arXiv subjects

Jianhao Chen

Publications and source records attributed to Jianhao Chen.

At least 19 recordsLinked to original sources

arXiv2026

NeuronTune: Fine-Grained Neuron Modulation for Balanced Safety-Utility Alignment in LLMs

Ensuring robust safety alignment while preserving utility is critical for the reliable deployment of Large Language Models (LLMs). However, current techniques fundamentally suffer from intertwined deficiencies: insufficient robustness against malicious attacks, frequent refusal of benign queries, degradation in generated text quality and general task performance, the former two reflecting tensions in robust safety and the latter constituting utility impairment. We attribute these limitations to the coarse-grained layer-wise interventions in existing methods. To resolve this, we propose NeuronTune, a fine-grained framework that pinpoints and modulates sparse neurons to achieve simultaneous safety-utility optimization. Our approach first pinpoints safety-critical and utility-preserving neurons across all layers via attack-aware attribution, then adapts meta-learning to adaptively modulate their activations. Crucially, the intervention scope of NeuronTune is dynamically controlled via neuron inclusion thresholds, providing a flexible mechanism to prioritize either security-critical or utility-priority requirements. Extensive experimental results demonstrate that our method outperforms existing state-of-the-art technologies, achieving superior model safety while maintaining excellent utility.

cs.LG↗

arXiv2026

Every Picture Tells a Dangerous Story: Memory-Augmented Multi-Agent Jailbreak Attacks on VLMs

Vision-Language Models (VLMs) expand the attack surface of safety-aligned systems by coupling visual perception with text generation. Existing multimodal jailbreak attacks primarily rely on crafted visual content, adversarial perturbations, or image-specific attack strategies, leaving the potential of reusable visual anchors in benign natural images largely unexplored. To address the problem, we introduce MemJack, a memory-augmented multi-agent framework for automated VLM red-teaming with natural images. MemJack combines visual-anchor discovery, visual-semantic camouflage, response evaluation, reflection-guided repair, and dynamic replanning within a closed-loop attack pipeline. Beyond attack generation, MemJack automatically transforms public images into attack anchors, enables the construction of MemJack-Bench, a dataset of over 113,000 interactive multimodal jailbreak trajectories for safety evaluation and defensive alignment. Extensive empirical evaluations across full, unmodified COCO val2017 images demonstrate that MemJack achieves a 71.48\% attack success rate (ASR) against Qwen3-VL-Plus, scaling to 90\% under extended budgets. Compared with representative multimodal jailbreak baselines under the same natural-image evaluation setting, MemJack achieves the highest ASR while requiring fewer mean rounds to success, demonstrating superior jailbreak effectiveness on VLMs. These results demonstrate that benign natural images can act as transferable jailbreak anchors and reveal substantial vulnerabilities in current safety-aligned VLMs.

cs.AI↗

arXiv2026

ContiGuard: A Framework for Continual Toxicity Detection Against Evolving Evasive Perturbations

Toxicity detection mitigates the dissemination of toxic content (e.g., hateful comments, posts, and messages within online social actions) to safeguard a healthy online social environment. However, malicious users persistently develop evasive perturbations to disguise toxic content and evade detectors. Traditional detectors or methods are static over time and are inadequate in addressing these evolving evasion tactics. Thus, continual learning emerges as a logical approach to dynamically update detection ability against evolving perturbations. Nevertheless, disparities across perturbations hinder the detector's continual learning on perturbed text. More importantly, perturbation-induced noises distort semantics to degrade comprehension and also impair critical feature learning to render detection sensitive to perturbations. These amplify the challenge of continual learning against evolving perturbations. In this work, we present ContiGuard, the first framework tailored for continual learning of the detector on time-evolving perturbed text (termed continual toxicity detection) to enable the detector to continually update capability and maintain sustained resilience against evolving perturbations. Specifically, to boost the comprehension, we present an LLM-powered semantic enriching strategy, where we dynamically incorporate possible meaning and toxicity-related clues excavated by LLM into the perturbed text to improve the comprehension. To mitigate non-critical features and amplify critical ones, we propose a discriminability-driven feature learning strategy, where we strengthen discriminative features while suppressing the less-discriminative ones to shape a robust classification boundary for detection...

