arXiv subjects

Zhihao Yang

Publications and source records attributed to Zhihao Yang.

At least 19 recordsLinked to original sources

arXiv2026

LLM2Cert: LLM Synthesis of Certificates with SMT/Lean Verification

Lyapunov functions and barrier certificates give state-space guarantees of stability and safety, but synthesizing and verifying them---especially for nonlinear, high-dimensional systems---remains an open challenge. Existing pipelines verify candidates with numerical solvers (sum-of-squares programs), interval methods, or satisfiability-modulo-theories (SMT) solvers, which often require relaxations, are conservative, or scale poorly with model dimension and nonlinearity. These limitations have also shaped benchmark design: standard test suites are dominated by low-dimensional polynomial systems chosen to suit what solvers can handle rather than what engineering applications require. We address both sides of this challenge. First, we introduce DynaCertBench, a benchmark of 105 instances: 80 non-redundant systems curated from the literature, extended with high-dimensional and control-constrained variants and new nonlinearly coupled systems, comprehensively spanning stability, region-of-attraction, forward-invariance, and safety tasks. Second, we present LLM2Cert, an LLM-driven framework that synthesizes candidate certificates and verifies them against their proof obligations: each obligation is routed to an SMT solver or directly to the Lean Proof-Repair agent. On DynaCertBench, LLM2Cert verifies 81 of 105 instances ($77.1\%$), a $17.1$ percentage point improvement over one-shot proposal without feedback. On the 81 fixed certificates, LLM2Cert verifies 22 that All-SMT leaves unresolved. Code and data are available at https://github.com/FormalCert-AI/llm2cert.

eess.SY↗

arXiv2026

Scaling in Games: Continued Pre-Training For Embodied Agents In Diverse Virtual Worlds

Scaling data, diversity and model capacity has improved embodied learning in physical environments, where diverse experiences remain grounded in shared physical laws. Whether these gains extend to virtual worlds with different visual styles, dynamics, controls, and objectives remains unclear. We investigate this question through a controlled study of continued pre-training (CPT) across three axes: data volume, diversity, and model size. Our 120 training configurations span four model sizes from 0.8B to 27B, six data budgets from 50 to 1,600 hours, and five mixtures ranging from Minecraft-only data to diverse multi-world corpora. We assess performance and generalization using three-level validation losses and gameplay benchmarks from in-corpus worlds to unseen worlds. Our results show: (1) more data improves in-corpus performance, diversity improves transfer to related worlds with limited gains in distant worlds, and model size improves data efficiency but generalizes worse to distant worlds under single-world training; (2) we observe an early interface phase with broad cross-world benefits and a later world phase with gains narrowed to in-corpus and related worlds, and fit an empirical scaling law relating validation loss to data volume, diversity and model size; and (3) our models are competitive with Minecraft specialists and frontier models in 3D worlds, and perform best in real-time settings. However, the limits of imitation and persistent gaps in distant worlds suggest that building generalist embodied agents requires richer training stages beyond continued pre-training.

cs.AI↗

arXiv2026

Evolving to Match Capabilities: Improving Harness Self-Evolution via Skill Adaptation

Harness self-evolution offers a practical route for LLM agents to improve from execution experience without updating model parameters. Its effectiveness depends on whether models can successfully act on the guidance provided by their harnesses. However, we find that harness benefits are not uniform across model capabilities: skills that improve stronger models can provide little benefit or even degrade performance for weaker ones. Across seven models and three benchmarks, we identify 103 task-skill pairs where skills improve GPT-5.6-luna by 31.1 points but degrade Gemma-4-E4B by 8.8 points. In 86% of failed weak-model runs, the violated requirement is already stated in the skill. This motivates skill adaptation: reformulating existing skills to match the executor's capabilities and make their guidance easier to execute. To achieve this, we diagnose failures in tool execution, context management, instruction following, and verification, and distill these diagnoses into a fixed meta-skill that guides adaptation during harness self-evolution. Guided by the meta-skill, a capable external rewriter, Kimi-K3, adapts skills without execution evidence, improving the three weak executors by 5.1-14.1 points over the original skills. We then incorporate this adaptation into a five-round self-evolution loop, where each model revises its own skill from execution trajectories and verifier feedback. Across seven executors, from Gemma-4-E4B to Kimi-K2.5, the meta-skill consistently improves final scores by 2.0-5.1 points over the same loop without the meta-skill. Our findings provide practical design guidance for developing more effective harness self-evolution methods.

