arXiv subjects

Wei He

Publications and source records attributed to Wei He.

At least 37 records · Page 2Linked to original sources

arXiv2026

Claw-SWE-Bench: A Benchmark for Evaluating OpenClaw-style Agent Harnesses on Coding Tasks

General-purpose agents such as OpenClaw are increasingly used as autonomous tool users, but their coding ability is difficult to measure under SWE-bench: a generic agent does not by itself satisfy the clean Docker workspace, patch, and prediction contract required for scoring. We introduce Claw-SWE-Bench, a multilingual SWE-bench-style benchmark and adapter protocol that makes heterogeneous agent harnesses, or claws, comparable under fair settings including a fixed prompt, runtime budget, workspace contract, patch extraction procedure, and evaluator. The full benchmark contains 350 GitHub issue-resolution instances across 8 languages and 43 repositories, drawn from SWE-bench-Multilingual and SWE-bench-Verified-Mini after future-commit cleanup. We also release Claw-SWE-Bench Lite for faster validation, which is an 80-instance subset selected by a cost-aware, rank-aware procedure over 17 calibration columns. On the full benchmark, OpenClaw with a minimal direct-diff adapter scores only $19.1\%$ Pass@1, whereas the full adapter reaches $73.4\%$ with the same GLM 5.1 backbone, showing that adapter design is essential for enabling OpenClaw-style harnesses to perform coding tasks effectively. Across an OpenClaw $\times$ nine-model sweep and a five-claw $\times$ two-model sweep, model choice changes Pass@1 by $29.4$ pp and harness choice by $27.4$ pp under fixed models; systems with similar accuracy can differ substantially in total API cost. Claw-SWE-Bench therefore treats harness and cost accounting as first-class axes of SWE-style coding-agent evaluation, providing both a full benchmark and a low-cost reference set for reproducible comparison. The data is available at https://github.com/opensquilla/claw-swe-bench and https://huggingface.co/datasets/TokenRhythm/Claw-SWE-Bench.

cs.LG↗

arXiv2026

Cycle-resolved explainability of energy storage impacts on whole-system cost and emissions

Grid-scale storage is not intrinsically cost-reducing or emissions-reducing: whether a storage cycle lowers or raises whole-system cost and CO2 depends on when, where, how long, what charges it, and what it displaces. Existing studies often assess storage value in aggregate or through asset-level market response, leaving a blind spot between system value and operational actions. Here we resolve this blind spot by attributing whole-system cost and CO2 consequences to individual charge-discharge cycles across present-day (2023) and planned (2030) UK and EU power systems. We show that aggregate indicators conceal sharply different cycle-level outcomes. In today's systems, commercially rational battery dispatch can reduce cost while increasing emissions, whereas long-duration energy storage more consistently captures infrequent cross-timescale co-benefit cycles that reduce both. Changing the dispatch objective redistributes storage operation within the existing opportunity set, but supply-side decarbonisation shifts that set itself, expanding co-benefit opportunities across technologies. This complementary structure shows that batteries serve recurrent opportunities, while long-duration storage captures stress-period co-benefits. The results identify gaps between market arrangements and whole-system value, and provide a basis for storage portfolio design, siting, and carbon-sensitive market reform that targets the operating regimes through which storage creates public value.

physics.soc-ph↗

arXiv2026

Low-Temperature Suppression of Intertwined Orders in La$_{1/3}$Sr$_{2/3}$FeO$_{3}$ Thin Films

The strong coupling between spin, charge, and lattice degrees of freedom in perovskite oxides leads to an array of exotic phenomena, giving these materials rich phase diagrams that can include coupled orders. This is exemplified by the A-site doped ferrite La$_{1/3}$Sr$_{2/3}$FeO$_{3}$ (LSFO), which exhibits a coupled paramagnetic-antiferromagnetic and charge ordering phase transition at $\sim$190 K that has been well studied in thin films, bulk, and polycrystalline samples. However, the low temperature behavior of LSFO thin films below $\sim$100 K has not been thoroughly explored. This work uses several X-ray scattering and spectroscopy techniques to directly probe LSFO's magnetic and charge order down to low temperature. Using resonant X-ray scattering, we observe a complete suppression of LSFO's known antiferromagnetic and charge order below $\sim$25 K. Further spectroscopy and coherent scattering measurements provide insight into LSFO's electronic structure and domain dynamics in this new low temperature phase, and we propose possible explanations for the observed order suppression based on reduced dimensionality of domains in our thin films. Our findings provide insight into the effects of competing interactions in strongly correlated materials, particularly those with coupled orders.

