Searcharxiv⌕ Search

arXiv subjects

Yong Xie

Publications and source records attributed to Yong Xie.

At least 19 recordsLinked to original sources

ReA-Bench: Benchmarking Responsible Agent Auditing

As agents perform tasks autonomously in local and cloud environments, auditing their execution trajectories can help service providers identify risks and improve agent harnesses and foundation models. Our analysis of production and red-team trajectories shows that some risks are repeatedly missed across auditing configurations. Overall recall obscures which risks are missed and whether higher recall comes with more false positives on legitimate behavior. Existing benchmarks offer limited insight into these failures. Generated trajectories can contain inconsistent tool calls and responses, conspicuous risky actions, or repetitive attack patterns. Grouping results by risk type also makes it difficult to tell whether auditors overlooked evidence, accepted an unsupported conclusion, or lacked information needed to make a judgment. We introduce ReA-Bench, a diagnostic benchmark for auditing responsible agent behavior. It organizes auditing challenges into six tiers: internal safety knowledge, contradictions within trajectories, evidence retrieval in long contexts, verification of sources and reasoning, indirect steering, and missing information. It includes challenging benign cases in which actions resemble risky ones but are consistent with the task and its authorization. It also provides paired trajectory versions to test whether auditors detect the same risks more reliably when the trajectory is shortened or missing information is supplied. Most samples are drawn from real production trajectories, with scenarios manually selected and adapted to preserve realistic tasks and risks that could plausibly arise during execution.

cs.AI↗

CausalVerify: End-to-End Verification of Causal Analyses by Language Models

Language models increasingly perform empirical analyses end to end, yet existing evaluations assess the written explanation or whether generated code executes, not whether the executed workflow recovers the intended causal estimand. We introduce CausalVerify, an execution-grounded benchmark for end-to-end causal analysis that follows a model from research-context interpretation to estimand recovery. It scores this workflow at four distinct layers: method recognition, design specification, executable implementation, and estimand recovery. It combines 259 real-paper contexts, 100 fixed-seed synthetic scenarios with executable reference estimates, and 23 paper-twin pairs in which a model commits to a design before seeing the data and its executed analysis is scored against a canonical estimator on the same realised dataset. Execution is not correctness. Among 426 model-written workflows that run without error, 15.5% fail verification, and a keyword score of the effect direction stated in the text is a poor proxy for recovery. In a 23-pair, nine-model paired study, replacing a model's committed design with the reference design and its execution conventions raises joint recovery of the point estimate and standard error from 15.0% to 51.5%; yet 48.5% of eligible seeds still fail under the reference design. The direction replicates on six pairs built afterwards under a frozen construction protocol, although on the two newest pairs the gain is confined to models from the family that built the references. Design specification is consequential but not sufficient: a plausible method and runnable code do not guarantee recovery, and even supplying the reference design and its conventions leaves substantial downstream failure. CausalVerify evaluates the executed workflow rather than its surface plausibility, and every reported number is recomputed from frozen artifacts by a single script.

cs.AI↗

Evidence-Grounded Auditing of Identification Assumptions in Climate-Policy Causal Evaluations

Difference-in-differences (DID) studies are widely used to evaluate climate policy, but assessing the evidence supporting their identification assumptions remains challenging. We introduce ARGUS, a structured language-model pipeline that audits reported evidence against an eleven-dimension assumption-implication-evidence rubric and abstains when relevant evidence cannot be retrieved. We evaluate ARGUS using injected flaws, economics papers, and a small pilot with reconciled labels. On the 11-flaw benchmark, ARGUS detects 73% of planted flaws, compared with 18% for a keyword-based pipeline. Across 26 economics papers, ARGUS abstains on about 40% of paper-dimension assessments for lack of retrievable evidence. In a five-paper pilot with labels reconciled by two annotators, it assigns a higher risk level than the labels on 25 of the 33 assessments it completes. A rule fixed before the labels arrived removes most of this in-sample; weighted agreement stays low. ARGUS provides evidence-linked risk reports that localize potential weaknesses for expert review, without adjudicating causal claims. Code and data: https://github.com/yonghongzhang-io/ARGUS

cs.CL↗

Modular and Cost-effective Scanning Photocurrent Microscopy System for Sub-micron characterization of 2D optoelectronic devices