cs.LG↗

arXiv2026

Video-Arena: A Scalable Agentic Video Benchmark for Spatiotemporal Capability

Understanding and tracking a dynamic world are fundamental to visual intelligence. Existing video benchmarks provide limited insight into whether models can maintain a coherent understanding of evolving world state. Building such a benchmark from real-world videos is challenging because the underlying world state is partially observable, making annotations hard to obtain. To close this gap, we introduce Video-Arena, a benchmark that turns competitive games into scalable, verifiable tests of how well models understand evolving world states from videos. Specifically, inspired by how humans understand dynamic events, we evaluate spatiotemporal world state understanding and tracking through three task categories: perceiving critical events, interpreting state updates, and tracking the resulting states. Our pipeline can generate videos and aligned reference labels from game records at scale. To keep evaluation affordable and reproducible, we curate a golden set of 180 tasks across 9 games and run evaluations under two protocols: native visual understanding and tool-assisted agentic analysis. The best-performing systems achieve F1 scores of and under the two protocols, respectively. The result highlights that native models struggle to perceive events reliably, while tool-assisted systems improve event recovery but still struggle to interpret state updates and track the resulting states consistently over time. Our project page is in https://videoarena-bench.github.io

cs.CV↗

arXiv2026

X-Ray Spectroscopy for Characterizing Stimulated Raman Scattering in Inertial Confinement Fusion

Accurate characterization of stimulated Raman scattering (SRS) remains a critical challenge in inertial confinement fusion (ICF), as SRS not only scatters laser energy but also generates suprathermal electrons that preheat the fuel and degrade implosion performance. Conventional backscatter diagnostics provide direct measurements of SRS but cannot collect the entire scattered-light signal, limiting the accurate characterization of SRS strength. Using a non-local thermodynamic equilibrium collisional-radiative model with a double-Maxwellian electron distribution, we systematically investigate how suprathermal electrons modify the titanium \textit{K}-shell x-ray emission spectra. The spectra exhibit high sensitivity to the suprathermal-electron fraction at relatively low bulk electron temperatures, making them particularly suitable for diagnosing suprathermal electrons during the early stage of ICF, when even a small suprathermal-electron population can compromise fuel compression. The calculated spectra reproduce experimental measurements from the Nova Laser Facility with good agreement, and the inferred suprathermal-electron fractions show improved consistency with independently measured SRS losses compared with the original spectral analysis by Glenzer~\href{https://doi.org/10.1103/PhysRevLett.81.365} {\text{[S. H. Glenzer \textit{et al}., Phys. Rev. Lett. \textbf{81}, 365(1998)]}}. These results demonstrate that \textit{K}-shell spectroscopy, combined with accurate NLTE collisional-radiative modeling, provides a reliable probe of SRS strength in laser-produced plasmas. Hence, this approach may be extended to spatially resolved diagnosis of SRS strength, offering a promising complement to conventional backscatter diagnostics.

physics.plasm-ph↗

arXiv2026

Dual Attention Residuals

Recent work extends Transformer residual pathways along two complementary axes: historical retrieval selects information from earlier depths, whereas multi-stream methods maintain multiple residual trajectories. These capabilities have largely been studied in isolation, and assigning an independent retriever to each stream still prevents one trajectory from influencing depth selection in another. We propose Dual Attention Residuals (DAR), which brings multi-stream interaction into historical retrieval through reciprocal cross-stream addressing. For each target stream, DAR computes depth weights from normalized states in the opposite stream and applies them to values from the target stream's own history. The retrieved states are combined for an unchanged Transformer branch and updated through constrained gated writes; a block-form variant operates on block-level histories to control overhead. Across dense models from 0.1B to 1B parameters and a 7B sparse-MoE model, DAR consistently improves validation loss over standard residual Transformers and Attention Residuals. Routing ablations show that the gain cannot be explained by an additional stream or value projection alone. Representation and intervention analyses further show that reciprocal cross-stream selection preserves depth-wise diversity and avoids the redundancy or functional imbalance observed in alternative two-stream designs.

