arXiv subjects

Yuan Wang

Publications and source records attributed to Yuan Wang.

At least 55 records · Page 3Linked to original sources

arXiv2026

What Resolve Rate Hides: Trajectory Structure Diagnostics for Coding Agents

Coding agents are ranked almost entirely by resolve rate: whether their final patch passes the target tests. Yet two agents can reach the same outcome through very different processes, and a single pass/fail label says nothing about why a run failed or why an accepted run spent extra steps, time, or tokens. This process evidence lives in the trajectory, which records a run's searches, reads, edits, tool calls, validation, and reversions. However, raw traces are heterogeneous and hard to compare across runs. We present TraceProbe, a trajectory-diagnostic framework that recovers what resolve rate hides. TraceProbe normalizes each raw run into a canonical nine-type action taxonomy with deterministic effect labels, then applies two rule-based modules: Insight names single-trajectory anti-patterns adapted from established debugging practice (e.g., search loops, verification skips), while Converge aligns pairs of runs and classifies where their behavior diverges under controlled references. Applying TraceProbe to 2,500 trajectories from five production settings on SWE-Bench Verified, we find that (i) file choice is too coarse to separate success from failure, whereas function selection and completion behavior localize it; (ii) Insight anti-patterns act mainly as corpus-level difficulty clues, with search loops the most stable; and (iii) even resolved runs differ in how quickly they reach relevant code and how much failed work they incur. Trajectory structure thus adds auditable diagnostic context to outcomes by localizing inspection targets, suggesting failure hypotheses, and prioritizing runs for review.

cs.SE↗

arXiv2026

MedStreamBench: A Time-Aware Benchmark for Streaming and Proactive Medical Video Understanding

Existing medical video benchmarks primarily evaluate whether a model produces the correct answer, but rarely assess whether it answers at the right time. In real clinical settings, AI systems must decide not only what to predict, but also when to answer, defer judgment, or proactively raise alerts. This creates a critical gap between benchmark evaluation and deployment requirements. We present MedStreamBench, a benchmark for time-aware medical video understanding. MedStreamBench integrates 22 medical datasets and 5,419 QA instances across four temporal settings: retrospective, present, future, and proactive. Unlike conventional benchmarks that assume full-video access, MedStreamBench restricts models to temporally bounded evidence windows and supports both single-turn and streaming evaluation. We further introduce a proactive monitoring setting that requires models to determine whether and when clinically relevant alerts should be triggered. Beyond answer correctness, MedStreamBench evaluates temporal behavior through responsiveness and post-evidence stability. Experiments on leading general-purpose and medical vision-language models reveal a substantial gap between offline recognition and temporally grounded decision-making, with performance dropping markedly in streaming and proactive settings. Our benchmark is available at https://huggingface.co/datasets/Venn2024/MedStreamBench.

cs.CV↗

arXiv2026

AtomiMed: Hierarchical Atomic Fact-Checking for Universal Clinical-Aware Medical Report Evaluation

Traditional metrics for Medical Report Generation (MRG) predominantly rely on surface-level n-gram overlap, which fails to capture clinical factual accuracy and often overlooks catastrophic diagnostic errors. We address this fundamental limitation by proposing \textbf{AtomiMed}, a universal, modality-agnostic evaluation framework that decomposes complex medical narratives into a standardized, multi-level hierarchy of Atomic Clinical Facts, encompassing Disease-level entities and Attribute-level descriptors, including location, morphology, and severity. By implementing an Agentic Cross-Verification loop between ground-truth and predicted reports, AtomiMed simulates a multi-radiologist peer-review process to verify clinical consistency, thus enabling the decoupled assessment of diagnostic detection and descriptive accuracy. To facilitate standardized evaluation, we introduce \textbf{MRGEvalKit}, an open-source toolkit for automated hierarchical extraction, and curate \textbf{OmniMRG-Bench}, a comprehensive multi-modal benchmark covering X-ray, CT, MRI, and Ultrasound. Extensive experiments on multiple expert-annotated reader studies demonstrate that AtomiMed achieves significantly higher correlation with human radiologist judgment compared to traditional and model-based metrics. Our code are release at https://github.com/Venn2336/MRGEvalkit

