. Envs-FORGE estimates seed pass rates, scores six projection--direction actions around a target learning frontier, and solves a per-seed mixed-integer linear program (MILP) to choose the action that conditions generation. The selected action drives synchronized rewriting of the instruction, fixtures, oracle solution, tests, and Docker environment; only gold-verified bundles enter RL training. The indexed MILP form also supports optional soft skill coverage for portfolio planning. On Qwen 3.5 35B, Envs-FORGE improves Pass@1 over Base by 9.2 percentage points on tb-core (40.0% to 49.2%) and 6.4 points on tb-2.0 (23.0% to 29.4%), exceeding the strongest fixed-recipe baseline by 2.4 and 2.1 points. It reaches 77.1% on SWE-bench Verified versus 73.4% for Base, and improves tb-core by 6.8--9.2 points across the evaluated 4B--35B models. All synthesis methods export 100 verified environments and use 2.27M--2.88M synthesis tokens, placing the comparison at the same downstream training-set size and the same operational scale. The source code is available at https://github.com/DataArcTech/DataArc-SynData-Toolkit/.

Large language models (LLMs) are increasingly used for code generation, yet they remain vulnerable to prompts that elicit insecure implementations. Existing defenses typically rely on predefined threat models or known vulnerability patterns, limiting their effectiveness against novel attacks. We propose CodeSIFT, a threat-model-agnostic detection method that leverages influence functions to identify batches of prompts that induce anomalous model behavior. Rather than detecting specific vulnerabilities, CodeSIFT measures the parameter-space influence of generated code and uses a statistical test to determine whether a candidate prompt set deviates from a benign reference distribution. To evaluate our approach, we introduce two benchmark datasets covering a variety of vulnerabilities. We evaluate CodeSIFT on three open-weight code LLMs ranging from 3B to 7B parameters, achieving AUROC scores of up to 0.98 at moderate-to-high injection rates, while maintaining well-calibrated false positive rates and substantially outperforming static analysis baselines. These results suggest that influence-function-based detection is a promising direction for identifying malicious code-generation prompts without requiring prior knowledge of the underlying attack class.

We introduce Mobius-v0, an architecture that comprises a globally shared Memory (FFN) that stores knowledge vectors and multiple Reasoners (Self-Attn) that iteratively achieve compositional reasoning. Using hidden states as cache and carrier, reasoners repeatedly query memory for required knowledge-vectors, while the knowledge is transmitted back to reasoning operators. Through this knowledge-reasoning-separation architecture, Mobius achieves better knowledge compression and reasoning efficiency. Built upon Mobius-v0 architecture: 1) Our 7B model trained-from-scratch achieves similar downstream score as a 7B Transformer baseline with 62.6% of baseline's training data. 2) Our Intern-S2-Mobius, continually-pretrained from Qwen3.5-35B, achieves similar downstream score while delivering nearly 4x end-to-end inference speedup.

Fine-grained robotic evaluation matters for understanding embodied models, going beyond binary success rates and rule-based process scores. We present PRM-as-a-Judge 1.5, a toolkit for robot process assessment that turns rollout videos into dense progress curves and derives multiple fine metrics. PRM-as-a-Judge 1.5 introduces three metrics, building on version 1.0, that characterize failure-side progress, post-drawdown recovery, and success-side execution quality, helping users understand embodied model capability. Based on the rollout videos from benchmarks, we perform a comprehensive assessment of the embodied models, providing some fine-grained metric results and key findings. We further introduce RoboPulse++ to evaluate the reliability of process reward models (PRM), providing evaluators with a more accurate testing platform. Moreover, we release a user-friendly assessment suite, including the benchmark, metric implementation, and visualization tools, to support reproducible manipulation process evaluation. We call on the community to rethink how robots are evaluated and establish transparent, procedural, and reproducible assessment as a foundation for the next generation of embodied intelligence.

Monocular temporal 3D detection aims to detect objects in 3D, given a monocular video. Query-based 3D detectors unify detection and cross-view association, but their learnable queries fit the spatial distribution of the training data (e.g., field-of-view). We show that this issue is especially severe when these models are applied to monocular video, hindering generalization to unseen datasets and environments. To address this limitation, we introduce MAGneT-3D, the first method for domain-generalized monocular temporal 3D object detection. Instead of relying on static learnable queries, we propose a Domain-Robust Anchor Generator (DRAG) approach that adaptively derives 3D proposals during inference. To further enable domain generalization, we propose a Temporal Refinement and Identity Merging (TRIM) strategy, reducing dependence on specific 3D proposals. To enable comprehensive domain-generalization evaluation, we establish a cross-dataset benchmark spanning nuScenes, Waymo, Lyft, and ONCE. Under zero-shot domain shifts, MAGneT-3D outperforms all baselines, improving NDS from 12.1% to 18.6% while also increasing in-domain accuracy.

On-policy distillation (OPD) offers a promising way to transfer reasoning capabilities from stronger teacher models, but applying it to long-context reasoning teachers and short-context students introduces practical challenges, including tokenizer mismatch, teacher-student distribution mismatch, response length explosion, and training instability. In this work, we study this setting by transferring proof-reasoning capabilities from the long-context reasoning model SU-01 to short-context student models. To handle tokenizer differences, we perform OPD in a shared text space and align only tokens that occupy identical text spans under the student and teacher tokenizers. To mitigate the problem of excessive generation length and frequent truncation, we introduce a student reference KL loss and mask the advantages of special termination tokens such as </think> and <|im_end|>. This strategy constrains the student from drifting excessively from its initial policy, thereby mitigating the teacher-student distribution mismatch problem and fostering steady length growth. Experiments on both same-family and different-family student models, including Qwen3, Qwen3.5, Intern-S2, GLM-4.7, Gemma-4, show consistent gains in mathematical reasoning, especially natural-language math proving. Notably, Intern-S2-Preview improves by 21.2 points on ProofBench, reaching 55.2 and surpassing Gemini-2.5-Pro. It also improves on science benchmarks such as HLE and HiPhO, suggesting that OPD transfers reasoning capabilities that generalize beyond the mathematical training domain.

Time series analysis in high-stakes domains relies on recurring data releases, where new observations can alter the evidence base and the validity of later conclusions. Existing time series QA benchmarks mostly rely on fixed snapshots, leaving temporal validity and cutoff-aware evidence use unevaluated. We introduce TimeSage-EV, a live benchmark for agentic time series analysis in evolving environments. It tracks 60 real institutional scenarios across 6 domains, comprising 1,485 scenario-period QA pairs from Feb 2023 to May 2026 and spanning monthly, weekly, daily, and irregular release cadences. At each period, large language model (LLM) agents receive time series data and source reports, while the withheld target release provides ground truth. TimeSage-EV evaluates state identification, data summarization, and outlook reasoning. Experiments with frontier LLM agents and TimeSage-1.0, a novel self-evolving agent with a reusable analytical skill library, reveal significant performance gaps across model tiers and recurring failures in temporal validity, exogenous context use, and adaptation. We release TimeSage-EV as a research resource with monthly updates, code, a leaderboard, and failure-mode analyses.

LiDAR bundle adjustment is widely utilized in mapping to construct globally consistent point cloud maps. In this paper, we propose the first fully parallel computing framework to accelerate LiDAR bundle adjustment for large-scale mapping, incorporating three key techniques. First, we design an adaptive, asynchronous data loading strategy to efficiently process large-scale point cloud datasets on memory-constrained GPUs. Secondly, we present a novel bottom-up voxelization method for extracting planar features, enabling fully parallelized pre-processing. Thirdly, we build upon a majorization-minimization formulation to accelerate compute-intensive tasks in the optimization via parallel computation, including the computation of residuals, Jacobian and Hessian matrices, and a parallel increment solver. To support our design, we provide both theoretical and experimental analysis of the time complexity of our approach. Extensive benchmarking on large-scale public datasets across various computational platforms validates the robustness and adaptability of our approach, achieving up to a tenfold improvement in computational efficiency while preserving mapping accuracy comparable to state-of-the-art methods. To benefit future research, the implementation code is available on GitHub.

Model merging combines trained models directly in weight space, offering a compute-efficient alternative to additional fine-tuning. Selecting merge parameters is nevertheless difficult because downstream evaluations are expensive, gradients are unavailable, and source capabilities can conflict. We formulate merge-parameter selection as a black-box multi-objective optimization problem and introduce MOBO-Merge, a merge-operator agnostic framework that uses multi-objective Bayesian optimization to approximate the Pareto front under a limited evaluation budget. We evaluate Qwen3-4B and Llama-3.1-8B in two-model instruction-math and three-model instruction-math-code settings using Linear, SLERP, TIES, and block-wise merge operators. On held-out benchmark partitions, MOBO-Merge obtains higher mean hypervolume than random search in 11 of 12 reported comparisons. The gain is small for one-dimensional Linear interpolation but substantially larger for several TIES, block-wise, and three-objective searches. No merge operator is uniformly best: TIES leads in three of four family-setting combinations, whereas Block-Linear 4x is strongest for the Llama three-model merge. These results show that multi-objective Bayesian optimization is valuable as a search layer for expressive merge parameterizations.

Temporal vision-language models (TVLMs) offer a reusable, prompt-based interface for surgical video understanding, yet, their robustness under clinically realistic acquisition artifacts in endoscopy remains insufficiently characterized. In practice, degradations such as defocus, haze, motion blur, noise, cautery smoke, and packet loss introduce structured distribution shifts which may compromise video-text alignment. We study the robustness of temporal VLMs under such shifts caused by corruptions in clip frames. We introduce Endo-C6, a compact corruption benchmark of six endoscopy-realistic perturbations evaluated at a fixed high severity, and apply it to public Gastrointestinal (GI) endoscopy and laparoscopic cholecystectomy videos. Under a standardized prompt protocol, we benchmark 3 recent surgical TVLM baselines and analyze robustness in both mean and worst-case settings, spanning 294 dataset-level evaluations. Finally, we present RobustEndoCLIP, obtained by few-shot parameter-efficient tuning with VeRA, outperforming existing TVLM baselines. Our findings show that off-the-shelf TVLMs can exhibit severe worst-case collapse under endoscopy-specific corruptions, whereas lightweight few-shot adaptation can substantially improve corrupted performance and robustness without changing the prompt-based interface. We expect Endo-C6 to support standardized robustness reporting and promote more reliable clinical vision-language systems.

Fugitive emissions from waste sites increasingly expose communities to toxic and odorous gases, yet public-health responses remain largely retrospective, with episodes investigated only after residents have been exposed. Here we show that the meteorological drivers of elevated hydrogen sulphide (HS) at a long-monitored European landfill, and the timescales over which they act, can be identified directly from routine monitoring data. We introduce CAIRN (Causal-Anchored Inference for Receptor Nowcasting), a machine-learning framework whose internal memory is matched to these measured timescales: a fast component tracking hour-scale wind-borne transport and a slow component tracking multi-hour weather changes. Trained to predict gas measurements, CAIRN operates using only routine weather variables and the calendar, without hand-engineered features. Its behaviour is consistent with the identified transport mechanisms, and the framework transfers unchanged to a second monitoring station and to co-emitted methane. Combining four such nowcasters produces a site-level, tiered alert aligned with WHO odour guidance that closely reproduces the alert generated by a direct sensor network and tracks an independent record of community odour complaints. Weather-driven nowcasting can therefore estimate community impact as an emission episode unfolds, providing public-health authorities with a validated, graded trigger for intervention and enabling exposure to be reduced during events rather than after them.

In today's fast-paced environment, the ability to swiftly access, understand, and act on data is no longer optional; it is essential. Yet most organizations remain data-rich but insight-poor, constrained by the complexity of querying, interpreting, and explaining enterprise-scale information. We present Polaris, a supervisor-led multi-agent framework for conversational enterprise analytics that bridges this gap. Polaris introduces Dynamic Task Coordination (DTC), a decision-theoretic orchestration layer that models agent-task assignment as adaptive bipartite matching, enabling real-time coordination, recovery, and optimization across specialized agents for querying, visualization, and reasoning. By coupling DTC with reason-first, ReAct-style agents, Polaris transforms natural-language queries into coherent analytical workflows that not only retrieve and visualize data but also explain the underlying "why." Evaluation on structured enterprise datasets demonstrates high semantic fidelity and answer relevancy, underscoring the potential of multi-agent orchestration to deliver trustworthy, end-to-end business intelligence at scale.

In large and heavy structures, vibrations arise during motion, posing significant challenges for precise manipulation. To accomplish the desired motion, control algorithms must effectively suppress these structural vibrations. In cutting edge projects, such as remote maintenance of future fusion energy reactors (tokamaks), the manipulation of this type of structure is defined as a crucial task. This paper presents a control strategy to suppress transverse vibrations in flexible payloads during motion using a collaborative payload manipulation approach. Two different industrial robot arms are arranged in a leader follower configuration for the manipulation strategy. The leader robot guides the motion with shaped velocity commands, while the follower robot ensures compliance with the estimated external forces applied by the leader on the payload through an admittance controller. Unlike existing methods, the proposed approach enables collaborative manipulation of heavier and larger flexible objects, addressing additional challenges such as vibration suppression and heterogeneous robot specifications. The dynamics of the leader follower payload system are modeled using an equivalent mass spring damper model, and it is shown that, with appropriate admittance parameters, the total energy of the system is passively dissipated. A stability proof is also provided. Numerical simulations validate the proposed method, and experimental results demonstrate its effectiveness.

Action anticipation (AA) aims to recognize ongoing human or humanoids actions from partial observations, enabling robots to predict intentions before the actions are completed. Although skeleton-based AA offers efficiency advantages, existing approaches assume that all action classes are seen during training, which limits their deployment in real-world scenarios where novel actions inevitably arise. To address this gap, we study the new task of Zero-Shot Skeleton-Based Action Anticipation (ZS-SkAA). This task requires recognizing unseen action classes using only limited early-stage skeleton sequences, combining the challenges of partial observations, temporal dynamics, and zero-shot generalization. To establish foundational research for ZS-SkAA, we introduce:(1) A baseline model comprising a spatio-temporal feature extractor and a mutual information estimation and maximization module. This baseline model explicitly aligns partial visual features with semantic class embeddings across modalities by estimating and maximizing their mutual information, enhancing generalization to unseen classes.(2) A benchmark protocol using the NTU RGB+D dataset, which is adapted for rigorous ZS-SkAA evaluation. Experiments demonstrate the effectiveness of our model as a strong baseline for ZS-SkAA, achieving high zero-shot accuracy on NTU RGB+D. This work establishes ZS-SkAA as a vital research direction for real-world systems requiring generalization to novel actions.

Operating teams of autonomous aircraft in dynamic, uncertain, and potentially adversarial environments requires safety protocols that are reliable yet selective, and allow agents to fly in close proximity while making progress toward mission objectives. We introduce adversarial time-to-collision (aTTC), a risk metric that quantifies, for a given agent, how quickly any surrounding agent could reach it assuming adversarial intent. We embed aTTC into the control barrier function (CBF) framework, defining the barrier directly in time rather than distance or velocity. The resulting aTTC-CBF is inherently anticipatory: agents modulate their own velocity based not on whether a peer is on a collision course, but on how quickly one could reach collision given its dynamical constraints. A differentiable neural-network surrogate makes the aTTC computable in real time within a standard CBF quadratic program. Across long time-horizon simulations of 3D independent-pursuit and formation-flight scenarios, the aTTC-CBF achieves up to twice the waypoint progress at half the collision rate of a higher-order distance-based CBF baseline.

Popular facts are memorised more deeply during pretraining and resist removal longer than rare ones, yet existing LLM unlearning methods apply uniform gradient pressure regardless of training-data frequency. We propose the AdaPop (Adaptive Popularity) method, which combines local token confidence with a per-fact popularity-dependent exponent derived from an external proxy (e.g., Wikidata sitelinks, LLM-as-Judge), and automates the forget-retain balance via a dual-ascent controller that adjusts the retain penalty each epoch. Across three model families and two benchmarks, AdaPop leaks ~5x less forgotten content than competing methods under paraphrased queries and ~1.6x less under adversarial reformulations. We support our analysis with internal metrics: under our method, forget-set hidden states move further from the pre-unlearning model's states than under other methods, while retain-set representations remain close.

Life science knowledge graphs make large collections of structured data available through SPARQL, but each resource uses its own schema, identifiers, and links. TogoMCP helps language model agents query these resources by providing curated Metadata Interoperability Exchange files. Creating and maintaining these files still requires language model assisted drafting, validation, and manual review. We study \emph{live schema grounding}, where an agent obtains the schema evidence needed for a question directly from the current endpoints. We present \textsc{autoschema}, a general framework for live schema grounding that requires no training. It inspects live schemas, maps entity names in a question to graph identifiers, explores relation paths, and finds possible connections between resources during iterative query construction. We use TogoMCP as our main comparison framework. We evaluate \textsc{autoschema} on Resource Focused Biomedical KGQA, Multi Resource Biomedical KGQA, Longitudinal Biomedical Semantic QA over BioASQ Task B, and Chemistry Knowledge Graph Transfer to a previously undocumented RDF graph. \textsc{autoschema} improves mean factoid accuracy over TogoMCP in the biomedical KGQA tasks and gives consistent gains in the longitudinal BioASQ evaluation. It also reduces iteration budget exhaustion and uses fewer tool calls on average in the core evaluation. The transfer study gives preliminary evidence that live schema grounding can support irregular and previously unseen graphs without first creating a curated schema file.

Autoformalization is commonly framed as translating natural-language mathematical statements into machine-verifiable formal languages such as Lean 4. However, faithful formalization requires more than translation. Models must map mathematical concepts to the complex hierarchy of types and definitions in formal libraries such as Mathlib, while ensuring that generated statements preserve the meaning of the source propositions. Existing approaches struggle because they rely heavily on the model's parametric memory for library-specific knowledge, while common data construction pipelines often resort to filtering single-pass outputs and lack mechanisms for feedback-driven revision. To address these challenges, we introduce MathForm, an autoformalization framework for constructing verified training data through Mathlib knowledge retrieval and verification-guided iterative refinement. Before generation, a retrieval planner gathers relevant definitions and existing formalizations from Mathlib to guide the formalization generator. Generated statements are then revised using compiler diagnostics and semantic-consistency feedback. Using this framework, we construct FormalVerse, a Lean 4 dataset containing approximately 367K verified examples across diverse mathematical domains and sources. We then train MathForm-8B through supervised fine-tuning followed by reinforcement learning. Across six benchmarks, MathForm-8B achieves average Pass@8 rates of 88.06% under Syntax Check (SC) and 72.37% under Consistency Check (CC), outperforming multiple specialized 32B autoformalizers. On the challenging FATE-H and FATE-X subsets, it attains CC pass rates of 63% and 37%, exceeding the strongest specialized baselines in both cases.

As large language models enter professional domains, they must satisfy domain constraints, include critical evidence, and provide complete reasoning rather than merely produce fluent responses. Existing post-training methods often rely on holistic preferences or outcome-level verification, while recent rubric-based methods usually generate rubrics independently for each query. In specialized domains, such unconstrained rubrics may omit critical requirements and vary across samples, hindering the diagnosis and targeted repair of persistent capability deficiencies. We propose APTER (Adaptive Post-Training with Expert-Grounded Rubrics), a framework that integrates structured domain knowledge into fine-grained evaluation, optimization, and diagnosis for specialized complex reasoning. First, expert-grounded rubric construction starts from an expert criteria framework built by domain experts, where each criterion represents a stable professional capability. For each query, APTER selects relevant criteria and instantiates them into query-level rubrics linked to their source criteria, turning reusable expert criteria into executable query-level supervision without reference answers. Second, adaptive post-training uses rubric verdicts as both optimization and criterion-level diagnostic signals. Aggregating low-scoring verdicts by criterion ID reveals persistent deficiencies and triggers targeted supervised fine-tuning updates during reinforcement learning. Experiments on mathematical reasoning and medical question answering show consistent gains across both domains. Across three model generations, APTER improves the mathematics and medical averages over the corresponding base models by up to 15.86 and 8.04 points, respectively. Code and rubric datasets are available at https://github.com/AntDT-APTER/APTER.

Hallucination is a major challenge for retrieval-augmented generation (RAG) systems in the legal domain, where ungrounded answers can lead to serious consequences. To better understand this problem, we conduct a fine-grained analysis of hallucination behavior in eight legal RAG systems across two legal corpora, the GDPR (in English) and a national civil law (in French). Using claim-level and answer-level evaluation, we report on hallucination density and severity, analyze performance across question categories and user personas, and validate our findings on an independent set of 142 legal-expert-authored questions. Our results show that hallucinations remain pervasive, ranging from less than 10% of responses for the best-performing systems to nearly half in the worst case. We further find that false-premise questions, containing incorrect assumptions that must be rejected, produce high hallucination rates on the manually-drafted questions.

Evolutionary feature construction has shown strong promise in symbolic regression by automatically discovering informative transformations of input features that enhance a simple base learner. However, existing approaches often lack explicit mechanisms to preserve important constructed features discovered during evolution, and valuable genetic material can be lost when genetic operators disrupt effective features. This paper introduces an adaptive protection mechanism that leverages feature importance metrics to selectively preserve constructed features during evolution. The mechanism provides stronger protection for more important constructed features while still allowing less important features to be modified and to incorporate useful building blocks from more important features. We evaluate the approach using multiple feature importance calculation methods and demonstrate its robustness across different base learners. Experimental results on 98 regression benchmark datasets show that the proposed mechanism consistently improves solution quality over baseline approaches, and experiments on two credit classification datasets demonstrate that the method also extends effectively to improve search effectiveness beyond symbolic regression.

Banks analyse sequential financial transaction data to perform many tasks, including fraud prevention, credit risk assessment and offer personalization. To improve the predictive accuracy of these tasks, Payments Foundation Models encode transaction sequence data as rich contextual embeddings, which can then be provided to task-specific models as features. However, these Foundation Models are not designed for flexible zero-shot reasoning across novel downstream prediction tasks, limiting their adaptability and utility. Existing LLM-based approaches to zero-shot prediction often fail to fully exploit the predictive signal within transaction data, while relying on costly text serialization or task-specific architectures that scale poorly. To address these limitations, we present the Multimodal Instruction Network for Transactions (MINT), a framework that connects a pretrained transaction sequence encoder to a decoder-only LLM through lightweight embedding injection, transaction-language alignment, and instruction tuning. We find that MINT achieves state-of-the-art predictive question-answering performance in both in-distribution and out-of-distribution questions, while substantially reducing input tokens, latency, and memory consumption compared to text-serialization baselines. Through comprehensive analyses of representations, alignment strategies, training data, and history length, we establish that compact transaction embeddings are a superior approach to transaction representation than text serialization for multimodal reasoning and zero-shot prediction tasks.

The key-value (KV) cache stores information from past tokens and is a major memory bottleneck in long-context inference. Existing quantization methods address this bottleneck by representing the KV cache uniformly with lower-precision data types and designing quantization schemes to minimize reconstruction error in the cache itself, without accounting for how that error propagates through attention mechanisms. We prove that, under a white-noise quantization model, the expected attention-aware distortion decomposes into additive key and value contributions that factor across tokens and channels. Building on transform coding and reverse water-filling, which are classical tools from signal processing and rate-distortion theory, we introduce Attention-Aware Transform Coding (AATC), which allocates bits over a calibration set to minimize attention-aware distortion. On Llama-3.1-8B-Instruct and Qwen-2.5-7B-Instruct, evaluated across LongBench, RULER, GSM8K, MMLU-Pro, and MATH-500, our method achieves near-lossless accuracy at approximately $5.8\times$ compression, whereas each baseline degrades in at least some settings.

Tabular anomaly detection is dominated by classical density-proxy methods (Isolation Forest, OCSVM, LOF), reconstruction-based detectors (Autoencoders, VAEs), and modern non-parametric scorers (COPOD, ECOD, Deep SVDD), all of which approximate the inlier distribution only indirectly; explicit energy-based models are largely absent. Motivated by the recent revival of EBMs in deep learning (e.g., Energy-Based Transformers, JEPA), we revisit the classical Deep Boltzmann Machine (DBM) for this task and hypothesize that its mean-field energy combines more effectively with a reconstruction-based score than same-lineage pairs do. We evaluate a two-hidden-layer DBM on two tabular benchmarks spanning distinct domains (UCI Bank Marketing and NSL-KDD) against eight classical and modern baselines across twenty random seeds. The DBM mean-field energy matches the strongest baseline (the Autoencoder) on Bank Marketing and statistically beats it on NSL-KDD, while significantly outperforming the remaining seven on both datasets. When fused with the Autoencoder via rank fusion, the DBM energy yields a statistically significant improvement on both datasets (AUROC=+0.014, p<0.01 on Bank Marketing; +0.002, p<0.001 on NSL-KDD); every non-DBM-derived base model instead fails to improve or significantly degrades the AE-paired ensemble. Our position is that classical EBMs, exemplified by the DBM, deserve a place in the tabular anomaly detection toolbox as a non-redundant complementary view to the reconstruction-based scores that dominate current practice.

Large language models (LLMs) have shown remarkable reasoning and generative capabilities, motivating their use as universal reasoning engines for perception. While modern approaches such as vision-language models (VLMs) have attempted to incorporate reasoning capabilities into visual sensing, the integration of LLMs with the millimeter-wave (mmWave) modality-despite its unique advantages under low light and occlusion-remains largely unexplored. The principal bottlenecks stem from the scarcity of radar language pairs, severe cross-dataset heterogeneity, and the absence of a foundational mmWave encoder. We address this gap through a minimal textualization interface that serializes each mmWave point cloud into concise natural language, allowing off-the-shelf LLMs to operate in a question answering (QA) setting. Building on this, we present mmWave-QA, the first benchmark for language-conditioned mmWave human perception. mmWave-QA aggregates heterogeneous public mmWave datasets and harmonizes them via calibration-aware preprocessing and global taxonomy alignment, while providing natural language QA. Spanning six scenarios and five QA tasks, the benchmark enables standardized evaluation across diverse mmWave hardware and experimental conditions, establishing a foundation for scalable research on mmWave-LLM integration. We further evaluate and analyze LLMs on our mmWave-QA, highlighting their zero-shot reasoning potential for radar perception, as well as their robustness under visual degradation.

Pixel-goal navigation specifies targets directly in the agent's camera view, but a target pixel provides neither metric depth nor traversability, making 3D goal grounding and collision-free continuous planning challenging. We present OccPlanner, a goal-aware occupancy-conditioned diffusion planner that grounds pixel goals in egocentric metric space and sequentially conditions the goal representation on temporal visual context and learned local 3D occupancy features. To provide occupancy supervision at scale, we introduce L3ROcc, which converts monocular RGB navigation videos into robot-centric local 3D occupancy annotations through geometric reconstruction and ray-based visibility reasoning. We train OccPlanner on InternData-N1 and evaluate it in closed-loop simulation across four unseen scene categories from InternScenes and two goal-distance ranges. In the 5-8 m setting, OccPlanner increases the average success rate (SR) over NavDP from 20.81% to 71.55% across the four categories, reaching 86.20% and 84.92% in cluttered-easy and cluttered-hard scenes, respectively. Real-world open-loop experiments on a Unitree Go2 further provide initial evidence of sim-to-real transfer and adaptation with L3ROcc-generated supervision.

