Searcharxiv⌕ Search

arXiv subjects

Rui Wang

Publications and source records attributed to Rui Wang.

At least 19 recordsLinked to original sources
arXiv2026

LLMCodec: Adapting Video Codecs for Efficient Weight Compression of Large Language Models

The rapid development of large language models(LLMs) has led to remarkable advances in natural language processing. However, the increasing scale of these models introduces substantial challenges in terms of storage, transmission, and deployment. Though great efforts have been devoted to model compression and quantization, existing methods often rely on fine-tuning or calibration data, which exhibit limited generalization across different tensor types. In this paper, we argue that video codecs offer a promising solution for LLM compression, due to their inherent compatibility with matrix structured data, configurable compression strategies, and the availability of highly optimized, off-the-shelf implementations. Therefore, we present LLMCodec, a video codec-based LLM compression method that integrates affine quantization with the recent VVC/H.266 video codec. Beyond VVC, we further compare a range of video codecs and encoding profiles to evaluate their impact on compression performance. Experiments on different models demonstrate the robustness and generality of LLMCodec. Notably, on LLaMA-3-8B at 2-bit precision, LLMCodec reduces perplexity by over 1.5x and improves downstream task accuracy by 21% compared with the existing method.

cs.MM↗
arXiv2026

Behind Video Generation: Observations and Insights on Multimodal Understanding from Kling

Video generation has progressed from isolated clips to coherent short-form audiovisual sequences, and the bottleneck is shifting from rendering to semantic control. This report summarizes the observations and insights on multimodal understanding gained from successive generations of Kling. We argue that generation quality is the projection of multimodal understanding through the entire data--conditioning--alignment--evaluation chain, and we examine this chain module by module. Kling-MLLM, a shared multimodal foundation model that unifies image, video, audio, synchronized audio--video, and interleaved inputs, is specialized for each downstream module. Because a DiT is trained on far fewer training samples than a language model, we treat captioning as a semantic contract that maximizes description density, and extend it to reference-grounded captioning that makes reference bindings explicit. Prompt enhancement translates user intent into this caption distribution: basic PE is derived from captions of compelling videos and refined with text-level and generator-in-the-loop reinforcement learning, while agentic PE refines the story and retrieves compelling examples. We show that a reliable text encoder comes from perceptual training rather than from stronger language ability, and that candidates can be screened from text alone before end-to-end training. Reward modeling pairs a general reward model for explainable, localized critiques with a decision reward model for fast multidimensional scoring, treating perceptual bandwidth as a first-class design parameter. Finally, we review task-specific automatic evaluation, focusing on protocols that expose scoring shortcuts and help trace failures to their upstream source. We hope these lessons offer practical guidance for building the understanding stack behind controllable video generation.

cs.CV↗
arXiv2026

Extended 2.25/8.60 GHz Observations of the Magnetar Swift J1818.0-1607

We present the long-term rotational evolution and radio emission properties of the magnetar Swift~J1818.0$-$1607, using simultaneous 2.25/8.60~GHz observations obtained with the Shanghai Tian Ma Radio Telescope from MJD~58936 to 59427. The average pulse profiles exhibit significant evolution with time and frequency, attributable to relative changes in the leading, central, and trailing emission components. The magnetar generally spins down with an average rate of $-1.65(4)\times10^{-11}\,\mathrm{s^{-2}}$, with the magnitude of the spin-down rate first increasing and then decreasing over the monitoring span. The overall flux density shows an initial rise followed by a decline, with sporadic variability superimposed; the variations at 2.25 and 8.60~GHz are broadly consistent across most epochs. The two-point spectral index increased from $α\simeq-1.8$ shortly after the outburst to near-zero values around MJD~59030--59050, followed by repeated decreases and increases. Despite this variability, 41 of the 54 two-point spectral-index measurements in the later campaign have $α>-1$. Three component-dependent types of mode changing are identified: Type~I, involving primarily intensity changes in the central component; Type~II, involving central and trailing components; and Type~III, involving leading and central components. All three classes include events showing a corresponding component response in the other observing band, with event-dependent response strengths and directions. Type~I variations are generally more pronounced at 8.60~GHz than at 2.25~GHz.