cs.CL↗

arXiv2026

Can a Small Model Learn to Look Before It Leaps? Dynamic Learning and Proactive Correction for Hallucination Detection

Hallucination in large language models (LLMs) remains a critical barrier to their safe deployment. For hallucination detection to be practical in real-world scenarios, the use of efficient small models is essential to ensure low latency and minimal resource consumption. However, existing methods rely on fixed verification strategies, where simply tuning small models to mimic fixed verification trajectories fails to capture the adaptability required for diverse hallucination patterns, thereby inducing planning instability. To address this limitation, we propose a ``Learning to Evaluate and Adaptively Plan'' (LEAP) framework, which shifts hallucination detection from fixed execution to dynamic strategy learning. Specifically, LEAP first employs a powerful teacher model to iteratively explore and refine verification strategies through a failure-driven loop. This dynamic planning capability is then distilled into an efficient student model, augmented by a novel proactive correction mechanism that enables the model to evaluate and optimize its verification strategy before execution. Experiments on three benchmarks demonstrate that LEAP outperforms state-of-the-art methods, offering an effective and scalable solution for reliable hallucination detection.

cs.AI↗

arXiv2026

Organizing, Orchestrating, and Benchmarking Agent Skills at Ecosystem Scale

The rapid proliferation of Claude agent skills has raised the central question of how to effectively leverage, manage, and scale the agent skill ecosystem. In this paper, we propose AgentSkillOS, the first principled framework for skill selection, orchestration, and ecosystem-level management. AgentSkillOS comprises two stages: (i) Manage Skills, which organizes skills into a capability tree via node-level recursive categorization for efficient discovery; and (ii) Solve Tasks, which retrieves, orchestrates, and executes multiple skills through DAG-based pipelines. To evaluate the agent's ability to invoke skills, we construct a benchmark of 30 artifact-rich tasks across five categories: data computation, document creation, motion video, visual design, and web interaction. We assess the quality of task outputs using LLM-based pairwise evaluation, and the results are aggregated via a Bradley-Terry model to produce unified quality scores. Experiments across three skill ecosystem scales (200 to 200K skills) show that tree-based retrieval effectively approximates oracle skill selection, and that DAG-based orchestration substantially outperforms native flat invocation even when given the identical skill set. Our findings confirm that structured composition is the key to unlocking skill potential. Our GitHub repository is available at:https://github.com/ynulihao/AgentSkillOS.

cs.CL↗

arXiv2026

Suppressed Rupture of Thin Metal Films via van der Waals Epitaxy

Ultrathin metal films exhibit liquid-like instabilities, rupturing via surface diffusion far below their melting points. This behavior constrains thermal budgets for advanced integrated circuits and emerging 2D-crystal devices. Here, we demonstrate that these instabilities can be fundamentally suppressed using graphene as a van der Waals (vdW) template. While conventional 20-nm-thick gold films break up into islands below 300 {\textdegree}C, templated films not only remain stable but also become structurally refined after annealing above 600 {\textdegree}C. This exceptional stability stems from a vdW-mediated crystallographic texture that reorganizes grain boundaries into a mechanically robust network. This mechanism significantly widens the processing window for nanoscale interconnects and enables high-temperature integration of metals with 2D-crystal technologies.

cond-mat.soft↗

arXiv2026

Bosonic phases across the superconductor-insulator transition in infinite-layer samarium nickelate

Superconductivity arises from the global phase coherence of Cooper pairs. Modulation of phase coherence leads to quantum phase transitions, serving as an important tool for studying unconventional superconductivity. Here, we demonstrate bosonic phases across the superconductor-insulator transition in infinite-layer nickelate superconducting films by the control of spatially periodic network patterns. Magnetoresistance oscillations with a periodicity of h/2e provide direct evidence of 2e Cooper pairing in nickelates. The phase transition is predominantly driven by enhanced superconducting fluctuations, and Cooper pairs are involved in charge transport across the transition. Notably, we observe two types of anomalous metallic phases, emerging respectively at finite magnetic fields and down to zero magnetic field. They can be characterized by bosonic excitations, suggesting the dynamic roles of vortices in the ground states. Our work establishes nickelates as a key platform for investigating the rich landscape of bosonic phases controlled via the phase coherence of Cooper pairs.