Mechanical ventilation is a critical life-support intervention, requiring dynamic adjustments to ventilator settings as a patient's condition evolves. While reinforcement learning (RL) offers a promising framework for optimizing these sequential decisions, standard approaches rely primarily on structured electronic health record (EHR) data, missing crucial clinical context recorded in free-text notes. Integrating longitudinal clinical notes into RL state spaces is challenging because notes are heavily inflated by temporal redundancy, such as copy-forward text, templating, and repetitive documentation, which dilutes time-local updates and degrades state representation quality. To address this, we propose a redundancy-aware multimodal state representation framework that explicitly removes duplicated note text over time before policy learning. We evaluate two computationally efficient temporal decomposition strategies for removing duplicated note text: (1) an embedding-space decomposition using singular value decomposition on local history subspaces, and (2) an interpretable sentence-level diff operation that filters out previously documented sentences before text encoding. Using real-world ICU data, we demonstrate that state representations constructed by stripping temporal note redundancy significantly outperform both structured-only and raw-note baselines across multiple off-policy evaluation methods (Model-Based Rollouts, Fitted Q-Evaluation, Weighted Importance Sampling, and Weighted Doubly Robust Evaluation). Our findings show that explicitly isolating new clinical information from repeated note text yields higher-quality state representations and directly improves RL performance for clinical decision support.

Recent training-free post-training quantization methods restore model accuracy through closed-form residual compensation. To constrain additional model storage overhead, several existing methods gate layer selection by goodness-of-fit, retaining only those layers whose compensation yields a positive residual fit score and discarding the rest. In this paper, we show that, under the low-bit W4A4 setting, this gating mechanism fails to distinguish poorly predictable quantization error from numerical solver failure. Rank-deficient input activations yield severely ill-conditioned or numerically singular Gram matrices, causing the closed-form solver to become unstable and produce spuriously negative fit scores. Consequently, existing goodness-of-fit gates misclassify affected layers as uncompensable and discard them. Many of these discarded layers can nevertheless provide substantial error recovery when their compensation is computed using a numerically stable solver. To address this problem, we propose a parameter-free truncated pseudoinverse solver which removes collapsed directions prior to inversion. On ViT-B with the W4A4 setting, our training-free method achieves 81.42\% top-1 accuracy, outperforming prior post-training methods and fine-tuning-based baselines. Combined with joint low-rank and quantization compression, the proposed method reaches a deployable operating point of 80.26\% accuracy at 54.7 MB, providing a well-balanced trade-off between model size and accuracy.

Visual on-policy distillation relies heavily on an informative teacher-student asymmetry, through either a larger, stronger teacher or privileged supervision, such as reference answers or ground-truth regions of interest. This raises a fundamental question: where can informative asymmetry come from when nothing privileged is available? We answer this by inverting where the asymmetry comes from. Rather than adding privileged information to the teacher, we subtract information from the student. This asymmetry creates the same effective learning signal for free as a teacher with access to information unavailable to the student, without ground-truth annotations, rewards, or a separate stronger teacher model. Building on this principle, we introduce Self-Supervised Visual On-Policy Distillation (S$^2$VOPD), a simple yet effective method that constructs on-policy learning signals from asymmetric augmented views. S$^2$VOPD distills the teacher's distribution conditioned on the original image on-policy into the student distribution conditioned on a strongly augmented view of the same image. We systematically explore a broad design space of visual augmentations and uncover that (1) asymmetry matters: all four augmentation families improve performance, while symmetric self-distillation degrades it; (2) strength matters: performance peaks at a moderate strength; and (3) the gap must remain task-consistent: augmentations that completely remove the question-relevant evidence can induce large but uninformative discrepancies. Across six fine-grained perception benchmarks, S$^2$VOPD improves Qwen3.5-4B from 70.7% to 77.4%, above all open-source models compared, up to Qwen3-VL at 235B, and surpasses GPT-5.4. While holding training data the same, it recovers 96% of the improvement achieved by methods with privileged information. Website is at https://williamium3000.github.io/s2vopd

Open-vocabulary object detection test-time adaptation (OVOD-TTA) aims to address the performance degradation that pre-trained base models suffer when encountering image-domain shifts. Existing source-free OVOD-TTA methods rely either on refined test-time information for re-scoring or on pseudo-labels for self-training, leading to significant accuracy degradation when initial predictions are poor. Meanwhile, most conventional source-domain estimation methods recover abstract, sparse representations suitable for the classification task, but fail to capture the dense, concrete features required for detection. To address these issues, we propose PISA, a novel source-free OVOD-TTA method that can be seamlessly integrated into open-vocabulary visual backbones. The core components of our method are the Corruption-Invariant Feature Extractor (CIFE), the Feature Alignment Module (FAM), and a multi-scale alignment framework (BAA). To capture detection-suitable features, we develop CIFE to exploit the invariance of CLIP's visual features across corrupted images, ensuring robustness against various corruptions. We further develop FAM and BAA for the pre-training and adaptation to transform the corruption-invariant features into pseudo-individual source-domain features that are close to the original source-domain features. In this way, dense and concrete pseudo-individual source-domain features are used for supervision instead of unreliable pseudo-label signals. Experiments on the corrupted VOC-C, COCO-C, and LVIS-C benchmarks across three base models demonstrate that PISA substantially improves both the localization precision and the category recognition accuracy of the original models. Notably, PISA achieves state-of-the-art performance without requiring access to source-domain data, surpassing existing methods by 3.92% in AP@50% on COCO-C.

Spatial perception and reasoning from visual observations require recovering geometric structure, establishing correspondences, and understanding spatial relations. Existing approaches typically address these capabilities separately using task-specific architectures or external geometric modules, limiting knowledge transfer among complementary representations of the same physical scene. We introduce SPARGen, a unified multimodal framework that casts 3D reconstruction, dense correspondence, and spatial reasoning as instruction-conditioned generation tasks. SPARGen serializes compact structured and linguistic outputs as token sequences while generating dense geometric fields in image-aligned forms, enabling spatial supervision to jointly shape shared representations within a native multimodal generative model. Experiments across benchmarks for 3D reconstruction, correspondence, and spatial reasoning show that SPARGen achieves competitive performance across heterogeneous spatial tasks within a single native multimodal generative framework.

Octree-based anchor Gaussian Splatting has emerged as a scalable representation for city-scale novel view synthesis, where multi-level anchors adaptively capture scene content from coarse building structures to fine architectural details. However, we identify a fundamental limitation in existing methods: cross-level feature isolation, where each level's anchor features are optimized independently with no inter-level communication, causing color drift on building facades and over-smoothing in textured regions. We present HiCo-GS, a high-fidelity reconstruction framework with two complementary modules. Cross-Level Context Aggregation (CLCA) enables bidirectional hierarchical prior injection by leveraging the octree's spatial containment structure to aggregate per-level context vectors into parent-self-child triplets, fused via a lightweight MLP with residual connection. Coarse-level structural priors flow down to inform fine-level anchors, while fine-level detail statistics feed back to prevent over-smoothing, at negligible computational overhead. Depth-Normal Geometric Consistency (DNGC) regularization enforces agreement between rendered normals and depth-derived normals through an alpha-weighted consistency loss, complemented by edge-aware smoothness losses with progressive warmup that exploit the strong planar priors ubiquitous in urban geometry to suppress floating artifacts. We further introduce the China-Pagoda dataset comprising 8 ancient Chinese pagodas with over 1,200 images each, featuring dense ornamental carvings, curved multi-layer eaves, and repetitive fine-grained textures. Extensive experiments on Mill19, UrbanScene3D, MatrixCity, and China-Pagoda demonstrate that HiCo-GS achieves state-of-the-art rendering quality and substantially cleaner geometry across real-world and synthetic urban benchmarks.Code: https://github.com/WZ-CS/HiCo-GS.

Autonomous pursuit-evasion is a fundamental challenge for Unmanned Aerial Vehicles (UAVs), requiring rapid decision-making under tightly coupled dynamics and continuously changing opponent behaviors. Traditional rule-based or differential-game approaches often struggle with high-dimensional aerial interactions and agile maneuvering. We present AgilePE, a complete system for autonomous UAV pursuit-evasion via self-play reinforcement learning. AgilePE integrates agile low-level control, competitive policy optimization, and sim-to-real deployment in a unified framework. The policy directly maps onboard state observations to Collective Thrust and Body Rates (CTBR) commands, enabling end-to-end agile maneuvering without intermediate trajectory planners or waypoint controllers. For training, we use competitive self-play with Prioritized Fictitious Self-Play (PFSP) and a diversified opponent pool, enabling agents to improve against historical policies while stabilizing optimization and reducing policy oscillation. This process leads to the emergence of sophisticated pursuit and evasion strategies. For real-world deployment, we develop a hardware-aligned simulation pipeline that models actuator-response dynamics, communication latency, and domain randomization. The learned policies transfer zero-shot to real quadrotors without task-specific tuning. Real-world experiments reproduce pursuit-evasion tactics observed in simulation, including rapid dodging and flanking, and demonstrate interactive two-agent zero-shot deployment.

Mobile GUI Agents powered by multimodal large language models (MLLMs) show promise in human-computer intelligence. However, current research primarily focuses on reactive task execution while lacking a comprehensive understanding-prediction-execution process for user intentions, which are the core requirements of active agents. In this paper, we propose the Act2Intention framework that builds an active mobile agent by integrating understanding, predicting user intentions, and executing decisions. First, we construct the Act2Intention Bench through data collection and validated generation, comprising 72,511 intentions and over 700,000 actions across 52 apps, thereby establishing the first benchmark for evaluating proactive agents via continuous intention-action trajectories. We further develop the Act2Intention Agent, achieving proactive services through Proactive-oriented Intention Understanding, Personalized Proactive Intention Prediction, and Experience-guided Intention Execution. Experimental results show that supervised fine-tuning on Act2Intention Bench yields absolute improvements of +32.0 Acc-S, +10.25 Acc-S, and +6.9 SSR points over non-fine-tuned counterparts under the same agent framework for intention understanding, prediction, and execution, respectively. This success underscores the necessity and value of the Act2Intention Bench, which establishes a standardized platform for developing and evaluating proactive agents and consequently paves the way for research on intention-driven human-computer interaction.

To mitigate attention dilution in high-entropy TLS 1.3 flows, we propose BGA, a noise-immune neural distillation framework for encrypted threat intelligence.The methodology first employs Analysis of Variance (ANOVA) to decouple high-discriminatory control-plane features - specifically industrial setpoints - from stochastic cryptographic noise. To resolve the extreme class imbalance within a corpus of 86,878 flow records, a Wasserstein GAN with Gradient Penalty (WGAN-GP) module, enforcing the 1-Lipschitz constraint, is integrated to synthesize high-fidelity minority samples, elevating the detection recall of rare Malicious State Command Injections(MSCI) attacks by 43.2%. At its core, the BGA architecture integrates Bidirectional Long Short-Term Memory (BiLSTM) for temporal dependency extraction and an Adaptive Gated Multi-Head Attention mechanism. This gated unit functions as a neural filter to dynamically suppress encryption artifacts while amplifying malicious signatures. Extensive evaluations on CIC-IDS-2018 and Edge-IIoT benchmarks demonstrate a performance ceiling exceeding 95.2% across all key metrics. Furthermore, noise-injection stress tests confirm BGAs superior structural resilience with a 8.57% performance margin over vanilla Transformers, while its ultra-low inference latency of 0.2820 ms (estimated 1.6920 ms via theoretical scaling for ARM) indicates a high potential for real-time feasibility on heterogeneous industrial edge gateways, providing a promising architectural baseline for future hardware implementation.

LeWM is a lightweight visual world model that learns latent dynamics end-to-end from pixels and ranks candidate action sequences by the distance between their predicted endpoints and the goal. However, LeWM has two limitations. First, during training, it learns local next-step transitions without evaluating complete trajectories relative to the task goal. Second, during planning, it ranks candidates solely by predicted endpoint distance. Because model predictions may differ from actual execution outcomes, the candidate whose predicted endpoint is closest to the goal may not perform best when executed in the environment. The evolution of the complete predicted trajectory can therefore provide complementary information beyond endpoint distance. To address these limitations, we propose Traj-LeWM, which retains LeWM's local-dynamics objective and endpoint score while introducing a goal-conditioned latent trajectory cost (LTC) that aggregates trajectory-level information as a complementary signal. During training, LTC-based trajectory-preference supervision complements next-step prediction in shaping the shared representation. During planning, LTC is combined with endpoint distance to incorporate intermediate-path information into candidate ranking. With joint endpoint-plus-LTC scoring, Traj-LeWM outperforms LeWM on Push-T, OGBench-Cube, Reacher, and Two-Room by $3$, $14$, $7$, and $7$ percentage points, respectively. Controlled experiments and ablations further verify the complementary roles of trajectory-level representation shaping and path-aware candidate ranking.

Production scheduling in complex manufacturing environments is challenging when sequence-dependent setup times, stochastic disturbances, and due-date constraints must be addressed simultaneously. While reinforcement learning (RL) methods have shown promising results in research, most studies rely on simplified benchmark processes, limiting their industrial relevance. This paper demonstrates the applicability of RL-based scheduling in an industry-inspired coating process that reflects practical complexities such as sequence-dependent setup times, machine breakdowns, and variable utilization. The open-source Digital Model Playground (DMPG), a discrete event simulation framework, is used to model the scenario and to train RL agents. Two standard algorithms, Deep Q-Networks and Proximal Policy Optimization, are benchmarked against conventional dispatching rules to illustrate feasibility and to provide a transparent testbed for further research. Results indicate that RL-based scheduling achieves balanced improvements across key performance indicators, with PPO delivering the most robust performance. The main contribution of this work is to bridge the gap between academic research and industrial practice by validating RL-based scheduling in a realistic, shareable scenario and by providing a reusable open-source framework for future studies.

Graph Neural Networks (GNNs) can solve prediction tasks by unintentionally exploiting shortcuts---that is, edges, nodes, and features that correlate with but are not causal for the prediction---which compromise their reliability in out-of-distribution tasks. We introduce XIGL, an architecture-agnostic human-in-the-loop strategy for removing such shortcuts from GNNs. Our key insight is twofold. On the one hand, reliance on shortcuts can be detected by inspecting GNN explanations. On the other hand, once made aware of such shortcuts, sufficiently expert users can provide tailored corrective feedback, which helps deconfound the model. XIGL supports any query strategy; however, since corrective feedback can be expensive to acquire, we develop an active learning strategy for prioritizing explanations that are more likely to display shortcut behavior, lowering annotation and cognitive costs. We showcase the effectiveness of XIGL, including both existing and proposed explanation-based strategies, on several GNN architectures. Our implementation is available online.

Lagrangian modeling is vital to fluid dynamics, as it characterizes particle transport and complements the Eulerian description.However, Lagrangian trajectories are less commonly available than Eulerian fields, while most neural operators are trained and evaluated primarily in the Eulerian representation. This mismatch motivates a new learning problem: can a model trained solely on Eulerian observations generalize zero-shot from Eulerian field prediction to Lagrangian particle rollout, without Lagrangian supervision or task-specific adaptation? To address this problem, we propose the Transferable Latent Operator (TLO), which learns a unified flow representation shared by Eulerian field prediction and Lagrangian particle rollout. TLO decouples latent flow evolution from coordinate-dependent decoding: querying the evolving latent representation at fixed spatial coordinates yields Eulerian fields, whereas querying velocities at particle positions and recursively updating these positions enables Lagrangian rollout. Across five fluid-dynamics benchmarks, TLO consistently outperforms existing neural operators in both Eulerian field prediction and zero-shot Lagrangian rollout, with further gains from limited Lagrangian fine-tuning.

Scientific simulations often produce scalar volumes faster than they can be stored, transferred, and loaded, while in situ reduction must use only a limited share of simulation resources. This work encodes scalar fields as anisotropic Gaussian primitives under a fixed budget. The complete primitive set is allocated analytically from local field structure, including position, orientation, and shape, then refined directly against the scalar field without densification, pruning, or count changes. The selected budget determines encoded storage before refinement and, together with the iteration schedule, provides a controllable refinement-time budget. In a controlled benchmark, truncation-aware field evaluation reduces encoding time by up to 51x; 1.4 million Gaussians encode a billion-voxel volume in at most four minutes on one desktop GPU, with reduced-iteration refinement completing in under one minute. Across five datasets spanning 2.1 million to 1.1 billion evaluated voxels, compression-useful configurations achieve 15.0-38.7 dB PSNR at compression ratios from 2.2x to over 40,000x. Pre-encoding structure statistics characterize fields for which one-shot allocation yields limited gains from additional capacity. Because primitives retain scalar attributes rather than baked appearance, a single compact model serves every subsequent visualization state - supporting post-hoc transfer-function, colormap, lighting, and viewpoint changes without re-encoding.

Autonomous LLM agents are increasingly deployed in complex real-world workflows, yet they remain vulnerable to runtime behavioral drift, a silent deviation from the original task that can lead to irreversible side effects on external systems. Existing approaches address drift at the prompt level but lack structured mechanisms for step-level detection, risk assessment, and recovery decision. Because the main task-executing agent is often a large and expensive model that cannot be re-trained on every deployment, this work targets a plug-and-play recovery module instead. It introduces a graph-based framework in which a single small language model is trained via reinforcement learning to specialize at each node of a recovery graph, external to the main agent. Each node has a precise role\,: drift classification, operation detection, risk evaluation, or final decision and the model learns to produce structured XML-formatted reasoning adapted to that role. Training combines rule-based structural rewards with an LLM-as-judge semantic-quality signal, so that the model is graded both on how it answers (schema and length) and on what it says. Experiments on the public AppWorld benchmark show that the method generally exploits information about the suspected drift onset to issue correct recovery decisions using a small language model. In addition, the trained small language model reliably respects the prescribed output schema and produces semantically appropriate content in each field according to its assigned node role.

Reasoning-intensive retrieval requires text representations to capture not only semantic similarity, but also the reasoning needed to determine relevance under a given retrieval instruction. Existing reasoning-enhanced embedding models improve retrieval by incorporating reasoning information into dense representations, yet their supervision is typically dominated by the final retrieval objective. As a result, latent reasoning trajectories may learn shortcut reasoning patterns that preserve retrieval performance without producing meaningful incremental retrieval gains. We propose Retrieval Grounding Latent Reasoning (RGLT), a latent reasoning framework for dense retrieval that explicitly connects intermediate latent transitions with retrieval improvements. RGLT performs non-autoregressive reasoning in hidden space through an instruction-conditioned latent reasoning trajectory constructed from silent tokens. It combines process-supervised explicit-to-implicit distillation with retrieval-grounded supervision, using stage-wise CoT reconstruction to shape intermediate latent states and retrieval-effect credit to optimize incremental retrieval gains across the latent reasoning trajectories. Experiments on reasoning-intensive retrieval benchmarks show that RGLT consistently outperforms strong baselines while preserving efficient embedding inference.

When forecasting hourly returns for 1,000 US equities, we observe an unexpected phenomenon: predictions become nearly flat and show poor stock ranking, as measured by cross-sectional correlation. We call this forecast collapse. Surprisingly, the phenomenon largely disappears when forecasting trading volume under the same setting. We investigate forecast collapse across time-series foundation models (TSFMs), twelve deep-learning forecasting models, and 97 public benchmark configurations, and find that it is closely tied to target predictability. We identify two distinct reasons behind it: low predictability limits the amplitude of calibrated point forecasts, while per-series objectives leave cross-series structure unidentified. These findings reveal a calibration-ranking tradeoff: optimizing squared error leads to flat predictions, whereas directly optimizing cross-sectional correlation improves ranking but can inflate forecast amplitude by more than an order of magnitude. To address this tradeoff, we introduce CalibRank, a simple objective that balances calibration and ranking. On Finance1K, CalibRank nearly triples cross-sectional correlation while keeping amplitude close to the target, and improves correlation on all tested models. Our results reveal a blind spot in conventional time-series evaluation: per-series metrics can hide failures in cross-series structure needed by downstream decisions.

The Shapes Constraint Language (SHACL) is a W3C recommendation to express syntactic constraints, called shapes, on RDF graphs. SHACL validators are used to test whether a given graph adheres to such a shape. However, RDF graphs often come with OWL ontologies, whose implicit knowledge needs to be taken into account. This is classically handled by first applying reasoning and then performing the constraint checking on the results, often using different technologies which makes the process inefficient and vulnerable for mistakes. To overcome this, we propose to internalise the OWL axioms in the SHACL constraints; we construct a rewriting which takes as input both shapes and an OWL EL$^-$ ontology -- a fragment of OWL EL restricting the usage of existential restrictions -- and produces SHACL constraints. This output can then be evaluated by any validator supporting SHACL core regardless of its reasoning support, while yielding the same results as the traditional approach. The implementation of our translation is evaluated both against applying state-of-the-art reasoners and validators consecutively, as against validators with built-in reasoning support. For our benchmark, we show that our approach is in general more efficient in finding violations compared to the sequential approach, thus providing a powerful tool which simplifies combining reasoning with validation.

The one-warehouse multi-store (OWMS) system is a fundamental inventory network in which a nonreplenishable warehouse allocates shared stock across multiple stores over time. Existing OWMS learning policies are built around a fixed target calibrated to the initial average resource rate, but such a fixed-target architecture cannot re-center after realized sales change the remaining resource available per future period. We develop Resource-Adaptive Primal-Dual Learning, a new learning framework that tracks the primal-dual resolving path with censored demand as the remaining-resource state evolves. In each period, the current resource rate indexes the target store allocations and dual variable, while censored sales provide gradient estimates for updating both. The analysis combines expected-sales geometry with a moving-target argument to yield logarithmic expected regret, improving on the state-of-the-art square-root-order guarantees of existing OWMS learning policies. The underlying design and analytical ideas may inform other online learning problems with depleting shared resources. Numerical experiments further demonstrate good finite-horizon performance of a practical variant across different horizon lengths and inventory regimes.

Cloud-local LLM inference systems have the potential to use the reasoning capability of large cloud models while protecting sensitive user data on personal devices. Cloud-bound requests must exclude personally identifiable information (PII) to prevent external data leakage. Existing privacy-preserving methods rely on prompt perturbation, entity masking, or model fine-tuning, but these approaches may distort contextual semantics or require additional training. This paper proposes P2Skill, a prompt-based skill distillation method in which a local small language model (SLM) autonomously performs decomposition, PII-aware routing, paraphrasing, and reconstruction by following the skill prompts. Skills are iteratively refined from execution failures by a cloud LLM, enabling the local SLM to generalize beyond memorized PII patterns, and therefore P2Skill requires no privacy-specific fine-tuning or learned auxiliary detectors. Evaluation on a four-domain benchmark shows that P2Skill achieves $1.69\times$ and $3.66\times$ higher privacy-preserved inference quality than previous baselines.

Safety classifiers deployed with large language models often fail for two reasons: their decisions reflect the policy learned during training rather than the deployer's desired policy, and their performance degrades as deployment traffic evolves. We present Regime-Conditional Verification (RCV), a lightweight wrapper that adapts an off-the-shelf safety classifier without retraining it. RCV estimates, from the classifier's internal representations, the probability that each prediction disagrees with the deployer's policy, and selectively corrects predictions likely to be wrong. The same correctness estimates also provide a label-free signal for detecting distribution shift, enabling a maintenance loop that updates the correctness estimation layer and resorts to classifier fine-tuning only when necessary. Across three off-the-shelf safety classifiers and two benchmark datasets, RCV improves adherence to the deployer's policy in every classifier-dataset combination, catching up to 0.81 of previously missed unsafe content without modifying the underlying classifier. In a deployment study with ten attack campaigns, each a harm category held out of RCV's training, RCV detects every campaign in a dedicated injection panel; in the maintenance census most drift episodes are repaired without updating the classifier, and the fine-tune is reserved for the residual episodes that repair does not restore.

Collaborative perception breaks through single-view limitations via multi-agent information exchange. However, multi-source noise such as pose errors and communication delays degrades fusion feature quality, constraining perception performance. Joint training of detection and BEV segmentation provides a natural remedy, where segmented road regions help constrain target distributions and detection bounding boxes help recover ambiguous segmentation boundaries. To this end, we propose a robust Collaborative perception framework with expert-driven Detection and bev Segmentation (CoDS). To address spatial inconsistency in fusion quality, we first introduce the Collaborative Reliability Map (CoRM) to explicitly quantify feature quality distribution. Based on CoRM, we design the Semantic Mixture-of-Experts (S-MoE) module to extract differentiated features for inconsistent feature demands. Finally, to further mitigate feature noise degradation, the Bidirectional Task Complementary Interaction (BTCI) refines task-aware features through bidirectional injection. Extensive experiments on OPV2V and V2V4Real datasets show that our CoDS surpasses existing baselines on both tasks and maintains stable robustness under multi-source noise. Code: https://github.com/JinlongW128/CoDS and https://openi.pcl.ac.cn/OpenAIDriving/CoDS.

Localized 3D stylization aims to modify the appearance of a specified object part while preserving the remaining surfaces. In large reconstruction models (LRMs), this task is challenging because style is injected into intermediate appearance representations before rendering, while compact triplane features are shared across target and non-target surfaces, causing style leakage and boundary ambiguity. We propose Owner3D, a training-free framework for localized 3D stylization that integrates localized appearance control directly into the LRM reconstruction process. Specifically, Owner3D introduces ownership-guided style writing to restrict reference-style injection to target regions, producing a single localized stylized triplane without additional training while avoiding separate global style and appearance representations. To resolve appearance ambiguity near semantic boundaries, we further introduce boundary dual slots that maintain separate local feature sources for target and non-target regions. Finally, a surface-first texture readout hierarchically combines surface, 3D, and triplane ownership evidence to robustly recover appearance under incomplete visibility. Experiments on a benchmark constructed from Google Scanned Objects and PartNet demonstrate that Owner3D consistently outperforms existing 3D stylization methods in target-region style fidelity and non-target appearance preservation, reducing appearance leakage by 86.4% and 89.9% compared with StyleSplat and LAENeRF, respectively.

Transition-state (TS) structures define the energetic barriers and mechanistic pathways of elementary chemical reactions, yet their identification remains computationally demanding because conventional saddle-point searches require expensive quantum-mechanical calculations. Recent machine-learning approaches have accelerated TS generation by predicting structures from reaction endpoint information, but they primarily learn geometric correspondence between endpoints and TSs, leaving the structural transformations underlying elementary reactions implicitly represented. To address this limitation, we introduce TransTS, a reaction-transformation-aware framework for generalizable TS generation from atom-mapped reactant-product pairs. TransTS explicitly learns atom-level structural transformations between reaction endpoints and integrates them with a unified atom-aligned geometric representation of reactants, TSs and products, enabling reaction-aware equivariant generation of TS geometries. TransTS is designed to provide reliable TS initial guesses for subsequent quantum-chemical refinement, where generated structures are evaluated not only by geometric similarity but also by their ability to converge to validated saddle points and recover the intended reaction pathways. Across IID and zero-shot OOD benchmarks, TransTS demonstrates improved TS initialization quality, with particularly strong generalization to unseen reaction distributions. On the challenging GDB-10-rxn and GDB-17-rxn OOD benchmarks, TransTS generates TS candidates that more frequently converge to validated saddle points and recover the intended elementary reactions after refinement than existing approaches under the same training regime. Scaling reaction coverage and model capacity further improves both geometric fidelity and refinement outcomes.