astro-ph.HE↗
arXiv2026

ToolSpace: Scaling Tool-Centric Environment Synthesis for Long-Horizon Agents

Large language models are deployed as long-horizon agents that compose diverse tools across MCP services into executable workflows. Existing MCP environment-scaling methods are data-centric: they first synthesize a database and scenario, then derive tools and tasks from them. As a result, tool interfaces may not match real ones, tool compositions are not grounded in execution, and tasks rarely span more than one environment. We propose ToolSpace, a tool-centric framework that reverses this order. ToolSpace mines data, state, and retry dependencies from adjacent calls in execution traces of real MCP services and open agent corpora, using deterministic rules rather than LLM judgment, and clusters the resulting weighted tool-composition graph into environments. Databases are synthesized only for state-mutating environments, while query and stateless tools replay recorded returns. ToolSpace then performs state-aware random walks with rollback and decompiles each chain into a natural-language task with assertion-based checks, keeping only samples that replay successfully from the initial state. Applied to AutomationBench and tau3-Bench Banking, the pipeline produces harder Pro versions on which 7 frontier models drop by up to 38.2 points. Across seven long-horizon tool-use benchmarks, ToolSpace raises Qwen3.6-35B-A3B from 23.0 to 41.0 average score (43.2 with RL) and GLM-5.2 from 44.6 to 56.6, on par with Claude Opus 5. Ablations show that execution-witnessed edges clearly outperform LLM-judged ones, and performance keeps improving with data and environment scale.

cs.CL↗
arXiv2026

Who pays for urban environment management? The role of the third-party service provider

The dominant contemporary paradigm of direct state-led urban environmental regulation entails significant risks, including high administrative burden, frequent conflicts with polluting entities, and economic vulnerabilities for those sources. To address these deficiencies, this paper integrates third-party service provider (TPSP) into China's existing management framework as a beneficial extension beyond conventional government intervention. Utilizing an evolutionary game approach, we compare two distinct institutional settings, a non-integration management mode and an integration management mode, and scrutinize the evolutionary dynamics across both two- and three-dimensional systems. To enhance system robustness, we embed a strategic behavior-based dynamic punishment mechanism that effectively eliminates the perturbing effects of varied penalty rules, thereby steering all stakeholders toward an ideal stable state. This comparative institutional analysis isolates the marginal contribution of third-party intermediation, specifies when outsourcing environmental services outperforms direct government enforcement, and reframes evolutionary game theory as a design device for institutional self-correction. Furthermore, a predefined variable importance analysis is conducted to highlight critical factors warranting attention in practical policy implementation. Extensive simulation experiments, carried out under diverse system-level and parameter-level conditions, validate the theoretical properties of our model. The findings offer concrete managerial insights to inform the formulation of more effective urban environmental supervision regulations.

math.OC↗
arXiv2026

CoworkOS:Scaling Agent Coworkers for Real-world Automation

Rapid advances in cowork agents are shifting human-AI collaboration toward real-world automation, delivering software, designs, and other artifacts. However, such productivity work demands both high-quality delivery on complex productivity tasks and continued monitoring and decision-making over long horizons, a combination that remains an open challenge. We introduce CoworkOS, a general-purpose harness for long-horizon work that separates task strategy from execution management: the model decides actions, delegation, and acceptance, while the runtime maintains state and feedback. Its three-layer architecture enables direct execution or delegation to Subagents, synchronously or asynchronously, under stable responsibilities. Through a shared application layer, humans and agents inspect and revise the same evolving documents, presentations, CAD models, and Blender scenes, so human-in-the-loop participation can enter both through conversation and through direct modification of the work itself. To sustain continuity across agents and sessions, CoworkOS further introduces hierarchical workspaces and persistent semantic memory, so working files, key decisions, and long-running Job logs remain shareable, reusable, and resumable after interruption. Across 12 benchmarks and 6 categories of real-world applications, CoworkOS proves highly competitive with frontier agent applications such as Codex and DeepSeek Harness. As a sustained-work case study, CoworkOS uses GLM-5.3 as its backbone to conduct RSI-driven automated post-training on Qwen3.6-35B-A3B. Over five rounds, the model's average score across eight benchmarks rose from 33.19 to 42.95 (+9.76). Looking ahead, we believe CoworkOS opens a meaningful path toward general cowork-agent intelligence.