Scanning photocurrent microscopy (SPCM) is a powerful technique for probing local optoelectronic phenomena in 2D semiconducting devices. However, commercial setups remain costly, complex and often lack flexibility and adaptability. In this work, we present a home-built SPCM platform built around the retrofitting of a conventional metallographic microscope by coupling it with different light sources (single-mode fiber-coupled lasers and multimode fiber-coupled high-power LEDs), a motorized XY stage, a digital camera and an electronic readout module. This system enables simultaneous acquisition of photocurrent and reflection intensity maps, requiring minimal modifications of the microscope. We reached sub-micron spatial resolution and high imaging fidelity by correlating photocurrent maps with reflection maps, optical micrographs and AFM topography data on different devices fabricated with different materials (InSe, MoS2, WSe2, Gr), on different substrates (Si/SiO2, compact disk). This work provides a reliable, accessible and reproducible high-performance SPCM platform that can be easily implemented in most laboratories for microscale optoelectronic characterization of 2D devices.

cond-mat.mtrl-sci↗

The Double Measurement Confound in Agent Benchmarks: De-Scaffolding, Ground-Truth Scoring, and Reliability Beyond the Mean

Agent benchmarks are increasingly used to compare large language models (LLMs) and guide deployment decisions, yet benchmark scores are meaningful only if they measure model capability rather than properties of the evaluation pipeline. We identify a double measurement confound: execution-critical decisions are performed by a fixed scaffold instead of the model, while the scorer evaluates outputs using criteria that may not reflect task correctness. We unify these issues within a measurement-theoretic framework that characterizes when benchmark scores can be interpreted as evidence of model capability, and instantiate it with an audit-and-repair protocol that (i) transfers execution-critical decisions from the scaffold to the model, (ii) replaces shape-based evaluation with seeded ground-truth scoring, and (iii) reports reliability beyond the mean through worst-case and tail-risk metrics. Experiments on ComtradeBench show that the joint intervention transforms a nearly flat leaderboard into a reliability spectrum that distinguishes both average performance and robustness across seeds. Applying the audit to existing benchmarks further shows that scorer validity is benchmark-specific, whereas scaffold ownership is an uncontrolled axis wherever we probed it. Our results suggest that benchmark scores should be interpreted together with their scaffolding level, scoring criterion, and reliability profile, providing a practical framework for more valid evaluation of LLM agents.

cs.SE↗

Oyster-II: Reinforcement Learning for Constructive Safety Alignment in Large Language Models

Large language models (LLMs) have demonstrated remarkable capabilities across diverse applications, yet ensuring their simultaneous safety, helpfulness, and trustworthiness remains a persistent challenge. Conventional refusal-oriented alignment strategies mitigate harmful content generation but systematically fail to serve legitimate user needs, often withholding information that could safely and constructively address the underlying intent of sensitive queries. Building upon the constructive safety paradigm pioneered by Oyster-I, which moves beyond blanket refusal toward thoughtful, response-oriented safety alignment, we identify two critical limitations of its Supervised Fine-Tuning (SFT)-based scheme: insufficient safety generalization to out-of-distribution scenarios and a phenomenon we term safety chain-of-thought (CoT) over-generalization, wherein safety-oriented reasoning patterns are excessively applied to benign queries, degrading helpfulness and user experience. To address these limitations, we propose Oyster-II, a reinforcement learning (RL)-based constructive safety alignment framework that adopts a Zero-RL paradigm combined with a multi-stage reinforcement learning strategy.Evaluated across extensive benchmarks, Oyster-II comprehensively surpasses both Qwen3-14B and its predecessor Oyster-I on safety dimensions, achieving cross-scale performance comparable to Qwen3-Max and Qwen3.5-397B.

cs.AI↗

MAD-OPD: Breaking the Ceiling in On-Policy Distillation via Multi-Agent Debate

On-policy distillation (OPD) trains a student on its own trajectories under token-level teacher supervision, but existing methods are capped by a single-teacher capability ceiling: when the teacher errs, the student inherits the error. OPD also remains largely unexplored in agentic tasks, where per-step errors compound across long trajectories and destabilize training. We propose MAD-OPD (Multi-Agent Debate-driven On-Policy Distillation), which breaks this ceiling by recasting the distillation teacher as a deliberative collective of teachers that debate over the student's on-policy state; the debate produces an emergent collective intelligence that supplies token-level supervision, with each teacher's contribution weighted by its post-debate confidence. To extend OPD to agentic tasks, we also introduce On-Policy Agentic Distillation (OPAD), which adds step-level sampling to stabilize training under multi-step error compounding. We additionally derive a task-adaptive divergence principle, selecting JSD (Jensen-Shannon divergence) for agentic stability and reverse KL (Kullback-Leibler) divergence for code generation, and verify it both theoretically and empirically. Across six teacher-student configurations (Qwen3 and Qwen3.5; 1.7B-14B students, 8B-32B teachers) and five agentic and code benchmarks, MAD-OPD ranks first across all six configurations; on the 14B+8B$\to$4B setting it lifts the agentic average by $+2.4\%$ and the code average by $+3.7\%$ over the stronger single-teacher OPD.