Scientific figures and tables encode essential experimental evidence, yet remain difficult for digital libraries and multimodal AI systems to retrieve and interpret. The ALD/E-ImageMiner benchmark and ICDAR 2026 Competition on Information Extraction from Atomic Layer Deposition/Etching Scientific Figures provide 1,951 figures from 205 publications, expert-annotated for classification, data table extraction, summarization, and visual question answering. In these companion proceedings, we present a forward-looking perspective on how the benchmark can guide future scientific-image challenges. We examine how its tasks probe capabilities from visual and quantitative reading to domain-grounded reasoning and evidential justification, and how Bloom-informed question design can support deeper scientific understanding. We propose "scientific conceptual understanding from images" as a long-term benchmark objective, with future directions including broader domains and figure types, contextual and cross-document synthesis, hypothesis evaluation, provenance, uncertainty, counterfactual grounding, and open-ended multimodal research. This perspective connects the ICDAR 2026 challenge to a broader agenda for machine-actionable scientific visual knowledge and verifiable multimodal scientific AI.

AI agents increasingly act on external systems through standardized tool-calling protocols such as the Model Context Protocol (MCP), yet no infrastructure layer constrains their actions to what a principal has verifiably authorized: authorization logic lives in application code, is neither signed nor independently auditable, and the resulting logs lack evidentiary value. We present Mandato, a governance proxy that enforces digitally signed mandates on agent actions at the protocol level. A mandate is a machine-readable, cryptographically signed authorization artifact specifying which tools an agent may invoke, under which parameter constraints and contextual conditions, for how long, and on whose behalf; the proxy evaluates every tool call against the applicable mandate chain, blocks non-conforming calls in line, and records every decision -- permit, deny, and the evidence for each -- in an append-only, hash-chained audit log designed for evidentiary use and periodically anchored via qualified timestamps. The mandate is deliberately modeled on the civil-law institution of delegation of authority, making the artifact legible to lawyers and auditors, not only to engineers. We give the mandate model and its decision semantics, the reference architecture as an MCP-transparent proxy with separated decision and enforcement points, and a mapping of the mechanism onto EU AI Act Articles 12 and 14, GDPR accountability, NIS2, and eIDAS 2, including a roadmap to qualified attestation through Qualified Trust Service Providers (QTSPs). We describe the implementation status of the reference system and a quantitative evaluation plan covering enforcement overhead, audit completeness, and tamper-evidence verification cost.

Conversational recommendation for e-commerce is increasingly mediated by large language models (LLMs), yet many real-world deployments operate under a stricter requirement: recommendations must be drawn only from a merchant's fixed catalog, without web search or unsupported product claims. In this setting, the main challenge is reliability under hard constraints: the system must satisfy user requirements, remain grounded in available inventory, and preserve preferences across multiple conversational turns. We present MACS (Multi-Agent Commerce System), a hybrid multi-agent framework for reliable conversational recommendation in fixed-catalog settings. MACS uses LLMs for language-facing tasks such as interpreting user requests, eliciting preferences, and generating responses, while correctness-critical operations, including product retrieval, hard-constraint filtering, brand exclusion, and progressive relaxation, are executed deterministically by the merchant agent. A session-persistent preference layer tracks constraints across turns, enabling consistent handling of budget overwrites and exclusion reversals. On a 140-query single-turn benchmark, MACS achieves the highest pass rate (87.1%) and perfect brand compliance (1.000). On a 10-scenario multi-turn benchmark, MACS achieves the strongest macro Pass@5 (72% vs. 56% GPT+Catalog / 52% Gemini+Catalog) with zero constraint drift. The advantage is sharpest on exclusion reversal (100% vs. 20% / 0%) and constraint accumulation (100% vs. 60% / 40%). Mean judged response quality is similar across systems (0.751 vs. 0.736). These results suggest that hybrid architectures combining deterministic constraint enforcement with session-persistent preference tracking provide stronger reliability-oriented performance than catalog-bound prompt-only baselines in the fixed-catalog merchant setting.

Machine learning is rapidly reshaping constitutive modeling, offers new ways to learn material behavior directly from experimental data, and challenges long-established modeling paradigms. But with a growing number of machine-learning-based approaches available, how do they compare in practice? In this paper, we use the classic experimental data of Treloar to benchmark popular frameworks for hyperelasticity: (Generalized-Invariant) Constitutive Artificial Neural Networks, Physics-Augmented Neural Networks, (Adaptive) Material Fingerprinting, and Efficient Unsupervised Constitutive Law Identification & Discovery. We compare their fitting performance, computational cost, hyperparameter sensitivity, and ease of implementation. Furthermore, we discuss the trade-offs between predictive accuracy and model complexity. The latter is assessed by quantifying both the number of material parameters in the discovered models and the computational time required to evaluate the constitutive model and its derivatives. The results show that all methods can reproduce the benchmark data remarkably well. Rather than identifying a single winner, we highlight the strengths and limitations of each approach and provide practical guidance for their use. The source code for all six methods, including the training and comparison scripts, as well as all results and data used in this study, is publicly available via https://doi.org/10.5281/zenodo.21915635.

Seismic facies segmentation has emerged as a significant challenge in geophysics, requiring robust methods and systems to effectively identify geologically analogous facies with limited labeled data. Although existing studies have shown promising results in 2D facies segmentation, they often preprocess the original 3D seismic volumes into sets of 2D slices, typically the inline and crossline directions, and treat this problem as a purely 2D segmentation task. This simplification introduces discontinuities across slices and fails to preserve the spatial and structural continuity in 3D seismic data, thus limiting the model's ability to learn coherent geological patterns. In this work, we present a comparative and reproducible benchmark for voxel-based 3D seismic facies segmentation, built upon publicly available seismic volumes including the Netherlands F3 and the Parihaka datasets, with standardized data splits and evaluation metrics. By evaluating the three representative families of modern 3D segmentation architectures, we establish strong baseline results that highlight the potential and remaining challenges for future research in this domain.

Authoritative scientific knowledge in geoscience remains largely trapped in legacy monographs and historical literature, where unstructured text and complex layouts hinder computational access. We introduce HERMES, a scalable multi-agent framework that extracts structured data from ultra-long scientific documents. Using a coordinating large language model, HERMES integrates domain constraints, validation rules and evidence tracing within a unified document-level extraction process that incorporates parsed text, tables, figures and captions. Applied to the 55-volume Treatise on Invertebrate Paleontology, the system produced a structured database of 32,277 fossil taxonomic entities and 451,878 attributes, released online at https://treatise.geolex.org. Extraction performance remained stable across fossil groups (average F1 scores of approximately 0.90 for entities and 0.91 for attributes), improving per-volume efficiency approximately sixfold relative to the tested fully manual baseline. Evaluation in palaeomagnetism and geochemistry, conducted without additional model training, demonstrated transfer across distinct geoscience domains. This work provides a practical pathway to transform historical scientific literature into FAIR-oriented structured data, offering a sustainable infrastructure for data-intensive disciplines and large-scale knowledge integration.

Time series forecasting with pretrained foundation models has demonstrated strong zero-shot capabilities. However, achieving optimal performance on time series with short or negligible historical data in domain-specific applications typically requires adaptation via either fine-tuning or RAG. While fine-tuning is effective, it incurs substantial computational costs. This work explores RAG within univariate time series (Retrieval Augmented Generation) as a more efficient alternative, in particular RAF (Retrieval Augmented Forecasting), and introduces RAEF (Retrieval-Augmented Extended Forecasting), a model-agnostic method built upon RAF. RAEF incorporates key refinements to the retrieval and aggregation mechanisms: (1) direct retrieval in input-space rather than embedding-space, reducing inference overhead, and (2) concatenation-based aggregation that preserves temporal structure instead of averaging. Empirical evaluation across multiple benchmark datasets demonstrates that RAEF outperforms RAF in both accuracy and inference overhead. Furthermore, comprehensive comparisons with zero-shot and fine-tuned foundation models show that RAEF achieves competitive or superior performance to fine-tuning while avoiding its computational burden, establishing it as a practical and scalable approach for domain adaptation in time series forecasting.

Robotic manipulation of flat objects is challenging due to the ungraspable configurations and strong variations in object geometry and material. Existing methods rely on heuristic pre-manipulation and are often evaluated in closed settings with limited generalization. We propose a unified framework that decouples the manipulation into a strategy generator and an action execution module. The strategy generator predicts appropriate manipulation strategies from object point clouds by learning strategy-centric, object-invariant representations via simulated data transformation and contrastive learning. Conditioned on the predicted strategy, the execution module decomposes long-horizon manipulation into reusable action primitives and dynamically composes them to generate stable trajectories. To enable systematic evaluation, we introduce FlatLab, a comprehensive simulation benchmark for robotic flat object manipulation. FlatLab provides high-fidelity physical simulation of diverse rigid and deformable flat objects, automated multi-modal data collection, and standardized task definitions and evaluation protocols. Experiments conducted in FlatLab demonstrate that our approach generalizes effectively to unseen objects and categories, outperforming existing baselines. The project page and the code are provided at https://flatlab-web.github.io/.

This paper integrates end-to-end Visual-Language-Action (VLA) models with agentic tool-use to propose Agentic Robot with Tool-use (ART). ART is a tool-injection framework that tunes any VLA model to leverage off-the-shelf tool modules for low-level vision, high-level affordance, and embodiment enhancement. Compared to vanilla VLA models with a whole continuous action solution space, ART reduces the complexity of the action solution space through tool-use, which not only improves generalizability across different tasks but also reduces data dependency. To demonstrate the advantages (high generalizability and low data dependency) of this framework, we first built a dataset of 30K tool-use trajectories and action demonstrations, which is much smaller than those used by baseline methods. We then designed a training regimen for long-trajectory tool-use reasoning in challenging environments. Experiments show that ART achieves a 20% higher success rate than mainstream baselines on simulation and real-world tasks, such as pick-and-place in the dark at novel viewpoints. Empirical results highlight the benefits of an agent-based approach: modular tool utilization enables more efficient training, lightweight deployment, and scalable integration of new tools. This design fosters robustness, adaptability, and extensibility, paving the way for the practical deployment of VLA systems in complex real-world scenarios.

Modern text-to-image diffusion models rely on classifier-free guidance (CFG) to achieve high image fidelity and text alignment. However, CFG typically applies a static, global scale across all timesteps, samples, and conditions -- a choice that is generally suboptimal and can introduce artifacts, as different states may benefit from different levels of guidance. While time-varying schedules are known to improve quality, designing them by hand is non-trivial and application-dependent. In this paper, we learn the guidance schedule as a function of diffusion time, conditioning and the current noisy sample, in order to better align sampled images with the text prompt. We frame this as a density ratio estimation problem: a discriminator is trained to estimate the time-dependent log-density ratio between the true and guided marginal distributions, while a lightweight generator network predicts the optimal, state-dependent guidance scale. Empirically, our approach outperforms both heuristic CFG schedules and prior methods for learning dynamic guidance on text-to-image generation benchmarks.

Skills have emerged as a practical and effective approach for enhancing LLM agents at inference time through structured packages of knowledge. However, existing evaluations largely measure whether skills improve aggregated task success, leaving a more fundamental question underexplored: \emph{\textbf{When do skills help, why do they work, and where do they fail?}} Through controlled experiments across various benchmarks, agent harnesses and LLMs, we isolate the effects of representation, outcome annotation, retrieval difficulty, and cross-framework robustness of skills. To further answer this question, we design a contrastive study that combines controlled quantitative experiments with paired trajectory analysis. We normalize 8,135 trial records from controlled experiments and retain 238 valid unique labels from 240 open-coded records. We consolidate these observations into a taxonomy of three high-level categories and twelve skill-use modes: skills work when noisy trajectories become procedural anchors that stabilize execution. Skills improve over Workflow Memory by 6.06 points in matched comparisons. Procedural anchoring accounts for 65.7\% of skill cases, versus 4.5\% for explicit knowledge injection, showing that skills stabilize action rather than inject missing facts. Retrieval is a separate bottleneck: as pools grow from 5 to 100, actual-use precision falls from 29.6\% to 3.3\%. Confusable distractors impair offline identification, yet downstream success remains stable; exact ground-truth invocation is neither sufficient nor necessary. Skills fail under brittle assumptions, incompatible contexts, or insufficient adaptation. These findings move evaluation beyond aggregate success rates and guide reliable self-evolving agents.

Existing multimodal RAG methods often flatten structured documents into isolated text and image units, weakening the source organization and local text-image logic needed for faithful evidence selection and placement. We propose HAM-RAG, a Hierarchy-Aware Multimodal RAG framework for structure-faithful interleaved generation. HAM-RAG uses document hierarchy as a grounding signal across retrieval and generation, contextualizing textual and visual evidence and preserving source position and local text-image relations in the prompt. We further introduce HAM-Bench, covering Wukong, Wiki, arXiv, and Recipe across game walkthroughs, web pages, scientific papers, and step-wise recipe documents. Across multiple backbones, HAM-RAG improves the main multimodal average by 17.3% over the strongest non-hierarchical baseline. On Wukong, HAM-RAG improves Img-CBS by 24.2% over the strongest non-hierarchical baseline, demonstrating substantially better local text-image alignment. The main experiments and ablation study together demonstrate that document hierarchy is a key grounding signal for faithful image selection, placement, and local text-image alignment. These findings highlight the value of hierarchy-aware grounding for reliable multimodal assistants that generate answers faithful to the source organization, procedural structure, and local text-image evidence of structured documents, such as technical manuals, maintenance guides, and industrial SOPs. The code is available at https://github.com/MCCodeAI/HAM-RAG.git.

We present a demonstration showcasing the Autonomy Testbed for Multi-purpose Orbiting Systems (ATMOS), a planar spacecraft-analog robot designed for hardware-in-the-loop evaluation of guidance and control strategies in microgravity-like conditions. Using ATMOS as the physical test platform, we investigate the design, analysis, and performance evaluation of control architectures for remotely operated spacecraft under round-trip communication delays. In this work, we develop and experimentally validate a control strategy that combines state prediction and trajectory tracking control to perform a docking maneuver, accounting for time-varying random communication latency between ground operators and the ATMOS system. The demonstration includes a long-distance remote control experiment between Seoul and Stockholm, introducing realistic intercontinental delays and variability. The results highlight the capability of ATMOS to support rapid, reliable, and cost-effective testing of spacecraft teleoperation concepts, establishing a first step toward robust validation of on-orbit operations in microgravity-like environments.

Dexterous manipulation is a fundamental capability for embodied intelligence, but scaling it remains difficult because robot demonstrations are expensive to collect and action spaces vary across embodiments. Policies trained on heterogeneous data can also entangle task-relevant visual cues with embodiment-specific appearance, limiting cross-embodiment generalization. We present AdvDex, a unified Vision-Language-Action framework for learning dexterous manipulation from human and robot demonstrations. First, we introduce OmniShare, a large-scale multimodal dataset of human manipulation demonstrations that provides high-quality kinematic supervision and tactile measurements while reducing reliance on robot teleoperation. Second, we propose the Joint-Aligned Action Space (JAAS), a canonical action representation comprising an $\mathrm{SE}(3)$ wrist pose and 15 finger joints, thereby functionally aligning human hands, dexterous robot hands, and parallel grippers. Finally, we use domain-adversarial learning to reduce embodiment-specific information in the learned visual representation. Experiments on hand-action prediction and real-world dexterous manipulation show consistent improvements over baselines, effective zero-shot human-to-robot skill transfer, generalization to unseen objects and environments, and data-efficient few-shot adaptation.

Benefiting from high temporal resolution and dynamic range, event-based local feature methods have attracted increasing attention. However, event sparsity, noise, and limited texture still hinder robust local feature learning. Deploying such methods on resource-constrained platforms such as unmanned aerial vehicles also requires balancing accuracy and energy efficiency. To address these challenges, this paper proposes \textbf{E-S2Feat}, a spiking neural network framework for event-based local feature detection and description. The framework jointly optimizes local feature learning from the perspectives of feature representation and selection. First, a module-specific spiking activation mechanism preserves fine-grained structural cues and discriminative information under low-bit, energy-efficient inference, thereby improving overall representation fidelity. Furthermore, a semantic-guided feature modulation mechanism leverages semantic priors to refine keypoint response distributions and enhance local descriptor discriminability, thereby guiding the model to extract local features with greater geometric stability and stronger discriminative capability. Experiments on the ECD and EDS datasets show that the proposed method significantly outperforms baseline methods such as SuperEvent in pose estimation accuracy. It also achieves accuracy comparable to its artificial neural network counterpart while delivering an approximately 4.8-fold improvement in theoretical computational energy efficiency. Visual-inertial odometry experiments on the TUM-VIE dataset further verify the effectiveness and practical application potential of the proposed method in complete SLAM systems.

Vision-language-action (VLA) models for autonomous driving jointly produce scene interpretation, language-based reasoning, and driving trajectories. Existing evaluations often use independently selected synthetic, simulated, and physical data, so measured performance gaps can be confounded by changes in scenario content rather than genuine domain sensitivity. We propose SSP (Synthetic-Simulation-Physical), an event-matched Syn2Sim2Phy evaluation framework that anchors cross-domain comparison to the same safety-critical interaction. Starting from a synthetic long-tail video, SSP builds a validated event specification that preserves road topology, participant roles, relative motion, conflict evolution, passing order, response constraints, and event phases. Platform-specific realizations are then constructed in CARLA and on a closed proving ground and are evaluated only after transfer audits confirm preservation of mandatory event properties. SSP maps heterogeneous outputs from OpenEMMA, LLaViDA, and Alpamayo-R1 into common semantic slots and a 1 s trajectory window to assess output validity, semantic accuracy, critical-interaction recognition, trajectory quality, and risk response. Across Cut-in and vulnerable-road-user crossing cases, the macro-averaged Integrated VLA Capability Scores are 0.259, 0.291, and 0.325 in the Synthetic, Simulation, and Physical domains, respectively, while the best domain varies by scenario. Alpamayo-R1, OpenEMMA, and LLaViDA obtain scores of 0.405, 0.338, and 0.131. SSP provides a reproducible scene-transfer chain and an evidence-qualified evaluation of VLA behavior without assuming that the Physical domain is universally superior.

Action-conditioned video world models require low-latency causal generation and reliable responses to game-native controls. Although causal distillation enables one- or few-step video synthesis, extending it to interactive world models remains challenging, as discrete keyboard states and continuous mouse motion must remain aligned with temporally compressed latent chunks during causal training and autoregressive rollout. We introduce ForgeWM, a progressive framework that transforms a bidirectional action-conditioned video generator into efficient few-step world models through domain adaptation, teacher-forced causal training, causal consistency distillation, and on-policy distribution matching with a bidirectional teacher. The resulting budget-specialized students operate at steady-state denoising budgets of 1, 2, and 4 steps. ForgeWM further supports a dual-path deployment protocol combining latency-critical interaction with optional replay-time refinement, where the one-step student re-noises and refines its saved draft. On paired Minecraft trajectories, ForgeWM leads the evaluated systems in Imaging Quality, reference-aligned motion-profile agreement, action-sign accuracy, and mouse-control accuracy, while achieving the lowest reference LPIPS; the same four-stage recipe transfers to gamepad-controlled FPS gameplay. Replay-time refinement matches four-step reference quality while remaining roughly three times closer to the experienced trajectory than regeneration from noise. These results demonstrate ForgeWM's effectiveness for controllable few-step video generation.

Live game commentary is scarce: it exists for professional esports broadcasts and almost nowhere else. We present a content-based video narration system that produces spoken, esports-style commentary for arbitrary gameplay recordings using a general-purpose vision-language model (VLM) and a text-to-speech back end, with no game-specific instrumentation, no engine telemetry, and no task-specific training. Three mechanisms carry the system. Temporal mosaic packing arranges nine uniformly sampled frames into a single 3x3 image, letting an image-native VLM reason about motion while consuming one image payload per segment instead of nine. Context-conditioned prompting replays the K most recent narrations as assistant-role history, suppressing the repetition that dominates per-segment captioning of static scenes. Duration-conditioned generation and elastic alignment constrain narration length in the prompt, then time-scale or symmetrically pad the synthesized audio so each utterance fills its segment slot exactly, giving frame-accurate muxing without a forced aligner. The implementation supports either cloud TTS or a 6-bit quantized 4B-parameter on-device TTS model on Apple silicon, making the speech stage fully local. We report a qualitative case study on real-time strategy footage, a cost model showing the mosaic reduces per-minute image payloads by 9x, and a candid account of observed failure modes - hallucinated game state, resolution loss from mosaicking, and prosody artifacts from time-scaling. We release the system as a reproducible baseline, with an evaluation protocol for the quantitative study a full version will report.

Understanding tens-of-minutes surgical videos requires long-horizon temporal reasoning, answering what happens before, after, or across stages of a procedure by grounding the question in visual evidence spread across time. Existing approaches handle this poorly: a one-shot vision-language model (VLM) compresses the whole procedure to fit its context window and loses the detail a "before" or "after" question depends on, while video agents that train the model where to look are data-hungry and transfer poorly to out-of-domain surgery. We build an agent harness that separates reasoning from perception and improves by evolving context rather than optimizing weights. A text-only orchestrator plans which evidence to gather and issues an auditable sequence of tool calls, while frozen vision-language sub-agents execute each call over the pixels, viewing, cropping, inspecting frames, and retrieving external knowledge. We further propose a gradient-free, reward-gated Heuristic Skill Distillation loop that mines the agent's own low-scoring traces and keeps a candidate skill only when it raises a validation reward, yielding reusable retrieval skills, notably directed re-look. Growing an external skill library rather than tuning weights, the loop adapts from only about 100 labeled examples, far fewer than supervised or reinforcement fine-tuning requires. To evaluate this agent, we introduce MedClawBench, a de-leaked, doctor-grounded benchmark of 1,123 questions over self-built long neurosurgery recordings and a held-out public lecture-video test split. Across both datasets and all four evaluation dimensions, our agent consistently outperforms one-shot VLMs and general video-agent frameworks, with the largest gains on the long, out-of-domain neurosurgery videos. Project page: https://fyycs.github.io/medclaw/.

Large Reasoning Models (LRMs) achieve strong performance on complex tasks through extended chain-of-thought generation, but incur substantial computational costs during inference. In production settings, batched inference is essential for high throughput, yet the existing training-free adaptive pruning methods we evaluate severely degrade in this regime. Because a batch must share a single pruning mask, these methods aggregate activations across samples and then apply threshold-based selection; the threshold, calibrated offline on unaggregated activations, no longer matches the aggregated distribution, so the realized sparsity ratio drifts and accuracy on reasoning tasks collapses under batched inference. In this work, we propose a training-free adaptive pruning method designed specifically for batched inference in LRMs, built on two components. First, we replace threshold-based selection with periodic top-k selection over the aggregated importance scores, which is unaffected by the shift that aggregation induces in the activation distribution, and which runs selection once per update period rather than at every token, preserving the speedup. Second, based on the observation that important neurons re-fire periodically during long reasoning generation, we introduce an activation memory that accumulates importance across update phases so that recurring neurons are retained. Experiments on diverse reasoning benchmarks demonstrate that our method outperforms the previous state-of-the-art adaptive pruning method by 39.7 percentage points in average accuracy at batch size 4 with 50% target sparsity on DeepSeek-R1-Distill-Qwen-7B, and reaches 1.40x speedup over dense inference at 50% actual sparsity.

Perceptual loss functions in Deep Neural Network (DNN) deblurring architectures improve the overall quality of restored images. However, few focus on explicitly targeting sharpness in the restorations. We conduct a subjective study of models trained with and without losses which explicitly target sharpness using a four-protocol approach, exploring preferred sharpness levels and effects on image quality. We introduce a novel dataset of images with uniform sharpness increments along with Difference Mean Opinion Scores (DMOS). Additionally, we propose a novel class of Sharpness Informed (SI) Image Quality Assessment (IQA) metrics which properly penalize over-sharpening. Our new SI-PSNR metric outperforms all other PSNR variants in terms of correlation statistics on IQA benchmarking datasets. We show that, on average, images restored using a sharpness-aware composite loss are preferred in 67% of binarized comparisons, as opposed to losses that do not explicitly target sharpness.

Nanbeige4.2-3B is a 3B-parameter agentic model built around a Looped Transformer (LT) that reuses one stack of layers for a second forward pass, adding effective depth without additional parameters. Evaluated on Apple Silicon (MPS), we identify five independent bugs which prevent the released checkpoint from running via Hugging Face transformers out of the box (including a silently-zeroed RoPE buffer and calls to removed transformers cache APIs). Furthermore, we show that fixing these bugs is still not sufficient for agentic tasks, due to the LT's layer-reuse strategy (which effectively doubles peak attention memory) used to achieve parameter efficiency. We thus introduce a chunked-prefill strategy which alleviates the incurred memory-capacity penalty, extending allowable context width by $2.7 \times$ on 32~GiB shared memory. However, even with the reduced memory overhead, we show that patches are required to render Nanbeige4.2-3B usable; resolving both system prompt and MPS-native memory bugs finally allows reliable evaluation on standard MCP and tool-calling benchmarks. On a subset of MCPMark, the debugged model completes up to 30\% of real agentic tasks (up from the original's 0\%), while, on BFCL, it is near-perfect at single tool calls (yet fails the majority of multi-tool tests). We release the patched checkpoint, system prompt optimizer, and evaluation harnesses at https://github.com/johnhalloran321/Nanbeige4.2-3B-mps-fix.

Quantum optical skyrmions, as topologically robust quantum information carriers, hold transformative potential for resilient high-dimensional quantum information networks. However, their practical exploitation was still restricted to a single quantum channel, which precludes the multiplexing essential for practical high-capacity quantum networks. Here, we utilize a metasurface to achieve multi-channel quantum state distribution of the polarization-entangled photon pairs, inducing a two-photon bunching effect in both the spin and spatial dimensions with compact flat optics. At the spatial bunching port, controlled manipulation of the spin-orbit interaction enables the generation of a tunable single-photon skyrmion pair. In contrast to any prior skyrmion generation, the single-photon skyrmions are mediated and topologically controlled by quantum measurement in multiple channels. Concurrently, during the amplitude and phase modulation process, both the skyrmion localization and the texture helicity can be precisely customized. The proposed tunable single-photon skyromions offer multidimensional controllability and topological stability provide a viable path toward noise-resilient high-dimensional quantum information processing.

The recent advances toward sixth-generation (6G) and beyond-6G networks have accelerated the need for intelligent resource management mechanisms capable of supporting heterogeneous services under shared infrastructures in network slicing. However, resource allocation in network slicing naturally forms a resource-coupled cooperative optimization problem with competing slice demands. Slices compete for limited resources to minimize individual latencies while coordinating to avoid conflicts and underutilization. Although multi-agent reinforcement learning (MARL) has shown promising performance in such settings, existing online formulations remain costly, unsafe, and difficult to deploy due to their reliance on environmental interactions and communication among agents. In this work, we present explainable artificial intelligence (XAI)-guided conservative decentralized execution (X-CODE). X-CODE is an explainable offline MARL that operates offline without environmental interaction, nor inter-agent communication. It exploits explainability-aware reward shaping to modify the relative preference among joint offline transitions during centralized training to improve decentralized resource-allocation behavior. In deployment, the agents operate independently without signaling exchange among the agents. Simulation results demonstrate that the proposed approach achieves zero observed resource-conflict events in the evaluated test episodes while minimizing per-slice latencies. Moreover, the proposed framework exhibits lower signaling overhead and reduces effective inference latency by 88 % under the considered communication-delay model compared to the online baselines. Source codes and datasets are available through: https://github.com/Eslam211/xcode-ran-slicing.