cond-mat.str-el↗

arXiv2026

Ultrafast Magneto-optical Fingerprints of Altermagnetism in MnTe

Recently identified altermagnets exhibit a distinctive dual-space nature: they possess spin-split electronic bands akin to ferromagnets in momentum space while maintaining the fully compensated magnetization of antiferromagnets in real space. This inherent duality, originating from the same crystal symmetry, gives rise to various intriguing physical phenomena unique to altermagnets. Consequently, a robust and efficient experimental signature capable of revealing this dual character is critically needed. The magneto-optical Kerr and Voigt effects, given their high sensitivity to ferromagnetism and antiferromagnetism, respectively, are ideally suited to probe this duality. Here, using time-resolved pump-probe magneto-optical measurements, we report the coexistence of pronounced Kerr and Voigt effects in the altermagnet MnTe. Combining the magnetization measurement and first-principles calculations, we demonstrate that the Kerr effect originates from the intrinsic Berry curvature of altermagnetism distribution in momentum space, while the Voigt effect arises from an anisotropic permittivity induced by the in-plane N\'eel order in real space, directly revealing the dual-space nature of altermagnets. Furthermore, the transient Kerr signal exhibits faster relaxation dynamics than the transient Voigt signal, underscoring their distinct origins in Berry curvature and N\'eel order, respectively. These findings establish transient magneto-optical responses as distinctive fingerprints of altermagnetism and position altermagnets as promising platforms for manipulating magneto-optical phenomena in ultrafast spin optoelectronics.

cond-mat.mtrl-sci↗

arXiv2026

Sibyl-AutoResearch: Autonomous Research Needs Self-Evolving Trial-and-Error Harnesses, Not Paper Generators

Autonomous research systems increasingly make the scientific workflow executable: agents can propose ideas, run code, inspect results, and draft papers. But executable workflows do not by themselves produce research judgment. We analyze where current systems lose trial experience: weak evidence becomes prose, pilot signals become broad claims, memory remains textual, and recurring process failures do not change later behavior. We introduce Sibyl-AutoResearch, a self-evolving AutoResearch framework built around Scientific Trial-and-Error Harnesses. A harness lets agents run bounded trials, preserve positive and negative outcomes, and route lessons into later planning, validation, claim scope, scheduling, critique, writing, and harness repair. We formalize this through two auditable conversion units: trial-to-behavior conversion, which links trial signals to later research actions, and trial-to-harness-behavior conversion, which links recurring process failures to system updates. We implement the framework in SIBYL, a file-backed autonomous research system that exposes the state, roles, memory, gates, and artifact traces needed to inspect these conversion paths. A retrospective audit identifies eight high-confidence conversion events, with a median latency of one iteration and a maximum latency of three iterations. A recovered-failure registry further shows how five naturally occurring failure classes, including duplicate results, stale numbers, and unsupported statistics, were blocked, downgraded, or routed into later repair. These traces do not establish a comparative performance claim; they show that the proposed conversion units are recoverable from realistic autonomous-research workspaces. The SIBYL framework and system are available at https://github.com/Sibyl-Research-Team/AutoResearch-SibylSystem.

cs.MA↗

arXiv2026

Self-Consistent Latent Reasoning: Long Latent Sequence Reasoning for Vision-Language Model