cs.CL↗

arXiv2026

CoGoal3D: Collaborative 3D Object Detection with 3D-Aware Fusion and Refinement

V2X collaborative object detection features overcoming the limitations of single-vehicle systems by aggregating environmental features from multiple collaborative agents. However, existing mainstream V2X perception methods mainly focus on 2D BEV object detection. When 3D detection task is concerned, inferior results are obtained because they ignore the 3D spatial misalignment caused by differing height and attitude among the collaborators. In this paper, we propose a novel collaborative 3D object detection framework called CoGoal3D, which extracts and refines the 3D feature gradually in a two-stage pipeline. In the first stage, a multiscale 3D-aware global fusion module is designed to mitigate the 3D spatial misalignment. The resulting proposals are then refined in the second stage with an auxiliary task of 3D point reconstruction. An effective multi-agent collaborative data augmentation strategy is further proposed to enrich the training data while minimizing information loss. Extensive experiments on public real-world datasets demonstrate that our CoGoal3D achieves new state-of-the-art performance, with 3D AP@0.7 improvements of 10.86%, 10.34%, and 10.18% on the DAIR-V2X, V2V4Real, and V2X-Real datasets, respectively. Code is available at https://github.com/Megalo-f/CoGoal3D.

cs.CV↗

arXiv2026

EVADE-Bench: Multimodal Benchmark for Evaluating and Enhancing Evasive Content Detection

E-commerce platforms increasingly rely on Large Language Models (LLMs) and Vision Language Models (VLMs) to detect illicit or misleading product content. However, these models remain vulnerable to evasive content, which refers to inputs that have been deliberately modified through techniques such as word splitting, euphemistic language, or image cropping to conceal policy violations while still conveying prohibited claims. Crucially, detecting such content requires a model to simultaneously master two capabilities: accurately comprehending complex rules, and correctly inferring the true intent behind deliberately obfuscated multimodal inputs. While prior work has separately explored LLM reasoning over complex rules and LLM-based detection of evasive content, no existing benchmark combines both within a unified evaluation framework. This gap is particularly consequential in e-commerce, where accurate moderation demands that both capabilities operate in concert. To address this gap, we introduce EVADE-Bench, the first expert-curated Chinese multimodal benchmark specifically designed to evaluate LLMs and VLMs on evasive content detection in real-world e-commerce scenarios. Our comprehensive evaluation of 26 open- and closed-source LLMs and VLMs reveals that even state-of-the-art models frequently misclassify evasive samples. We further demonstrate that clearer rule categorization significantly improves model prediction consistency and reduces false predictions, highlighting the critical role of benchmark design in enabling reliable evaluation. To explore paths for performance improvement, we investigate the feasibility of multi-agent decomposition for multimodal reasoning, wherein visual description and logical inference are decoupled into separate agents, and find that this strategy yields notable accuracy gains.

cs.CL↗

arXiv2026

Techno-economic Analysis of Light Isotope-enriched Elements for Lightweighting Applications

Lightweighting is critical to mass-sensitive applications such as aircraft and space transportation. Conventional lightweight strategies often rely on new designs of materials and structures. An alternative approach is to enrich the lightest stable isotopes in an element to reduce the elements atomic mass while having little effect on structural and chemical properties. However, the economic feasibility of this concept remains unclear. Here we present a techno-economic analysis of light isotope-enriched elements for lightweighting applications by estimating isotope enrichment cost and the economic gain from mass reduction. The enrichment cost is scaled from established large-scale processes. Twelve common aerospace-relevant elements are considered, including Li, B, C, Mg, Cl, Ti, Ni, Fe, Cu, Zn, Mo, and Sn. We find that nine elements, especially Li, B, Zn, Ni, Mo, and Sn, show potentially attractive economic benefit at moderate enrichment levels, whereas C, Mg, and Fe provide little or no benefit. With the optimized enrichment levels, an Airbus A380 is expected to save approximately USD 700 K over a 30-year operational lifetime, a SpaceX Falcon 9 could save USD 516 K, and a SpaceX Starship is expected to save USD 2.37 million over its whole lifetime. While the exact enrichment cost needs to be further investigated, these results provide an initial screening of promising candidate elements and highlight isotopic mass reduction as a potential drop-in lightweighting strategy.