cs.CE↗

arXiv2026

A Universal Dance of Galactic Disks: Ubiquitous Precession and Its Implications

Precession is a very common phenomenon for small-scale astronomical objects. However, the precession of galactic disks, occurring on a scale larger than kilo-parsec, has barely been studied in the literature. Quantifying this precession in observations remains challenging due to the lack of high-resolution dynamical data. Cosmological simulations, where gravitational interactions are self-consistently modeled, offer a unique avenue for investigating disk precession. Leveraging the IllustrisTNG simulations, we trace the evolution of spin orientation in Milky Way-like galaxies over cosmic time. We find that disk precession is ubiquitous in galaxies and significantly affects galaxy evolution. The precession is driven by the external tidal torque originating from the anisotropic matter distribution within $30\ \mathrm{kpc}$, and is violent at $\mathrm{z} > 1$ and becomes gentler but significant at $\mathrm{z} \sim 0$, when the disks are considered dynamically settled. Disk precession can induce significant cold gas warp, which is often observed in the Milky Way and nearby galaxies. We predict that the Milky Way is precessing at a rate of $\simeq3-10$ degrees per billion years at current epoch based on its observed warp. Violent precession can heat the orbits of stars, which may eventually produce prolate elliptical galaxies. The tidal torque from central galaxies can cause the precession of nearby satellite galaxies and causes their disks to point towards the centrals, which explains the observational radial alignment. We also find that the precession of accreted cold gas stream, regulated by the galaxies' torque, is crucial for the evolution of disk galaxies.

astro-ph.GA↗

arXiv2026

Deep Research in Physical Sciences: A Multi-Agent Framework and Comprehensive Benchmark

Deep research agents are Large Language Model (LLM)-based systems designed for autonomous, multi-step scientific reasoning, and they hold immense potential for accelerating research in the physical sciences. However, comprehensive and in-depth evaluations of their capabilities within this domain remain lacking. To address this gap, we introduce PhySciBench, a benchmark highly relevant to physical science research, comprising 200 expert-curated questions, balanced between physics and chemistry, across six task categories that reflect real-world scientific workflows. Evaluations of state-of-the-art models and agent systems on PhySciBench reveal limited performance; even the strongest baseline, Gemini Deep Research, achieves an accuracy of only 33.5%. Analysis of failure cases identifies three recurrent deficiencies: fragility in extended reasoning chains, limited knowledge transfer across steps, and a lack of physics-grounded self-verification. Motivated by these findings, we develop DelveAgent, a modular multi-agent framework equipped with an adaptive planning loop, dual-granularity memory, and a hierarchical physics-grounded reflection mechanism. Across four scientific benchmarks, DelveAgent improves accuracy by up to 7.5 percentage points while reducing inference costs to approximately one-third of the strongest baseline. These results establish the significance of PhySciBench as a critical benchmark for evaluating AI systems in the physical sciences and demonstrate that architectural specialization can effectively enhance the reliability of autonomous scientific research. Our data and code are publicly available at https://github.com/yigengjiang/physci-deepresearch.

physics.comp-ph↗

arXiv2026

Beyond Artifacts: Towards Generalizable Synthetic Song Detection via Music-Intrinsic Features

The rapid advancement of AI music generators highlights the urgent need for reliable Synthetic Song Detection (SSD). Existing SSD methods often rely on low-level artifacts or fixed feature assumptions, struggling to capture generator-agnostic cues. To address this, we propose Sofia (Synthetic-song detection framework via music features), a flexible framework that models music-intrinsic attributes via feature-specific experts and an adaptive Mixture-of-Experts (MoE) module. By configuring Sofia with representative Vocal, Audio-effect, Global structure features, and their combinations, we present their individual and complementary contributions. To comprehensively evaluate our framework, we further construct MUSIC8K, a challenging benchmark featuring lastest emerging generators and realistic audio perturbations. Experiments show that Sofia learns generator-agnostic representations from music-intrinsic features, improving the F1 score by 18.5 points over the strongest baseline on MUSIC8K-O while maintaining strong robustness.