In language reasoning, longer chains of thought consistently yield better performance, which naturally suggests that visual latent reasoning may likewise benefit from longer latent sequences. However, we discover a counterintuitive phenomenon: the performance of existing latent visual reasoning methods systematically degrades as the latent sequence grows longer. We reveal the root cause: Information Gain Collapse -- autoregressive generation makes each step highly dependent on prior outputs, so subsequent tokens can barely introduce new information. We further identify that heavily pooled ($\geq 128\times$) image embeddings used as supervision targets provide no more signal than meaningless placeholders. Motivated by these insights, we propose SCOLAR (Self-COnsistent LAtent Reasoning), which introduces a lightweight detransformer that leverages the LLM's full-sequence hidden states to generate auxiliary visual tokens in a single shot, with each token independently anchored to the original visual space. Combined with three-stage SFT and ALPO reinforcement learning, SCOLAR extends acceptable latent CoT length by over $30\times$, achieves state-of-the-art among open-source models on real-world reasoning benchmarks (+14.12% over backbone), and demonstrates strong out-of-distribution generalization.

cs.CV↗

arXiv2026

SenseBench: A Benchmark for Remote Sensing Low-Level Visual Perception and Description in Large Vision-Language Models

Low-level visual perception underpins reliable remote sensing (RS) image analysis, yet current image quality assessment (IQA) methods output uninterpretable scalar scores rather than characterizing physics-driven RS degradations, deviating markedly from the diagnostic needs of RS experts. While Vision-Language Models (VLMs) present a compelling alternative by delivering language-grounded IQA, their visual priors are heavily biased toward ground-level natural images. Consequently, whether VLMs can overcome this domain gap to perceive and articulate RS artifacts remains insufficiently studied. To bridge this gap, we propose \textbf{SenseBench}, the first dedicated diagnostic benchmark for RS low-level visual perception and description. Driven by a physics-based hierarchical taxonomy that unifies both non-reference and reference-based paradigms, SenseBench features over 10K meticulously curated instances across 6 major and 22 fine-grained RS degradation categories. Specifically, two complementary protocols are designed for evaluation: objective low-level visual \textit{perception} and subjective diagnostic \textit{description}. Comprehensive evaluation of 29 state-of-the-art VLMs reveals not only skewed domain priors and multi-distortion collapse, but also \textit{fluency illusion} and a \textit{perception-description inversion} effect. We hope SenseBench provides a robust evaluation testbed and high-quality diagnostic data to advance the development of VLMs in RS low-level perception. Code and datasets are available \href{https://github.com/Zhong-Chenchen/SenseBench}{\textcolor{blue}{here}}.

cs.CV↗

arXiv2026

Compound effects of traffic and climate on electric vehicle HVAC energy consumption: a spatiotemporal framework with city-level attribution

Real-world electric vehicle (EV) energy consumption can deviate by 20-40% from rated values, driven by ambient temperature, traffic congestion, and route characteristics. Existing studies treat these factors in isolation or as static loads, leaving the compound effect of co-varying climate and traffic on HVAC energy unquantified and per-route attribution unavailable. We develop a spatiotemporal simulation framework that couples traffic-aware driving speed, time- and location-specific ambient temperature, and physics-based submodels (cabin HVAC, traction, battery thermal management) at the segment level, paired with a regression-based decomposition that attributes HVAC variability to temperature and trip-duration components on a per-route basis. Applied through a factorial design across seven UK cities and eight radial routes from Manchester, the framework shows total energy varying by 14\% across cities while HVAC energy varies by up to 89\%, making cabin thermal management the primary differentiator under winter conditions. Trip duration, set by traffic and road type, is frequently the dominant driver of HVAC variability: in London, 83\% of above-average HVAC energy is attributable to congestion-extended trip time rather than to temperature. The decomposition yields a closed-form HVAC model from three inputs (ambient temperature, average speed, trip distance), with physically interpretable coefficients and straightforward transfer to other vehicles or regions through three coefficient re-fits. EV range variability is substantially shaped by traffic and road-network characteristics, with implications for route planning, infrastructure design, and energy equity.