cond-mat.mtrl-sci↗

arXiv2026

FlowPIE: Test-Time Scientific Idea Evolution with Flow-Guided Literature Exploration

Scientific idea generation (SIG) is critical to AI-driven autonomous research, yet existing approaches are often constrained by a static retrieval-then-generation paradigm, leading to homogeneous and insufficiently divergent ideas. In this work, we propose FlowPIE, a tightly coupled retrieval-generation framework that treats literature exploration and idea generation as a co-evolving process. FlowPIE expands literature trajectories via a flow-guided Monte Carlo Tree Search (MCTS) inspired by GFlowNets, using the quality of current ideas assessed by an LLM-based generative reward model (GRM) as a supervised signal to guide adaptive retrieval and construct a diverse, high-quality initial population. Based on this population, FlowPIE models idea generation as a test-time idea evolution process, applying selection, crossover, and mutation with the isolation island paradigm and GRM-based fitness computation to incorporate cross-domain knowledge. It effectively mitigates the information cocoons arising from over-reliance on parametric knowledge and static literature. Extensive evaluations demonstrate that FlowPIE consistently produces ideas with higher novelty, feasibility and diversity compared to strong LLM-based and agent-based frameworks, while enabling reward scaling during test time.

cs.AI↗

arXiv2026

LaDy: Lagrangian-Dynamic Informed Network for Skeleton-based Action Segmentation via Spatial-Temporal Modulation

Skeleton-based Temporal Action Segmentation (STAS) aims to densely parse untrimmed skeletal sequences into frame-level action categories. However, existing methods, while proficient at capturing spatio-temporal kinematics, neglect the underlying physical dynamics that govern human motion. This oversight limits inter-class discriminability between actions with similar kinematics but distinct dynamic intents, and hinders precise boundary localization where dynamic force profiles shift. To address these, we propose the Lagrangian-Dynamic Informed Network (LaDy), a framework integrating principles of Lagrangian dynamics into the segmentation process. Specifically, LaDy first computes generalized coordinates from joint positions and then estimates Lagrangian terms under physical constraints to explicitly synthesize the generalized forces. To further ensure physical coherence, our Energy Consistency Loss enforces the work-energy theorem, aligning kinetic energy change with the work done by the net force. The learned dynamics then drive a Spatio-Temporal Modulation module: Spatially, generalized forces are fused with spatial representations to provide more discriminative semantics. Temporally, salient dynamic signals are constructed for temporal gating, thereby significantly enhancing boundary awareness. Experiments on challenging datasets show that LaDy achieves state-of-the-art performance, validating the integration of physical dynamics for action segmentation. Code is available at https://github.com/HaoyuJi/LaDy.

cs.CV↗

arXiv2026

Spectral Scalpel: Amplifying Adjacent Action Discrepancy via Frequency-Selective Filtering for Skeleton-Based Action Segmentation

Skeleton-based Temporal Action Segmentation (STAS) seeks to densely segment and classify diverse actions within long, untrimmed skeletal motion sequences. However, existing STAS methodologies face challenges of limited inter-class discriminability and blurred segmentation boundaries, primarily due to insufficient distinction of spatio-temporal patterns between adjacent actions. To address these limitations, we propose Spectral Scalpel, a frequency-selective filtering framework aimed at suppressing shared frequency components between adjacent distinct actions while amplifying their action-specific frequencies, thereby enhancing inter-action discrepancies and sharpening transition boundaries. Specifically, Spectral Scalpel employs adaptive multi-scale spectral filters as scalpels to edit frequency spectra, coupled with a discrepancy loss between adjacent actions serving as the surgical objective. This design amplifies representational disparities between neighboring actions, effectively mitigating boundary localization ambiguities and inter-class confusion. Furthermore, complementing long-term temporal modeling, we introduce a frequency-aware channel mixer to strengthen channel evolution by aggregating spectra across channels. This work presents a novel paradigm for STAS that extends conventional spatio-temporal modeling by incorporating frequency-domain analysis. Extensive experiments on five public datasets demonstrate that Spectral Scalpel achieves state-of-the-art performance. Code is available at https://github.com/HaoyuJi/SpecScalpel.

cs.CV↗

arXiv2026

RuCL: Stratified Rubric-Based Curriculum Learning for Multimodal Large Language Model Reasoning

