Search arXiv⌕ Search

arXiv subjects

Chen Qian

Publications and source records attributed to Chen Qian.

At least 19 recordsLinked to original sources
arXiv2026

Defense-as-Skill: Evolving Runtime Guard Skill for Skill-Augmented Agents

Skill-augmented agents load reusable skills as persistent runtime context, improving task performance but also giving malicious skills a durable channel for steering future actions. Such skills may leak secrets, corrupt code, bypass approvals, or stage data for exfiltration only after a concrete user task and workspace state make the unsafe action appear useful. This makes pre-install vetting insufficient and calls for runtime, task-conditioned protection. We propose Defense-as-Skill, a defense paradigm that implements the runtime guard itself as an installable, inspectable, and editable skill. Our guard, SkillSonar, runs alongside untrusted task skills and checks sensitive actions against the user's task boundary, routing each action to an allow, replan, or confirmation decision without modifying the underlying agent runtime. To study this setting, we construct SCOPE-R, a task-conditioned dataset covering 6 risk families and 21 sub-categories, with 206 attack-confirmed malicious instances and 43 benign tasks. We then improve SkillSonar on the SCOPE-R training subset using runtime guard-skill evolution, a Monte-Carlo Tree Search procedure that evolves the on-disk guard skill from feedback on the rollouts. Across Claude Code and OpenClaw, the evolved guard substantially reduces attack success while maintaining a favorable safety-utility trade-off; on repeated GLM-5 runs, ID ASR drops from 0.482 to 0.104 and OOD ASR from 0.606 to 0.115, with gains transferring across victim models, held-out risk families, and external benchmarks. These results show that runtime guards can be skill-native, transparent, and evolvable policy artifacts for skill-augmented agents.

cs.CR↗
arXiv2026

KVCompass: A Discovery Plane for Internet-Scale Prefix Cache Sharing in LLM Serving

KV cache reuse is one of the most effective ways to reduce LLM inference cost, but today its benefits are largely confined to a single cluster. When a provider serves the same model across geographically distributed regions, each region may repeatedly prefill the same popular prefixes, wasting GPU time and increasing time-to-first-token (TTFT). Exploiting reuse across regions involves discovering which worker currently holds a reusable KV cache for each request's prefix. We show that this KV cache discovery problem is structurally harder than conventional object lookup. Unlike hash-placement mechanisms such as DHTs, where an object's location is a deterministic function of its key, KV caches are created on the workers that serve requests, and their holder sets change with subsequent requests and KV cache evictions. KV cache discovery therefore calls for decoupling KV cache identity from placement. We present KVCompass, a discovery plane for Internet-scale KV cache sharing. The discovery index supports variable-length prefix lookup with constant-time local table reads at the edge, allowing each request to be forwarded to the worker that best balances KV cache reuse against load and network distance. Evaluation includes a 64-worker geo-distributed simulation and a 14-worker three-continent real-machine testbed. Evaluation shows that KVCompass achieves $1.3-14\times$ lower P99 TTFT than the baselines, up to $3.7\times$ lower P99 end-to-end latency, up to $4.7\times$ the throughput, and a peak load factor within 2% of the Oracle floor.

cs.DC↗
arXiv2026

Chemical-Potential-Dependent Superconducting Spectroscopy on Bi2Te3/NbSe2 Heterostructures

High-resolution scanning tunneling spectroscopy performed at dilution refrigerator temperatures was employed to investigate the superconducting properties of topological insulator/superconductor (Bi2Te3/NbSe2) heterostructures. Coherence peaks with inner kinks or inner peaks, deviating from the conventional single s-wave spectral line shape, were observed in both 5 quintuple-layer (5QL) and 7QL Bi2Te3 thin films. Based on the combination of chemical-potential-dependent spectroscopic measurements and model simulations, we conclude that these spectral features arise from contributions of bulk and surface states in Bi2Te3 by selectively proximity-coupled to a single superconducting gap from NbSe2, rather than inheriting the complicated superconductor's order parameter of NbSe2.