Reasoning segmentation requires multimodal large language models (MLLMs) to translate implicit instructions into precise pixel-level masks. MLLMs encode an image as visual tokens, each of which merges a group of image patches. In remote sensing images, small targets, thin structures, and adjacent instances can occupy different parts of the same visual token. Assigning a single binary mask label to such a token loses its internal spatial structure, causing nearby targets to merge and object boundaries to become coarse. To bridge this representational gap, we introduce FIRM, a Fine-grained Intra-token Representation of Masks. For each visual token, FIRM predicts a mask code that specifies an $r\times r$ binary sub-cell pattern rather than a single foreground/background label. Given a target identified by the MLLM, the complete grid of mask codes is predicted in one mask pass. Fixed lookup converts the predicted codes into a discrete sub-cell mask, while marginalizing the code distribution yields a soft structural field. To further recover fine-grained boundaries within each sub-cell, we introduce a lightweight continuous renderer that refines this field using pre-merge visual features and image details. Across five reasoning and referring segmentation benchmarks on satellite and UAV images, FIRM achieves leading results, including $70.5/80.5$ gIoU/cIoU on LaSeRS and a $3.0$-point average gain on EarthReason. These results demonstrate the value of explicitly representing intra-token mask patterns for fine-grained MLLM segmentation.

As large language model inference shifts toward lower precision, post-training quantization (PTQ) becomes increasingly brittle, making quantization-aware training (QAT) essential for preserving model quality. However, QAT computes the loss and surrogate gradients using a lossy reconstruction of latent full-precision weights, while applying updates to the latent weights themselves. This mismatch can lead to suboptimal training trajectories and a higher loss floor. Second-order PTQ methods mitigate a similar gap by minimizing loss-aware reconstruction error, but doing it once for a frozen model can take hours; repeating this process throughout QAT as the weights evolve is impractical. We introduce QUASAR, a QAT method that continuously performs lightweight, loss-aware reconstruction in the training loop to lower the loss floor and improve the resulting low-bit model. At each training step, QUASAR uses the exponential moving average of squared gradients as online saliency estimates, searches over a small set of clipping ranges, and fits affine dequantizers via saliency-weighted least squares. Our analysis shows that the loss-aware reconstruction error is the only reconstruction-dependent term in the QAT convergence bound and controls the loss of the final quantized model, establishing QUASAR's objective as a principled optimization target. QUASAR modifies only the training procedure and supports standard deployment formats, including integer quantization and NVFP4, with no inference-time changes or overhead. Across Qwen3 and Llama-3.1, QUASAR achieves the lowest held-out KL divergence among competitive QAT methods at 2, 3, and 4 bits, reducing KL by at least 10% at 3 and 4 bits and by 29% at 2 bits. At 2 bits, it improves average accuracy across eight tasks by 3.5-4.3 percentage points over strong QAT and PTQ baselines.

Federated learning (FL) enables collaborative model training across distributed devices without sharing raw data; however, it faces significant communication bottlenecks and channel impairments in practice. Conventional network layer treatments either idealize the channel as error free or apply equal error protection (EEP) to transmitted model updates, failing to account for the inherently unequal importance of quantization bits within a single local model. To address this limitation, we propose a cross layer polar code based FL scheme that leverages the unequal error protection (UEP) property of polar codes under finite block lengths. Specifically, the proposed design selectively protects more significant quantization bits, thereby mitigating the detrimental effects of channel noise. We further provide a rigorous convergence analysis of the proposed scheme, deriving an upper bound on the convergence gap, which we then jointly optimize over the number of quantization bits and the polar code block length across all training iterations. Experimental results demonstrate that both constant and variable block length configurations of our polar code based scheme consistently achieve substantial performance gains over uncoded and LDPC-based EEP benchmarks, with the advantage becoming increasingly pronounced as the channel quality deteriorating. These findings confirm the efficacy of our cross-layer design in enhancing FL robustness and efficiency under realistic channel conditions.

High-quality creative writing data for large language models (LLMs) remains dominated by story-centric data, limiting models' ability to follow the structural and functional conventions of diverse creative formats. We propose an attribute-guided genre expansion framework for scaling creative writing data beyond story generation. By separating thematic breadth from genre-form control, our framework leverages human-authored story prompts as diverse creative seeds, while utilizing manually curated genre attributes to enforce distinct structural, stylistic, and formatting conventions. We combine these to prompt strong LLMs for genre-faithful query-response pairs, which are then quality-filtered. Applying this framework, we construct the Multi-Genre Collection, a 50K-example corpus spanning 13 creative genres, including story, rap, lyrics, scripts, game design, character design, and other creative formats. Experiments across out-of-distribution writing benchmarks and held-out genre diagnostics demonstrate that models fine-tuned on our data consistently surpass not only base models and writing-specialized baselines, but also models trained on existing writing corpora. Genre-count ablations further indicate that controlled genre expansion, rather than story-centric scaling alone, is a key driver of robust creative writing capability.

Neural representations for videos (NeRV) have shown strong reconstruction fidelity by storing video-specific information in network weights. However, existing formulations typically require either costly per-video optimization or video-specific weight generation, making it difficult to scale to efficient amortized video representation. We propose CoANeRV, a coordinate-aware token-space framework that adapts the broader token-conditioned neural-field paradigm to amortized video representation. CoANeRV forms compact video tokens in one feed-forward pass and uses a shared coordinate-conditioned decoder to reconstruct continuous spatio-temporal queries, avoiding per-video decoder optimization or generation while retaining coordinate-level reconstruction flexibility. To make token-space reconstruction effective, CoANeRV introduces a coordinate-aware decoding architecture that aligns spatio-temporal queries with video tokens through axis-adaptive positional encoding and temperature-modulated cross-attention. Block-wise coordinate querying further reduces peak attention memory, making high-resolution reconstruction practical. Experiments on diverse video datasets show that CoANeRV consistently improves reconstruction quality over prior feed-forward NeRV and INR baselines, reduces peak memory compared with attention-based coordinate decoders, and provides efficient amortized encoding without per-video optimization. These results support the proposed video-specific combination of feed-forward token formation, spatio-temporal coordinate retrieval, and memory-bounded dense querying. The code is available at https://github.com/jialong2023/CoANeRV.

Iterative Generative Models (IGMs) span autoregressive and diffusion paradigms, and hybrid variants that couple them can achieve remarkable image-generation fidelity. However, their iterative inference incurs substantial computational overhead, making Post-training Quantization (PTQ) appealing for acceleration, while directly applying vanilla PTQ to hybrid IGMs can trigger model collapse. By analyzing these failures, we identify two critical challenges: Excessive Outliers (EOs) in the activations create an irreconcilable trade-off between preserving normal precision and covering EOs, resulting in severe degradation in generation quality; Amplified Anomalies (AAs) arising unpredictably from minor quantization errors, create a mismatch between calibration and inference, thus iteratively triggering model collapse. To address these challenges, we introduce HyGenQ, a PTQ framework for hybrid IGMs. HyGenQ comprises Hierarchical Cluster Decoupling (HCD) and Scaling Recalibration (SR). HCD identifies and decouples outlier channels via a multi-stage clustering process, effectively isolating EOs while maintaining normal value precision, thereby alleviating performance degradation. SR scales AAs beyond Gaussian Bound, thereby avoiding model collapse caused by aggressive truncation. Extensive experiments demonstrate that HyGenQ successfully quantizes representative hybrid IGMs to 8-bit precision (W8A8), significantly outperforming existing baselines and validating its robustness across different model families.

Large language models have made natural language interfaces to databases (NLIDB) newly credible, but LLM text-to-SQL systems fail in a way that matters for deployment: a hallucinated column or a mis-aggregated total yields a fluent wrong answer, indistinguishable at the point of use from a right one. Where the consumer cannot inspect the generated query, as in enterprise AI deployments and operational dashboards, and increasingly where the consumer is a tool-using agent rather than a person, accuracy alone is insufficient: nothing marks which answers to distrust. This is a reliability problem before it is an accuracy problem. We propose an architectural pattern for such systems, a trusted kernel with a generative shell, resting on one invariant: a component that can fabricate may influence which question the system answers, never which value it returns. A generative shell interprets underspecified input and phrases replies; a deterministic kernel matches fully specified questions against a bounded set of answerable question shapes and compiles them to queries by deterministic execution. The two meet at a confirmation the user reads before any value is computed, and requests the kernel cannot express are declined rather than approximated. We call this structural abstention, and distinguish it from the statistical abstention of selective prediction and calibrated confidence: refusal here needs no confidence estimate, because unanswerable requests are unrepresentable. We specify the pattern implementation-independently, give a five-decision recipe and work it across three domains, extend the invariant from returned values to the actions of agentic systems, and report a two-year production case study alongside two generative alternatives, a fine-tuned parser and a tool-retrieval agent. We close against enterprise and reliability benchmarks published since.

The request-to-handoff gap has three coupled sources: ambiguity about the behavior intended at request time, physical-state drift accumulated during action generation, and residual incompatibility when the new action finally assumes control. BICPO-VLA addresses them in sequence. First, an instruction-aware causal history encoder identifies the behavior supported by the command and current task progress. Second, sequential Haar subspace generation decomposes each action chunk into complementary pairwise scaffold and residual coefficients, enabling two specialized generation stages followed by exact reconstruction. By reducing iterative refinement in the original action space, it shortens the interval over which the robot continues moving before the new chunk becomes available. Finally, BICPO rolls the known outgoing actions to the actual handoff state and applies reference-relative Flow-DPO among behaviorally matched candidates, adapting the generated chunk to the remaining request-to-handoff mismatch without changing its intended behavior.

Open-vocabulary 3D scene graphs provide compact semantic memory for language-guided navigation, but mapped objects are often exposed through a single fused feature or committed semantic label. Such commitment can remove minority yet task-relevant hypotheses from the task-time interface. We present OpenBelief-Nav, an evidence-preserving object memory that retains observation-level phrases, reliability cues, and frame-mask provenance while maintaining separate aggregate geometric and visual representations. Semantically related phrases are consolidated into a vocabulary-independent object belief from which task-specific readouts perform fixed-vocabulary projection or free-form retrieval. On five ScanNet200 and eight Replica scenes, full-belief projection achieves mIoU scores of 0.2742 and 0.2912, compared with 0.2393 and 0.2701 for a matched early-commit readout. Across 78 HM3D-YCB navigation trials, consensus and early-commit retrieval each achieve 60/78 successes, compared with 58/78 for belief-weighted retrieval and 55/78 for DualMap. Across 20 Unitree G1 runs organized as 10 matched evaluation cases, a correction policy permitting at most two verified candidate attempts improves target-confirmation success from 6/10 to 8/10 relative to top-1-only execution. Code will be released upon acceptance at https://openbelief-nav.github.io/.

Detecting distributional changes in high dimension is difficult when neither the pre-change nor post-change density is parametrically specified. We introduce a representation-based approach that retains all degree-at-most-two density information while replacing density estimation by matrix mean estimation. For observations in $[-1,1]^d$, a symmetric feature matrix $H_2(X)\in\R^{(d+1)\times(d+1)}$ is constructed so that $M(f)=\E_f H_2(X)$ is an isometric encoding of the degree-two orthogonal projection of the density. We scan matrix CUSUMs after rank-$r$ truncation, exploiting the low rank of the projected jump rather than sparsity of individual coordinates. The resulting \LRD{} estimator has a tent-shaped population objective and a nonasymptotic operator-norm analysis whose leading stochastic term scales as $\sqrt{rd\log(nd)}$. For multiple changes, we give a seeded narrowest-over-threshold procedure and prove exact recovery by an induction that preserves an isolating interval for every undetected change. A cross-fitted scalar refinement learns the changing low-rank direction on one fold and localizes on the other, attaining $\widetilde O_{\Pp}(κ^{-2})$ error; a matching Le Cam lower bound shows optimality up to logarithms. A geometrically $β$-mixing extension follows from a dependent matrix Bernstein inequality. Experiments with ambient dimension up to $200$, a three-change $d=100$ sequence, and a $128$-feature human-activity benchmark show that the method remains computationally practical and accurately detects pure dependence changes that are invisible to mean CUSUMs.

We formulate first-order optimization as a minimum-time control problem. The iterate is the state, the update, a combination of the gradients observed so far, is the control, and the points with gradient norm at most a prescribed tolerance form the target set. For a fixed objective and start, the minimum number of oracle queries needed to reach the target is a value function: it measures the complexity of the instance, not the worst case over a class. On a strongly convex quadratic, the conjugate gradient iterates emerge from the discrete Pontryagin conditions, and the value is a controllability index. Beyond the quadratic, a Hessian-generated reachable span replaces the controllability matrix, and feasibility becomes a reachability question: whether a critical point can be reached at all, and in how many steps, is read from the span. Curvature is a resource: a checkable condition certifies that an instance needs fewer steps than the controllability index of its quadratic model at the minimizer, and the gap can grow without bound with the dimension. The value is thus a benchmark for the intrinsic difficulty of each instance, against which any first-order method can be measured.

EV-WM represents candidate quality with feature and event scores, but these scores do not explicitly record an unmet task predicate, a route label for an available correction mechanism, or a post-correction acceptance result. We present Onto-EV-WM, an ontology-grounded diagnosis and verification-gated correction interface layered above EV-WM rather than a replacement world-model architecture. The implemented task-local TBox defines entity types, predicate signatures, and constraints; source-specific grounding maps predicted or simulator-observed states to task ABoxes; and deterministic rules retain each missing predicate and its arguments when assigning a route label. Learned or heuristic proposers remain separate from this symbolic interface; native task predicates determine acceptance, and the bounded protocol determines whether a failed verification is retried. In the aligned PointMaze evaluation, EV-WM and Onto-EV-WM both report 94% success, with mean final-state distances of 0.90573 and 0.61177, respectively; the separately budgeted search reaches 100% success. On LIBERO-Goal, the ontology represents failed task conditions as typed records, retains their predicate arguments, and associates them with the declared source/joint correction route and predicate-gated acceptance; the complete configuration reports 93.8% corrected-window success on seed 0 and 94.05 +- 0.30% across four evaluation-sampling seeds. On the fixed 10,030-task LIBERO-Plus registry, Onto-EV-WM succeeds on 8,526 tasks (85.00%), with suite-level success rates of 65.98% for LIBERO-10, 91.39% for LIBERO-Goal, and 91.38% for both LIBERO-Object and LIBERO-Spatial. These numbers report the performance of the complete ontology-grounded configurations under the tested simulator protocols; an ontology-only causal share is not measured separately, and real-robot recovery is not evaluated.

Large language model (LLM) agents are evolving from conversational assistants into autonomous systems that execute long-horizon tasks through reasoning, tool use, code generation, and workspace manipulation. As agents increasingly operate over persistent environments and multi-step workflows, they face challenges analogous to those addressed by transactional database systems: reliable execution, consistent outcomes, safe concurrency, and durable state management. We introduce the concept of an agentic transaction and propose an ACID-compliant agent system framework that reinterprets the classical ACID properties for agent execution through four semantic guarantees: Semantic Atomicity, Semantic Consistency, Semantic Isolation, and Semantic Durability. Together, these properties provide a principled foundation for building reliable agent systems despite model uncertainty and dynamic execution environments. To instantiate this framework, we develop an ACID-compliant data agent that realizes these guarantees through transactional exploration-execution-validation cycles, transactional skill hubs, confidence divergence-based validation, semantic dependency-aware isolation, and transaction-aware semantic state management. Experimental results on widely used benchmarks show that our system achieves a 10.6% improvement over state-of-the-art agents, including Claude Code. This work opens a broader research agenda on extending transactional principles and system architectures toward building trustworthy, scalable, and self-evolving AI agent systems.

Neural networks for seismic fault segmentation are often borrowed from computer vision and medical imaging domains where they train under relatively much larger labeled data resources. Optimizing their architecture under tight labeled data budgets as are common in geophysical applications is not a trivial problem. Manually designing data-optimal architectures is time-consuming while classical neural architecture search (NAS) is restricted to hand-crafted search spaces and large compute budgets. We present an agentic NAS system in which a panel of three large language models (Claude, GPT-5.1, and Gemini~2.5~Pro) debates each candidate architecture to unanimous consensus, authors the complete PyTorch implementation, cross-reviews it, and submits it to an automated validate-train-score loop with a hard 450K parameter budget, keep-or-revert lineage, and a memory of failed mechanisms. Operating on source code rather than a predefined operation menu, the search ran on a single consumer GPU and trained only eight candidates. It discovered \ours{}: a 425K-parameter encoder-decoder with a strip-pooling bottleneck, squeeze-and-excitation gating, an asymmetric one-conv decoder, and a feature-pyramid fusion neck. Trained under a protocol identical to all baselines on sections derived from the Thebe fault dataset, it attains the highest F1 (0.578) and IoU of all models tested while being the smallest, outperforming a published-capacity U-Net (31M parameters, F1 0.484), DeepLabV3-ResNet50 (39.6M, 0.516), an Attention U-Net(1.83M, 0.502). The search cost 101 LLM calls ($\sim$1.15M input / 0.39M output tokens) and roughly one GPU-day, making consensus-gated LLM panels a practical, low-cost route to domain-specific architecture discovery.

The rapid adoption of AI coding assistants and autonomous agentic development systems has coincided with major changes in the pace and structure of open-source software engineering. Yet empirical longitudinal evidence of these changes at the team level remains limited. We present a descriptive longitudinal analysis of seven engineering metrics: pull request (PR) throughput, cycle time, contributor diversity, PR comment density, merge rate, new-author participation, and PR size. Metrics were computed from all merged PRs in two high-velocity AI infrastructure repositories, vLLM (February 2023-June 2026; 18,290 PRs) and SGLang (January 2024-June 2026; 14,938 PRs). We segment development into four eras aligned with major changes in AI-assisted software development and examine human- and bot-authored activities. Both projects show substantial increases in development velocity and AI-developer collaboration signals. PR throughput increased 21x in vLLM and 17.9x in SGLang, while bot-authored PRs accounted for less than 0.2% of this growth, indicating that the increase was overwhelmingly human-driven. In the latest era, median cycle time was 1.04 days for vLLM and 0.62 days for SGLang, while P90 cycle times reached 16.8 and 14.3 days, respectively. Monthly unique authors increased steadily in both projects, suggesting broader contributor participation. PR comment density increased 4.2x in vLLM and 3.8x in SGLang, with bot comments contributing an estimated 15-20% of the increase. In contrast, PR size remained relatively stable across eras. Overall, AI-assisted development is associated with higher throughput, broader contributor participation, and increased AI-developer collaboration signals in high-velocity open-source software development.

Code translation must preserve executable behavior across many programming languages, yet neural code translation has largely focused on a few popular languages such as C++, Java, and Python. This leaves a niche, many-to-many setting where parallel supervision is sparse, producing plausible but non-executable translations. We address this setting with preference-based reinforcement learning driven by execution-based supervision. Our pipeline firstly expands verifiable seed Python programs into a multilingual pool of execution-validated codes. Using the pool, a base LLM generates translation candidates across language pairs, which we label by their execution outcomes. The resulting preferences are used to train a reward model that scores cross-language translation quality. Finally, we optimize our base LLMs with GRPO over 600 directed language pairs (25 x 24) using the reward model as a signal. To evaluate the niche translation capability, we introduce HumanEval-X++, an execution-based benchmark that extends HumanEval-X to a broad many-to-many language space. We evaluate our approach using Qwen-3.5 4B and 9B models. On HumanEval-X++ and existing benchmarks, it yields consistent gains over the untrained baselines. In particular, the 4B model achieves an average improvement of 13% across all languages on HumanEval-X++, with a gain of 21% on mid-tier languages. Our study establishes a reliable approach of data generation, training, and benchmarking, paving the way toward further bootstrapping the quality of many-to-many translation for programming languages.

Machine-learned physical surrogate models have become promising alternatives to mesh-based numerical solvers. Among them, graph neural networks (GNNs) are well suited for representing simulation meshes and learning nodal state evolution through message passing. However, conventional flat message passing becomes inefficient on large, high-fidelity meshes because information propagates only one hop per layer, requiring deep processors for long-range interactions and increasing computational cost, memory usage, and the risk of over-smoothing. To address this limitation, we propose Hierarchical Interpolating MeshGraphNets (HI-MGN), a multiscale extension of MeshGraphNets for efficient long-range communication on unstructured meshes. HI-MGN replaces the flat processor with a hierarchical multiscale processor that coarsens graphs using farthest-point sampling and Voronoi partitioning while preserving the original mesh topology. Message passing on coarse graphs enables information to travel over larger geometric distances with fewer layers, and a learned graph interpolation network reconstructs fine-resolution features. Across three structural and fluid benchmarks, HI-MGN achieves improved accuracy compared with MeshGraphNets and the Bi-Stride Multi-Scale GNN while reducing training time and peak memory usage. The results show that topology-aware hierarchical message passing and learned coarse-to-fine interpolation provide an effective and practical framework for scalable mesh-based physics surrogate modeling.

Machine learning is increasingly used to learn structure property relationships from spectroscopic and diffraction data, yet its adoption in materials discovery is often limited by poor interpretability of model predictions. Although attention mechanisms are frequently treated as inherently explainable, unregularized attention can yield unstable, fragmented, or intensity driven attribution patterns that obscure the physical origin of these relationships. Here we introduce SPEAR (Structure Property Explainability with Attention Regularization), a framework that constrains attention distributions during training to improve their stability, selectivity, and physical interpretability. SPEAR augments attention based regression with a learnable temperature that controls attention concentration and a smoothness penalty that enforces coherence across neighboring spectral positions, treating attention as a learnable explanatory object rather than a post hoc visualization. Using synthetic spectral benchmarks with known generative structure, we show that attention regularization produces smooth, contiguous attribution profiles aligned with causal features while preserving predictive accuracy. Applied to experimental X ray diffraction data from a combinatorial rare earth zirconate thin film library, the regularized model selectively emphasizes physically relevant diffraction features and decouples feature importance from raw peak intensity. The reflection it identified prompted a reassessment of our earlier structural analysis, revealing a correlation between the 220 peak position, the tetragonal distortion that accommodates cation size disorder, and the local thermal conductivity. Attention regularization therefore provides a principled training constraint for explainable structure property regression, yielding mechanistically meaningful explanations without sacrificing predictive performance.

Human speech production is constrained by physiology, giving rise to characteristic temporal structure on acoustic signals. We hypothesise that these constraints manifest as structured trajectory dynamics in the latent space of Self-Supervised Learning (SSL) models, and that synthetic speech violates them detectably. To test this hypothesis, we train a causal Long Short-Term Memory (LSTM) next-frame predictor on bonafide speech only (Stage 1), using the deepfake-specialised SSL backbone Wav2Vec2-Large-AntiDeepfake, and compare against a static global-average-pooling baseline using identical features, thus isolating the contribution of temporal modelling. A supervised Stage 2, which trains a Multi-Layer Perceptron on the frozen LSTM internal states using labelled data, is included to characterise the role of spoof supervision. Our system achieves competitive or state-of-the-art performance across six benchmarks: ASVspoof 2019/2021, Codecfake, In-the-Wild, MLAAD-EN, and Deepfake-Eval-2024, including best published EER on ASVspoof 2021 (0.75\%) and, notably, Stage 1 trained on bonafide speech only surpasses the published supervised baseline from the same backbone on DE2024 (30.35\%). On near-domain benchmarks, static and dynamic approaches perform comparably. On harder cross-corpus benchmarks with diverse synthesis methods, trajectory dynamics provide substantial gains, confirming that temporal physiological constraints carry detection signal beyond utterance-level statistics.

Transient thermal grating (TTG) measurements probe thermal transport on submicrometer spatial and subnanosecond temporal scales, where departures from purely diffusive behavior have been reported in a variety of solid-state systems. Within the Mori-Zwanzig projection framework, such effects can be interpreted as manifestations of memory arising from microscopic degrees of freedom that remain unresolved at the observational scale. Motivated by this perspective, we formulate a forward model for TTG response in which heat transport is described by a linear evolution operator with memory. Microscopic transport processes are systematically coarse-grained into a non-Markovian transport operator, whose finite-memory form provides a general description of TTG dynamics, while diffusive and long-memory behaviors emerge as asymptotic limits of the same underlying framework. Within this formulation, the experimentally observed TTG transient is interpreted as the response of an effective transport operator rather than as a direct signature of a particular microscopic transport mechanism. The primary outcome of the inversion procedure is the determination of effective operator parameters, while microscopic transport mechanisms are inferred only through a subsequent, material-specific modeling step. The proposed Mori-Zwanzig inspired framework therefore establishes a systematic route from TTG measurements to effective transport operators and, subsequently, to coarse-grained descriptions of microscopic transport dynamics.

We examine the interplay between ordinal, preference-based solution concepts in games and the long-run behavior of game dynamics, asking in particular to what extent the combinatorial data of a game -- its preference graph -- determine the outcomes of no-regret learning dynamics -- such as follow-the-regularized-leader (FTRL). In one direction, we show that the skeleton of every dynamically stable set (i.e. the set of pure profiles it contains) must also be preferentially stable, that is, it must be closed under profitable deviations. We then ask the converse question: when do preferences determine the long-run behavior of the players' learning dynamics? We begin by showing that preferences characterize asymptotic stability in the case of subgames -- i.e. subsets of pure profiles obtained by restricting players' action sets. Beyond this case however, the equivalence between dynamic and preferential stability collapses: concretely, we construct a three-player game with a preferentially stable set whose span is dynamically unstable, showing in this way that preferences do not suffice as a criterion of dynamic stability. We then bridge this gap via the notion of resilience under aggregate deviations, an easy-to-check payoff-based condition that guarantees asymptotic stability of arbitrary spans of pure strategies.

This paper presents an event-driven learning and benchmarking framework for the Dynamic Multi-Depot Vehicle Routing Problem with progressively revealed requests and evolving vehicle states. Masked MLP and Transformer policies are trained through behavior cloning and proximal policy optimization. Deterministic feasibility masking prevents invalid vehicle--request assignments, while fixed-prefix/flexible-suffix route commitments protect completed, active, and near-term decisions and separately measure vehicle reassignment and resequencing. The learned policies are compared with dynamic insertion heuristics and time-limited rolling-horizon optimization. In a 20-scenario policy benchmark, all methods completed every request without invalid actions, but nearest feasible achieved the lowest mean objective and outperformed the learned policies in routing quality, waiting time, stability, makespan, and runtime. Across five independent training runs, PPO had little average effect on the MLP and improved the Transformer on average, although with greater seed variability. Under the common protocol, nearest feasible achieved the lowest combined objective and route disruption, whereas rolling horizon achieved the lowest waiting times and makespan at substantially higher computational cost. The learned policies retained millisecond-level decisions and transferred to instances with up to 80 requests without retraining, but did not outperform the strongest heuristic. No single method was best across routing efficiency, service responsiveness, stability, and online computation.