cs.AI↗
arXiv2026

Denoising Is Stein: Moment Access and a Resource Hierarchy for Learning Hidden Structure in Diffusion Models

We study score learning from denoising pairs when the data contain a hidden non-Gaussian component. If covariance reveals the component, its subspace can be recovered directly. If covariance is isotropic, the first distinguishing moment has order $k^\ast$, and the learner must extract that moment from clean points and their noisy views. Correlations in the denoising loss are Stein discrepancies against a baseline; for polynomial fields they determine exactly the corresponding moment differences. Averaging fresh noisy views conditional on a clean point gives an exact moment query on the clean distribution. In the covariance-matched regime, orientation-independent dictionaries satisfying an $L^4/L^2$ condition need order $d^{k^\ast-1}$ features, and the stated dot-product kernel ridge models need the same power in samples. Small-step online SGD on one unit has this exponent up to logarithms. A two-stage learner uses the curvature of a smoothed denoising loss to recover the hidden direction, then sharpens it and fits a scalar readout. For every even $k^\ast$, it obtains a fixed score gain over the Gaussian baseline from $O(d^{k^\ast/2}\log d)$ clean points. In the covariance-visible regime, the frozen-feature threshold occurs at $d^J/J!$, while projection onto principal components makes the fitted-model, score-risk and generated-law distributions independent of ambient dimension. In the covariance-matched experiments, the measured partial-trace overlaps are consistent with the proposed spectral transition. A width-8 network trained with clean-point reuse shows a $d^2$-like training-set threshold over $d=64$--$192$, with fixed validation and calibration overhead.

cs.LG↗
arXiv2026

From Fragmentation to Fusion: A Survey and Roadmap for Omnimodal Embeddings

Embedding models underpin virtually every modern AI system, from search engines to recommendation systems. Despite shared mathematical foundations, research on embedding models remains fragmented by modality: the communities develop largely in isolation, seldom exchanging insights. We argue that this fragmentation obscures a deeper convergence. Driven by contrastive learning, Transformer architectures, and a granularity shift, embedding research across modalities is converging toward a common set of principles, even as modality-specific challenges remain underexplored. To reason across this convergence without conflating it, we propose a framework of three progressively harder capabilities: coverage, comparability, and compositionality. Three paths toward unification satisfy these in turn: transformation broadens coverage, alignment enables comparability, and joint embedding, the terminus we advocate, aims at compositionality. This position paper traces the resulting evolution, providing the first unified, cross-cutting treatment of omnimodal embedding research.

cs.CL↗
arXiv2026

SocialArena: Evaluating Social Intelligence in Large Language Models

We introduce SocialArena, a benchmark for systematically defining and evaluating the social intelligence of large language models (LLMs). Although LLMs are increasingly deployed in socially interactive applications such as mental health counseling, companionship chatbots, and educational tutoring, existing evaluations remain fragmented focusing on isolated capabilities or single scenarios and evaluation reliability is largely unverified. This work makes three main contributions to address these limitations. (1) We first propose a unified three-tier framework that defines social intelligence along three capability dimensions: role-playing ability, mental state reasoning, and behavioral reasoning. To operationalize this definition, we design a role-playing dialogue paradigm with explicit character thinking for evaluation, enabling simultaneous assessment of these capabilities in a manner that mirrors realistic human interactions. (2) To ensure diversity across social scenarios and thereby prevent evaluation conclusions from being confined to specific scenarios, we introduce the U-TRIMF framework, which guides us in synthesizing social contexts with broad coverage. (3) To obtain reliable evaluation, we invest $132.4K to recruit and train human annotators and adopt a human-annotated arena-style protocol, where annotators compare model responses and provide preference labels along with natural-language critiques. We hope that SocialArena can provide valuable insights and guidance for future LLM improvements.