cond-mat.supr-con↗
arXiv2026

Collaborative Learning and Reasoning over Model Slices

Large language models (LLMs) excel across diverse tasks, benefiting from knowledge priors, in-context reasoning, and test-time computation. Despite these strengths, single-model reasoning remains vulnerable to error propagation, as later steps depend on the model's earlier outputs, leaving reasoning blind spots difficult to independently reassess. To address this limitation, we develop \textit{collaborative learning}, encompassing the construction of heterogeneous collaborators and the learning of effective cross-model collaboration. Specifically, \textit{model slicing} replaces a large model with smaller, heterogeneous models under a comparable parameter budget, enabling prior reasoning to be reassessed across distinct model parameterizations. \textit{Pyramid supervision} combines protocol distillation with step-level supervision to refine reasoning and interaction decisions throughout collaboration. We evaluate our approach across four task categories: multi-hop question answering, mathematical reasoning, code generation, and open-ended constrained generation. The trained heterogeneous models achieve comparable or better task performance than the original large model and established multi-agent baselines, including reinforcement-learning-based methods, while also achieving higher inference efficiency than the multi-agent baselines and exhibiting differentiated collaboration patterns across tasks. These findings show that learned collaboration turns model heterogeneity into complementary reasoning capabilities.

cs.AI↗
arXiv2026

What Does Multi-Agent Debate Actually Change?

Multi-agent debate, in which several LLMs exchange arguments before producing an answer, raises a basic question: does expressed disagreement reflect changes in the members' own positions? No single signal can settle this question, so we organize the analysis around five questions: (A) does the debater say it disagrees; (B) does its reply text actually argue; (C) does its own position change after each debate turn; (D) how much, quantitatively, does the position change; and (E) how do members' final positions compare with their initial ones? We evaluate two- and three-member committees on 50 curated opinion questions from GlobalOpinionQA, using same-model, same-family, and mixed-family configurations with friendly, neutral, and hostile instructions assigned at each turn. (A) Tone changes reported agreement: the share of replies reporting strong agreement is 70.7-96.3% under friendly instructions, compared with 9.5-19.3% under hostile ones, varying with model choice. (B) Self-reports broadly align with text judgments, but consistency varies by agreement level and model choice. (C) Position changes depend on the interaction: after a peer's leaning-disagree reply, members reporting strong agreement switch options more often than those reporting leaning disagreement. (D) For open-weight members, the probability of the option a member already holds stays near saturation, even after a peer's pushback, while endorsement of its earlier position text drops after a peer's argument relative to neutral filler, more for Qwen3.8-27B than Inkling. (E) The selected option is unchanged from members' initial to final positions in over 90% of comparisons in every configuration. Taken together, expressed disagreement need not translate into position revision, either within individual exchanges or over a complete debate.

cs.AI↗
arXiv2026

MASBench: Benchmarking LLM-based Multi-Agent Collaboration under Partial Observability

Large language models (LLMs) have progressively evolved into the core of autonomous agents. Building on this progress, LLM-based multi-agent systems (MAS) coordinate multiple agents into a synergistic team to accomplish complex tasks that exceed the capabilities of individual agents. The effectiveness of such systems depends not only on the agents themselves, but also on how collaboration mechanisms are designed and organized. Note that real-world collaboration is typically partially observable, where each agent can only access partial information about the environment due to physical or privacy-related constraints. However, many existing multi-agent benchmarks assume global observability, and leave limited support for systematically evaluating collaboration mechanisms. To bridge this gap, we introduce MASBench, a multi-agent collaboration benchmark designed under partially observable constraints. It is organized into three progressive task categories: Reasoning, Scheduling, and Game. Through this structure, we progressively evaluate three representative collaboration mechanisms: Protocol, Memory, and Routing. MASBench further provides deterministic evaluation metrics, including performance score, communication cost, and cost effectiveness, to characterize both collaboration outcomes and communication overhead. Experiments across diverse LLM backbones and mechanism configurations offer empirical guidance for effective MAS design. Code is available at: https://github.com/BUPT-GAMMA/MASBench