physics.soc-ph↗

arXiv2026

Food-Embedded Cold Energy Flows in Decentralised Solar Cold Chains

Reliable cold storage is needed to reduce meat loss in informal food systems, but conventional cold-chain expansion is difficult where electricity supply is weak and battery-based solar refrigeration is costly. This study develops an hourly techno-economic optimisation framework for decentralised solar-powered cold storage in interconnected open-air meat markets. Using five meat markets in Abuja, Nigeria, the model combines field-derived cooling demand, solar photovoltaic generation, refrigeration, battery storage, phase change material thermal storage, and directed inter-market meat flows. A key feature is that pre-chilled meat is represented as a carrier of product-embodied cooling credit, while phase change material storage remains a stationary cold-side storage component at each market. This allows cooling supplied at one market to reduce the sensible cooling load required at another market. Results show that shifting part of the storage function from battery storage to phase change material thermal storage reduces battery capacity by approximately 67\% and lowers total system cost by up to 15\% compared with battery-only systems. Allowing inter-market cooling-credit exchange further reduces total system cost by 8\% and aggregate phase change material storage capacity by 35\%, mainly by reallocating refrigeration and storage requirements across connected markets. The findings show that product flows can change where cooling services are required, allowing refrigeration and storage capacity to be coordinated across connected sites. Accounting for the product-mediated redistribution of cooling demand extends decentralised energy-system planning beyond isolated demand nodes and may inform cluster-level cooling infrastructure design in other infrastructure-constrained food networks.

physics.soc-ph↗

arXiv2026

Gated Multimodal Learning for Interpretable Property Energy Performance Prediction and Retrofit Scenario Analysis

Achieving resilient and sustainable cities requires scalable approaches to decarbonising residential buildings, which account for about 20% of UK greenhouse gas emissions and 25% of energy-related emissions in the European Union. Energy Performance Certificates (EPCs) support regulation and retrofit planning, but their reliance on on-site inspections limits timely city-scale assessment. This study introduces a gated multimodal model to predict Standard Assessment Procedure (SAP) energy efficiency and Environmental Impact (EI) scores by integrating EPC tabular variables, assessor-written free text, and Geographic Information System (GIS)-derived spatial features describing footprint geometry, height, area, and orientation. Sample-wise gating learns property-specific modality weights, while an auxiliary band classification head stabilises training. In a Westminster, London case study, the model predicts SAP and EI scores with MAEs of 4.03 and 4.76 points and R2 values of 0.757 and 0.748, respectively, achieving a mean MAE of 4.39. Ablation results show that full multimodal fusion outperforms unimodal and bimodal baselines for both score prediction and band-level classification. Interpretability analyses provide decision-relevant evidence: gating weights indicate strong reliance on assessor text; SHAP highlights main fuel, built form, and construction age band; text occlusion prioritises roof and wall fields; and spatial attribution is dominated by height and footprint area, with sensitivity to footprint shape. The validated framework is further applied to retrofit scenarios for wall insulation, roof insulation, and window glazing upgrades, indicating projected improvements in SAP, EI, annual energy cost, and equivalent CO2 emissions. Overall, the framework provides scalable property-level evidence for retrofit screening, intervention prioritisation, and net-zero housing transitions.

cs.LG↗

arXiv2026

Hidden States Know Where Reasoning Diverges: Credit Assignment via Span-Level Wasserstein Distance

Group Relative Policy Optimization (GRPO) performs coarse-grained credit assignment in reinforcement learning with verifiable rewards (RLVR) by assigning the same advantage to all tokens in a rollout. Process reward models can provide finer-grained supervision, but they require step-level annotation or additional reward modeling. We show that hidden-state distributions contain a useful signal for local reasoning quality that can be extracted using only outcome-level correctness labels available in RLVR. Specifically, within each GRPO group, the Wasserstein distance between span-level hidden state distributions of correct and incorrect rollouts increases around regions where their local reasoning quality diverges. This association holds both across examples and within individual trajectories, suggesting that hidden-state distributional divergence can serve as a self-supervision signal for fine-grained credit assignment. We formalize this observation with a separation theorem showing that, under mild structural assumptions, post-divergence spans have larger Wasserstein distances than pre-divergence spans whenever the population-level distributional gap exceeds finite-sample noise. Motivated by this result, we propose \textbf{S}pan-level \textbf{H}idden state \textbf{E}nabled \textbf{A}dvantage \textbf{R}eweighting (SHEAR), which modifies GRPO by using span-level Wasserstein distances to scale token-level advantages, amplifying updates on tokens whose hidden states are more separated from the opposing group. The method requires no additional model and only minimal changes to the training pipeline. Experiments on five mathematical reasoning benchmarks and five code generation benchmarks show improvements over standard GRPO and strong performance relative to supervised process reward models, while requiring no additional annotation or reward model training.