cs.AI↗
arXiv2026

Contact Imitation Interface: Transferring Dexterous Skills by Showing How Contact Unfolds

Robot-free human demonstrations offer a promising route to scaling dexterous manipulation data. The key challenge is to preserve the demonstrator's natural dexterity while producing contact-grounded observations and actions that transfer to robot hands. We present Contact Imitation Interface (CIMI), a framework with contact-grounded demonstration and learning interfaces. The demonstration interface combines shared contact geometry and fingertip RGB views with contact-alignment-first retargeting and reprojection of wider scene views. The learning interface introduces a hierarchical policy in which a contact-surface planner predicts relative contact-surface trajectories and a controller coordinates wrist and finger commands to improve contact alignment. CIMI combines contact-grounded observations, retargeting, and action representations, improving mean final-stage success by at least 50 percentage points over baseline policies across six hand-task pairs on LEAP and Wuji. Hardware designs and software are publicly available.

cs.RO↗
arXiv2026

Unconventional SOC-activated Altermagnetic Spin Splitting and Half-Angle Locking of Spin Transport

Conventional altermagnetic spin splitting arises from staggered exchange and sublattice-dependent anisotropic hopping without requiring spin-orbit coupling (SOC). Here we derive a low-energy effective model showing that SOC and staggered exchange can jointly activate additional spin splitting on momentum planes where the nonrelativistic contribution vanishes. In hexagonal $α$-MnTe, this SOC-activated contribution produces d-wave spin splitting near $Γ$ with out-of-plane spin polarization. Its leading angular pattern counter-rotates by half the in-plane Néel-vector rotation angle. First-principles spin-transport calculations reveal a corresponding approximate half-angle locking between the Néel-vector orientation and the electric-field directions maximizing the signed longitudinal and transverse spin conductivities. Analytical transport calculations based on the effective model trace this locking to the same angular coupling between momentum and the Néel vector that governs the spin-splitting pattern. Our findings connect the microscopic mechanism of SOC-activated altermagnetic spin splitting to directional spin transport, providing a theoretical basis for probing Néel-vector orientation and controlling spin-current generation.

cond-mat.mtrl-sci↗
arXiv2026

JevAdvBench: A Benchmark and Black-Box Attacks for Reinforcement Learning for Calibrated Decisions Models

Typed decision models (TDMs), such as Jev, answer a question about an input with a probability, an option, or an ordinal level. Web services execute these decisions automatically above a confidence threshold. An attacker who writes part of the input can change a decision that the service then executes. Current adversarial benchmarks cannot measure this risk, because they judge generated text. A TDM writes no text, and its wrong decisions have a valid format. We present JevAdvBench, to our knowledge the first unified benchmark for the adversarial robustness of TDMs. It contains 862 questions with audited answers and twelve black-box attack methods that edit the input or the options by adaptive search. Because a changed decision is not always wrong, a review keeps only 5,586 answer-preserving variants, and we attack only correct decisions. The attacks flip 46.3% of Jev's correct decisions, while identical re-runs flip only 0.17%. A flawed argument without a command flips 77.2%, close to direct prompt injection at 92.0%. The same variants flip 34.0% to 51.7% of correct decisions on eight open TDMs. At a 0.8 threshold, 23.5% of the decisions that the service still executes are wrong. On Agentgateway, the Jev guardrail rejects a jailbreak request but accepts it with an inserted argument. Web services should therefore treat TDM inputs as untrusted and should not act on confidence alone. Project website: https://JevAdvBench.github.io/JevAdvBench/

cs.CR↗
arXiv2026

Hessian-Similar Optimization with Capped Participation: Lower Bounds and Virtual Masters