cs.AI↗
arXiv2026

Text-Centric Post-Training for Omni-Modal Reasoning

Improving joint audio-visual reasoning in Omni Large Language Models typically incurs substantial data construction and training costs. Our diagnostics reveal multi-hop reasoning difficulties despite correct answers to all corresponding single-hop questions and suggest partial decoupling in the local optimization of perception and reasoning objectives. This motivates post-training with different emphases on these capabilities. Text-only reasoning training yields gains across data sources, model scales, and families. With the best-performing text-only configuration, supervised fine-tuning followed by reinforcement learning (RL) raises Qwen2.5-Omni-7B's geometric mean of nine reasoning scores by 25.83% over the base model, outperforming the complete native audio-visual route with 56.6% fewer GPU-hours. Training on data synthesized entirely by a text-only LLM raises this geometric mean by 21.01% without audio-visual data in construction or training. However, text-only training degrades perception. We therefore propose a text-centric post-training paradigm: text-only training provides the main reasoning optimization, and reduced-data native audio-visual RL then refines perception. Refinement uses about 90% fewer input tokens than full-data audio-visual RL, restores perception above the base level, and retains 93.5% of the best-performing text-only pipeline's reasoning gain.

cs.CL↗
arXiv2026

OctoNest: Adaptive Cross-Device Execution through Stateful Control

Computer use agents are expanding from single-device operation toward cross-device systems that coordinate tasks across heterogeneous environments. Execution conditions are often only partially known at planning time and revealed through interaction. Failures may require intra-device modality switching or inter-device reassignment; failing to distinguish these cases can lead to repeated failures or premature termination. However, existing systems primarily scale up single-device agents without sufficiently distinguishing device-level and modality-specific execution conditions. We propose OctoNest, which coordinates stateful cross-device orchestration and iterative device-local modality control. Device Agents refine subtasks and select modalities, while an Orchestrator uses execution feedback to revise plans and device assignments. We also introduce CAPEBench, comprising 158 instances from 23 cross-device seed tasks with controlled perturbations. OctoNest leads all three quality metrics, improving Perfect Pass over the strongest baseline by 18.35 percentage points and reducing token cost per perfect pass by 39.8\%. Further analyses support the complementary roles of local refinement and global revision and demonstrate CAPEBench's ability to distinguish control limitations under changing execution conditions.

cs.CL↗
arXiv2026

CoEvoke: Evoking Multi-Agent Collaboration for Trajectory Collection

Training multi-agent collaboration requires trajectories of checking, feedback, and repair, yet task-solving rollouts may end in immediate approval or repeat unsuccessful attempts without useful correction. We introduce CoEvoke, a cross-domain framework that evokes these interactions during trajectory collection. Controlled perturbations create recoverable difficulties, while restricted reference-based guidance directs investigation. The resulting dataset contains 4,567 reconstructable trajectories over 4,130 software-engineering and reasoning tasks, linking public interactions to evidence, decisions, and task-native outcomes. Separate experiments over 2,287 same-task pairs show feedback increases of 47.72-100 percentage points across seven perturbation groups; guidance has smaller, model-dependent effects. With identical training tasks, decision-record counts, and updates, perturbation-enriched judge training improves ACPBench-Hard success from 35.38% to 42.12% (+6.73 points). SWE verifier training gains 7.60-9.40 points over interface-matched untrained references. Workflow and post-training evaluations preserve task-solving topology with unperturbed inputs and no private-reference guidance. These results support designing collection conditions to obtain corrective supervision that improves collaborative decisions with frozen solving agents.

cs.LG↗
arXiv2026

Rethinking Visual Token Compression for Video Large Language Models: A Simple Yet Strong Baseline