cs.CL↗

arXiv2026

More Than Meets the Eye: Measuring the Semiotic Gap in Vision-Language Models via Semantic Anchorage

Vision-Language Models (VLMs) excel at photorealistic generation, yet often struggle to represent abstract meaning such as idiomatic interpretations of noun compounds. To study whether high visual fidelity interferes with idiomatic compositionality under visual abstraction, we introduce DIVA, a controlled benchmark that replaces high-fidelity visual detail with schematic iconicity by generating paired, sense-anchored visualizations for literal and idiomatic readings. We further propose Semantic Alignment Gap ($\Delta$), an architecture-agnostic metric that quantifies divergence between literal and idiomatic visual grounding. We additionally introduce a directional signed bias $b(t)$ to separately measure the direction and strength of literal preference. Evaluating 8 recent VLMs, we reveal a consistent Literal Superiority Bias: model scale alone does not resolve literal preference, and increased visual fidelity is associated with weaker symbolic alignment, suggesting cognitive interference from hyper-realistic imagery. Our findings suggest that improving compositional understanding requires iconographic abstraction of visual input and anchoring interpretation and generation in intended meaning.

cs.CL↗

arXiv2026

Neural Network-Based Adaptive Event-Triggered Control for Dual-Arm Unmanned Aerial Manipulator Systems

This paper investigates the control problem of dual-arm unmanned aerial manipulator systems (DAUAMs). Strong coupling between the dual-arm and the multirotor platform, together with unmodeled dynamics and external disturbances, poses significant challenges to stable and accurate operation. An adaptive event-triggered control scheme with neural network-based approximation is proposed to address these issues while explicitly considering communication constraints. First, a dynamic model of the DAUAM system is derived, and a command-filter-based backstepping framework with error compensation is constructed. Then, a neural network is employed to approximate external frictions, and an event-triggered mechanism is designed to reduce the transmission frequency of control updates, thereby alleviating communication and energy burdens. Lyapunov-based analysis shows that all closed-loop signals remain bounded and that the tracking error converges to a neighborhood of the desired trajectory within a fixed time. Finally, experiments on a self-built DAUAM platform demonstrate that the proposed approach achieves accurate trajectory tracking.

cs.RO↗

arXiv2026

AgentV-RL: Scaling Reward Modeling with Agentic Verifier

Verifiers have been demonstrated to enhance LLM reasoning via test-time scaling (TTS). Yet, they face significant challenges in complex domains. Error propagation from incorrect intermediate reasoning can lead to false positives for seemingly plausible solutions, while lacking external grounding makes verifiers unreliable on computation or knowledge-intensive tasks. To address these challenges, we propose Agentic Verifier, a framework that transforms reward modeling into a multi-turn, tool-augmented deliberative process. We introduce complementary forward and backward agents: one traces solutions from premises to conclusions, while the other re-checks conclusions against their underlying premises. This bidirectional process enables a comprehensive, reliable, and interpretable assessment of solutions. To facilitate practical deployment, we propose AgentV-RL. Through proactive exploration and reinforcement learning, the verifier autonomously interleaves tool-use with internal reasoning. Extensive experiments show that Agentic Verifier yields consistent performance gains under both parallel and sequential TTS. Notably, our 4B variant surpasses state-of-the-art ORMs by 25.2%, positioning it as a promising paradigm for agentic reward modeling.

cs.CL↗

arXiv2026

Realistic Detector Geometry Modeling and Its Impact on Event Reconstruction in JUNO

JUNO is designed to determine the neutrino mass ordering with an energy resolution of 3% at 1 MeV. In the real detector, however, deformations of the central stainless-steel structure during installation lead to deviations of the photomultiplier tube (PMT) positions from their design values. Based on the limited survey data of the PMTs and the stainless-steel truss, we perform a correlation analysis of the measured points and propose a method to predict the positions of all PMTs. Using the resulting realistic geometry, we demonstrate that the detector deformation has a negligible effect on the energy reconstruction. In contrast, inaccuracies in the assumed geometry can introduce vertex biases of up to 40 mm. Incorporating the realistic geometry into the calibration-based PMT response model removes this bias and preserves the stability of the reconstruction algorithms.