What is the round complexity of Hessian-similar optimization when each round can contact at most $s$ of $m$ clients? Prior rates permit a free master or full participation and do not characterize strict caps. We count every one-vector query and reply. A hidden-client chain gives a lower bound for zero-respecting local algorithms with adaptive cohorts and random stopping: on strongly convex quadratics, the required rounds are $Ω((m/s)\sqrt{\min\{δ_a,\barδ/\sqrt m\}/μ}\,\log(1/ε))$ for $s\leq(m-2)/2$ in the high-accuracy, high-dissimilarity regime. Here $μ$ is strong convexity, while $δ_a$ and $\barδ$ bound anchor and average Hessian variation. To remove dependence on a poor anchor, virtual-master SVRS uses independent estimator and proximal cohorts and averages the proximal replies. With Catalyst, it reaches relative gap $ε$ in $\tilde O(m/s+(\sqrt{m/s}+m^{3/4}/s)\sqrt{\barδ/μ})$ expected rounds on quadratics with strongly convex clients, independent of $δ_a$. Combining this method with charged analyses of existing algorithms matches the lower bound's polynomial dependence for $s\leq\sqrt m$ and $s=Ω(m)$; the worst gap is $O(m^{1/8})$. A held-out 12-seed synthetic study finds fewer rounds than tuned SVRP in nine of ten dip-family incomplete-cohort cells, versus two of 30 other cells. The lower bounds concern zero-respecting algorithms and do not cover unrestricted messages.

cs.LG↗
arXiv2026

A Statistical Framework for Understanding Causal Effects that Vary by Treatment Initiation Time in EHR-based Studies

Standard practice in electronic health record (EHR)-based studies evaluating the comparative effectiveness of bariatric surgery relative to no surgery is to estimate and report a constant treatment effect across calendar time. However, real-world treatment strategies can evolve, particularly when comparators include standard of care or surgical procedures where techniques may improve, making it clinically important to ascertain whether efficacy of bariatric surgery has changed over time. Efforts to determine whether treatment efficacy itself is evolving are complicated by changing patient populations, with potential covariate shift in key effect modifiers. Through a comprehensive analysis of EHR data from Kaiser Permanente comparing two bariatric surgical procedures to standard of care, we develop a statistical framework to estimate calendar time-specific average treatment effects and describe both how and why effects vary across treatment initiation time in EHR-based studies. Our approach projects doubly robust, time-specific treatment effect estimates onto candidate marginal structural models and uses model selection to best describe how effects vary by calendar time. We further introduce a novel summary metric, based on standardization, to quantify the degree to which observed effect changes may be attributed to covariate shift in the population receiving treatment, as opposed to changes in underlying treatment efficacy.

stat.ME↗
arXiv2026

Spark-Audio 1.0 Technical Report

This paper introduces Spark-Audio 1.0, a speech foundation model trained on domestically produced computing infrastructure. The model accepts text and speech inputs and generates text outputs. It supports multilingual (99 languages) and multi-dialect (202 dialects) speech recognition, speech translation, audio perception and understanding, and both text- and speech-based dialogue, marking a shift from merely ``hearing clearly'' to genuinely ``understanding'' speech. Architecturally, Spark-Audio 1.0 couples a 0.65B-parameter dense audio encoder with a 30B-A3B Mixture-of-Experts (MoE) large language model (LLM). The audio encoder is pretrained with representation-enhanced BEST-RQ and then refined through multi-task supervised training. The audio--LLM is trained through staged pretraining, followed by post-training that combines supervised fine-tuning with on-policy distillation. In total, the model is trained on 13 million hours of audio data together with large-scale text data. We evaluate Spark-Audio 1.0 on mainstream audio and text benchmarks, as well as on in-house test sets drawn from real-world application scenarios. Despite using substantially less audio training data than Qwen3.5-Omni-Flash, Spark-Audio 1.0 achieves comparable performance on most audio and text tasks, outperforms it in multilingual speech recognition, spoken dialogue, and real-world application scenarios, and approaches the performance of larger closed-source models. Nevertheless, there remains room for improvement in instruction following and complex audio reasoning. Finally, the successful training of Spark-Audio 1.0 on domestic hardware, including Huawei Ascend 910B and Hygon BW1000 chips, demonstrates the feasibility of a domestic technical pathway for building speech foundation models.

eess.AS↗
arXiv2026

Towards Adversarially Robust Multigraph Neural Networks for Financial Crime Analysis