Video Large Language Models (Video LLMs) have achieved remarkable progress in video understanding, but their inference efficiency is constrained by the large number of visual tokens produced by long videos. Recent video token compression methods increasingly introduce sophisticated strategies for token selection, pruning, and merging. This raises a fundamental question: how much of compression performance can be obtained by simply preserving the structure encoded in the visual representations? We investigate this question with SimpleCluster, a simple and training-free baseline that performs position-aware cross-frame clustering in the visual feature space and represents each cluster using the mean of its original visual features. Extensive experiments across four video understanding benchmarks and three representative Video LLMs show that SimpleCluster achieves competitive or superior performance over recent compression methods across a wide range of token retention ratios, with particularly strong robustness under extremely low retention rates (e.g., 1%). To understand this behavior, we analyze the feature space preserved by different compression methods in terms of local approximation fidelity and global coverage. The results show that stronger downstream performance is consistently associated with better preservation of the original visual feature distribution, especially its global coverage. These findings highlight feature-space preservation as an important consideration for video token compression under highly constrained token budgets. Our code is available at https://github.com/xiaozhang79/SimpleCluster.

cs.CV↗
arXiv2026

LogicEnvGen: Task-Logic Driven Generation of Diverse Simulated Environments for Embodied AI

Simulated environments play an essential role in embodied AI, functionally analogous to test cases in software engineering. However, existing environment generation methods often emphasize visual realism (e.g., object diversity and layout coherence), overlooking a crucial aspect: logical diversity from the testing perspective. This limits the comprehensive evaluation of embodied agent adaptability and planning robustness across distinct simulated environments. To bridge this gap, we propose LogicEnvGen, a novel method driven by Large Language Models (LLMs) that adopts a top-down paradigm to generate logically diverse simulated environments as test cases for agents. Given an agent task, LogicEnvGen first analyzes its execution logic to construct decision-tree-structured behavior plans and then synthesizes a set of logical trajectories. Subsequently, it adopts a heuristic algorithm to refine the trajectory set, reducing redundant simulation. For each logical trajectory, which represents a potential task situation, LogicEnvGen correspondingly instantiates a concrete simulated environment. Furthermore, we introduce LogicEnvEval, a novel benchmark for simulated environment generation, with four quantitative metrics. Experimental results verify the lack of logical diversity in baselines and demonstrate that LogicEnvGen achieves 1.08-2.67x greater diversity, significantly improving the performance in revealing agent faults by 3.34%-72.00%.

cs.RO↗
arXiv2026

VIDEAS: Distilling Explicit Action Semantics from Demonstration Videos for World Models via Prior-Guided Simulation

World models learn internal representations of environment dynamics to predict future states, enabling agents to optimize action plans without physical interactions. However, developing world models that genuinely internalize underlying causal physical laws to explicitly reason about action preconditions and subsequent state transitions remains an open challenge. In this paper, we propose VIDEAS, a data distillation framework that transforms continuous physical dynamics from operational videos into explicit action semantics for foundation models. Specifically, it deconstructs visual demonstrations into discrete action trajectories and utilizes advanced vision-language models (VLMs) to extract structured knowledge encapsulating action preconditions and effects. To ensure physical consistency, we introduce a prior-guided trajectory simulation mechanism grounded within a text-based environment to rigorously validate the extracted knowledge. Notably, we incorporate negative trajectories to enrich knowledge completeness and enhance data diversity to mitigate cognitive bias. Furthermore, we present VIDEAS-WM, an 8B/9B-parameter suite of language-based world models trained on 34K high-quality samples derived from AgiBot-World dataset. Extensive experiments demonstrate that VIDEAS-WM establishes state-of-the-art performance in high-level embodied action semantic reasoning, exhibiting profound physical understanding and robust generalization across unseen scenarios.