cs.CL↗

Identification and Structural Characterization of Twisted Atomically Thin Bilayer Materials by Deep Learning

Two-dimensional materials are expected to play an important role in next-generation electronics and optoelectronic devices. Recently, twisted bilayer graphene and transition metal dichalcogenides have attracted significant attention due to their unique physical properties and potential applications. In this study we describe the use of optical microscopy to collect the color space of chemical vapor deposition (CVD) molybdenum disulfide ($\mbox{MoS}_2$), and the application of a semantic segmentation convolutional neural network (CNN) to accurately and rapidly identify thicknesses of $\mbox{MoS}_2$ flakes. A second CNN model is trained to provide precise predictions on the twist angle of CVD-grown bilayer flakes. This model harnessed a dataset comprising over 10,000 synthetic images, encompassing geometries spanning from hexagonal to triangular shapes. Subsequent validation of the deep learning predictions on twist angles was executed through the second harmonic generation and Raman spectroscopy. Our results introduce a scalable methodology for automated inspection of twisted atomically thin CVD-grown bilayer.

cond-mat.mtrl-sci↗

Building an Affordable Self-Driving Lab: Practical Machine Learning Experiments for Physics Education Using Internet-of-Things

Machine learning (ML) is transforming modern physics research, but practical, hands-on experience with ML techniques remains limited due to cost and complexity barriers. To address this gap, we introduce an affordable, autonomous, Internet-of-Things (IoT)-enabled experimental platform designed specifically for applied physics education. Utilizing an Arduino microcontroller, a customizable multi-wavelength light emitting diode (LED) array, and photosensors, our setup generates diverse, real-time optical datasets ideal for training and evaluating foundational ML algorithms, including traversal methods, Bayesian inference, and deep learning. The platform facilitates a closed-loop, self-driving experimental workflow, encompassing automated data collection, preprocessing, model training, and validation. Through systematic performance comparisons, we demonstrate the superior ability of deep learning to capture complex nonlinear relationships compared to traversal and Bayesian methods. At approximately $60, this open-source IoT platform provides an accessible, practical pathway for students to master advanced ML concepts, promoting deeper conceptual insights and essential technical skills required for the next generation of physicists and engineers.

physics.ed-ph↗

Toward Full Autonomous Laboratory Instrumentation Control with Large Language Models

The control of complex laboratory instrumentation often requires significant programming expertise, creating a barrier for researchers lacking computational skills. This work explores the potential of large language models (LLMs), such as ChatGPT, and LLM-based artificial intelligence (AI) agents to enable efficient programming and automation of scientific equipment. Through a case study involving the implementation of a setup that can be used as a single-pixel camera or a scanning photocurrent microscope, we demonstrate how ChatGPT can facilitate the creation of custom scripts for instrumentation control, significantly reducing the technical barrier for experimental customization. Building on this capability, we further illustrate how LLM-assisted tools can be extended into autonomous AI agents capable of independently operating laboratory instruments and iteratively refining control strategies. This approach underscores the transformative role of LLM-based tools and AI agents in democratizing laboratory automation and accelerating scientific progress.

cs.AI↗

PC-CrossDiff: Point-Cluster Dual-Level Cross-Modal Differential Attention for Unified 3D Referring and Segmentation

3D Visual Grounding (3DVG) aims to localize the referent of natural language referring expressions through two core tasks: Referring Expression Comprehension (3DREC) and Segmentation (3DRES). While existing methods achieve high accuracy in simple, single-object scenes, they suffer from severe performance degradation in complex, multi-object scenes that are common in real-world settings, hindering practical deployment. Existing methods face two key challenges in complex, multi-object scenes: inadequate parsing of implicit localization cues critical for disambiguating visually similar objects, and ineffective suppression of dynamic spatial interference from co-occurring objects, resulting in degraded grounding accuracy. To address these challenges, we propose PC-CrossDiff, a unified dual-task framework with a dual-level cross-modal differential attention architecture for 3DREC and 3DRES. Specifically, the framework introduces: (i) Point-Level Differential Attention (PLDA) modules that apply bidirectional differential attention between text and point clouds, adaptively extracting implicit localization cues via learnable weights to improve discriminative representation; (ii) Cluster-Level Differential Attention (CLDA) modules that establish a hierarchical attention mechanism to adaptively enhance localization-relevant spatial relationships while suppressing ambiguous or irrelevant spatial relations through a localization-aware differential attention block. Our method achieves state-of-the-art performance on the ScanRefer, NR3D, and SR3D benchmarks. Notably, on the Implicit subsets of ScanRefer, it improves the Overall@0.50 score by +10.16% for the 3DREC task, highlighting its strong ability to parse implicit spatial cues.