cond-mat.supr-con↗

arXiv2026

OpenMic: A Multi-Agent-Based Stand-Up Comedy Generation System

Chinese stand-up comedy generation goes beyond plain text generation, requiring culturally grounded humor, precise timing, stage-performance cues, and implicit multi-step reasoning. Moreover, commonly used Chinese humor datasets are often better suited for humor understanding and evaluation than for long-form stand-up generation, making direct supervision misaligned with the target task. To address these challenges, we present OpenMic, an end-to-end multi-agent system built on AutoGen that transforms a user-provided life topic into a 3-5 minute Chinese stand-up performance and further produces a narrated comedy video. OpenMic orchestrates multiple specialized agents in a multi-round iterative loop-planning to jointly optimize humor, timing, and performability. To mitigate the dataset-task mismatch, we augment generation with retrieval-augmented generation (RAG) for material grounding and idea expansion, and we fine-tune a dedicated JokeWriter to better internalize stand-up-specific setup-punchline structures and long-range callbacks.

cs.AI↗

arXiv2025

RAJ-PGA: Reasoning-Activated Jailbreak and Principle-Guided Alignment Framework for Large Reasoning Models

Large Reasoning Models (LRMs) face a distinct safety vulnerability: their internal reasoning chains may generate harmful content even when the final output appears benign. To address this overlooked risk, we first propose a novel attack paradigm, Reasoning-Activated Jailbreak (RAJ) via Concretization, which demonstrates that refining malicious prompts to be more specific can trigger step-by-step logical reasoning that overrides the model's safety protocols. To systematically mitigate this vulnerability, we further develop a scalable framework for constructing high-quality safety alignment datasets. This framework first leverages the RAJ attack to elicit challenging harmful reasoning chains from LRMs, then transforms these high-risk traces into safe, constructive, and educational responses through a tailored Principle-Guided Alignment (PGA) mechanism. Then, we introduce the PGA dataset, a verified alignment dataset containing 3,989 samples using our proposed method. Extensive experiments show that fine-tuning LRMs with PGA dataset significantly enhances model safety, achieving up to a 29.5% improvement in defense success rates across multiple jailbreak benchmarks. Critically, our approach not only defends against sophisticated reasoning-based attacks but also preserves, even enhances, the model's general reasoning capabilities. This work provides a scalable and effective pathway for safety alignment in reasoning-intensive AI systems, addressing the core trade-off between safety and functional performance.

cs.AI↗

arXiv2025

ICL-Router: In-Context Learned Model Representations for LLM Routing

Large language models (LLMs) often exhibit complementary strengths. Model routing harnesses these strengths by dynamically directing each query to the most suitable model, given a candidate model pool. However, routing performance relies on accurate model representations, and adding new models typically requires retraining, limiting scalability. To address these challenges, we propose a novel routing method using in-context vectors to represent model capabilities. The method proceeds in two stages. First, queries are embedded and projected into vectors, with a projector and LLM-based router trained to reconstruct the original queries, aligning vector representations with the router's semantic space. Second, each candidate model is profiled on a query set, and the router learns -- based on in-context vectors of query and model performance -- to predict whether each model can correctly answer new queries. Extensive experiments demonstrate that our method achieves state-of-the-art routing performance in both in-distribution and out-of-distribution tasks. Moreover, our method allows for seamless integration of new models without retraining the router. The code is available at https://github.com/lalalamdbf/ICL-Router.

cs.LG↗

arXiv2025

Beyond GPT-5: Making LLMs Cheaper and Better via Performance-Efficiency Optimized Routing

Balancing performance and efficiency is a central challenge in large language model (LLM) advancement. GPT-5 addresses this with test-time routing, dynamically assigning queries to either an efficient or a high-capacity model during inference. In this work, we present Avengers-Pro, a test-time routing framework that ensembles LLMs of varying capacities and efficiencies, providing a unified solution for all performance-efficiency tradeoffs. The Avengers-Pro embeds and clusters incoming queries, then routes each to the most suitable model based on a performance-efficiency score. Across 6 challenging benchmarks and 8 leading models -- including GPT-5-medium, Gemini-2.5-pro, and Claude-opus-4.1 -- Avengers-Pro achieves state-of-the-art results: by varying a performance-efficiency trade-off parameter, it can surpass the strongest single model (GPT-5-medium) by +7% in average accuracy. Moreover, it can match the average accuracy of the strongest single model at 27% lower cost, and reach ~90% of that performance at 63% lower cost. Last but not least, it achieves a Pareto frontier, consistently yielding the highest accuracy for any given cost, and the lowest cost for any given accuracy, among all single models. Code is available at https://github.com/ZhangYiqun018/AvengersPro.