cs.RO↗
arXiv2026

Tessera: Demand-Driven KV Cache Management for Retrieval-Augmented LLM Serving

RAG and retrieval-based agent memory both inject retrieved content into LLM prompts, as document chunks and recalled memory records, respectively. The same content can recur across requests at different prompt positions or after different preceding contexts, preventing reuse through conventional prefix caching. Our characterization finds that records recurring outside the matching prefix account for over 70% of injected memory tokens in agent-memory workloads. Composable KV-reuse methods enable reuse in such cases, but online serving introduces a management problem: a recurring unit's KV states may not yet exist, may have been evicted, or may reside on another node. We present Tessera, a disaggregated serving system that makes retrieval the control plane for KV reuse. By exposing the context units needed before model execution, retrieval allows Tessera to combine current demand with retrieval history, KV residency, and generation load to coordinate cache management and request routing. Generation nodes concurrently prepare locally cached, remotely cached, and missing states, while retaining newly computed states off the request's critical path. Across RAG and agent-memory workloads, Tessera lowers mean TTFT by up to 3.6x over SGLang and LMCache with EPIC at matched request rates, and sustains low TTFT at rates where the baselines saturate, while matching the answer quality of the underlying composition policy.

cs.DC↗
arXiv2026

Thinking with Cameras: Active Visual Reasoning via Dynamic Viewpoint Control for Surveillance Video Understanding

Large vision-language models (LVLMs) have recently achieved remarkable progress in general-purpose video understanding. However, their application to real-world surveillance remains challenging due to the lack of large-scale domain-specific datasets and the limitation of passive observation from fixed viewpoints. In surveillance scenarios, critical visual evidence can be easily missed when targets are distant, small, occluded, or move beyond the current camera view. In this work, we introduce CamVLM, a new framework for Thinking with Cameras, which enables LVLMs to actively acquire visual evidence in real-world surveillance by continuously controlling camera viewpoints. We first construct CCTV-Anomaly, a large-scale surveillance video understanding dataset containing 14,133 videos across 10 anomaly categories, with detailed captions and event annotations. We further formulate viewpoint control as an active visual perception problem and build CamTrack-53K, an object-centric viewpoint trajectory dataset for learning camera actions. Moreover, we propose a reinforcement learning based viewpoint policy optimization framework, which models camera control as a sequential decision-making process and learns long-horizon observation strategies beyond supervised trajectory imitation. Extensive experiments demonstrate that CamVLM achieves state-of-the-art performance under both passive observation and dynamic viewpoint settings, validating the effectiveness of active camera-based reasoning for surveillance video understanding. Our datasets, model, and code will be available at https://github.com/xiaozhang79/CamVLM.

cs.CV↗
arXiv2026

TabSieve: Explicit In-Table Evidence Selection for Tabular Prediction

Tabular prediction can benefit from in-table rows as few-shot evidence, yet existing tabular models typically perform instance-wise inference and LLM-based prompting is often brittle. Models do not consistently leverage relevant rows, and noisy context can degrade performance. To address this challenge, we propose TabSieve, a select-then-predict framework that makes evidence usage explicit and auditable. Given a table and a query row, TabSieve first selects a small set of informative rows as evidence and then predicts the missing target conditioned on the selected evidence. To enable this capability, we construct TabSieve-SFT-40K by synthesizing high-quality reasoning trajectories from 331 real tables using a strong teacher model with strict filtering. Furthermore, we introduce TAB-GRPO, a reinforcement learning recipe that jointly optimizes evidence selection and prediction correctness with separate rewards, and stabilizes mixed regression and classification training via dynamic task-advantage balancing. Experiments on a held-out benchmark of 75 classification and 52 regression tables show that TabSieve consistently improves performance across shot budgets, with average gains of 2.92% on classification and 4.45% on regression over the second-best baseline. Further analysis indicates that TabSieve concentrates more attention on the selected evidence, which improves robustness to noisy context.

cs.LG↗
arXiv2026

Semord: Learned Semantic-Preserving Placement and Low-Fanout Routing for Distributed Vector Search