cs.SD↗

arXiv2026

Generative modelling powered by room-temperature polariton condensates

Generative modelling requires efficient stochastic nonlinear transformations and physical platforms that can naturally realise them. We experimentally demonstrate that nonlinear optical systems operating in the strong light-matter coupling regime can serve as physical transformation layers for conditional generative modelling. Specifically, we develop a workflow in which room-temperature exciton-polariton condensates formed in organic dye microcavities act as a physical stochastic transform within a generative adversarial network and enable conditional digit-to-image translation. By using the nonlinear many-body dynamics and intrinsic stochasticity of polariton condensates, the workflow outperforms baseline approaches based on digitally injected perturbations. We find that polariton-enabled sampling via generative adversarial network (Polariton GAN) yields improved inception score, digit preservation accuracy and structural similarity compared with both digital sampling and laser-based systems. We further show that spatially correlated output variations can naturally regularise adversarial training and enhance output diversity. Our results establish polariton condensation as a new computational resource for generative modelling, opening a pathway towards physics-enhanced machine learning systems.

cond-mat.dis-nn↗

arXiv2026

LENS: A Staged Design for Interaction Granularity in Sequential CTR Prediction

In sequential CTR prediction, a central design question is at what granularity the target should interact with the user behaviour sequence. Existing models mainly follow two routes. Raw-item architectures such as DIN let the target score each item in the sequence directly. This relies on well-trained item embeddings and becomes brittle for sparse items. Latent-query architectures such as HyFormer, MixFormer, and OneTrans build query representations by combining the target with other information. This is more robust across item-density regimes but blunter: target-specific control is diluted. We propose LENS to restore target-specific control within these coarser bottlenecks. LENS has two modules: a Target-Conditioned Query Gate (TCQG) for query activation and a Target-Conditioned Position Bias (TCPB) for history retrieval. We further introduce Query-Specific Position Bias (QueryPos), a simple static position-aware reference for latent-query backbones. Across three representative latent-query backbones and four datasets, the combined QueryPos+LENS design achieves positive total-gain point estimates in all twelve evaluated backbone--dataset cells. We also identify a density-dependent conditioning rule: as item density decreases, the optimal condition source shifts from item-only to item-plus-sequence.

cs.IR↗

arXiv2026

MotionGPT-2: A General-Purpose Motion-Language Model for Motion Generation and Understanding

Generating lifelike human motions from descriptive texts has experienced remarkable research focus in the recent years, propelled by the emerging requirements of digital humans.Despite impressive advances, existing approaches are often constrained by limited control modalities, task specificity, and focus solely on body motion representations.In this paper, we present MotionGPT-2, a unified Large Motion-Language Model (LMLM) that addresses these limitations. MotionGPT-2 accommodates multiple motion-relevant tasks and supporting multimodal control conditions through pre-trained Large Language Models (LLMs). It quantizes multimodal inputs-such as text and single-frame poses-into discrete, LLM-interpretable tokens, seamlessly integrating them into the LLM's vocabulary. These tokens are then organized into unified prompts, guiding the LLM to generate motion outputs through a pretraining-then-finetuning paradigm. We also show that the proposed MotionGPT-2 is highly adaptable to the challenging 3D holistic motion generation task, enabled by the innovative motion discretization framework, Part-Aware VQVAE, which ensures fine-grained representations of body and hand movements. Extensive experiments and visualizations validate the effectiveness of our method, demonstrating the adaptability of MotionGPT-2 across motion generation, motion captioning, and generalized motion completion tasks.

cs.CV↗

arXiv2026

Enhanced evidence of $X(7200)$ and improved measurements of $X(6900)$ parameters from a combined LHCb-ATLAS-CMS analysis