Computational approaches to drug discovery involve multiple sub-problems, and among them, drug-target binding affinity prediction plays an important role. Despite recent advances, accurately predicting binding affinity remains an open research area. The major objective of our paper is to perform a comprehensive review and comparative analysis of recent machine learning methods for drug-target binding affinity prediction, with a focus on identifying strengths, limitations, and research gaps. We review representative recent deep learning approaches that use common benchmark datasets and evaluation metrics, covering a range of neural network architectures and representation strategies. In addition, we analyze seven widely used benchmark datasets and commonly adopted evaluation metrics for drug-target binding affinity prediction. Our analysis indicates that although many methods report strong performance on standard benchmarks, their effectiveness is often influenced by dataset bias and limited evaluation settings. Furthermore, most methods exhibit reduced performance in cold-start scenarios, highlighting challenges in generalization. We identify several limitations of current approaches, including dataset imbalance, the lack of standardized evaluation, limited real-world applicability, and challenges in cold-start scenarios. We also discuss future research directions, including better dataset design, more robust evaluation methods, improved handling of cold-start problems, and the integration of multimodal representations.

Positron emission tomography (PET) imaging suffers from limited spatial resolution and low signal-to-noise ratio, which can compromise quantitative accuracy and lesion detectability. Deep learning-based denoising methods have demonstrated strong potential for improving PET image quality. However, their practical deployment in real-world settings remains challenging, often requiring multiple specialized models and expert interventions, such as identifying motion-induced misregistration artifacts, estimating noise levels to select an appropriate denoiser, and performing lesion-focused quantitative assessment after denoising. Recent advances in vision-language models (VLMs) for image quality understanding and large language models (LLMs) for contextual reasoning provide new opportunities for automated, decision-driven workflows. Inspired by expert workflows for PET image quality enhancement, we propose an VLM- and LLM-driven multi-agent PET denoising framework that dynamically assesses image quality and lesion status, autonomously selects optimal denoising models and parameters, and enables closed-loop feedback with rollback mechanisms. Experiments were conducted on Siemens Biograph Vision Quadra PET/CT data with 1/20 and 1/50 low-dose settings. Individual module evaluations demonstrated the reliability of the agentic components, while the complete framework achieved higher PSNR and SSIM than UNet, GAN, and DDPM baselines at both dose levels. These preliminary results demonstrate the feasibility of using a closed-loop multi-agent framework to adapt PET denoising strategies to different image conditions.

Macro placement significantly affects a chip's post-route performance, power, and area (PPA). Most placement methods optimize half-perimeter wirelength (HPWL) as the primary objective. However, recent benchmarking shows a near-zero correlation between HPWL and post-route timing metrics such as the worst negative slack (WNS) and total negative slack (TNS). As a result, all six evaluated artificial intelligence (AI) placers degraded PPA relative to the hierarchical baseline. Recent efforts have tried to train cross-stage predictors to close this gap. However, existing methods focus on macro-only representations and use pre-route metrics as training labels. A label fidelity study of ten circuits at four design flow stages reveals that HPWL and pre-route timing poorly reflect final post-route timing rankings. In contrast, post-global-routing achieves the best balance between final timing fidelity and label generation cost-effectiveness. Based on this finding, PPAPlace is a timing-driven differentiable surrogate predicting post-route PPA from macro and standard-cell placements. The surrogate is a dual-stream predictor that combines graph attention over the chip netlist with spatial convolution over the placement grid. It is trained on post-global-routing labels. The predicted WNS and TNS gradients flow end-to-end back to cell coordinates. PPAPlace exploits these gradients in two ways: as a co-objective injected into an analytical placer's optimization loop (PPAPlace-CoOpt), and as a post-placement refinement step that adjusts macro positions via projected gradient descent (PPAPlace-Refine). On five ChiPBench test circuits excluded from training, PPAPlace improves average WNS and TNS by 22\% and 51\% over the hierarchical baseline while preserving power and routability, using the same predictor without test-circuit retraining. Code is available at https://github.com/ValleyC/PPAPlace.

AI agents increasingly act on their users' behalf, handling tasks such as scheduling meetings, comparing offers, and haggling over prices. These principal-driven tasks routinely place the agent across from a counterpart (another user's agent, a seller, a recruiter) whose goals may conflict with its principal's. Yet the dispositions that make an assistant pleasant can make it a poor delegate: a friendly, helpful frontier model may disclose its principal's private information unprompted and concede at the first sign of resistance. We present SocialRL, a general recipe that trains social reasoning directly, and apply it to a 4B model across six domains: Deal-or-No-Deal, CaSiNo, Craigslist, Job Interview, Calendar, and Marketplace. Every domain is trained in-domain under the same recipe, and every policy is evaluated on all six. We find that (1) in-domain training reaches the frontier: on held-out scenarios the 4B matches or exceeds the GPT-5 family per domain, closing 73-122% of the baseline-to-frontier gap on the negotiation games, with 78% of buyer openings anchoring below target versus 3% untrained; (2) cross-domain transfer follows game structure: structurally paired games lift each other, a broad multi-issue donor lifts nearly all domains, and structurally isolated games transfer nothing; (3) guided by this transfer structure, two strategies, cascade RL and multi-teacher on-policy distillation (OPD), consolidate the per-domain specialists into a single unified 4B that reaches 0.627 average utility across all six environments, matching or exceeding GPT-4.1 (0.625), GPT-5.1 (0.619), and GPT-5.2 (0.613); (4) an explicit theory-of-mind scaffold helps only through training: distilling the ToM trace, rather than actions alone, lifts utility on every environment and generalizes better across them, and of the two ToM skills, only next-action prediction predicts negotiation outcomes.

Large language model (LLM) chatbots are increasingly used to answer clinical questions with citations to relevant clinical studies. Prior research has largely focused on citation fabrication, leaving a gap in evaluating the quality of retrieved studies and the factors driving their selection. In this study, we evaluated three general-purpose LLM chatbots: Claude Sonnet 5, Gemini 3.1 Pro, and ChatGPT GPT-5.5. We prompted the models with clinical questions adapted from 20 review questions in Issues 6 and 7 of the 2026 Cochrane Database of Systematic Reviews, simulating patient, clinician, and evidence-synthesis researcher roles. Each chatbot was queried under each user role with four independent repetitions, yielding 720 responses. Each chatbot was asked to support its answers with primary clinical citations, which we benchmarked against the included and excluded study sets of the Cochrane reviews. On average, a chatbot response retrieved 39.2% $\pm$ 29.8% of Cochrane included studies, while citing 5.0% $\pm$ 9.4% of excluded studies. Recall of Cochrane included studies varied significantly by model and user role. ChatGPT achieved higher recall than Claude or Gemini (63.1% $\pm$ 29.5% vs. 37.0% $\pm$ 23.8% vs. 17.3% $\pm$ 13.1%; $p=2.0\times10^{-5}$). The researcher role yielded higher recall than the clinician or patient roles (42.8% $\pm$ 30.8% vs. 38.6% $\pm$ 28.9% vs. 36.1% $\pm$ 29.3%; $p=2.0\times10^{-5}$). Controlling for publication year, citations per year, and open-access status, sample size was the only independently significant predictor of retrieval (odds ratio 1.80 per 1-unit increase in log sample size, 95% CI 1.37-2.36, $p=2.34\times10^{-5}$). These findings suggest that while LLM chatbots can retrieve some studies identified by expert reviewers, their performance varies by model and user role, and they exhibit a bias toward clinical trials with larger sample sizes.

Top-1 accuracy on ImageNet-1k remains the most commonly reported metric in visual recognition. Quality issues with the dataset have been repeatedly reported, yet the original 2012 noisy labels are still predominantly used. The paper presents a comprehensive effort, which goes well beyond prior correction attempts, towards obtaining accurate and complete ImageNet-1k validation set annotations. The result, ReImageNet, includes multilabel correction, object localization, revised class definitions, and semantic attributes (text-recognition, rendition, reflection, crowd, dominant). The reannotation reveals that approximately 12% of the original ImageNet-1k labels are incorrect, 33.3% of images are multilabel and 3.8% contain no object from an ImageNet-1k class. With the new labels, top-1 accuracy increases by up to 1.2% for supervised models and by 5-6% for MLLMs. We argue that annotation at ImageNet scale cannot realistically be completed in one pass, as errors and definitional issues are discovered only through annotating, and we build our pipeline around repeated refinement and error checking. We observed that human and LLM collaboration with appropriate tooling represents the current quality ceiling for annotation at this scale. ImageNet-1k issues propagate into its derivative test sets, indicating that the problem is structural rather than specific to any single benchmark. All annotations, class definitions, guidelines, and analysis code have been publicly released. Project page: https://vrg.fel.cvut.cz/reimagenet Annotations: https://huggingface.co/datasets/vrg-prague/ReImageNet Code: https://github.com/klarajanouskova/ImageNet

Analog IC layout design remains a labor-intensive iterative process dominated by simulation-driven refinement. Although end-to-end layout generators accelerate initial placement and routing, they still require experts to manually tune layout optimization parameters with repeated post-layout simulations for stringent design specifications. While Bayesian Optimization (BO) is widely adopted for parameter tuning in analog IC design, at the layout level it typically requires hundreds to thousands of evaluations, each involving costly parasitic extraction and post-layout simulation, which makes it impractical. Recently, Large Language Models (LLMs) have demonstrated potential in improving the sample efficiency of such simulation-driven tuning. However, their restricted access to geometric layout context and design-specific heuristics limits their ability to manipulate the layout optimization process. In this paper, we propose a simulation-aware LLM multi-agent framework that performs in-context policy improvement (ICPI) by iteratively updating layout optimization parameters exposed by an analog layout generator through an act-observe-reflect loop on compact structured layout representations. Experiments on real-world analog circuits show that, with only tens of post-layout simulations, our approach improves post-layout performance over the generator's built-in heuristics and BO-based tuning method.

Vision-language models (VLMs) remain unreliable on chart questions that require reasoning over visual quantities, and this weakness is usually attributed to a reasoning deficit and addressed with more reasoning supervision. We ask whether the difficulty lies in reasoning itself, or in the simpler skills that reasoning operates on: reading the plotted elements (\emph{perception}), locating them and binding them to their labels (\emph{grounding}), and performing single-step computations such as ranking, totals, and differences (\emph{simple reasoning}). We introduce \textbf{ChartProbe}, a diagnostic framework whose probes are generated directly from the code that renders each chart, so every gold answer is exact by construction, needs no human annotation, and attributes each failure to a single skill. ChartProbe enables an intervention prior work does not attempt: instead of synthesizing complex-reasoning data, we withhold complex questions and reasoning traces entirely, fine-tune on one simple skill at a time, and measure transfer to held-out complex-reasoning questions. Across three open-weight VLMs, supervising the simpler skills alone produces large gains on complex-reasoning questions the model never trained on: where these skills are weak and the model can be taught to read the image, training them recovers much of complex reasoning at no reasoning-data cost. The gains hold across three out-of-distribution settings: an unseen chart type (pie charts), a human-written benchmark disjoint from our images and templates (ChartQA), and a non-chart visual domain (CLEVR). Complex visual reasoning can therefore improve without complex-reasoning supervision.

Which reasoning behaviors are associated with correct answers in reasoning models, and does reasoning-oriented training amplify those behaviors? This distinction is important because reasoning-oriented training can make traces look more deliberative without amplifying the behaviors most tied to model correctness. We quantify this mismatch with Behavioral Lift, a metric that measures how much correctness changes when a behavior is present versus absent in a model's reasoning trace. Across 15 models and 6 benchmarks spanning text-only and vision-language reasoning, we annotate 15,282 traces with a taxonomy whose core behaviors are defined for both LLM and VLM traces. We find evidence for an Amplification-Lift Gap, in which thinking models strongly amplify self-correction, hypothesis testing, and uncertainty acknowledgment, while the highest-lift behaviors are confidence calibration, knowledge alignment, and self-awareness. Confidence calibration is among the strongest positive signals of correctness in both modalities, yet is barely amplified; uncertainty acknowledgment is amplified by 3--7$\times$, yet is weakly or negatively associated with correctness. We find that reasoning-oriented training does not preferentially amplify the highest-Lift behaviors, motivating process-level objectives that reward calibrated and grounded reasoning rather than surface form alone.

Two GPU kernels implementing the same scaled INT8 GEMM interface are usually treated as interchangeable. We test that assumption: holding the checkpoint, prompts, hardware, inference engine, decoding, and quantization configuration fixed, we swap only the INT8 linear kernel (CUTLASS versus Triton) inside vLLM. At 1.7B each arm reproduces itself bit-for-bit across cold restarts, yet the arms agree on no sequence in any end-to-end comparison we ran (0/8, 0/16, and 0/64). What makes this more than a benchmark discrepancy is an integer alibi: for shared INT8 operands under a verified no-overflow bound, the INT32 dot product is exact and order-independent, so the accumulator cannot be the source of any difference. Feeding both kernels identical operands from every linear layer of Qwen3-1.7B and 8B (196 and 252 layers), we find bit-identical outputs under power-of-two scales, confirming a pinned prediction list 196/196 and 252/252 (pre-registered at 1.7B, pinned but not blind at 8B), and observed differences of at most one bfloat16 spacing under the checkpoints' real scales. This localizes the divergence to scale application and output rounding after the exact accumulator. Applied as a probe checkpoint, the same intervention restores end-to-end bitwise agreement (8/8 and 16/16 sequences). Cross-implementation FP8 GEMM shows a different signature: both the prevalence and the magnitude of differences grow with reduction depth, while the INT8 fraction stays at parts per million and within one spacing over a 64x range of K. Teacher-forced replay ties layers to tokens: flips concentrate at small logit margins, which predict flip risk with ROC-AUC 0.94 on 16,384 positions. We will release the pre-registration, per-layer predictions, manifests with kernel-selection evidence, and a conformance procedure that turns these controls into a concrete check for kernel interchangeability.

Large pre-trained models have become foundational components of modern machine learning systems. Yet adapting these models to novel categories typically requires examples from the target distribution. In many domains, however, such data are unavailable. Zero-shot learning (ZSL) permits recognition under these limitations through relying on auxiliary semantic information such as textual descriptions. We introduce CAST (Closed-form Analytic Semantic Transfer), a training-free, image-free framework for extending a pre-trained classifier to previously unseen classes through weight injection. We provide a theoretical foundation for CAST and derive a finite-sample error decomposition that identifies the \emph{semantic extrapolation residual} $ρ_u$. The residual is a computable, model-agnostic measure and provides a principled criterion for dataset curation and benchmark design. Experiments on standard zero-shot learning benchmarks demonstrate that CAST matches or exceeds existing image-free approaches and approaches the performance of few-shot adaptation methods, while requiring neither iterative optimization nor examples from the target distribution.

Multiband Fano resonances are demonstrated in the near-infrared (near-IR) using an all-dielectric metasurface whose unit cell consists of four silicon nanoblocks on a glass substrate. An in-plane asymmetry triggers symmetry-protected quasi-bound states in the continuum (QBICs), producing multiple high-Q resonances. Their origin is identified through multipolar decomposition of the scattering cross section and field distributions at the resonances. The strong field localization at these resonances enables efficient multiband third-harmonic (TH) generation in the ultraviolet (UV), with a maximum simulated conversion efficiency of $8.5 \times 10^{-3}$ at a peak pump intensity of $1.6\,\mathrm{GW/cm^{2}}$. The metasurface is fabricated in symmetric and asymmetric configurations, and its linear and nonlinear responses are measured under normal incidence. A TH conversion efficiency of $1.2 \times 10^{-6}$ is obtained at a peak pump intensity of $3.25\,\mathrm{GW/cm^{2}}$. These results establish a route to multiband photonic devices, including multiwavelength lasers, multiband harmonic generation, and single-photon sources for quantum photonics.

In LLM-based code generation, Non-Functional Requirements (NFRs) are often specified as terse one-line phrases. We ask whether grounding those specifications in ISO/IEC 25010 Quality Model, either as rich natural-language prose (NL-rich) or as structured JSON (Structured), improves code generated on HumanEval/HumanEval-ET compared to a RobuNFR-style one-line baseline (NL-simple). We evaluate four NFRs (performance, error handling, code smell, readability) with ten prompt variations per condition under a fixed model snapshot and paired non-parametric analysis. Primary finding: ISO-grounded enrichment improves static quality proxies (unreadability density falls across all four NFRs (e.g., Performance 0.88 -> 0.69 for NL-rich)) and reduces sensitivity to prompt wording, but does not reliably improve functional correctness; for error handling, extended-test pass rate decreases, suggesting tension between defensive coding patterns and exact-output benchmarks. Secondary finding: when ISO content is held constant, NL-rich and Structured differ negligibly in correctness (|delta| <= 0.023), indicating that semantic content matters more than JSON-vs-prose format. Practitioners should invest in standard-grounded NFR content rather than serialization form. A fully traceable replication package is provided.

Advances in diffusion-based generative models have motivated the use of synthetic image generation to alleviate data scarcity in vision tasks. While this strategy has shown promise in natural image benchmarks such as ImageNet, its effectiveness in sparse, high-variance real-world domains remains unclear. In this work, we focus on domains where images differ substantially from common image datasets and additional data are expensive to obtain. Against non-generative data augmentation baselines, we evaluate the downstream classifier performance improvements yielded by two schools of generative sparse data extension: distribution modeling and sample perturbation. Across five trauma classification tasks using subject-wise train--validation splits, no generative approach consistently outperforms a strong non-generative baseline. Feature-space analysis reveals recurring failure modes: memorization or collapse, distributional drift, and generation of visually plausible but simplified canonical instances that are easier to classify than real data.

Object-goal navigation requires an agent to reason over object relationships and prioritize target-relevant objects for efficient decision making in unseen environments. While existing graph-based methods incorporate target-awareness at the feature or attention level, they remain permutation-invariant and lack an explicit mechanism to control information propagation order, limiting their ability to model target-dependent importance and long-range dependencies. In contrast, Graph-Mamba highlights that node prioritization through sequence ordering is critical for effective global reasoning. In this work, we investigate the node prioritization mechanism in Graph-Mamba and study its role in object navigation. We propose Graph-MambaNav, a target-aware spatial-temporal graph encoding framework that introduces a heuristic ordering over objects based on their relevance to the target, allowing more informative objects to be processed later to aggregate richer context. Both node ordering and edge weights are initialized from LLM-derived commonsense object relationships, providing a unified prior for structured reasoning. A spatial module integrates local message passing with global GraphMamba-based selective scanning, while a temporal module applies Mamba-based sequence modeling over object-wise temporal orders, allowing selective aggregation of historical context for long-range temporal reasoning. Experiments on AI2-THOR and RoboTHOR demonstrate improved navigation performance with generalization, and additional real-world robot deployment further validates the effectiveness of our proposed approach.

This paper describes the University of Florida Gators submission to the WMT26 Low-Resource Indic Language Translation shared task. We adapt the retrieval-augmented many-shot translation pipeline from our AmericasNLP 2026 system to translate between English and eleven North-Eastern Indian languages in both directions. At inference time, BM25 retrieves the most similar parallel examples from a language-specific training bank, and Gemini 2.5 Flash translates the input conditioned on these examples. No model fine-tuning is involved. Training banks combine official WMT26 data with publicly available corpora such as Samanantar and prior WMT shared task releases. A grid search over retrieval count r and development exemplar count d across all 22 language-direction pairs selects the best configuration for each submission.

Autonomous systems can fail in rare and heterogeneous ways, making real-world failure discovery difficult under limited testing budgets. Although cheaper proxies such as simulators, lower-fidelity systems, or related policies can be sampled extensively to find failures, proxy failures often do not transfer to the real world due to sim-to-real and system-to-system gaps. The key challenge is therefore to effectively leverage proxy system information for accurate prediction of severe target system failures. We propose an adaptive failure discovery method that combines proxy evaluations with limited target system results to guide scenario selection for target system testing. Our method learns a local predictor of target risk by correcting proxy failure signals using control-variate-inspired residual modeling. To find failures that are both likely and diverse, we combine this predictor with a support-aware mutual-information objective that favors realistic, well-supported regions while expanding coverage across failure modes. Across autonomous driving, manipulation, and quadruped velocity-tracking tasks, our method discovers up to 2$\times$ as many failures as random sampling and active-learning baselines, including severe and diverse failures missed by competing methods.

Deposition training requires attorneys to manage dynamic witness behavior, yet legal-AI evaluations largely focus on factual accuracy, reasoning, or response-level plausibility. We introduce WitnessSim, a deposition simulator driven by controllable legal personas. We use an evaluation framework separating behavioral realism from pedagogical usefulness. We assess realism through adversarial testing, blinded attorney comparison, and analysis of longitudinal behavioral trajectories. WitnessSim generally maintained plausible behavioral boundaries, and attorneys did not systematically prefer either original testimony or WitnessSim generated testimony. Pedagogical tests showed that witness behavior changed meaningfully in response to question form and attorney intervention without uniformly collapsing the assigned persona. Together, these results showcase a model of behavioral fidelity in legal simulations, and provide a framework for evaluating its performance.

Computer-aided detection (CADe) systems for colonoscopy promise to reduce clinical miss rates, yet reliable real-world deployment remains elusive. This translational gap stems in part from a structural flaw in model development: the reliance on curated datasets that under-represent the long negative stretches and procedure-related artifacts characteristic of routine examinations. Training and evaluating architectures strictly on these lesion-centric benchmarks creates an illusion of success, since such benchmarks cannot capture clinically crucial metrics. To expose this gap, we establish TRUE-Colon, a standardized benchmarking protocol that measures key deployment characteristics alongside localization accuracy, and evaluate four real-time architectures (Faster R-CNN, YOLOv8, YOLOv11, RT-DETR) across curated benchmarks (SUN, PICCOLO) and 60 unedited, full-length procedures (REAL-Colon). We observe a consistent transfer asymmetry: models trained strictly on curated clips suffer a severe performance collapse when evaluated on full procedures, whereas procedure-trained models substantially improve rejection of non-polyp content on REAL-Colon, and largely retain their accuracy on curated benchmarks. Beyond transferability, we find that the Transformer detector attains the strongest sensitivity and the earliest, most persistent detections, while the convolutional detectors stay competitive at a higher throughput. Together, these results indicate that both training and benchmarking for deployable CADe should shift from curated, lesion-centric clips toward full-procedure data and deployment-relevant operating points. Source code is available at https://github.com/sdoerrich97/true-colon.

Automatically generating textbook-grounded assessment items can reduce science teachers' workload, but existing retrieval-augmented generation (RAG) systems rely on flat retrieval, support only single-question generation, lack safeguards against weak evidence, and are ill-suited to low-resource, board-exam-structured curricula. We address these limitations with TeachMateGPT, a multi-agent system contributing four advances to curriculum-grounded science-assessment authoring. (i) COPE, a hierarchical knowledge base replacing token-window chunking with a multi-resolution index that segments documents along syllabus structure and links them at three granularities via a traversable graph-based lineage, matching evidence to each topic's instructional level. (ii) A staged, fail-closed agent pipeline replacing one-shot retrieve-then-generate: routing gates search, retrieval fuses dense and lexical evidence under a coverage gate that withholds generation on insufficient evidence, and specialist agents draft objective and constructed-response items. (iii) SAVER, a source-attributed verification protocol scoring faithfulness, relevance, and hallucination risk against retrieved evidence, applying stricter grounding checks across each creative question's four sub-parts, paired with teacher-in-the-loop evaluation rather than automatic filtering. (iv) NCTB-SciGen8, a curriculum-grounded dataset of 198 items (143 multiple-choice, 55 creative questions) spanning all 14 chapters of the NCTB Class 8 science textbook, produced by the pipeline and rated by three practicing teachers. TeachMateGPT raises faithfulness (0.68 $\rightarrow$ 0.96) and answer relevancy (0.60 $\rightarrow$ 0.89) over a vanilla RAG baseline.

Existing defenses against hallucination in retrieval-augmented and multi-agent pipelines remain partial: evidence is trusted despite modality disagreement, debate verifies an aggregate report rather than individual claims, and such verification occurs only after drafting, leaving inter-agent errors undetected until the final text. To close this gap, we present CLAIR-Fin, a nine-agent framework that decomposes each question into atomic claims maintained in a typed Financial Claim Ledger. Each claim is resolved through Asymmetric Evidence Authority, which conditions evidence trust on claim type rather than treating all modalities as equally reliable; Chain-of-Custody Verification, which checks grounding at the hand-off between drafting and adversarial review rather than only at the pipeline's exit; an Adaptive Rebuttal Cycle, which routes contested claims through adversarial debate whose depth scales with what that debate finds; and a terminal entailment audit paired with a continuous Hallucination Risk Index that distinguishes claims that passed scrutiny from claims never contested. We evaluate CLAIR-Fin on BB-FinQA-X, a 500-question cross-modal financial evaluation set built from Bangladesh Bank Annual Report material, stratified by query type, format, and difficulty. Relative to a single-pass retrieval-augmented generation baseline, it raises faithfulness ($0.780 \rightarrow 0.889$) while abstaining on 5.4% of questions when evidence is insufficient rather than forcing an unsupported response, and it exceeds stronger retrieval-strategy baselines such as HyDE and Graph-RAG on faithfulness ($\leq 0.874$).

Medical image segmentation is still largely treated as a vision-only problem, although clinical interpretation often relies on textual knowledge of anatomy, location, appearance, and surrounding context. Existing text-guided segmentation methods within the Vision-Language Model (VLM) paradigm often use language only as a late conditioning signal, limiting its influence on visual representation learning. We introduce MedPlex (Medical Plexus of Vision and Language), an end-to-end VLM framework that makes text guidance a continuous, clinically grounded component of segmentation learning. Through Bi-Fusion (Bidirectional Fusion), visual and textual representations evolve jointly across the encoding hierarchy. MedPlex further introduces class-level and region-level concept alignment to organize the shared representation at complementary granularities. Class-level alignment anchors each anatomical target to an aggregated clinical concept profile, while region-level alignment preserves individual concepts, such as shape, location, appearance, and texture, through class-specific visual evidence. In this way, language provides structured supervision throughout the encoder rather than serving only as a late-stage cue. MedPlex achieves state-of-the-art performance across CT and MR benchmarks for multi-organ, cardiac substructure, and tumor segmentation, including settings with real free-text clinical supervision. Code: https://github.com/rafiibnsultan/MedPlex.

A central goal of robot learning is to enable robots to execute rich instructions specified at runtime. Large-scale language-conditioned policies have made substantial progress toward this goal, yet still struggle with temporal structure and safety constraints. Linear Temporal Logic (LTL) provides a powerful language to express complex, non-Markovian instructions. However, guiding learned manipulation policies toward LTL satisfaction remains challenging because modern policies generate short-horizon action chunks and replan in closed loop, while almost all LTL specifications are evaluated over long-horizon trajectories. In this paper, we introduce hint$^2$, a method for guiding short-horizon policies toward satisfying complex LTL specifications at inference time using hierarchical world models. Our key idea is to derive two separate guidance objectives using each world model's abstraction level. A high-level model predicts future action-induced transitions in task-relevant atomic propositions to guide progress through the LTL automaton, while a low-level dynamics model predicts immediate state evolution for accurate local safety guidance. Our results show that hint$^2$ overcomes the limitations of current LTL-guided diffusion methods, outperforms existing inference-time steering methods in CALVIN, and successfully completes instructions with complex liveness and safety constraints more elegantly than language-conditioned alternatives. Finally, we demonstrate that hint$^2$ can handle complex instructions on a real UR5e manipulator.