Vector databases are increasingly deployed in distributed settings where different users, sites, or domains maintain vector data. Existing vector databases rely on a coordinator to record which shards store which parts of the vector space and to route each query to those shards. In a decentralized setting, peers may join, leave, or move data without a trusted node tracking every change, and outdated routing information can therefore send queries to the wrong peers or require contacting many peers, reducing vector retrieval recall and increasing network latency. We present Semord, a decentralized vector search overlay system that achieves high recall by routing each ANN query to a small set of relevant peers, without relying on a centralized coordinator. Semord addresses this problem by making semantic locality routable: 1) We propose VHash to place semantically related vectors near each other in the overlay key space while avoiding load imbalance, so that each query only needs to contact a small neighborhood of peers for distributed local ANN ranking. 2) We design VecDHT, a communication protocol that maintains decentralized routing, region metadata, churn resilience, and VHash updates under membership and workload changes. Our extensive experiments on a real testbed show that Semord improves recall by more than 15% and reduces contacted peers by over 60% compared with decentralized baselines. Semord also approaches the recall and latency of a centralized oracle baseline while reducing peak peer-local ANN index memory by more than 2X. Controlled large-scale simulations further show that Semord scales across real-world embedding workloads and remains robust under churn for scoped vector retrieval as a decentralized overlay.

cs.NI↗
arXiv2026

Nonlinear Magneto-Optical Probing of Time-Reversal Symmetry Breaking

Solid-state harmonic generation provides a nonlinear probe of symmetries encoded in electronic wave functions. In the subgap and weak-injection regime, time reversal pairs the harmonic responses driven by fields of opposite ellipticity, strongly suppressing elliptical dichroism in time-reversal-symmetric crystals. We show that, in a magnetic crystal, spin-orbit coupling transfers time-reversal-symmetry breaking from the spin sector to the orbital wave functions and lifts this pairing through the geometric phases of the electric-dipole current. Semiconductor-Bloch-equation calculations for centrosymmetric bilayer Cr2Ge2Te6 predict pronounced third-harmonic elliptical dichroism that reverses with the magnetization. Under linearly polarized driving, SOC-induced geometric-phase accumulation generates a nonlinear transverse current and strongly enhances the harmonic rotation and ellipticity. These results identify the geometric phase as a key microscopic contribution to the nonlinear magneto-optical response. This work establishes helicity-resolved harmonic emission and nonlinear polarimetry as complementary probes of spin-orbit-coupled magnetic order.

physics.optics↗
arXiv2026

VLA-Precision: Asymmetric Co-Bootstrapping for Efficient Real-World Online RL of Vision-Language-Action Models

Pretrained vision-language-action (VLA) models enable broad manipulation but remain unreliable in tasks demanding precision and repeatability. Applying real-world online reinforcement learning (RL) to VLA post-training enables autonomous trial-and-error improvement beyond demonstrations alone, but exposes two bottlenecks: 1) unreliable value signals can induce policy drift; 2) large-VLA overhead constrains throughput and sample efficiency. To address these challenges, we present VLA-Precision, an efficient real-world online RL framework featuring the Asymmetric Co-Bootstrapping (ACoB) algorithm and the ACoB-Stream architecture. Specifically, ACoB establishes asymmetric co-bootstrapping across timescales: early intervention-guided behavioral learning rapidly improves policy performance while enhancing online experience quality. As autonomous experience accumulates, global return propagation and local preference ranking progressively calibrate value estimates, yielding relative action advantages for reference-regularized policy improvement while suppressing drift. To enable ACoB on large VLAs, we develop ACoB-Stream, a closed-loop experience--policy architecture that establishes invariant-state decoupling and on-demand streaming as design principles, delivering up to 10.9$\times$ improvements in throughput and computational efficiency. Extensive evaluations on nine high-precision chemistry tasks across four categories and four robot embodiments show that VLA-Precision achieves 98.3\% mean success rate in 45.8 min/task, with 27.6 s episodes running at 1.2$\times$ and 1.8$\times$ the speeds of VLA and RL baselines. Resources are available at https://vla-precision.github.io.

cs.RO↗