Financial transaction networks are naturally represented as multigraphs, where repeated transactions between the same pair of accounts are modeled as distinct edges. Recently proposed multigraph neural network architectures leverage this structure explicitly to achieve a strong performance on anti-money laundering (AML) tasks. However, their robustness to adversaries that manipulate transaction structures to evade detection remains largely unexplored. We present the first scalable framework for structural attacks and adversarial training of multigraph neural networks using node-edge incidence matrices, adapting projected gradient descent and projected randomized block coordinate descent methods to optimize structural perturbations over such matrices. We further introduce a cycle-based regularizer that penalizes changes in the global counts of directed 2- and 3-cycles, providing a structural measure of attack plausibility. Experiments on synthetic pattern-detection and realistic AML datasets show that our learned attacks substantially outperform non-learnable perturbation baselines in reducing detection performance (PR-AUC), while adversarial training consistently improves robustness against these attacks. These results demonstrate that multigraph neural networks are vulnerable to learned structural attacks and that adversarial training provides an effective defense. Our code is available at https://github.com/denheijmans/robust_multignn.

cs.LG↗
arXiv2026

Detecting Misbehaviors of Large Vision-Language Models by Evidential Uncertainty Quantification

%Large vision-language models (LVLMs) have shown substantial advances in multimodal understanding and generation. However, when presented with incompetent or adversarial inputs, they frequently produce unreliable or even harmful content, such as fact hallucinations or dangerous instructions. This misalignment with human expectations, referred to as \emph{misbehaviors} of LVLMs, raises serious concerns for deployment in critical applications. These misbehaviors are found to stem from epistemic uncertainty, specifically either conflicting internal knowledge or the absence of supporting information. However, existing uncertainty quantification methods, which typically capture only overall epistemic uncertainty, have shown limited effectiveness in identifying such issues. To address this gap, we propose Evidential Uncertainty Quantification (EUQ), a fine-grained method that captures both information conflict and ignorance for effective detection of LVLM misbehaviors. In particular, we interpret features from the model output head as either supporting (positive) or opposing (negative) evidence. Leveraging Evidence Theory, we model and aggregate this evidence to quantify internal conflict and knowledge gaps within a single forward pass. %We extensively evaluate our method across four categories of misbehavior, including hallucinations, jailbreaks, adversarial vulnerabilities, and out-of-distribution (OOD) failures, using state-of-the-art LVLMs, and find that EUQ consistently outperforms strong baselines, showing that hallucinations correspond to high internal conflict and OOD failures to high ignorance. Furthermore, layer-wise evidential uncertainty dynamics analysis helps interpret the evolution of internal representations from a new perspective. The source code is available at https://github.com/HT86159/EUQ.

cs.LG↗
arXiv2026

Orientation-Dependent Universality of Spin-Transport Fluctuations in Two-Dimensional Altermagnets

Universal conductance fluctuations are conventionally organized by symmetry class, but crystalline anisotropy introduces additional structure in spin transport. Here we demonstrate orientation-dependent universal spin-transport fluctuations in disordered two-dimensional $d$-wave altermagnets without spin-orbit coupling. Rotating the altermagnetic orientation changes both longitudinal spin-conductance and transverse spin-Hall conductance fluctuations, while each conserved-spin sector remains in class AI. Altermagnetic orientation controls both the fluctuation amplitudes and the development of disorder-insensitive plateaus. These plateaus become flatter and less sensitive to the altermagnetic strength over the studied parameter ranges as the orientation approaches $45^\circ$, with representative fluctuation amplitudes of approximately $0.60e/(4π)$ for the longitudinal spin conductance and $0.40\,e/(4π)$ for the spin-Hall conductance. The longitudinal fluctuation amplitude correlates strongly with anisotropy of the clean-band Fermi-velocity tensor. Spin-sector covariances explain how the conserved channels combine into total fluctuations. Our results extend anisotropy-sensitive universal-fluctuation physics to conserved-spin altermagnetic transport and establish crystalline orientation as a control parameter within a fixed symmetry class.

cond-mat.mes-hall↗