Physics-constrained diffusion for high-contrast, sharp-interface multiphase fields faces three coupled difficulties. At coefficient jumps, expanded pointwise strong-form PDE residuals contain singular gradient terms that can penalize physical interfaces. Under extreme contrast, low-magnitude phases may fall below the diffusion noise floor and be erased, misscaled, or generated with negative coefficients, while a global likelihood scale allows high-magnitude phases to dominate supervision. We therefore propose Multiphase-Diff, which makes three corresponding contributions: (i) a conservative flux residual that avoids differentiating discontinuous coefficients and enforces discrete conservation; (ii) an analytic bijective representation that maps low-amplitude signals to order-one latent scales and guarantees coefficient positivity through exponential decoding; and (iii) a Jacobi-preconditioned likelihood that normalizes local residual scales for balanced supervision. Experiments on three complementary multiphase benchmarks demonstrate the superiority of Multiphase-Diff over seven baselines in both physical and distributional fidelity and its robustness across phase contrasts and compositions, establishing its effectiveness for scientific sample generation in this challenging regime.

Transforming multimodal sources into condensed and structured media outputs can be fundamentally conceptualized as a long-horizon agentic process centered on a model-harness system. While an ideal harness system should align with human design priors and accumulate reusable experience through empirical exploration to drive recursive self-improvement, existing paradigms remain static and fall short of this capability. In this paper, we present AutoDesign, a framework that aligns with human design priors, where a meta-harness optimizer guides a code agent to recursively improve harness based on rollout feedback. To instantiate and evaluate this framework, we focus on the academic paper-to-poster generation task and introduce PosterBench, comprising a 100-paper Main Track spanning five disciplines and PosterBench-mini, a shared 10-paper subset for controlled evaluation. On the PosterBench Main Track, AutoDesign achieves the highest score of 78.32, surpassing the closed-source commercial system Claude Design by 7.45 points. Across seven controlled code-agent-model configurations, integrating the learned DesignHarness consistently improves performance, increasing the average PosterBench Score from 54.99 to 67.39 (+12.4%). In a fully autonomous long-horizon loop, it executes 253 tool calls and 11 editing turns within 40 minutes for under $3, reaching average conference-poster quality in human evaluation. A system-blind human study further demonstrates that AutoDesign achieves the highest human preference among evaluated systems.

Recent advances in foundation models have enabled AI scientists to automate increasingly complete research workflows, from hypothesis generation and code execution to manuscript preparation. Yet workflow coverage alone does not provide access to the full evidence on which scientific discovery depends. Existing systems typically reason over text, code, labels, or precomputed summaries, leaving scientifically decisive spatial, temporal, cross-channel, and procedural relations unavailable to the agent. We introduce OmniScientist, an end-to-end, omni-modal AI scientist that conducts multidisciplinary research directly from heterogeneous raw evidence. A perception layer and 3 autonomous agents for ideation, experiment, and writeup operate within a deterministic pipeline, allowing observations to shape research questions, experimental decisions, and final claims throughout the research lifecycle. By running idea, rigour, and claim checks in code, the system enforces novelty screening, statistical validity, execution provenance, and numerical traceability. We evaluate OmniScientist on 36 real-data cases spanning 5 discipline families, 4 families of scientific evidence, and modalities including images, signals, audio, video, 3-D structures, trajectories, tables, formulae, and graphs. The system completes the full path from raw data to a compiled manuscript in all 36 cases and achieves a mean overall paper score of 6.3 with the reference reasoning backbone. In paired comparisons against a blind variant that receives only precomputed scalar features, direct perception improves all 7 evaluation dimensions and wins 85% of head-to-head judgments. These results show that lifecycle-wide perception is essential for evidence-grounded scientific discovery and provides a practical path toward broadly capable AI scientists.

Humanoid motion tracking is central to teleoperation and whole-body imitation, yet evaluation often disagrees with what people perceive in videos. Kinematic errors average per-frame pose differences but miss the physical artifacts that matter most, particularly unstable support and incorrect contacts such as foot skating and mistimed touch-downs. Meanwhile, widely used test suites are small and lack the diversity needed to stress contact-rich, long-horizon behaviors. We introduce HumanTracker to make humanoid tracking evaluation both perceptually aligned and scalable. The HumanTracker benchmark contains approximately 153 hours of optical motion trajectories from multiple professional performers, organized into four motion families with text labels for fine-grained diagnosis. We further propose HumanScore, a preference-aligned metric trained on 12K motion pairs containing 24K motions. Across representative state-of-the-art trackers, HumanScore better predicts human preferences and reveals contact and stability failures that kinematic metrics often miss.

Video world models simulate future states conditioned on current observations and user actions. Recent systems have demonstrated impressive video consistency and action controllability over long sequences. However, fairly comparing these interactive models remains challenging. In practice, a human player typically evaluates a world model by pursuing long-horizon objectives through interaction. For example, a user may turn around 360 degrees to see whether the environment remains consistent, or walk into the water and inspect whether realistic water ripples are generated. The action sequence required to achieve the same objective may vary substantially between models, making fixed action-conditioned evaluation unsuitable for cross-model comparison. To address this, we employ multi-modal Agent Players to interact with world models toward specified long-horizon objectives. Building on this paradigm, we introduce PlayWorld, a benchmark providing 171 scenarios, each with a specified objective. To evaluate performance thoroughly, we assess models along four core dimensions: geometry consistency, interaction fidelity, out-of-sight evolution, and insight evolution. In addition, we incorporate basic ability metrics for video quality and controllability. Experiments across nine state-of-the-art world models reveal that current models remain unreliable on long-horizon interactive objectives, particularly in maintaining spatial consistency and persistent state evolution. Code and data are available at https://github.com/kxding/PlayWorld.

The orbital angular momentum (OAM) of light is a discrete, unbounded degree of freedom that underpins mode-multiplexed communications and high-dimensional quantum photonics. Yet, dynamic OAM control remains dependent on bulky free-space optics or cascaded architectures that separate switching from wavefront shaping, hindering nanoscale integration. Here, we engineer artificial van der Waals crystals from rhombohedrally stacked (3R) MoS$_2$, in which spatial control of the local crystal orientation imprints a nonlinear geometric phase onto the second-harmonic (SH) field, enabling background-free generation of SH vortex beams in an ultrathin (46 nm) van der Waals platform. Leveraging the C$_{3v}$ symmetry of 3R-MoS$_2$, we demonstrate monolithic, all-optical switching with sub-optical-cycle precision between Hermite-Gauss-like and Laguerre-Gaussian vortex SH beams with opposite topological charges ($l=\pm1$). Our results establish artificial 3R-MoS$_2$ crystals as a monolithic platform for the generation and all-optical reconfiguration of nonlinear structured light at the nanoscale, advancing active nanophotonic sources for integrated classical and quantum photonic technologies.

Interactive world models must support persistent memory, responsive interaction, and long-horizon generation, yet these requirements place conflicting demands on the model. Maintaining history in the denoiser context or key-value cache incurs growing cost, forcing a trade-off between session length and retained memory, while low-latency interaction relies on few-step generation whose capabilities are bounded by its teacher. Evoke addresses both limitations by externalizing persistent world state and redesigning the teacher for long-horizon interactive generation. Scene geometry is maintained in an external, camera-indexed world state bank, from which only view-relevant information is retrieved, keeping the denoiser context bounded as the session grows. Rather than treating the teacher as a fixed generator, we design it for long-horizon supervision: its sparse attention combines chunk-wise grouping, retrieval of selected distant frames, and a linear-attention global state, yielding linear growth in memory and compute while enabling supervision over long horizons. Such supervision exposes content drift that stays locally plausible within short windows, while per-chunk conditioning enables prompt changes and event control throughout the sequence. A 30-second distribution-matching objective, applied under self-forced rollouts, transfers both capabilities to a three-step student that uses no classifier-free guidance, improving resistance to long-term drift while preserving responsive conditioning. With bounded context and recurrent external memory, Evoke supports open-ended, continuously evolving generation; on a single H200 at $384\times 640$, each $1.5\,\mathrm{s}$ chunk is generated in $2.11\,\mathrm{s}$. As a three-step world model, Evoke achieves state-of-the-art performance on WBench while remaining competitive on VBench-Long and VBench-2.0.

Modern language models are trained on heterogeneous web-scale text corpora. Consequently, studying knowledge and skill acquisition is difficult, as prior exposure to related content is hard to characterize. To address this challenge, we introduce LITTLECURRICULUM, a curated 88B-token pretraining corpus tailored to U.S. elementary school material, explicitly excluding concepts, facts, and vocabulary taught above Grade 5. Training a 5B-parameter LLM from scratch on LITTLECURRICULUM yields LITTLELEARNER, a model with sufficient language competence for open-ended evaluation, yet with clear knowledge and capability boundaries mapped to interpretable curriculum guidelines. We release LITTLECURRICULUM and LITTLELEARNER as a developmentally restricted sandbox to study how models acquire, represent, and use data under a well-defined training scope. We illustrate the sandbox's utility in a first suite of experiments on injecting new knowledge through post-training and in-context learning. These methods let LITTLELEARNER better utilize existing knowledge, but do not raise out-of-scope capabilities. Our findings underscore the value of this controlled environment for future investigations.

Part-aware 3D generation aims to create digital assets that are coherent as complete objects while exposing structural parts for editing, material assignment, animation, and reuse. Existing methods impose this structure outside the native generation loop: segmentation-based methods partition an already generated shape, while additive methods synthesize parts from predefined layouts, boxes, or tokens and then reconcile them into a whole. The former preserves the generated geometry but fixes the object before part boundaries are determined; the latter exposes part cardinality but often leaves shared boundaries vulnerable to gaps, interpenetrations, and material discontinuities. In this paper, we propose SCULPT, a framework that addresses these challenges through subtractive composition. Given a complete object represented in a structured 3D latent space, SCULPT iteratively applies a joint split predictor to generate one extracted part together with the remaining object. The predictor performs a coupled denoising process conditioned on both the image and the current 3D state, so the extracted part and updated remainder are generated together rather than reconciled after generation. The joint split predictor processes both outputs on the union of their native sparse 3D supports, allowing neighboring supports to overlap rather than imposing a disjoint voxel partition. The rollout ends when the remainder support becomes empty or reaches a fixed safety cap, allowing the number of generated parts to adapt to each object within that bound. Extensive experiments demonstrate state-of-the-art geometry on PartObjaverse while preserving strong complete-object reconstruction after part assembly. Results on four dataset images, one text-to-image-generated input, and one real-world photograph further show fine-grained textured part decomposition beyond the benchmark.

In this paper, we formalize a joint communication-control strategy optimization (JCCO) problem in multi-agent linear systems with quadratic costs, under the common-information-based (CIB) framework from decentralized stochastic control. For computational tractability, we focus on such JCCO problems with partially nested (PN) information structures (ISs). In particular, with a baseline communication protocol that leads to a PN IS, we establish a series of conditions under which the partial nestedness is preserved under the (additional) communication strategies to be optimized, while violating them may cause nonlinearity of the optimal strategies in general, with open-loop communication strategies. We then develop a dynamic-programming-based approach to compute the optimal control strategies of JCCO with open-loop communication strategies, which yields a set of closed-form Riccati Equations. As a byproduct of independent interest, such an approach also offers a way to solve decentralized linear-quadratic control with PN ISs and output feedback, under the CIB framework. Finally, we extend such an approach to JCCOs with closed-loop communication strategies, yielding a more tractable dynamic program than an infinite-dimensional CIB-belief-based one.

Many clinical prediction models treat post-intervention outcomes as a one-step mapping from baseline measurements to a future endpoint. However, recovery after a procedure often unfolds as an irregular trajectory: clinical observations, medication changes, repeat interventions, and physiological measurements are recorded asynchronously and can change risk assessment over time. We propose an intervention-aware clinical world model that represents each patient with a structured latent state and evolves it through time-ordered post-intervention events. The model first encodes baseline imaging into a 3D spatial latent state. It then updates this state using procedural context, static covariates, elapsed time, and peri-event physiological embeddings. Follow-up imaging provides training-only supervision through a latent forecasting objective. We apply the framework to atrial fibrillation ablation. During the 90-day recovery window, irregular post-procedure records provide clinically meaningful evidence for long-term recurrence risk. In repeated internal cross-validation on DECAAF-II, our model achieves AUROC 0.756 and AUPRC 0.777 for recurrence prediction. It also achieves a scar-extent MAE of 2.971 percentage points without requiring follow-up MRI intensities at inference. The learned state supports recurrence-risk queries at different horizons and retrospective input editing of blanking-period records.

Current large language model development relies on massive, often non-permissible datasets, creating a high barrier for researchers committed to open-source and ethically sourced data. We introduce Mimir v1, a 1-billion-parameter language model based on the Hierarchical Reasoning Model (HRM) architecture, that is trained from scratch and delivers highly competitive performance for English and sets a new state of the art for Danish using only permissible post-training data. Trained on a mixture of 161 datasets, Mimir v1 outperforms the original HRM-Text 1B and competes with larger frontier models like Qwen 3.5 4B and Gemma 4 E2B, tested across 20 benchmarks for English, Math & Code and Danish. The model is available on the Hugging Face Hub: https://huggingface.co/danish-foundation-models/DFM-Mimir

Tabular-to-image methods have emerged as novel approaches to leverage the high predictive performance of convolutional neural networks and vision transformers. They convert tabular data into image representations, mapping each feature at a fixed pixel location derived from a dimensionality-reduction method (e.g., t-SNE, UMAP, PCA). However, they encode only the marginal value of each feature and discard information about feature relationships. We propose TabSOM, a tabular-to-image encoding built on the Self-Organizing Map (SOM), which provides: (i) a spatial layout in which every input feature occupies a fixed canvas position derived from its component plane via collision-free Hungarian assignment; and (ii) a graph that captures pairwise feature relationships derived from the SOM component planes. The resulting image stacks two multi-scale node channels: one encodes feature values at fixed scales, while the other encodes pairwise feature interactions as spatial connections between related features. Two SOM-derived interpretability approaches are introduced: a prototype-inspired partial dependence plot and a class--separation importance score. Benchmarked against twelve existing tabular-to-image methods across public binary-classification datasets, TabSOM ranks first or second on every dataset and achieves the lowest variance of any method evaluated. Interpretability obtained with TabSOM was validated against Random Forest, XGBoost, and SHAP, the class-separation score shows reasonable agreement with established baselines on the top-ranked features while capturing complementary structural information from input data. These results demonstrate that TabSOM provides an effective and interpretable approach for applying deep learning architectures to tabular data, bridging the performance--interpretability gap in this domain.

Repeated in-situ evaluation of ocean-glider planners requires scarce vehicles, operators, deployment and recovery resources, and ocean conditions that cannot be reset for competing algorithms. We present a guided, installation-free browser-native digital test range that transforms a selected region into a reproducible four-dimensional experiment. The system leads users from regional domain selection through mission-scoped bathymetry, time/depth forcing, science objectives, optional task decomposition, route specification, current-advected execution, observation generation, and scoring. Its primary contribution is a common plan-to-observation contract unifying vehicle, sensing, and evaluator assumptions across manual routes, transparent built-in algorithms, and imported classical or learned-planner outputs, while exported artifacts form dataset-ready records. A controlled Observing System Simulation Experiment (OSSE) evaluates five classical planners in two episodes, three deterministic seeds, and a calibrated 60-hour horizon. All 54 missions completed and recovered without hard violations, while planner rankings and dive-policy effects revealed operational-scientific tradeoffs. An authentic public deployment supplied a field-referenced audit to scope current kinematic boundaries. Separately, source-locked GliderFlight 1.2.0 achieved native-to-browser parity through Pyodide/WebAssembly, establishing a pathway for high-fidelity multi-tier simulation. The resulting operational space is scientifically traceable and component-qualified for mission-scale pre-deployment experimentation.

Scientific discovery increasingly requires AI systems that can reason over scientific evidence of heterogeneous modalities, interact with scientific tools and environments, and sustain progress across long task horizons. We present Intern-S2-Preview, a series of scientific agentic foundation models designed to support multimodal scientific understanding, reasoning, generation, and long-horizon tasks. The training pipeline begins with scientific multimodal pre-training over rendered scientific documents, interleaved image-text data, and diverse scientific corpora. Starting from the pretrained checkpoint, we apply a unified post-training pipeline consisting of supervised fine-tuning, scalable multi-task reinforcement learning (RL), black- and white-box agentic RL, and on-policy distillation. This pipeline is supported by practical techniques that improve rollout and training stability and efficiency, including partial rollout with off-policy correction, adaptive length regularization, online speculative decoding, robust multi-task optimization, and trace-aware experience assembly for agentic tasks. At the architecture level, Intern-S2-Preview-397B extends time series modelling from efficient long-sequence understanding to numerical forecasting, while Memory Decoder is studied as a separate memory-augmented path for rapid scientific specialization without modifying the frozen 397B backbone. Evaluations across scientific, multimodal, agentic, and general-purpose benchmarks show that Intern-S2-Preview-397B achieves competitive or leading results in multiple settings. The time series modules improve scientific signal understanding and forecasting on SciTS, while the separate Intern-MemDec-4B extension improves the Biology-Instructions average score from 56.92 to 60.32 without modifying the frozen 397B backbone.

Snapshot Compressive Imaging (SCI) offers an efficient solution for high-speed video acquisition and, under exposure-time camera--scene relative motion, multi-view scene capture by compressing temporal or spatial information into a single 2D measurement. While recent studies have explored SCI for 3D scene reconstruction, existing methods struggle with significant challenges due to information loss, limited viewpoint diversity, and the computational burden of jointly optimizing 3D representations and camera poses. In this work, we propose a novel framework that reconstructs high-quality 3D scenes from a single SCI measurement by leveraging 3D Gaussian Splatting (3DGS) and the powerful priors of large-scale vision foundation models (VFMs). Our primary reconstruction combines measurement-derived 3D VFM initialization with SCI-aware Gaussian optimization. After coarse-stage convergence, an auxiliary 2D VFM provides pseudo-view supervision at synthesized viewpoints for local appearance refinement. To further address the instability caused by ambiguous SCI supervision during 3DGS optimization, we introduce Opacity-Guided Splitting and Growth Regulation (OSGR), an SCI-specific densification strategy that augments split candidates using local opacity statistics, discourages loss-compensating opacity inflation through mean-opacity regulation, and bounds representation growth with explicit candidate-ratio and Gaussian-count constraints. Extensive experiments across multiple benchmarks demonstrate that our method achieves the strongest overall performance, combining leading reconstruction quality and robustness to viewpoint variation with competitive computational efficiency.

Efficiently retrieving relevant clips from large-scale driving logs is essential for data curation, model development, and safety analysis. Structured and rule-based retrieval systems can explicitly target driving events, but typically require expert-defined rules, auxiliary data, and multi-stage perception pipelines. Multimodal embedding models offer a simpler and more efficient alternative by representing each video with a single searchable vector. However, general-purpose models often rely on shortcuts from static scene context and struggle to distinguish motion-centric events, such as turning left versus right or accelerating versus decelerating. In this work, we study how to adapt a general-purpose multimodal embedding model to driving-video retrieval. We first fine-tune Qwen3-VL-Embedding on paired clips and reasoning traces from nuReasoning using an InfoNCE objective. While this stage substantially improves overall retrieval, caption supervision alone remains insufficient for fine-grained motion understanding. We therefore introduce TraVEL (Trajectory-Guided Video Embedding Learning), a motion-aware fine-tuning framework that uses ego-trajectory similarity as a reward within Group Relative Policy Optimization. Trajectories serve only as privileged training supervision; retrieval still operates on single-vector video embeddings without ego poses, expert rules, or auxiliary perception outputs. We further construct a driving-video retrieval benchmark from nuReasoning. Experiments show that TraVEL improves motion-centric retrieval across model scales: relative to SFT, it raises longitudinal and lateral mAP by 9.8 and 4.7 points at 2B, with corresponding gains of 7.2 and 1.5 points at 8B. TraVEL thus combines physically grounded supervision with efficient embedding-based search.

We present \textbf{DreamX-Phi 1.0}, an action-conditioned video world model for robotic manipulation that, given an observed frame, a language instruction, and a prescribed action sequence comprising end-effector poses and gripper states, predicts the resulting future observations. Yet realism alone does not guarantee faithfulness: a convincing rollout can still move the wrong arm or lose the manipulated object. To ensure the prediction respects each arm's commanded path, we inject per-arm $\mathrm{SE}(3)$ transformations into attention via \textbf{PRoPE-style geometric encoding}, preserving arm identity and rigid-motion structure. Action control alone does not fully constrain scene geometry or the evolution of small manipulated objects. We therefore add a lightweight \textbf{depth branch} for scene-level geometry and use \textbf{SAM3 masks} with a frozen \textbf{V-JEPA teacher} to maintain object consistency throughout grasping. We further distill the multi-step generator into a few-step student via distribution-matching distillation for efficient deployment. At the time of writing, \model{} achieves first place on Track~1 and second place on Track~2 of the WorldArena~2.0 Challenge. Our model and code will be publicly available.

When asked about entities outside their knowledge boundary, LLMs routinely fabricate plausible-sounding details rather than backing off to safer, more general claims. We frame this failure through a Gricean lens: a cooperative speaker who is uncertain about a referent retreats up the specificity hierarchy, trading informativeness for truthfulness. We ask whether LLMs have the ingredients to perform this retreat. Using a T-REx-based benchmark that varies entity familiarity and referent specificity, we probe models to answer two questions: (i) do their activations encode whether a referent falls inside the knowledge boundary, and (ii) do they anticipate the specificity of the referent they are about to generate? We find that the answer to both is yes, but the two signals are not reconciled in generation. Models overwhelmingly prefer specific referents even when the entity is unknown to them, and do so even when offered correct generic alternatives. The substrate for a Gricean retreat is present, but the policy that would act on it is not. We position our findings as a first step toward Gricean alignment, training or steering objectives that couple knowledge-boundary awareness to referent-specificity during generation.

We present our submission to Task 3 of the Gen$μ$ 2.0 Challenge on visual concept unlearning. Building on MapRoute, we introduce task-specific training objectives, richer concept representations, and semantic routing for concept-specific mapper selection. Our approach improves robust concept removal while preserving unrelated and semantically adjacent concepts. On the official benchmark, evaluated using the Erasing-Retention-Robustness (ERR) metric on Stable Diffusion v1.4, our method outperforms the state-of-the-art baseline by 12.1\% on average across the five concept categories, achieving substantial gains.

We present Multi-Agent Reasoning and Coordination (MARC), an open-source framework that replaces monolithic LLM prompting with deterministic multi-agent orchestration for clinical reasoning. MARC coordinates role-specialized agents for extraction, reasoning, answer generation, and evaluation, with explicit context passing and traceable intermediate outputs, enabling stage-wise failure attribution. We additionally introduce a Decomposer module that generates task-specific agent prompts from a plain-language description, eliminating manual prompt engineering. The framework supports both API-based and local CPU-compatible deployments and is entirely configurable via YAML, without code modifications. MARC is designed to be model-agnostic, interpretable, and accessible to clinical domain experts without programming expertise. The full framework is available at https://github.com/Penn-RAIL/MARC-v1.

Vision-language-action models (VLAs) are moving rapidly towards deployment as general-purpose manipulation policies, but we currently lack basic tools for understanding what these models represent internally or for monitoring them at runtime. Leveraging ideas from mechanistic interpretability, we probe the residual stream of $π_{0.5}$ and find that task progress, the normalized time remaining in a trajectory, is linearly readable from the activations. We find that this signal is present in the pretrained PaliGemma backbone prior to training on any robot-specific data. A single linear probe generalizes to unseen tasks and varies under language counterfactuals when trained on multi-prompt data, but does not enable meaningful steering of the policy. These properties make the signal directly useful for instrumenting deployed VLAs. We use the probe as a simple label-free OOD detector, which detects stalled task progress, and find it competitive with state-of-the-art methods. Our results suggest that VLAs have rich, linearly readable internal representations of semantic quantities like task progress, and that learning to read these signals offers a lightweight, interpretable path toward monitoring deployed visuomotor policies.

Analog circuit design is a time-consuming, iterative process in a nonlinear and high-dimensional design space that relies heavily on expert intuition. Among recent developments, LLMs have introduced a promising approach by bringing natural language reasoning to circuit design tasks. The majority of conventional LLM-based approaches provide fragmented solutions that focus either only on sizing or topology generation. These methods require adding specific technical knowledge manually, which is inefficient and prone to hallucinations during circuit sizing. Moreover, the inherent trade-off in meeting different specs makes current approaches iterative and tedious. Another shortcoming is the inability to create innovative topologies, which may lead to sub-optimal designs due to reliance on conventional topologies. In this paper, we present AaLLM, an open-source end-to-end multi-agent LLM workflow that takes user specs as input and outputs the appropriate netlist, encompassing both topology generation and circuit sizing. AaLLM automates the creation of a relevant knowledge base from research papers and textbooks to combat tedious manual data collection. A RAG model is implemented to emulate circuit design expertise using this knowledge base. Moreover, AaLLM uses a novel tri-agent feedback system comprising a Designer that determines circuit component values, a Critic that scrutinizes these values, and an Evaluator that minimizes circuit sizing iterations by arbitrating between the other two agents. AaLLM-generated novel topologies achieve a figure of merit (FoM) comparable to that of known topologies, and up to 3x higher for certain circuits. Testing on several circuit topologies, our results show a 3x - 4.5x decrease in the number of SPICE calls at inference when compared to SOTA multi-agent LLM pipelines. The results also show a 40x decrease in wall-clock time compared to existing approaches.

Modern image classification models excel when trained on single task-specific datasets but often struggle to generalize across domains and difficulty levels. We propose ARMDIL, an Adaptive Router for Multi-Domain Image classification with LLMs. ARMDIL is an ensemble that uses a multimodal large language model (MLLM) agent to dynamically route each image to the most suitable vision backbone. Our diverse ensemble employs convolutional neural networks (ResNets), self-supervised representation learners (SSL), and vision-language models (VLMs), each trained on a unified label space constructed from multiple image datasets with differing distributions and characteristics. Empirical evaluations illuminate the distinct capabilities and vulnerabilities of each architecture across disparate visual domains. Crucially, we show that ARMDIL effectively navigates these trade-offs, performing competitively with specialized training-based routers. Furthermore, it drastically improves adaptability by allowing new information to be integrated via simple prompt modifications, while enhancing interpretability through natural language reasoning traces. These advances in cross-dataset image classification pave the way for more reliable general-purpose vision systems such as AI assistants and autonomous robots.

Contact-rich manipulation failures are often detected only after the robot has committed to contact. This is especially limiting in wrist-camera setups: close gripper--object views help observe contact, but a poor approach may already push, miss, slip, or disturb the object before conventional detectors react. We introduce \emph{ContactGuard}, a pre-contact execution monitor for chunked visuomotor policies. Given the policy's planned action chunk, ContactGuard predicts its short-horizon consequence in latent visual space and aborts if the predicted future latent indicates likely failure. Its latent world model is trained from unlabelled robot trajectories to predict compact multi-view visual embeddings under planned actions, avoiding pixel-level video prediction. A lightweight failure probe is then trained from a small labelled set of pre-contact clips. At deployment, ContactGuard anchors prediction before an imminent contact event, rolls the model forward under the policy's own actions, and verifies the predicted post-contact latent. Across real-world contact-rich manipulation tasks, ContactGuard predicts failure more accurately than direct and corrupted-action ablations, and transfers to live robot as a pre-contact abort signal without modifying the underlying policy.