cs.CL↗

arXiv2025

Learning Compact Representations of LLM Abilities via Item Response Theory

Recent years have witnessed a surge in the number of large language models (LLMs), yet efficiently managing and utilizing these vast resources remains a significant challenge. In this work, we explore how to learn compact representations of LLM abilities that can facilitate downstream tasks, such as model routing and performance prediction on new benchmarks. We frame this problem as estimating the probability that a given model will correctly answer a specific query. Inspired by the item response theory (IRT) in psychometrics, we model this probability as a function of three key factors: (i) the model's multi-skill ability vector, (2) the query's discrimination vector that separates models of differing skills, and (3) the query's difficulty scalar. To learn these parameters jointly, we introduce a Mixture-of-Experts (MoE) network that couples model- and query-level embeddings. Extensive experiments demonstrate that our approach leads to state-of-the-art performance in both model routing and benchmark accuracy prediction. Moreover, analysis validates that the learned parameters encode meaningful, interpretable information about model capabilities and query characteristics.

cs.AI↗

arXiv2025

Longitudinal magneto-thermal conductivity and magneto-Seebeck of itinerant antiferromagnetic BaMn$_2$Bi$_2$

Thermal transport, generally mediated by the direct microscopic exchange of kinetic energy via lattice phonons, can also be modified by contributions from additional quasiparticles, such as electrons and magnons. However, a comprehensive understanding of the magnon influence has yet to be realized and remains an active research area. The most significant roadblock has been a lack of available materials in which these three quasiparticles can be clearly identified and quantitatively examined in order to provide an intrinsic understanding, not only of their independent contributions to thermal conductivity but also of the cross-correlated interactions among them. Itinerant antiferromagnetic (AFM) BaMn$_{2}$Bi$_{2}$ with PT symmetry exhibits Anderson metal-insulator localization, which can be tuned into the metallic regime via an applied magnetic field due to its unique electron-magnon interactions. We identify itinerant AFM BaMn$_{2}$Bi$_{2}$ as an ideal material for scientific investigations into how these quasiparticles participate in thermal conductivity. Here, we present the direct contribution of electrons, phonons, and magnons to thermal conductivity, as well as their interspecies interactions, supported by detailed analyses conducted in the framework of the Boltzmann transport formalism. The comparison of the magneto-thermal conductivity and magneto-electrical conductivity, as well as the magneto-Seebeck effect of itinerant antiferromagnetic BaMn$_{2}$Bi$_{2}$, gives unique insight into how magnons participate in longitudinal thermal-associated phenomena.

cond-mat.str-el↗

arXiv2025

CAVGAN: Unifying Jailbreak and Defense of LLMs via Generative Adversarial Attacks on their Internal Representations

Security alignment enables the Large Language Model (LLM) to gain the protection against malicious queries, but various jailbreak attack methods reveal the vulnerability of this security mechanism. Previous studies have isolated LLM jailbreak attacks and defenses. We analyze the security protection mechanism of the LLM, and propose a framework that combines attack and defense. Our method is based on the linearly separable property of LLM intermediate layer embedding, as well as the essence of jailbreak attack, which aims to embed harmful problems and transfer them to the safe area. We utilize generative adversarial network (GAN) to learn the security judgment boundary inside the LLM to achieve efficient jailbreak attack and defense. The experimental results indicate that our method achieves an average jailbreak success rate of 88.85\% across three popular LLMs, while the defense success rate on the state-of-the-art jailbreak dataset reaches an average of 84.17\%. This not only validates the effectiveness of our approach but also sheds light on the internal security mechanisms of LLMs, offering new insights for enhancing model security The code and data are available at https://github.com/NLPGM/CAVGAN.

cs.CR↗

arXiv2025

Sliding two-dimensional superconductivity and charge-density-wave state in a bulk crystal