We report stronger evidence for the $X(7200)$ state and markedly improved measurements of the $X(6900)$ resonance parameters based on a combined analysis of the di-$J/ψ$ mass spectrum using published data from LHCb, ATLAS, and CMS. Through simultaneous fits to the datasets from all three experiments, we observe the $X(6900)$ with overwhelming significance ($>12σ$) and determine its mass and width with improved precision. For the $X(7200)$, we find consistent signals across multiple interference models, with significances ranging from $3.7σ$ to $6.6σ$; in the best-fit model (the CMS three-resonance scheme), the significance reaches $6.6σ$, providing substantially stronger evidence for this state. Our results underscore the essential role of interference effects in fully charmed tetraquark spectroscopy and offer new constraints on their production mechanisms at the LHC.

hep-ex↗

arXiv2026

Bridging the Detection-to-Abstention Gap in Reasoning Models under Insufficient Information

We highlight a failure mode of large reasoning models on questions with insufficient information: models may recognize that a problem is under-specified, yet still continue reasoning and produce unsupported final answers instead of abstaining. We formalize this mismatch as the detection-to-abstention gap, where detected insufficiency fails to translate into final abstention. This gap is especially concerning in high-risk domains such as medical AI, where answers based on incomplete evidence can be more harmful than refusal. To close this gap, we propose Judge-Then-Solve (JTS), a trajectory-level reasoning-control framework that trains models to make an explicit answerability commitment before solution generation. Rather than treating abstention as a final-answer style, JTS casts it as a control decision: the model either proceeds to solve or terminates early based on its answerability judgment. We instantiate this policy through supervised warm-up and missing-premise reinforcement learning with consistency and length-shaping rewards. Experiments on dense and MoE reasoning models show that JTS substantially improves reliable abstention across datasets and pushes Abstention@Detection (A@D) to near-saturation, indicating that models not only detect missing information but also act on that detection. By terminating unanswerable trajectories immediately after the answerability judgment, JTS reduces unnecessary reasoning and improves inference efficiency when continued deliberation would amplify unsupported assumptions. We also observe that missing-premise training can alter reasoning behavior on difficult but answerable problems, reducing unproductive self-reflection. These results suggest that abstention under insufficient information is a key form of reasoning control for deploying reasoning models safely and efficiently.

cs.AI↗

arXiv2026

GLM-5V-Turbo: Toward a Native Foundation Model for Multimodal Agents

We present GLM-5V-Turbo, a step toward native foundation models for multimodal agents. As foundation models are increasingly deployed in real environments, agentic capability depends not only on language reasoning, but also on the ability to perceive, interpret, and act over heterogeneous contexts such as images, videos, webpages, documents, GUIs. GLM-5V-Turbo is built around this objective: multimodal perception is integrated as a core component of reasoning, planning, tool use, and execution, rather than as an auxiliary interface to a language model. This report summarizes the main improvements behind GLM-5V-Turbo across model design, multimodal training, reinforcement learning, toolchain expansion, and integration with agent frameworks. These developments lead to strong performance in multimodal coding, visual tool use, and framework-based agentic tasks, while preserving competitive text-only coding capability. More importantly, our development process offers practical insights for building multimodal agents, highlighting the central role of multimodal perception, hierarchical optimization, and reliable end-to-end verification.

cs.CV↗

arXiv2026

Think, then Score: Decoupled Reasoning and Scoring for Video Reward Modeling

Recent advances in generative video models are increasingly driven by post-training and test-time scaling, both of which critically depend on the quality of video reward models (RMs). An ideal reward model should predict accurate rewards that align with human preferences across diverse scenarios. However, existing paradigms face a fundamental dilemma: \textit{Discriminative RMs} regress rewards directly on features extracted by multimodal large language models (MLLMs) without explicit reasoning, making them prone to shortcut learning and heavily reliant on massive data scaling for generalization. In contrast, \textit{Generative RMs} with Chain-of-Thought (CoT) reasoning exhibit superior interpretability and generalization potential, as they leverage fine-grained semantic supervision to internalize the rationales behind human preferences. However, they suffer from inherent optimization bottlenecks due to the coupling of reasoning and scoring within a single autoregressive inference chain. To harness the generalization benefits of CoT reasoning while mitigating the training instability of coupled reasoning and scoring, we introduce DeScore, a training-efficient and generalizable video reward model. DeScore employs a decoupled ``think-then-score'' paradigm: an MLLM first generates an explicit CoT, followed by a dedicated discriminative scoring module consisting of a learnable query token and a regression head that predicts the final reward. DeScore is optimized via a two-stage framework: (1) a discriminative cold start incorporating a random mask mechanism to ensure robust scoring capabilities, and (2) a dual-objective reinforcement learning stage that independently refines CoT reasoning quality and calibrates the final reward, ensuring that higher-quality reasoning directly translates to superior model performance.