Instruction-tuned language models achieve strong performance across a range of generation tasks, but have also recently been shown to exhibit verbalized overconfidence. In question answering, verbalized model overconfidence may be associated with the consistency of the generated supporting rationales. In this paper, we study whether corresponding changes in the lexical diversity of generated answer rationales accompany changes in model confidence induced by instruction tuning. We evaluate three matched base and instruction-tuned models across question-answering benchmarks and find that instruction tuning consistently alters answer confidence, despite limited changes in predictive accuracy and decreases in likelihood-based calibration. Secondly, we observe a non-uniform effect of instruction tuning on rationale diversity: cross-rationale diversity consistently decreases, whereas surface-level lexical diversity varies in both direction and magnitude across models and benchmarks. Finally, we find that these differences persist after controlling for answer selection and rationale length, confirming that confidence and rationale diversity capture distinct effects of instruction tuning.

Transformer-based language models achieve strong performance but incur substantial inference cost due to repeated high-dimensional matrix multiplications. We propose Reduced Matrix Multiplication (RMM), a training-free, input-adaptive inference method that reduces Transformer matrix products by selecting informative slices along their contraction dimensions, without modifying model weights. Under a simple retention-ratio control, RMM provides a smooth and predictable accuracy-efficiency trade-off. Across language models ranging from 1B to 70B parameters, we find that reduction tolerance depends on the model family, task, component, and retention ratio, although it often improves with model scale. Under moderate reduction, RMM remains robust across the evaluated discriminative, autoregressive generation, and long-context settings. We further show that the same principle extends to multimodal vision-language inference. Mechanistic ablations reveal a structural asymmetry within Transformers: attention-side computations are substantially more reducible than MLP components. Finally, wall-clock benchmarks with custom kernels on an NVIDIA A100 show that these computational savings can translate into practical runtime gains, especially at longer sequence lengths. Together, these results position RMM as a scalable direction for input-adaptive inference-time optimization.

Visual bottlenecks that focus policy inputs on regions of interest (ROIs) can improve data-efficient visuomotor learning by separating where to look from how to act. Many ROI interfaces rely on external spatial labels, such as gaze, object classes, or affordance annotations. Label-free alternatives often derive crops from trajectories by detecting gripper or motion events and centering a fixed crop at the projected end-effector. Such action-derived crops are useful spatial priors that require no additional labels, but they encode fixed choices about event timing, proxy points, and crop scale. When the visual evidence needed for control lies away from the end-effector or changes continuously with task progress, these crops can become misaligned. We propose Seeker, a task- and state-conditioned readout that learns attention from action. Starting from frozen DINOv3 features, Seeker iteratively updates a query with gathered visual evidence, producing progression-aware ROIs solely from action supervision. The learned ROI serves as a spatial interface for RGB cropping, mask-guided background augmentation, and point-cloud filtering. In simulation and the real world, Seeker improves data efficiency and robustness over no-crop, augmentation, and action-derived crop baselines. On real robots, Seeker raises average in-domain success from the best baseline's 48.3% to 76.7% and success under lighting/background shifts from 20.0% to 60.0%.

We consider the problem of autonomously learning robot skills under a limited practice budget for sequential tasks. We propose an active skill learning algorithm, \emph{Deliberate Practice (DP)}, that computes a provably \emph{budget-optimal} allocation---practicing skills that maximize expected cumulative reward while being learnable within the budget. DP estimates both the time needed to master skills and the cumulative reward of the task plans that the skills unlock. Computing a budget-optimal allocation is challenging as it requires reasoning about combinatorially many skill plans over a large practice budget. Our key contribution is a bilinear program that can compute this exactly using off-the-shelf solvers. Through simulated and real-world experiments on long-horizon manipulation tasks, we show that our approach allows robots to optimally use limited practice time to acquire useful policies and improve long-horizon planning.

Reinforcement learning improves autonomous-driving vision-language-action (VLA) models by evaluating trajectories sampled from the current policy. Group relative policy optimization (GRPO) learns from reward differences within each rollout group. When all sampled trajectories are poor, this relative signal can rank failures without identifying behavior outside the failed region. We introduce FIRE-VLA, a failure-informed self-evolution framework that converts such unresolved failures into privileged supervision for the next policy. Low-reward, low-diversity groups trigger self-distillation from a frozen round-start copy of the same model. Teacher and student have the same parameter scale, but only the teacher observes the hidden future trajectory. Supervision follows the student's generated prefix and is restricted to answer tokens, while GRPO remains active for every group. The updated policy supplies the teacher for the next round, allowing the routed failure distribution to change with the policy without requiring a larger external teacher. Starting from the same Qwen2.5-VL-3B SFT checkpoint, the comparison matches student rollout and policy-update counts. On 6,019 examples from 150 held-out nuScenes scenes, FIRE-VLA retains comparable single-sample planning, reduces G=4 mean L2 from 1.848 to 1.500 m, and lowers evaluation-persistent failure prevalence from 13.03% to 11.20%. The reduction in mean error arises mainly from rare severe rollouts rather than uniform improvement across ordinary trajectories.

6G networks will not be serving as communication infrastructures only; rather, they are expected to evolve into intelligent systems, where thousands of autonomous artificial intelligence (AI) agents are interconnected. The agents are deployed across a wide range of platforms including low Earth orbit (LEO) satellites, high-altitude platforms (HAPs), unmanned aerial vehicles (UAVs), edge servers, and terrestrial devices. These agents continuously observe their environment and exchange information. Semantic communication provides an efficient mechanism for exchanging meaningful information instead of raw data. However, its effectiveness depends on the communicating agents having sufficiently aligned beliefs to correctly interpret and decode the transmitted messages. This assumption becomes difficult to satisfy in the 6G network where heterogeneous AI models operate under diverse computational constraints and continuously acquire different knowledge from their local environments. This article presents a heterogeneity-aware belief synchronization framework for 6G AI-native networks. It uses latent translation models deployed on multi-access edge computing (MEC) servers. These models translate belief updates from one agent to agent-specific knowledge without requiring joint training and a homogeneous architecture of models. By exchanging compact belief updates through a latent translation model only when necessary, the framework preserves privacy, reduces synchronization cost, and minimizes local knowledge drift. We validate the framework through a case study on a multi-layered terrestrial/non-terrestrial network. Results demonstrate that it maintains low synchronization cost, measured by the number of parameters transmitted, and low belief alignment error across the heterogeneous agents in the case study.

Dexterous manipulation promises substantially richer robot interaction with the physical world, but learning these behaviours remains constrained by the difficulty of collecting consistent, complete-task demonstrations. Unlike parallel-jaw manipulation, dexterous tasks require the operator to coordinate arm motion with precise, contact-rich finger behaviour throughout the task. We introduce NestDex, a nested policy-learning framework that reduces this burden by using learned hand skills to assist demonstration collection. The operator controls the arm and regulates the active hand skill through a single-DoF clutch, rather than directly specifying the full finger trajectory. The inner hand policy adapts its motion from the latest proprioceptive history, while a vision-language selector activates the appropriate skill for each task stage. The resulting demonstrations train a separate outer visuomotor policy that controls both the arm and hand without the inner policies at deployment. A hand-action variational autoencoder provides compact hand-action targets while retaining arm commands in joint space. Across real-world dexterous manipulation experiments, NestDex improves demonstration reliability and efficiency, and the resulting empirical evaluations support effective autonomous policy learning. Video Demo are available at project website https://aus.bot/research/nestdex.

LLM-based agents increasingly rely on external memory to support long-horizon reasoning and interaction. However, the main bottleneck is not simply storing past experience, but recovering the right set of evidence when relevant information is distributed across many interactions. Existing approaches struggle with this access problem. Full-context methods require noisy long-context search, flat retrieval often returns isolated and incomplete records, and graph-based memory systems can be expensive to construct while compressing rich event context. We introduce RippleMem, a long-term memory system that replaces one-shot retrieval with adaptive associative recollection. Inspired by cue-dependent episodic retrieval and associative completion, RippleMem stores interaction history as cue-rich episodic memory units and organizes them in an event-centric memory graph. Given a query, it first recalls relevant memory anchors through hybrid cues, then expands from these anchors along semantic and structural associations to recover missing supporting evidence. In this way, initially recalled memories serve not only as answer context, but also as cues for completing the evidence needed to answer. Experiments on LoCoMo and LongMemEval-S show that RippleMem achieves the best overall performance across evaluated settings, improving LLM-as-a-Judge accuracy by 3.95% on LoCoMo and up to 11.87% on LongMemEval-S, while reducing graph construction cost by about 30x.

LLM benchmark scores can be precise even when the observation protocol does not identify the behavioral property they are intended to measure. In a controlled, solver-grounded setting, we formalize a protocol-level identifiability audit over a finite behavioral policy class: given policies H, observation support O, and estimand $τ$, we test whether O separates every pair with different $τ$. The audit requires zero model calls and resolves our diagnostic case: base-only observation collapses seven frozen deterministic policies into one equivalence class; full support yields seven classes and no cross-estimand collisions; every leave-one-out support retains a constructive collision witness. Empirically, both constrained-generation variants have pair-validity 1.0, yet base accuracy and selective-response fidelity diverge - 0.620 versus 0.324 across six balanced oracle-transition directions (cluster-bootstrap 95% CI [0.600, 0.642] vs. [0.304, 0.345]) - and the gap recurs on a second deterministic source (0.646 vs. 0.331). The audit also synthesizes a minimum identifying support $O^*$ for the frozen policy class: two cells instead of the full 36-cell tensor. This case shows how evaluation-design validity can be checked structurally before model inference and why base correctness does not determine intervention-response fidelity.

We study a large language model (LLM) service in which a provider chooses a per-token price and a default reasoning-token allocation, while a user may accept the default, customize the allocation, or exit. Larger allocations can improve accuracy but increase token cost and latency. We model this interaction as a Stackelberg game and derive the user's unique optimal customized allocation in closed form. For any price, the acceptable defaults form either an empty set or a compact interval. We characterize the provider's optimal default through a three-regime rule, reduce equilibrium computation to a one-dimensional price optimization, and prove the existence of the equilibrium. We further show that defaults affect the implemented reasoning allocation only when users value the convenience of avoiding customization; otherwise, every service-providing outcome implements the user's optimal customized allocation. Experiments with two compact open-weight reasoning models on five mathematics and science benchmarks support the accuracy-token model and show how model and task characteristics determine equilibrium prices, defaults, and reasoning allocations.

Recent advances in teleoperation have enabled robotic manipulators to perform dexterous, human-arm-like motions. However, human operators may fail to avoid suddenly appearing obstacles promptly and effectively, particularly under network latency or limited attention, thereby creating safety risks. To address this issue, we propose a lightweight and modular framework for proactive collision avoidance, operating directly at the end-effector velocity-command level. After preprocessing the point cloud, the framework first predicts potential collisions based on time-to-collision (TTC) with integrated overshoot protection, and subsequently rotates the relative-velocity vector using Rodrigues' rotation formula. The deflection changes only the direction of the relative velocity while preserving its magnitude, thereby mitigating the deadlock problem commonly encountered by conventional artificial potential field (APF) methods. The prediction module compensates for point-cloud processing latency introduced by complex teleoperation pipelines, while the lightweight design enables the high-frequency control required for teleoperation. Simulations across diverse scenarios show that the proposed method achieves a higher end-effector collision avoidance rate than the baseline methods. Experiments on a physical robotic system further validate its collision-avoidance effectiveness.

Existing vision-language model (VLM) benchmarks emphasize perception and reasoning accuracy (how well VLMs describe and reason about what they see in an image), with limited attention to behavioral reliability under uncertainty (how they behave when visual evidence is missing or misleading). We introduce SciFigBench, a diagnostic VLM benchmark for scientific figure understanding that jointly evaluates perception, reasoning, and behavioral reliability under uncertainty. It contains 250 figures with high-quality human annotations across three evaluation aspects, totaling 600+ hours of annotation effort. We further extend these figures via image transformations, reasoning questions, resistance probes, caption-bias probes, and confirmed selective-blur targets, producing over 34,000 evaluation setups for stress testing. We further propose the Admittance-Resistance-Inductance (A-R-I) framework to evaluate whether models acknowledge insufficient evidence, resist misleading context, and infer cautiously from partial information. Our results reveal substantial behavioral differences among models. GPT-5.2 achieves the highest description quality (MQM 91.6) with strong reasoning accuracy (78.4%), yet hallucinates unreadable content in 96% of cases, whereas Gemini 3.1 Pro, a comparably capable model (MQM 90.2, reasoning 81.0%), admits uncertainty in 71% of such cases and achieves the strongest resistance score (0.91). These findings show that high perception and reasoning accuracy alone do not guarantee behavioral reliability, a dimension critical for deployment in scientific workflows.

SiN photonic crystal (PhC) cavities are promising platforms for nonlinear and quantum photonics because of their wide transparency window, CMOS compatibility, and negligible two-photon absorption. However, realizing high-Q/V cavities remains challenging because of the relatively low refractive index of SiN. Here, we employ inverse design to optimize a two-dimensional SiN PhC cavity and experimentally demonstrate a quality factor of approximately 80,000, the highest reported for a near-stoichiometric SiN 2D PhC cavity. Furthermore, both second- and third-harmonic generation are observed from the same cavity, providing experimental evidence of strong optical confinement and large Q/V. Our results establish inverse-designed SiN PhC cavities as a promising platform for nonlinear photonics and future heterogeneous integrated photonic devices.

Multi-round retrieval-augmented generation (RAG) must decide when to stop searching as evidence accumulates. Because the deployed policy is determined by the first STOP on each trajectory, this is a sequential selection problem rather than an independent state-classification task. We adapt S2G-RAG's structured sufficiency-and-gap judgment to a frozen Search-R1 pipeline and train a Qwen3.5-2B judge on 3,009 states from 900 disjoint HotpotQA questions. Search-R1's reasoner, retriever, corpus, prompt, and search budget remain unchanged, while the judge checkpoint and stopping threshold are selected on grouped validation and frozen before confirmatory evaluation. On the confirmatory test set, the resulting policy reduces retrieval calls by 77 (3.70\%) relative to Native Search-R1, while Official Exact Match decreases by 0.625 percentage points. Thus, the trained S2G-style structured judge reduces retrieval while broadly preserving answer accuracy. The result does not imply unchanged or improved accuracy, safe stopping, or lower total inference cost.

Aerial physical interaction requires aerial manipulation platforms to generate large interaction forces and torques while maintaining precise end-effector control. However, conventional underactuated aerial manipulators suffer from strong position-attitude coupling, whereas fully actuated platform designs often face structural complexity, limited payload capacity, and insufficient torque output. This paper presents FAM-DQ, a dual-quadrotor based fully actuated aerial manipulator designed for high-torque physical interaction tasks. By mounting two quadrotor propulsion modules at the ends of a central frame through passive joints, while using a gear-driven servo to regulate the pointing direction, FAM-DQ achieves decoupled $6$-DoF end-effector control with omnidirectional manipulation capability and enhanced torque output. Experiments including trajectory tracking, attitude tracking, static torque measurement, and screw driving validate the proposed design. FAM-DQ achieves a maximum torque of $1.019~\mathrm{N}\cdot\mathrm{m}$ with a total mass of $0.447~\mathrm{kg}$, corresponding to a torque-to-mass ratio of $2.28~\mathrm{N}\cdot\mathrm{m/kg}$.

Squeezed light underpins quantum-enhanced sensing and continuous-variable quantum information processing, and integrated photonics offers a route to producing it at scale. Universal to these applications are squeezed-light generation and measurement. Importantly, quantum measurements serve not only as readout but also as active operations in quantum-state evolution. However, integrating squeezed-light generation and photodetection on the same photonic chip has remained challenging because they impose fundamentally conflicting material requirements: low optical loss to preserve quantum correlations, but efficient photon absorption for photodetection. Here, we demonstrate squeezed-light generation, routing, and balanced homodyne detection integrated on a single photonic chip through heterogeneous integration. A two-mode squeezed quantum microcomb comprising 34 quantum modes is measured with approximately 3 dB squeezing. Our work establishes a scalable architecture for fully integrated squeezed-light quantum photonic systems, unifying quantum-state generation, processing, and detection on a single chip.

As large language models (LLMs) are increasingly applied for emotional companionship, evaluating their behavior and capabilities in intimate relationships has become a pressing issue. However, existing assessments primarily characterize general personality traits, providing limited insight into model behavior within intimate and emotionally sensitive contexts. Therefore, we introduce adult attachment theory into LLM evaluation and use the Experiences in Close Relationships-Revised (ECR-R) scale to characterize attachment anxiety and avoidance. To evaluate emotional companionship capabilities of LLMs in realistic interaction scenarios, we present an emotional companionship benchmark, ECBench, spanning four scenarios including emotional support, collaborative tasks, conflict resolution, and social guidance, across friendship and romantic relationships. ECBench is utilized to assess model behavior using 11 dialogue-quality metrics and three evaluation methods. We evaluate the attachment tendencies of 32 LLMs and select representative models to investigate how these tendencies manifest in contextualized multi-turn interactions and whether they can be shaped through prompting. Our study provides a theoretical lens from psychology, along with practical tools to understand and select LLMs for emotional companionship.

When visual evidence is occluded or chaotic, models should abstain. In this paper, we show that Vision-Language Models (VLMs) can internally distinguish when abstention is required, but fail to express it anyway. We introduce TRAPSBench, a procedurally generated video benchmark of 1,404 matched physics pairs in which a single targeted change renders the outcome undeterminable from the visual evidence. Furthermore, we introduce Penalized Epistemic Calibration Score (PECS), a new robust metric that requires models to both answer correctly when the outcome is knowable, and abstain when the outcome is not. Across 16 VLMs spanning five families, spontaneous restraint is poor: the best PECS is 0.292. The bottleneck is expression, not perception: linear probes decode answerability from hidden states at up to 0.91 AUROC across physics domains; steering a single-layer void direction causally induces or suppresses abstention. Our results replicate across three open-weight families (Qwen, Gemma, LLaVA). The failure is also more pronounced in visual than textual uncertainty: models detect textual impossibility about 4x more readily than missing visual evidence. Closing this representation--output gap likely requires output-stage interventions.

Multilingual retrieval-augmented generation (mRAG) equips large language models with access to globally distributed external knowledge for complex multilingual question answering. Recent approaches either translate retrieved documents into English or the query language to bridge the cross-lingual semantic gap, or decompose a complex query into sub-questions and aggregate the intermediate reasoning process. However, both lines of work suffer from two limitations. First, one-size-fits-all translation alignment, blanket translation discards culturally and linguistically native information unique to the target language, introduces translation noise, and inflates system cost. Second, greedy decomposition and aggregation, uncontrolled decomposition produces redundant sub-questions that compound errors during step-wise reasoning, and the final aggregation over reasoning paths further amplifies these errors. We address both with our method Syfer, a synthesizer-folding framework for multilingual multi-hop question answering that defers translation rather than applying it by default. Syfer first invokes a format-constrained decomposer to produce a sub-question graph in the original language, followed by a decomposition-quality check; when the check passes, sub-questions are answered sequentially under a retrieve-then-answer policy in the target language, and the English translation pathway with bilingual sub-question graph alignment is activated only when the check fails. Experiments across multiple languages show that Syfer attains competitive accuracy while striking a favourable balance between performance and computational cost.

With the rapid advancement of large language models (LLMs), research idea generation has attracted increasing attention. Existing approaches enable LLMs to retrieve relevant literature and propose novel ideas for research areas. However, current evaluation practices for idea generation remain fragmented and lack objective standards, often relying on direct LLM scoring, which limits their ability to provide unified and reliable assessments across a coherent distribution of generated ideas. To address this challenge, we propose LigBench, an automated evaluation benchmark that enables fine-grained and reliable evaluation of AI research ideas, consistently applicable across different generation distributions. In addition, we introduce PAIR-IQ, a dataset tailored for training pairwise idea judgment models and serving as an auxiliary reference to support more objective comparative evaluation. Extensive experiments demonstrate that LigBench achieves stable and interpretable evaluations, significantly improving alignment with expert judgments. Furthermore, models trained on PAIR-IQ exhibit enhanced ranking accuracy and robustness, establishing a principled standard for scalable and objective research idea assessment.

Long-horizon surgical robot manipulation is challenging because task rewards are sparse, while meaningful interaction changes occur at irregular intervals. Existing world-model agents typically imagine at primitive-step resolution, leaving variable-duration task progress implicit. Manually specified stages can provide intermediate structure, but their task specific boundaries are difficult to align with state-dependent interaction transitions. We propose S2-HWM, a Sparse Event-Structured Hierarchical World Model that learns sparse event evidence from primitive latent trajectories to coordinate an event-level manager and a primitive-step worker. The event evidence schedules manager goal updates, and each selected latent goal conditions the worker's primitive actions until the next update. The learned event evidence also forms variable-duration segments for an Event Transition Model (ETM), which predicts the next?boundary stochastic state, segment duration, and accumulated segment reward. Chaining these event-level predictions provides a variable-duration continuation beyond the primitive imagination horizon for manager learning, while the worker retains primitive-step actor-critic learning. On a SurRoL-based PegTransfer task, S2-HWM achieves a success rate of 98.7%, outperforming the flat GAS DreamerV3 baseline by 22.7 percentage points.

Training and evaluating spatial reasoning in embodied agents requires diverse environments that are both geometrically faithful and semantically queryable. Synthetic simulators offer ground truth semantics but sacrifice realism; simulators based on reconstructions of real-world environments have realistic appearance but lack ground truth semantics by default. We propose using Semantic Radiance Fields (SRF) as simulators for spatial reasoning agents. SRFs are a representation that unifies these requirements by lifting 2D semantic segmentations from pretrained vision models into a 3D radiance field that jointly encodes geometry, appearance, and per-class semantic identity. The resulting fields are reconstructed from posed RGB captures of real scenes and support novel-view synthesis, semantic and free-space queries within a single grounded representation. This enables the efficient generation of diverse real-world environments to train and evaluate spatial reasoning models. As an example application, we outline an SRF-driven simulator for an orchard apple-reaching task, in which the radiance field supplies camera rendering, semantic ground truth, and occupancy queries to a physics engine.

In expert-parallel (EP) MoE serving, every layer synchronizes at the slowest GPU. Dispatchers balance token counts (EPLB, LPLB, UltraEP) or activated-expert counts (METRO), assuming expert time is linear in one. Measurements on two datacenter GPU generations show it is neither: below $n^* \approx 156$--$168$ tokens, HBM weight streaming dominates---cost attaches to $activated replicas$, not tokens; above it, grouped GEMM rounds tokens to 128-tile $M$-tiles, so $splitting$ an expert adds padded compute. A max-affine profile $t=\max(a+bG,\,c+βN)$ captures both regimes. Realistic decode batches hold hot experts in the linear regime and cold in the flat $simultaneously$; recorded batches show proxy dispatches differ by $1.4$--$1.6\times$ in modeled block time (p95 up to $1.7\times$), and $which$ proxy wins flips with the regime. We formalize per-batch dispatch as a fixed-charge makespan problem---NP-hard on two fully replicated GPUs, polynomial in degenerate limits---and present TEMPO, a makespan-aware dispatcher solving it in milliseconds off the critical path; its SGLang integration runs out-of-process and fuses dispatch with count collection into one in-graph kernel. Anchored by an 8-GPU Testbed A microbenchmark, TEMPO stays within $1\%$ of the best fixed baseline everywhere and wins by up to $15.5\%$ where regimes mix. End-to-end on Testbed B, Qwen3-235B (inside the win region) gains $4$--$6\%$ throughput and cuts p99 latency by $\sim 15.6\%$; DeepSeek-V3 (outside, communication-dominated) shows only mechanism cost. A phase diagram, not a universal win, is the claim: it predicts both outcomes before deployment.

Large-scale manipulation data is essential for robot learning, yet collecting robot demonstrations remains expensive and difficult to scale. Meanwhile, abundant egocentric human manipulation videos provide rich behavioral experiences, but transferring them across embodiments remains challenging due to differences between human hands and robotic end-effectors. Recent advances in video world models offer a promising pathway to synthesize robot-centric manipulation videos from human observations, while their cross-embodiment transfer capability remains largely unexplored. Therefore, we introduce H2R-Bench, a benchmark for evaluating cross-embodiment human-to-robot manipulation video generation, where models transform egocentric human demonstrations into robot manipulation videos under specified embodiments. Each benchmark instance contains a human demonstration video, target embodiment constraints, and source-grounded annotations covering task goals, action events, functional contacts, and object responses. H2R-Bench evaluates generated videos through five dimensions, including goal-state completion, action-event completion, functional contact transfer, embodiment correctness, and general video quality. We benchmark eleven state-of-the-art video generation models across six manipulation families and two robot embodiments. Our evaluation reveals that current video world models remain limited in human-to-robot manipulation transfer: even leading models often fail in embodiment consistency, functional interaction, and task execution. H2R-Bench provides a systematic diagnostic framework for evaluating whether video world models can bridge the human-to-robot embodiment gap and convert human manipulation observations into robot-centric training resources.

Enabling agents to learn from experience and internalize it into their policy has become a central problem in self-evolving AI. On-policy self-distillation (OPSD) offers an effective pathway by using a privileged self-teacher to provide dense supervision on the student's own trajectories; however, existing methods still rely heavily on designer-specified privileged artifacts (e.g., answers, feedback, skills, or trajectories), limiting the end-to-end learnability and scalability required for continual self-improvement. In this work, we introduce Latent On-Policy Self-Distillation (LOPD), which, rather than proposing another hand-crafted OPSD variant with a newly prescribed form of privileged context, makes the teacher's privileged context itself learnable end-to-end from experience. Technically, LOPD retrieves relevant experiences and composes them into continuous latent tokens that condition a self-teacher, while the student generates trajectories from the task and interaction history and receives dense token-level supervision at every visited prefix. We further introduce a privileged-margin objective to stabilize and regulate the learning of latent context. Empirically, LOPD demonstrates (I) strong performance, outperforming RLVR and representative OPSD methods including OPSD, SDPO, and Skill-SD across both agentic tool use and code generation; and (II) high learning efficiency, surpassing GRPO and Skill-SD with less than 30% of their rollout budget. Ablation studies further provide direct evidence that making privileged context learnable is necessary for realizing these gains. Together, these results position LOPD as a step toward a more scalable and self-directed paradigm for agent evolution.

The emerging Internet of Agents enables LLM-powered agents to discover peers, invoke tools, and delegate tasks across organizational boundaries. Existing protocols increasingly define how agents exchange messages, but not how an agent proves its identity, authorization, advertised capabilities, or accountability after delegation. We present InterSAGE, a trust-native protocol suite that supplies this missing security substrate alongside, rather than in place of, communication protocols. InterSAGE comprises four layers: Persistent Identity, Discovery, Trust Negotiation, and Accountability. Its four core primitives are: (1) Agent Identity Cards that bind developer, code package, operator, and deployment context; (2) capability-aware discovery using DID-bound Verifiable Credential manifests; (3) trust negotiation combining monotonic capability attenuation with two-tier access control; and (4) kernel-mediated cryptographic audit trails that bind usage, delegation, and execution traces to agent identity without a consensus ledger. InterSAGE is designed to complement MCP, A2A, ANP, and AG-UI, allowing communication protocols to evolve independently while keeping trust semantics explicit, portable, and verifiable. We compare InterSAGE with more than 50 efforts spanning agent protocols, decentralized identity, OAuth/OIDC extensions, zero-trust governance, delegation, and audit architectures. We show that no prior architecture jointly enforces persistent identity, capability-aware discovery, trust negotiation, and accountability as a unified four-layer trust substrate for secure agent interoperability.