cs.CV↗

Balanced Anomaly-guided Ego-graph Diffusion Model for Inductive Graph Anomaly Detection

Graph anomaly detection (GAD) is crucial in applications like fraud detection and cybersecurity. Despite recent advancements using graph neural networks (GNNs), two major challenges persist. At the model level, most methods adopt a transductive learning paradigm, which assumes static graph structures, making them unsuitable for dynamic, evolving networks. At the data level, the extreme class imbalance, where anomalous nodes are rare, leads to biased models that fail to generalize to unseen anomalies. These challenges are interdependent: static transductive frameworks limit effective data augmentation, while imbalance exacerbates model distortion in inductive learning settings. To address these challenges, we propose a novel data-centric framework that integrates dynamic graph modeling with balanced anomaly synthesis. Our framework features: (1) a discrete ego-graph diffusion model, which captures the local topology of anomalies to generate ego-graphs aligned with anomalous structural distribution, and (2) a curriculum anomaly augmentation mechanism, which dynamically adjusts synthetic data generation during training, focusing on underrepresented anomaly patterns to improve detection and generalization. Experiments on five datasets demonstrate that the effectiveness of our framework.

cs.LG↗

Controlled Automatic Task-Specific Synthetic Data Generation for Hallucination Detection

We present a novel approach to automatically generate non-trivial task-specific synthetic datasets for hallucination detection. Our approach features a two-step generation-selection pipeline, using hallucination pattern guidance and a language style alignment during generation. Hallucination pattern guidance leverages the most important task-specific hallucination patterns while language style alignment aligns the style of the synthetic dataset with benchmark text. To obtain robust supervised detectors from synthetic datasets, we also adopt a data mixture strategy to improve performance robustness and generalization. Our results on three datasets show that our generated hallucination text is more closely aligned with non-hallucinated text versus baselines, to train hallucination detectors with better generalization. Our hallucination detectors trained on synthetic datasets outperform in-context-learning (ICL)-based detectors by a large margin of 32%. Our extensive experiments confirm the benefits of our approach with cross-task and cross-generator generalization. Our data-mixture-based training further improves the generalization and robustness of hallucination detection.

cs.CV↗

Efficient Continual Pre-training for Building Domain Specific Large Language Models

Large language models (LLMs) have demonstrated remarkable open-domain capabilities. LLMs tailored for a domain are typically trained entirely on domain corpus to excel at handling domain-specific tasks. In this work, we explore an alternative strategy of continual pre-training as a means to develop domain-specific LLMs over an existing open-domain LLM. We introduce FinPythia-6.9B, developed through domain-adaptive continual pre-training on the financial domain. Continual pre-trained FinPythia showcases consistent improvements on financial tasks over the original foundational model. We further explore simple but effective data selection strategies for continual pre-training. Our data selection strategies outperform vanilla continual pre-training's performance with just 10% of corpus size and cost, without any degradation on open-domain standard tasks. Our work proposes an alternative solution to building domain-specific LLMs cost-effectively.

cs.CL↗

Target Refocusing via Attention Redistribution for Open-Vocabulary Semantic Segmentation: An Explainability Perspective

Open-vocabulary semantic segmentation (OVSS) employs pixel-level vision-language alignment to associate category-related prompts with corresponding pixels. A key challenge is enhancing the multimodal dense prediction capability, specifically this pixel-level multimodal alignment. Although existing methods achieve promising results by leveraging CLIP's vision-language alignment, they rarely investigate the performance boundaries of CLIP for dense prediction from an interpretability mechanisms perspective. In this work, we systematically investigate CLIP's internal mechanisms and identify a critical phenomenon: analogous to human distraction, CLIP diverts significant attention resources from target regions to irrelevant tokens. Our analysis reveals that these tokens arise from dimension-specific over-activation; filtering them enhances CLIP's dense prediction performance. Consequently, we propose ReFocusing CLIP (RF-CLIP), a training-free approach that emulates human distraction-refocusing behavior to redirect attention from distraction tokens back to target regions, thereby refining CLIP's multimodal alignment granularity. Our method achieves SOTA performance on eight benchmarks while maintaining high inference efficiency.