Superconductivity in the two-dimensional (2D) limit is a fertile ground for exotic quantum phenomena-many of which remain elusive in their 3D counterparts. While studies of 2D superconductivity have predominantly focused on mono- or few-layer systems, we demonstrate an alternative route-interlayer sliding in bulk crystals. Through a precisely controlled growth strategy, we engineer interlayer sliding in bulk 3R-NbSe2, deliberately disrupting [001] mirror symmetry and drastically suppressing interlayer coupling. Remarkably, this structural manipulation stabilizes Ising-type superconductivity coexisting with an unconventional charge-density-wave (CDW) state akin to that of monolayer 2H-NbSe2. The sliding phase exhibits a pronounced suppression of the upper critical field at low temperatures, revealing a delicate competition between Ising and Rashba spin-orbit coupling (SOC) in the globally noncentrosymmetric lattice. Intriguingly, the superconducting state displays two-fold symmetry, a signature that may arise from asymmetric SOC or a multi-component pairing order parameter. Our work establishes interlayer sliding as a symmetry-breaking tool to promote 2D superconductivity in bulk materials-without resorting to extrinsic intercalation or doping. More broadly, this approach sets a paradigm for unlocking hidden quantum states in layered materials, offering a new dimension in design of quantum matter.

cond-mat.supr-con↗

arXiv2025

Conflict Detection for Temporal Knowledge Graphs:A Fast Constraint Mining Algorithm and New Benchmarks

Temporal facts, which are used to describe events that occur during specific time periods, have become a topic of increased interest in the field of knowledge graph (KG) research. In terms of quality management, the introduction of time restrictions brings new challenges to maintaining the temporal consistency of KGs. Previous studies rely on manually enumerated temporal constraints to detect conflicts, which are labor-intensive and may have granularity issues. To address this problem, we start from the common pattern of temporal facts and propose a pattern-based temporal constraint mining method, PaTeCon. Unlike previous studies, PaTeCon uses graph patterns and statistical information relevant to the given KG to automatically generate temporal constraints, without the need for human experts. In this paper, we illustrate how this method can be optimized to achieve significant speed improvement. We also annotate Wikidata and Freebase to build two new benchmarks for conflict detection. Extensive experiments demonstrate that our pattern-based automatic constraint mining approach is highly effective in generating valuable temporal constraints.

cs.AI↗

arXiv2025

Stop Overvaluing Multi-Agent Debate -- We Must Rethink Evaluation and Embrace Model Heterogeneity

Multi-agent debate (MAD) has gained significant attention as a promising line of research to improve the factual accuracy and reasoning capabilities of large language models (LLMs). Despite its conceptual appeal, current MAD research suffers from critical limitations in evaluation practices, including limited benchmark coverage, weak baseline comparisons, and inconsistent setups. This paper presents a systematic evaluation of 5 representative MAD methods across 9 benchmarks using 4 foundational models. Surprisingly, our findings reveal that MAD often fail to outperform simple single-agent baselines such as Chain-of-Thought and Self-Consistency, even when consuming significantly more inference-time computation. To advance MAD research, we further explore the role of model heterogeneity and find it as a universal antidote to consistently improve current MAD frameworks. Based on our findings, we argue that the field must stop overvaluing MAD in its current form; for true advancement, we must critically rethink evaluation paradigms and actively embrace model heterogeneity as a core design principle.

cs.CL↗

arXiv2025

Aligning VLM Assistants with Personalized Situated Cognition

Vision-language models (VLMs) aligned with general human objectives, such as being harmless and hallucination-free, have become valuable assistants of humans in managing visual tasks. However, people with diversified backgrounds have different cognition even in the same situation. Consequently, they may have personalized expectations for VLM assistants. This highlights the urgent need to align VLM assistants with personalized situated cognition for real-world assistance. To study this problem, we first simplify it by characterizing individuals based on the sociological concept of Role-Set. Then, we propose to evaluate the individuals' actions to examine whether the personalized alignment is achieved. Further, we construct a benchmark named PCogAlignBench, which includes 18k instances and 20 individuals with different Role-Sets. Finally, we present a framework called PCogAlign, which constructs a cognition-aware and action-based reward model for personalized alignment. Experimental results and human evaluations demonstrate the reliability of the PCogAlignBench and the effectiveness of our proposed PCogAlign. We will open-source the constructed benchmark and code at https://github.com/NLPGM/PCogAlign.

cs.AI↗