cs.CV↗

arXiv2026

Can Multimodal Large Language Models Understand Pathologic Movements? A Pilot Study on Seizure Semiology

Multimodal Large Language Models (MLLMs) have demonstrated robust capabilities in recognizing everyday human activities, yet their potential for analyzing clinically significant involuntary movements in neurological disorders remains largely unexplored. This pilot study evaluates the capability of MLLMs for automated recognition of pathological movements in seizure videos. We assessed the zero-shot performance of state-of-the-art MLLMs on 20 ILAE-defined semiological features across 90 clinical seizure recordings. MLLMs outperformed fine-tuned Convolutional Neural Network (CNN) and Vision Transformer (ViT) baseline models on 13 of 18 features without task-specific training, demonstrating particular strength in recognizing salient postural and contextual features while struggling with subtle, high-frequency movements. Feature-targeted signal enhancement (facial cropping, pose estimation, audio denoising) improved performance on 10 of 20 features. Expert evaluation showed that 94.3 percent of MLLM-generated explanations for correctly predicted cases achieved at least 60 percent faithfulness scores, aligning with epileptologist reasoning. These findings demonstrate the potential of adapting general-purpose MLLMs for specialized clinical video analysis through targeted preprocessing strategies, offering a path toward interpretable, efficient diagnostic assistance. Our code is publicly available at https://github.com/LinaZhangUCLA/PathMotionMLLM.

cs.CV↗

arXiv2026

How Code Representation Shapes False-Positive Dynamics in Cross-Language LLM Vulnerability Detection

How code representation format shapes false positive behaviour in cross-language LLM vulnerability detection remains poorly understood. We systematically vary training intensity and code representation format, comparing raw source text with pruned Abstract Syntax Trees at both training time and inference time, across two 8B-parameter LLMs (Qwen3-8B and Llama 3.1-8B-Instruct) fine-tuned on C/C++ data from the NIST Juliet Test Suite (v1.3) and evaluated on Java (OWASP Benchmark v1.2) and Python (BenchmarkPython v0.1). Cross-language FPR reflects the joint effect of training-time and inference-time representation, not either alone. Text fine-tuning drives FPR upward monotonically (Qwen3-8B: 0.763 zero-shot, 0.866 pilot, 1.000 full-scale) while F1 remains stable (0.637-0.688), masking the collapse. We argue surface-cue memorisation is the primary mechanism: text fine-tuning encodes C/C++-specific API names and syntactic idioms as vulnerability triggers that fire indiscriminately on target-language code. A cross-representation probe, applying text-trained weights to AST-encoded input without retraining, isolates this: Qwen3-8B FPR drops from 0.866 to 0.583, and 37.2% of false positives revert to true negatives under AST input alone. Direct AST fine-tuning does not preserve the benefit (FPR at least 0.970), as flat linearisation introduces structural surface cues of its own. The pattern replicates across both model families. On BenchmarkPython the AST probe yields FPR=0.554, within 2.9 percentage points of the Java result, despite maximal surface-syntax differences, substantially weakening a domain-shift explanation. These findings motivate a pre-deployment consistency gate, running alerts through both text and AST paths, as a retraining-free filter for false-positive-sensitive settings, at the cost of reduced recall.

cs.CR↗

arXiv2026

U-ViLAR: Uncertainty-Aware Visual Localization for Autonomous Driving via Differentiable Association and Registration