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a prevailing paradigm for enhancing reasoning in Multimodal Large Language Models (MLLMs). However, relying solely on outcome supervision risks reward hacking, where models learn spurious reasoning patterns to satisfy final answer checks. While recent rubric-based approaches offer fine-grained supervision signals, they suffer from high computational costs of instance-level generation and inefficient training dynamics caused by treating all rubrics as equally learnable. In this paper, we propose Stratified Rubric-based Curriculum Learning (RuCL), a novel framework that reformulates curriculum learning by shifting the focus from data selection to reward design. RuCL generates generalized rubrics for broad applicability and stratifies them based on the model's competence. By dynamically adjusting rubric weights during training, RuCL guides the model from mastering foundational perception to tackling advanced logical reasoning. Extensive experiments on various visual reasoning benchmarks show that RuCL yields a remarkable +7.83% average improvement over the Qwen2.5-VL-7B model, achieving a state-of-the-art accuracy of 60.06%.

cs.CL↗

arXiv2026

AgentConductor: Topology Evolution for Multi-Agent Competition-Level Code Generation

Large language model(LLM)-driven multi-agent systems(MAS) coordinate specialized agents through predefined interaction topologies and have shown promise for complex tasks such as competition-level code generation. Recent studies demonstrate that carefully designed multi-agent workflows and communication graphs can significantly improve code generation performance by leveraging collaborative reasoning. However, existing methods neither adapt topology density to task difficulty nor iteratively refine the topology within an instance using execution feedback, which leads to redundant communication and performance bottlenecks. To address these issues, we propose AgentConductor: a reinforcement learning-optimized MAS with an LLM-based orchestrator agent as its core, which enables end-to-end feedback-driven dynamic generation of interaction topologies. For each query, AgentConductor infers agent roles and task difficulty, then constructs a task-adapted, density-aware layered directed acyclic graph (DAG) topology, underpinned by two key innovations. First, we design a novel topological density function that captures communication-aware mathematical characterizations of multi-agent interactions. Second, we adopt difficulty interval partitioning to avoid excessive pruning for precise topological density upper bound measurement per difficulty level and finer-grained control. Empirically, across three competition-level and two foundational code datasets, AgentConductor achieves state-of-the-art accuracy, outperforming the strongest baseline by up to 14.6% in pass@1 accuracy, 13% in density reduction, and 68% in token cost reduction.

cs.MA↗

arXiv2026

Structuring Reasoning for Complex Rules Beyond Flat Representations

Large language models (LLMs) face significant challenges when processing complex rule systems, as they typically treat interdependent rules as unstructured textual data rather than as logically organized frameworks. This limitation results in reasoning divergence, where models often overlook critical rule dependencies essential for accurate interpretation. Although existing approaches such as Chain-of-Thought (CoT) reasoning have shown promise, they lack systematic methodologies for structured rule processing and are particularly susceptible to error propagation through sequential reasoning chains. To address these limitations, we propose the Dynamic Adjudication Template (DAT), a novel framework inspired by expert human reasoning processes. DAT structures the inference mechanism into three methodical stages: qualitative analysis, evidence gathering, and adjudication. During the qualitative analysis phase, the model comprehensively evaluates the contextual landscape. The subsequent evidence gathering phase involves the targeted extraction of pertinent information based on predefined template elements ([placeholder]), followed by systematic verification against applicable rules. Finally, in the adjudication phase, the model synthesizes these validated components to formulate a comprehensive judgment. Empirical results demonstrate that DAT consistently outperforms conventional CoT approaches in complex rule-based tasks. Notably, DAT enables smaller language models to match, and in some cases exceed, the performance of significantly larger LLMs, highlighting its efficiency and effectiveness in managing intricate rule systems.

cs.AI↗

arXiv2026

Hardwired-Neurons Language Processing Units as General-Purpose Cognitive Substrates