physics.ins-det↗

arXiv2026

Graph-Based Fraud Detection with Dual-Path Graph Filtering

Fraud detection on graph data can be viewed as a demanding task that requires distinguishing between different types of nodes. Because graph neural networks (GNNs) are naturally suited for processing information encoded in graph form through their message-passing operations, methods based on GNN models have increasingly attracted attention in the fraud detection domain. However, fraud graphs inherently exhibit relation camouflage, high heterophily, and class imbalance, causing most GNNs to underperform in fraud detection tasks. To address these challenges, this paper proposes a Graph-Based Fraud Detection Model with Dual-Path Graph Filtering (DPF-GFD). DPF-GFD first applies a beta wavelet-based operator to the original graph to capture key structural patterns. It then constructs a similarity graph from distance-based node representations and applies an improved low-pass filter. The embeddings from the original and similarity graphs are fused through supervised representation learning to obtain node features, which are finally used by an ensemble tree model to assess the fraud risk of unlabeled nodes. Unlike existing single-graph smoothing approaches, DPF-GFD introduces a frequency-complementary dual-path filtering paradigm tailored for fraud detection, explicitly decoupling structural anomaly modeling and feature similarity modeling. This design enables more discriminative and stable node representations in highly heterophilous and imbalanced fraud graphs. Comprehensive experiments on four real-world financial fraud detection datasets demonstrate the effectiveness of our proposed method.

cs.LG↗

arXiv2026

SAGE: A Service Agent Graph-guided Evaluation Benchmark

The development of Large Language Models (LLMs) has catalyzed automation in customer service, yet benchmarking their performance remains challenging. Existing benchmarks predominantly rely on static paradigms and single-dimensional metrics, failing to account for diverse user behaviors or the strict adherence to structured Standard Operating Procedures (SOPs) required in real-world deployments. To bridge this gap, we propose SAGE (Service Agent Graph-guided Evaluation), a universal multi-agent benchmark for automated, dual-axis assessment. SAGE formalizes unstructured SOPs into Dynamic Dialogue Graphs, enabling precise verification of logical compliance and comprehensive path coverage. We introduce an Adversarial Intent Taxonomy and a modular Extension Mechanism, enabling low-cost deployment across domains and facilitating automated dialogue data synthesis. Evaluation is conducted via a framework where Judge Agents and a Rule Engine analyze interactions between User and Service Agents to generate deterministic ground truth. Extensive experiments on 27 LLMs across 6 industrial scenarios reveal a significant ``Execution Gap'' where models accurately classify intents but fail to derive correct subsequent actions. We also observe ``Empathy Resilience'', a phenomenon where models maintain polite conversational facades despite underlying logical failures under high adversarial intensity. Code and resources are available at https://anonymous.4open.science/r/SAGE-Bench-4CD3/.

cs.AI↗

arXiv2026

ONE-SHOT: Compositional Human-Environment Video Synthesis via Spatial-Decoupled Motion Injection and Hybrid Context Integration

Recent advances in Video Foundation Models (VFMs) have revolutionized human-centric video synthesis, yet fine-grained and independent editing of subjects and scenes remains a critical challenge. Recent attempts to incorporate richer environment control through rigid 3D geometric compositions often encounter a stark trade-off between precise control and generative flexibility. Furthermore, the heavy 3D pre-processing still limits practical scalability. In this paper, we propose ONE-SHOT, a parameter-efficient framework for compositional human-environment video generation. Our key insight is to factorize the generative process into disentangled signals. Specifically, we introduce a canonical-space injection mechanism that decouples human dynamics from environmental cues via cross-attention. We also propose Dynamic-Grounded-RoPE, a novel positional embedding strategy that establishes spatial correspondences between disparate spatial domains without any heuristic 3D alignments. To support long-horizon synthesis, we introduce a Hybrid Context Integration mechanism to maintain subject and scene consistency across minute-level generations. Experiments demonstrate that our method significantly outperforms state-of-the-art methods, offering superior structural control and creative diversity for video synthesis. Our project has been available on: https://martayang.github.io/ONE-SHOT/.

cs.CV↗