Human-to-robot (H2R) object handover is a fundamental capability for human-robot collaboration, yet progress is hindered by the scarcity of large-scale, human-centric datasets and the significant sim-to-real gap. To address these challenges, we introduce Hand2Bot, an RGB-D video dataset that provides rich contextual information such as body posture and facial expressions, specifically collected for handover scenarios with real-world noise patterns. We further propose PassGen, a generative pipeline that leverages stable video diffusion and an Intention-Aware Temporal Face Encoder to synthesize realistic handover sequences while ensuring hand-object consistency. To bridge the sim-to-real gap, we implement a morphology-based depth editing strategy that replicates realistic sensor noise found in physical depth maps. Experimental evaluations demonstrate that our framework achieves high intention identification accuracy and low false trigger rates in both ablation studies and real-world deployment on a physical robot platform. Our results confirm that training on PassGen allows for robust zero-shot transfer and earlier intention anticipation compared to traditional hand-centric baselines, effectively enabling socially aware robotic behavior in shared workspaces.

Outcome-driven reinforcement learning offers a scalable way to post-train vision-language-action (VLA) policies from sparse task-success feedback. In common GRPO-based VLA post-training, one rollout-level advantage is applied to every action in the trajectory. A rollout that completes several valid stages but fails later can therefore penalize the actions that produced its earlier progress. We call this trajectory-level credit aliasing. Temporal GRPO addresses this problem by constructing detectable task stages, aligning each rollout with stage-specific action intervals, and comparing only rollouts that have entered the same stage. The resulting stage advantages are applied to their corresponding intervals in a single policy update. On RoboTwin 2.0, Temporal GRPO improves task success and sample efficiency, with consistent gains across task horizons. Controlled updates on LIBERO-Long preserve shared prerequisite stages and concentrate improvement at the first stage where rollout outcomes diverge.

Understanding hand-object interaction from egocentric vision is essential for modeling how people physically engage with the surrounding world. Yet reasoning about physically grounded interaction requires estimating the forces acting on hands and objects, beyond localizing contact. We present EgoPHI, the first method that jointly estimates dense contact maps and 3D force distributions on hand and object meshes from a single monocular RGB image and object geometry. To address the lack of scalable ground-truth force annotations, we introduce a physics-based simulation pipeline that augments existing hand-object datasets with dense per-vertex force supervision. EgoPHI then learns dense 3D contact and force on interacting hand and articulated object meshes, extending vision-based force estimation beyond image-space or planar settings. Our evaluation on in-distribution and out-of-distribution benchmarks shows that EgoPHI improves force estimation over existing approaches while generalizing to unseen datasets. To evaluate sim-to-real transfer, we constructed two physical objects that capture dense object contact and force magnitude and used them to record a dataset of interactions from eight participants across diverse touch and grasp types. Our results demonstrate that EgoPHI recovers meaningful 3D contact and force distributions in simulated, out-of-distribution, and real-world settings, advancing egocentric hand-object understanding from contact localization toward physically grounded interaction reasoning.

Retrieval-augmented generation treats an external corpus as inference evidence, allowing injected documents to promote attacker-chosen claims. Existing detectors depend on trusted references, specific attack artifacts, or global thresholds sensitive to corpus topology. We present RAGSieve, a self-referenced detection framework that constructs its reference from the inspected system. RAGSieve-Query (RSQ) performs query-local contrast, scoring top-five candidates against ranks 6-20 of the same retrieval to detect answer-anchor concentration and carrier transitions. RAGSieve-Graph (RSG) performs corpus-local contrast, comparing each document's semantically similar but lexically distinct neighbors with its local baseline to detect coordinated density before queries arrive. Across three QA datasets and six poisoning constructions, RSQ achieves 95.2% AUROC and detects 82.2% of poison at 5% clean-document removal, versus 81.1%/52.5% for GMTP. RSG achieves 93.3%/79.8%, versus 79.4%/37.6% for CleanBase. Joint deployment reduces attack success from 67.4% to 14.0% while retaining 41.3% F1 on unpoisoned retrieval, demonstrating practical protection at both corpus ingestion and query time without poison labels or trusted corpora. Source code is available at https://github.com/XrazyMee/RAGSieve.

Knowledge-base question answering (KBQA) systems rely on effective retrieval and reasoning mechanisms to generate accurate answers from external knowledge sources. However, developing reliable KBQA systems for low-resource languages such as Bangla remains challenging due to limited retrieval-focused research, scarce language resources, and difficulties in grounding generated responses in external knowledge. In this work, we propose HybridRAG-BN, a retrieval-augmented framework for Bangla KBQA that integrates hybrid retrieval using BM25 and BGE-M3, answer generation using the GGUF version of Gemma-4-31B-Instruct, and a LoRA-fine-tuned Gemma-4-31B-Instruct model for answer verification and refinement. To further improve robustness, the framework incorporates a post-processing stage that addresses unresolved cases through fallback answer replacement and DuckDuckGo-assisted retrieval. Experimental results demonstrate the effectiveness of the proposed framework, achieving token-level F1 scores of 0.71654 and 0.72912 on the public and private leaderboards, respectively, securing first place in the competition.

Heterogeneous USV cooperative pursuit in constrained port waterways requires evader interception under navigation, traffic, and role constraints. This paper proposes OGR-MARL, an option-guided residual multi-agent reinforcement learning framework that is decoupled from a specific MARL algorithm. OGR-MARL integrates shared evader belief, role-conditioned option targets, adaptive rule penalties, and residual policy learning, allowing different MARL algorithms to learn corrective actions on top of rule-guided behaviors rather than exploring constrained port environments from scratch. We instantiate OGR-MARL with representative continuous-control MARL backbones, including MADDPG, MATD3, MAPPO, and MASAC, yielding OGR-MADDPG, OGR-MATD3, OGR-MAPPO, and OGR-MASAC. Experiments in an abstract Xiazhimen port-waterway scenario show that the OGR-MASAC instantiation achieves a 75.0% capture rate, promising mission-effective rule compliance, and the best heterogeneous coordination among the tested methods. Without retraining, zero-shot transfer to a QGIS/AIS-informed Xiazhimen map achieves promising results, demonstrating the generalization potential of OGR-MARL in more complex port scenarios.

Long-horizon LLM agents must preserve information from past interactions to support future tasks. Existing memory systems typically rely on eager consolidation, invoking LLMs after each interaction to extract, summarize, or update memories. This design makes memory construction increasingly costly as conversations grow. Coarse summarization can reduce construction cost but risks discarding fine-grained contextual evidence, whereas larger retrieval contexts or multi-hop LLM reasoning shift the overhead to query time. We present LycheeMemory V2, an efficient long-term memory framework that replaces turn-level consolidation with semantic segment-level consolidation. Instead of consolidating every interaction, LycheeMemory batches multiple exchanges into segments and encodes each finalized segment into context-independent typed memory records. Segment-level batching lowers LLM encoding frequency, while semantic boundary detection helps preserve coherent event-level and temporal evidence compared with fixed-window batching. The resulting records are organized with lightweight structured indexes for query-planned evidence retrieval. Experiments using GPT-4.1-Mini show that LycheeMemory achieves state-of-the-art performance, reaching 89.22% on LoCoMo and 92.20% on LongMemEval-S. Compared with A-Mem, it reduces construction tokens by 86.0% on LoCoMo and 75.9% on LongMemEval-S without increasing query-time token usage. More broadly, our results suggest that the accuracy--cost trade-off of long-term agent memory depends not only on what information is retained, but also on the granularity at which it is consolidated.

Long-form research reports generated by large language models drift, contradict themselves, and lose provenance: the same metric appears with different values, and rumor is quoted as confidently as an audited filing. We present a two-tier agentic system that separates a maintained, point-in-time knowledge library from report writing. A deterministic "librarian" ingests timestamped sources into a trust-tiered ontology, layering evidence cards, an authoritative metric ledger, and a claim graph into an always-current source of truth, not per-query RAG over raw chunks. A portable multi-agent "writer" runtime then composes a contradiction-free, evidence-grounded report at any knowledge cutoff T, reading only evidence with as_of <= T (no look-ahead); red-team verdicts flow back into the librarian. We evaluate on a self-collected, public corpus of 6,130 sources yielding 555,926 evidence cards (SEC EDGAR filings across 295 issuers and 11 sectors, U.S. Bureau of Labor Statistics releases, and Wikipedia). From the one library we compose four point-in-time reports on distinct theses and run eight reproducible experiments, whose headline metrics come from a deterministic quality-control gate, itself validated by defect-injection meta-evaluation at recall 1.0 and precision 1.0. A shared metric ledger removes 6,845 cross-section contradictions to zero. Tier-first selection is correct on 22/22 gold cases where a popularity-first baseline scores only 9/22; trust tiering leaks zero media-sourced numbers, and no government statistic displaces a company's own filing. A red-team refutation propagates back and self-corrects a later run with zero manual edits. Replay exhibits zero look-ahead violations across seven cutoffs while the library grows from 235,373 to 555,312 cards. Difficulty-tiered model routing exceeds the all-Opus quality ceiling while running 3.7x faster than serial.

Fast and wide-range wavelength switching is desirable for optical communications and photonic systems that are frequency-agile. However, thermo-optic (TO)-tuned integrated lasers often have limited switching times and tuning rates. This study demonstrates a hybrid-integrated III-V/Si3N4 external-cavity laser (ECL), combining a dual-microring Vernier filter with thermal pumping to give wide-range and fast wavelength control. The ECL provides single-mode static lasing wavelength tuning in the 1486-1614 nm range. Impulsive thermal pumping that is applied through microheaters with shorter duration and higher amplitude accelerates the switching time. A simple first-order thermal fit reproduces the measurements well, indicating that the TO-tuning dynamics are highly predictable. Consequently, two-step pulse thermal pumping is applied to the on-chip microheaters to exploit the initial quasi-linear heating transient and sustain the target wavelength at a subsequent equilibrium. The results show that 101 and 104 nm red- and blue-shift switches are achieved with quasi-linear tuning rates of 8.91 and 9.68 nm/us, respectively. This approach provides a practical route toward fast wavelength switching in TO-tuned ECLs, potentially extending their applicability within frequency-agile systems, such as wavelength-division multiplexed transceivers.

The performance of large language model (LLM)-based multi-agent systems (MAS) largely depends on effective communication topologies. Existing topology generation methods, however, typically learn communication topologies through black-box optimization driven solely by task-level rewards. While effective, such optimization provides little insight into why particular communication edges are selected, making it difficult to identify the critical communication subgraphs responsible for successful collaboration. To address this limitation, we propose E2-Explainer, a model-agnostic framework for providing interpretable explanations of communication topologies produced by arbitrary topology generators. Specifically, we formulate topology explanation as a causal attribution problem that identifies compact communication subgraphs supported by edge-level evidence of task preservation. We obtain this evidence with a Granger-style objective that measures how masking each communication channel changes the task outcome and the stability of the final response. The resulting budgeted subgraphs are then distilled into an amortized explainer, enabling efficient post-hoc explanation without repeated edge-level evaluations at deployment. Extensive experiments on multiple reasoning and coding benchmarks demonstrate that E2-Explainer identifies critical communication subgraphs that preserve successful collaboration. These subgraphs can also be executed directly to prune redundant communication edges, substantially reducing communication costs while maintaining competitive task performance.

Compositional reliability bounds for multi-agent systems multiply component reliabilities, a step licensed by a conditional-independence assumption that is routinely stated and rarely tested. We test it. Two instances of one model, in a two-agent handoff, co-fail on 90.0% of the missions on which either fails (log OR 6.66, 95% CI [6.38, 7.00]; phi 0.916), in a preregistered evaluation of 18,000 missions scored by deterministic code with no LLM judge. Substituting a different model reduces the association in six of six contrasts; substituting a different vendor, model already different, does not -- a registered hypothesis reported as a null. The error is signed and runs against the operator: positive dependence inflates joint failure above the independence product, so redundancy is over-credited exactly when components share a model. The assumption-free alternative is often vacuous, and fitting a dependence model is worse: we prove a bootstrap bound on a fitted model's functional loses coverage of the truth as n grows, the identification gap being O(1) while the bootstrap haircut is O(n^{-1/2}). More data makes such a certificate worse, with no visible symptom. We give a finite-sample certificate assuming no dependence structure: a linear program over the joint, over a Bonferroni-Clopper-Pearson box around measured co-execution moments. It is sound, sharp for the information supplied, and monotone in the moment family. Enriching ten moment functionals to fourteen narrows the identified interval by 85.7% and lifts the certified floor from 0.2455 to 0.4116. A companion anytime-valid certificate holds type-I error at 0.0471 under optional stopping. Common dependence statistics are marginal-bounded and can reverse an apparent ordering of conditions when the compared agents fail at different rates. Contracts, scoring code, analysis scripts, and the preregistration are released.

Cultural evaluation of large language models (LLMs) often focuses on high-resource standard languages, leaving regional culture and dialect communities underrepresented. We introduce BavGround, a benchmark for evaluating Bavarian regional cultural grounding and dialect competence across English, German and Bavarian. BavGround contains 206 multiple-choice source questions across eight cultural domains per language, yielding 618 multi-parallel instances, with items covering both broadly accessible cultural knowledge and source-grounded regional knowledge from journalism, historical sources, and specialist literature. We evaluate fifteen 7B-10B open-weight instruction-tuned models and one closed-model reference. Strong multilingual models perform best overall, but performance drops on Bavarian items and source-grounded questions, indicating persistent difficulty with dialectal and localized cultural knowledge. We further show that conclusions depend strongly on evaluation protocol: raw answer-letter scoring, shuffled-letter scoring, option-text likelihood, generated-answer parsing, and semantic matching can produce different absolute scores and rankings, especially for regionally adapted models. Finally, an exploratory analysis of GENBA-10B checkpoints suggests that continued pretraining improves answer-content likelihood unevenly across domains, while dialect competence remains comparatively weak. BavGround supports localized, protocol-aware evaluation of cultural representation in LLMs.

Agent-memory systems increasingly buy retrieval quality with structure, transforming raw conversation histories into summaries, embeddings, trees, or knowledge graphs before any question is asked. We ask how much of that benefit comes from the structure itself, rather than from competent retrieval over the raw history. We present ReFind, an agent-controlled search interface that builds no semantic structure at all: it leaves the conversation archive unmodified, indexes it lexically at turn granularity, and combines a generic iterative keyword-search loop with four chat-native controls grounded in empirical refinding work: session-aware rank fusion, local context expansion, temporal narrowing, and skipping already-inspected sessions. A separate reasoning stage answers from the collected evidence. Across a broad suite of conversational-memory tasks (single- and multi-hop QA, event ordering, and fact consolidation), roughly 2,800 questions on precise-retrieval and fact-tracking capabilities evaluated under the incremental multi-turn setting of MemoryAgentBench, ReFind attains the highest mean accuracy (58.2) of any system compared, above the strongest graph- and tree-based memory systems (HippoRAG 2, 53.2), all under a GPT-4o-mini backbone matched to every reused baseline. Controlled comparisons to single-shot BM25, a matched generic-agentic BM25 control, component removals, and agentic dense/hybrid variants separately support the roles of agent control, chat-native controls, and lexical retrieval. On LongMemEval-S/M, the same interface reaches 93.2 +/- 3.3 and 89.3 +/- 6.0 with GPT-5-mini. The results indicate that for precise, evidence-grounded questions over chat archives, much of the benefit credited to elaborate memory structures is recoverable by giving an agent controllable search over the unmodified record, with no LLM-based index construction at all.

Should you replace your text-embedding pipeline with a large language model? We answer this with a controlled, cost-aware comparison of ten LLMs across six families and 26 embedding models (118M to 14B parameters) on 37 tasks spanning classification, semantic textual similarity (STS), clustering, pair classification, and retrieval. In aggregate the two paradigms are effectively tied: the best LLM (Gemini 3.1 Pro, 77.6) and the best embedding model (77.2) differ by 0.4 points. Their strengths differ by task: LLMs lead on reasoning-heavy retrieval, embedding models lead on classification, and the two match on clustering, STS, and pair classification. Reaching that parity is expensive. An LLM costs up to 1,431x more than an embedding model of comparable quality (USD 154 vs. USD 0.11 per benchmark pass), and the open LLMs tested process tokens 2.5 to 736x more slowly on the same GPU. Reasoning tokens account for 28 to 81% of LLM inference cost; lower reasoning budgets preserve or improve retrieval quality for most models in our ablation. The Pareto frontier contains the leading embedding models and one LLM, Gemini 3.1 Pro. These results support a division of labour: use embedding models for similarity, classification, and clustering, and reserve LLMs for reasoning-intensive retrieval. Our code, datasets, and results are publicly available at https://github.com/embeddings-benchmark/embedders-dilemma.

Vision-Language Navigation (VLN) for humanoid robots poses challenges existing benchmarks fail to address: bipedal locomotion imposes physical constraints absent from wheeled agents, humanoid morphologies vary across platforms, and egocentric observations are distorted by locomotion-induced camera dynamics. We present HumanoidVLN, a physics-grounded simulator and benchmark for VLN across diverse humanoid embodiments. Built on NVIDIA Isaac Sim, our platform supports an extensible set of humanoid configurations, demonstrated on four robots (Unitree G1, Unitree H1, Internal-A, Internal-B) spanning 10-12 lower-body DoF and heights from 1.17m to 1.80m, via a hierarchical control stack combining a reinforcement learning locomotion policy with interchangeable PD or MPC path trackers. New robots and VLN models integrate with minimal effort; we demonstrate compatibility with NaVILA, DualVLN, StreamVLN, and JanusVLN. Environments are drawn from artist-designed scenes and 3D Gaussian Splatting reconstructions, filtered for navigable areas exceeding 100 square meters. Instructions are generated by a dual generator-reviewer plus paraphraser multi-agent pipeline with human-in-the-loop verification, yielding 933 collision-aware reference episodes, each paired with one fine-grained instruction and three coarse-grained stylistic variants (formal, natural, casual). Across four models and four embodiments, JanusVLN achieves the highest mean success rate of 43.55% and nDTW of 48.38. In a 20-episode sim-to-real pilot with DualVLN and the Unitree G1, navigation errors correlate strongly (r=0.935), with a mean absolute difference of 0.68m and mean trajectory similarity of 0.782 (+/-0.188) nDTW. These results highlight the interaction between VLN models, controllers, and humanoid embodiments under physical execution. Code, benchmark, and data will be released upon acceptance at https://humanoid-vln.github.io/.

Autonomous driving requires planning under both semantic constraints and predictive dynamics. Existing end-to-end driving approaches, however, typically emphasize only one side of this requirement: Vision-Language-Action (VLA) models exploit VLM priors for semantic reasoning, while World Action Models (WAMs) provide future-aware prediction through generative world modeling. This naturally motivates a unified planner that can leverage both semantic priors and predictive dynamics. However, we find that a naive combination through joint token-level attention suffers from an attention-allocation mismatch, where semantic shortcuts dominate the shared attention space and suppress predictive dynamics. Inspired by neuroscience evidence that complex behavior arises from coordination among functionally specialized systems, we propose BrainWAM, a structured action-space coordination framework that converts semantic reasoning and predictive world modeling into two specialized action-oriented pathways, and aligns them at the level of compact action representations. We further introduce an asynchronous rectified-flow inference strategy with decoupled video and action denoising, which shortens inference latency while preserving planning-relevant predictive context. BrainWAM reaches state-of-the-art performance on both NAVSIM v1 (89.5 PDMS) and NAVSIM v2 (89.6 EPDMS), consistently outperforming VLA-only or WAM-only methods, highlighting BrainWAM as a practical and promising direction for autonomous driving systems.

Retrieval can identify a past trajectory that may matter, yet it does not specify how an acting agent should use that trajectory after users, entities, constraints, or environment state have changed. We identify this post-retrieval reuse step as a distinct bottleneck for long-horizon trajectory memory and formulate an evaluation framework that holds candidate retrieval, target state, model, decoding, and tool budget fixed while varying the support delivered to the agent. We instantiate the framework with query-conditioned reuse (QCR), a deliberately simple target-bound note that records a reusable procedure, bindings to recover, applicability conditions, and verification requirements. QCR serves to test the reuse hypothesis rather than to claim a universally preferred memory format. Across 2,391 target instances in WebArena, WorkArena, and AppWorld, QCR reaches 62.3% average Success, 10.7 points above Full Trajectory, while using 48.9% fewer online tokens. Summary reranking selects a reusable memory for 94.8% of targets, placing end-task Success within 1.8 points of an oracle reusable selector. Analyses by trajectory length and source--target binding shift show that direct trajectory injection loses much of its utility as traces grow longer or source-specific values change, whereas target-bound support preserves a larger share of the measured gain. The resulting framework separates retrieval quality from the problem of turning retrieved experience into safe, useful support for a new task.

We study recursive self-improvement at the level of quantitative-investment research: whether an autonomous system can use evidence from earlier experiments to improve the hypotheses and candidates proposed in later iterations. We present AQuA, which comprises two separate language-model-driven research systems: one for symbolic factor discovery and one for trainable model development. The two systems do not share agents, memories, candidate spaces, or research state. Instead, each independently closes its own research loop by retaining validated evidence and using it to guide subsequent proposals. In this bounded sense, both systems implement recursive self-improvement at the level of the research process. Each system also uses its own sealed sandbox, which fixes the data splits, feature and label definitions, and evaluator while allowing the model to act only through constrained factor expressions or configuration diffs. The factor system, a manager-mediated multi-agent pipeline, discovers and combines factors into a signal that reaches a combined information coefficient of about $0.190$ on a crypto universe. The model system, a config-driven loop over a hybrid time-series architecture, reaches a per-stock information coefficient of $+0.0843$ on US equities and converts it into a threshold long/short strategy with a held-out Sharpe of up to $+2.50$ at a two-leg cost. The strategy is positive in every year from 2021 to 2025.

Large language models often fail when answer options require combining atomic judgments under explicit logical operators, even when they judge the individual atoms correctly. We study compound options connected by AND, OR, and NEITHER/NOR, introducing a framework that decomposes each option into atomic answers and scores contrastive hypotheses about each one, so the model never sees a compound option. An operator-constrained integer linear program then composes the calibrated scores into a single prediction. We evaluate on LOGICAL-COMMONSENSEQA and introduce LOGICAL-SATA, a reading-comprehension benchmark derived from SATA-Bench. Our framework improves Macro-F1 from 48.3 to 77.0 on the human-validated LOGICAL-COMMONSENSEQA split and from 47.0 to 75.6 on LOGICAL-SATA, with the largest gains on NEITHER/NOR.

FastThaiG2P provides sub-millisecond Thai grapheme-to-phoneme conversion for text-to-speech pipelines (International Phonetic Alphabet and Kokoro-TTS conventions) using a PyThaiNLP-tokenized, extensible dictionary and normalization rules for common Central Thai speech. The approach achieves an average latency of 0.15 ms per utterance on a benchmark of 27,242 synthetically generated utterances, of which 30\% is spent on tokenization, 12\% on normalization, and 58\% on out-of-vocabulary fallbacks (0.5\% OOV rate). To demonstrate its effectiveness, we used FastThaiG2P to phonemize Som-TTS, an open dataset containing 20 hours of grapheme-and-audio pairs, then trained an 82M-parameter StyleTTS 2 model based on a Kokoro-TTS recipe. The resulting model vocalizes intelligible Thai speech suitable for prototyping and development at 0.25 real-time factor (4x real-time) with ONNX inference on CPU.

Signal temporal logic (STL) provides a formal language for specifying real-time properties of real-valued observations, along with a quantitative robustness score for monitoring satisfaction. Control synthesis from STL specifications is of interest since manual controller design becomes infeasible as real-world systems grow in complexity. Moreover, many modern autonomous and AI-enabled systems lack accurate and complete system models, which makes optimization-based synthesis approaches unsuitable and motivates learning-based control. Prior work uses STL robustness scores as rewards in reinforcement learning (RL) to obtain control policies satisfying given specifications; however, robustness depends on execution history, leading to intractable state space expansion for general long-horizon specifications with arbitrarily nested temporal operators. This work introduces a novel automata-based approach that provides an efficient memory mechanism and associated Markovian rewards suitable for RL frameworks. Our approach constructs a timed alternating automaton from the given STL specifications, augments the state space with automaton locations and clock valuations, and derives rewards from the automaton acceptance condition. We empirically demonstrate that our approach learns policies that achieve higher robustness scores and satisfaction rates than those learned by existing approaches using robustness-based rewards.

Hierarchical open-vocabulary object navigation (OVON) requires agents to follow free-form instructions that may specify targets through scene-, room-, region-, and instance-level cues in unseen environments. Although recent work LangMap has formalized this setting, reliably solving it under partial observations remains challenging: spatial grounding requires persistent environment-level evidence, whereas target verification requires clear and discriminative candidate views. We present SAP-Nav, a fully online, zero-shot framework that addresses both requirements through active perception. SAP-Nav incrementally constructs a Queryable Spatial-Semantic Representation from actively acquired room views, enabling spatial semantic queries from any explored location. It further employs Active Viewpoint Verification to assess whether the current observation provides sufficient evidence and, when necessary, reposition the agent to a more informative viewpoint before verifying candidates against category and attribute constraints. Although designed for hierarchical OVON, SAP-Nav supports both hierarchical and standard category-level OVON without task-specific training or precomputed scene maps. Experiments on LangMap and HM3D-OVON show that SAP-Nav achieves the overall best performance, including a 12.2% improvement in SR over training-based methods on region-level navigation. Real-world robot experiments further demonstrate its practical feasibility. Code will be made publicly available upon acceptance.

Embodied agents must often identify and interact with objects based on their function rather than their identity, requiring them to actively acquire observations that reveal discriminative functional evidence. Existing affordance grounding methods operate from fixed viewpoints and lack mechanisms for deciding where to look when functional cues are occluded or incomplete. We introduce Active Functional Affordance Grounding, a new task in which an agent sequentially explores a scene to identify and spatially ground an object satisfying a functional query. To address this problem, we propose FUSE, an adaptive semantic-geometric evidence acquisition framework that combines explicit uncertainty-driven exploration with a learned amortized planner to efficiently select informative viewpoints. We further introduce a Habitat-based benchmark for evaluating active functional grounding. Experiments show that FUSE achieves the highest observed non-oracle grounding performance while reducing computation by 1.33x relative to fully explicit exploration, and remains effective across multiple affordance knowledge sources.

Strategic reasoning in Large Language Models (LLMs) within long-horizon environments is often limited by inconsistent subgoals. In these settings, finite attention resources prevent the model from maintaining strategic coherence over thousands of steps. This limitation leads to strategic drift, where localized decisions fail to sustain a coherent trajectory across reasoning. To address this, we introduce EpicStar, a framework that enables agents to learn memory as policy to tackle long-horizon reasoning. Specifically, the agent maintains a bank of successful past episodes as a heuristic alongside a working memory to track short-term environmental changes. During inference, a dynamic gating mechanism determines whether to execute a retrieved action directly or to perform new reasoning through a contextual fusion of the retrieved episodes and current working memory. Utilizing StarCraft II as the testbed, we evaluated EpicStar against diverse opponent styles. It significantly outperforms baseline methods, achieving higher win rates while consuming an order of magnitude fewer tokens, and it maintains this advantage consistently across difficulty levels and opponent strategies. Our findings provide compelling evidence that structured cross-episode memory is essential for enabling LLM agents to perform robust, long-term strategic execution in dynamic, autonomous settings.