The rapid advancement of Large Language Models (LLMs) has established language as a core general-purpose cognitive substrate, driving the demand for specialized Language Processing Units (LPUs) tailored for LLM inference. To overcome the growing energy consumption of LLM inference systems, this paper proposes a Hardwired-Neurons Language Processing Unit (HNLPU), which physically hardwires LLM weight parameters into the computational fabric, achieving several orders of magnitude computational efficiency improvement by extreme specialization. However, a significant challenge still lies in the scale of modern LLMs. A straightforward hardwiring of gpt-oss 120 B would require fabricating photomask sets valued at over 6 billion dollars, rendering this straightforward solution economically impractical. Addressing this challenge, we propose the novel Metal-Embedding methodology. Instead of embedding weights in a 2D grid of silicon device cells, Metal-Embedding embeds weight parameters into the 3D topology of metal wires. This brings two benefits: (1) a 15x increase in density, and (2) 60 out of 70 photomask layers are homogeneous across chips, including all EUV photomasks. In total, Metal-Embedding reduced the photomask cost by 112x, bringing the Non-Recurring Engineering (NRE) cost of HNLPU into an economically viable range. Experimental results show that HNLPU achieved 249,960 tokens/s (5,555x/85x that of GPU/WSE), 36 tokens/J (1,047x/283x that of GPU/WSE), 13,232 mm2 total die area, $59.46 M-123.5 M estimated NRE at 5 nm technology. Analysis shows that HNLPU achieved 41.7-80.4x improvement in cost-effectiveness and 357x reduction in carbon footprint compared to OpenAI-scale H100 clusters, under an annual weight updating assumption.

cs.AR↗

arXiv2026

Improving the accuracy and generalizability of molecular property regression models with a substructure-substitution-rule-informed framework

Artificial Intelligence (AI)-aided drug discovery is an active research field, yet AI models often exhibit poor accuracy in regression tasks for molecular property prediction, and perform catastrophically poorly for out-of-distribution (OOD) molecules. Here, we present MolRuleLoss, a substructure-substitution-rule-informed framework that improves the accuracy and generalizability of multiple molecular property regression models (MPRMs) such as GEM and UniMol for diverse molecular property prediction tasks. MolRuleLoss incorporates partial derivative constraints for substructure substitution rules (SSRs) into an MPRM's loss function. When using GEM models for predicting lipophilicity, water solubility, and solvation-free energy (using lipophilicity, ESOL, and freeSolv datasets from MoleculeNet), the root mean squared error (RMSE) values with and without MolRuleLoss were 0.587 vs. 0.660, 0.777 vs. 0.798, and 1.252 vs. 1.877, respectively, representing 2.6-33.3% performance improvements. We show that both the number and the quality of SSRs contribute to the magnitude of prediction accuracy gains obtained upon adding MolRuleLoss to an MPRM. MolRuleLoss improved the generalizability of MPRMs for "activity cliff" molecules in a lipophilicity prediction task and improved the generalizability of MPRMs for OOD molecules in a melting point prediction task. In a molecular weight prediction task for OOD molecules, MolRuleLoss reduced the RMSE value of a GEM model from 29.507 to 0.007. We also provide a formal demonstration that the upper bound of the variation for property change of SSRs is positively correlated with an MPRM's error. Together, we show that using the MolRuleLoss framework as a bolt-on boosts the prediction accuracy and generalizability of multiple MPRMs, supporting diverse applications in areas like cheminformatics and AI-aided drug discovery.

cs.LG↗

arXiv2025

Training Report of TeleChat3-MoE

TeleChat3-MoE is the latest series of TeleChat large language models, featuring a Mixture-of-Experts (MoE) architecture with parameter counts ranging from 105 billion to over one trillion,trained end-to-end on Ascend NPU cluster. This technical report mainly presents the underlying training infrastructure that enables reliable and efficient scaling to frontier model sizes. We detail systematic methodologies for operator-level and end-to-end numerical accuracy verification, ensuring consistency across hardware platforms and distributed parallelism strategies. Furthermore, we introduce a suite of performance optimizations, including interleaved pipeline scheduling, attention-aware data scheduling for long-sequence training,hierarchical and overlapped communication for expert parallelism, and DVM-based operator fusion. A systematic parallelization framework, leveraging analytical estimation and integer linear programming, is also proposed to optimize multi-dimensional parallelism configurations. Additionally, we present methodological approaches to cluster-level optimizations, addressing host- and device-bound bottlenecks during large-scale training tasks. These infrastructure advancements yield significant throughput improvements and near-linear scaling on clusters comprising thousands of devices, providing a robust foundation for large-scale language model development on hardware ecosystems.

cs.CL↗