Accurate localization using visual information is a critical yet challenging task, especially in urban environments where nearby buildings and construction sites significantly degrade GNSS (Global Navigation Satellite System) signal quality. This issue underscores the importance of visual localization techniques in scenarios where GNSS signals are unreliable. This paper proposes U-ViLAR, a novel uncertainty-aware visual localization framework designed to address these challenges while enabling adaptive localization using high-definition (HD) maps or navigation maps. Specifically, our method first extracts features from the input visual data and maps them into Bird's-Eye-View (BEV) space to enhance spatial consistency with the map input. Subsequently, we introduce: a) Perceptual Uncertainty-guided Association, which mitigates errors caused by perception uncertainty, and b) Localization Uncertainty-guided Registration, which reduces errors introduced by localization uncertainty. By effectively balancing the coarse-grained large-scale localization capability of association with the fine-grained precise localization capability of registration, our approach achieves robust and accurate localization. Experimental results demonstrate that our method achieves state-of-the-art performance across multiple localization tasks. Furthermore, our model has undergone rigorous testing on large-scale autonomous driving fleets and has demonstrated stable performance in various challenging urban scenarios.

cs.CV↗

arXiv2026

How is a gas sensor poisoned by volatile methylsiloxanes?

Volatile methyl siloxanes (VMSs), widely present in consumer and industrial products, have attracted increasing concerns due to their persistence, bioaccumulation behavior, and adverse health effects. Beyond their environmental implications, VMSs also pose operational challenges for sensing technologies because they readily decompose on sensing materials to form silicon-based compounds (e.g., silica and silane) that irreversibly impair sensing performance, a phenomenon commonly known as siloxane poisoning. Despite its prevalence, the mechanistic basis of this deactivation remains poorly understood. Herein, we present the first comprehensive theoretical study of siloxane-induced poisoning in catalytic gas sensors. Guided by our self-developed AI Agent, Digital Sensor Platform (DigSen), we first identify siloxane poisoning as a previously overlooked yet high-impact research direction. Using hexamethyldisiloxane (HMDS) as a model compound, we then conducted first-principles calculations to uncover decomposition pathways across noble metal surfaces. Strikingly, a descriptor-based microkinetic volcano model is developed to capture the trade-off between sensing activity and resistance to poisoning, enabling predictive identification of anti-poisoning candidates. These insights not only elucidate the origin of siloxane poisoning but also demonstrate how AI-driven discovery, mechanistic theory, and experiments can be integrated into a closed-loop framework for catalytic sensor design. More broadly, this AI-guided paradigm represents a generalizable strategy for materials digital discovery, offering a transferable methodology that extends well beyond siloxane systems to diverse classes of materials challenges.

physics.chem-ph↗

arXiv2026

How Far Are Video Models from True Multimodal Reasoning?

Despite remarkable progress toward general-purpose video models, a critical question remains unanswered: how far are these models from achieving true multimodal reasoning? Existing benchmarks fail to address this question rigorously, as they remain constrained by straightforward task designs and fragmented evaluation metrics that neglect complex multimodal reasoning. To bridge this gap, we introduce CLVG-Bench, an evaluation framework designed to probe video models' zero-shot reasoning capabilities via Context Learning in Video Generation. CLVG-Bench comprises more than 1,000 high-quality, manually annotated metadata across 6 categories and 47 subcategories, covering complex scenarios including physical simulation, logical reasoning, and interactive contexts. To enable rigorous and scalable assessment, we further propose an Adaptive Video Evaluator (AVE) that aligns with human expert perception using minimal annotations, delivering interpretable textual feedback across diverse video context tasks. Extensive experiments reveal a striking answer to our central question: while state-of-the-art (SOTA) video models, such as Seedance 2.0, demonstrate competence on certain understanding and reasoning subtasks, they fall substantially short with logically grounded and interactive generation tasks (achieving success rates <25% and ~0%, respectively), exposing multimodal reasoning and physical grounding as critical bottlenecks. By systematically quantifying these limitations, the proposed method provides actionable feedbacks and a clear roadmap toward truly robust, general-purpose video models. CLVG-Bench and code are released here.

cs.CV↗