cs.CV↗

Med-GLIP: Advancing Medical Language-Image Pre-training with Large-scale Grounded Dataset

Medical image grounding aims to align natural language phrases with specific regions in medical images, serving as a foundational task for intelligent diagnosis, visual question answering (VQA), and automated report generation (MRG). However, existing research is constrained by limited modality coverage, coarse-grained annotations, and the absence of a unified, generalizable grounding framework. To address these challenges, we construct a large-scale medical grounding dataset Med-GLIP-5M comprising over 5.3 million region-level annotations across seven imaging modalities, covering diverse anatomical structures and pathological findings. The dataset supports both segmentation and grounding tasks with hierarchical region labels, ranging from organ-level boundaries to fine-grained lesions. Based on this foundation, we propose Med-GLIP, a modality-aware grounding framework trained on Med-GLIP-5M. Rather than relying on explicitly designed expert modules, Med-GLIP implicitly acquires hierarchical semantic understanding from diverse training data -- enabling it to recognize multi-granularity structures, such as distinguishing lungs from pneumonia lesions. Extensive experiments demonstrate that Med-GLIP consistently outperforms state-of-the-art baselines across multiple grounding benchmarks. Furthermore, integrating its spatial outputs into downstream tasks, including medical VQA and report generation, leads to substantial performance gains. Our dataset will be released soon.

cs.CV↗

Oyster-I: Beyond Refusal -- Constructive Safety Alignment for Responsible Language Models

Large language models (LLMs) typically deploy safety mechanisms to prevent harmful content generation. Most current approaches focus narrowly on risks posed by malicious actors, often framing risks as adversarial events and relying on defensive refusals. However, in real-world settings, risks also come from non-malicious users seeking help while under psychological distress (e.g., self-harm intentions). In such cases, the model's response can strongly influence the user's next actions. Simple refusals may lead them to repeat, escalate, or move to unsafe platforms, creating worse outcomes. We introduce Constructive Safety Alignment (CSA), a human-centric paradigm that protects against malicious misuse while actively guiding vulnerable users toward safe and helpful results. Implemented in Oyster-I (Oy1), CSA combines game-theoretic anticipation of user reactions, fine-grained risk boundary discovery, and interpretable reasoning control, turning safety into a trust-building process. Oy1 achieves state-of-the-art safety among open models while retaining high general capabilities. On our Constructive Benchmark, it shows strong constructive engagement, close to GPT-5, and unmatched robustness on the Strata-Sword jailbreak dataset, nearing GPT-o1 levels. By shifting from refusal-first to guidance-first safety, CSA redefines the model-user relationship, aiming for systems that are not just safe, but meaningfully helpful. We release Oy1, code, and the benchmark to support responsible, user-centered AI.

cs.AI↗

Strata-Sword: A Hierarchical Safety Evaluation towards LLMs based on Reasoning Complexity of Jailbreak Instructions

Large language models (LLMs) have gained widespread recognition for their superior comprehension and have been deployed across numerous domains. Building on Chain-of-Thought (CoT) ideology, Large Reasoning models (LRMs) further exhibit strong reasoning skills, enabling them to infer user intent more accurately and respond appropriately. However, both LLMs and LRMs face the potential safety risks under jailbreak attacks, which raise concerns about their safety capabilities. Current safety evaluation methods often focus on the content dimensions, or simply aggregate different attack methods, lacking consideration of the complexity. In fact, instructions of different complexity can reflect the different safety capabilities of the model: simple instructions can reflect the basic values of the model, while complex instructions can reflect the model's ability to deal with deeper safety risks. Therefore, a comprehensive benchmark needs to be established to evaluate the safety performance of the model in the face of instructions of varying complexity, which can provide a better understanding of the safety boundaries of the LLMs. Thus, this paper first quantifies "Reasoning Complexity" as an evaluable safety dimension and categorizes 15 jailbreak attack methods into three different levels according to the reasoning complexity, establishing a hierarchical Chinese-English jailbreak safety benchmark for systematically evaluating the safety performance of LLMs. Meanwhile, to fully utilize unique language characteristics, we first propose some Chinese jailbreak attack methods, including the Chinese Character Disassembly attack, Lantern Riddle attack, and Acrostic Poem attack. A series of experiments indicate that current LLMs and LRMs show different safety boundaries under different reasoning complexity, which provides a new perspective to develop safer LLMs and LRMs.

cs.CY↗