Search arXiv⌕ Search

arXiv subjects

Bing Luo

Publications and source records attributed to Bing Luo.

At least 19 recordsLinked to original sources
arXiv2026

Towards Self-Evolving Narrative Systems: Real-Time Adaptation and Long-Term Optimization with LLMs

Traditional interactive narrative systems offer strong authorial control and structural coherence but often restrict player agency. In contrast, LLM-driven narratives enable open-ended generation and greater flexibility, yet frequently suffer from inconsistency and diminished narrative control. To navigate this trade-off, we propose a hybrid narrative framework that dynamically balances pre-authored content with real-time LLM generation via semantic similarity alignment. Additionally, sustaining narrative quality over extended gameplay remains a core challenge in LLM-based storytelling. To address this, we introduce a self-evolving, feedback-driven optimization framework that leverages in-context learning to iteratively refine prompts, eliminating the need for manual prompt engineering. Through a case study and participant-based evaluation, we show that our framework enhances narrative coherence while preserving player agency, effectively aligning generated content with both authorial intent and evolving player behavior.

cs.CL↗
arXiv2026

When Stored Evidence Stops Being Usable: Scale-Conditioned Evaluation of Agent Memory

Memory-agent evaluations report fixed-snapshot accuracy or retrieval quality, but these scores do not show whether evidence remains usable as irrelevant sessions (sessions not annotated as task-relevant evidence for the query) accumulate. We present a scale-conditioned evaluation protocol for agent memory under evidence-preserving growth: for each query, task evidence is held fixed while irrelevant sessions are added. The protocol logs agent-memory trajectories and reports four diagnostics: budget-compliant reliability, tail memory-call burden, failure-regime decomposition, and the usable-scale boundary where reliability falls below the target. Applied to LongMemEval and LoCoMo across flat, planar, and hierarchical memory interfaces, the protocol shows reliability loss is not a single phenomenon. On LongMemEval, HippoRAG stays within the two-call budget but loses 16-20 percentage points in budget-compliant reliability as irrelevant sessions are added; LiCoMemory's observed failures depend strongly on the agent, with Qwen3-8B exceeding the budget while Qwen3-32B and Qwen3-235B remain reliable in the tested range. The result supports a framework for making scalable-memory claims conditional on agent, interface, scale range, and interaction budget. Code and evaluation artifacts: https://github.com/AndyLu666/memory-at-scale

cs.AI↗
arXiv2026

ACED: Reference-Anchored Aggregation of Natural-Language and Executable Evidence for Mathematical Reasoning

Natural-language self-consistency and executable programs provide complementary test-time evidence for mathematical reasoning, yet either channel may support the same semantic error. Thus we propose ACED, a reference-anchored framework that treats the self-consistency answer as the default and executable outputs as supporting evidence rather than interchangeable votes. ACED converts source-specific answer counts into smoothed supports, combines them through a weighted logarithmic pool, and overrides the reference only when a calibrated relative-evidence margin is exceeded. The same rule supports both candidate selection and admission of executable-only answers. Across five math word-problem benchmarks, ACED matches or outperforms the corresponding baselines. On GSM8K, ACED-CANDIDATE recovers 89.7% of the attainable candidate-set gain with only three damaging overrides, while ACED-ALL reaches 86.9% accuracy versus 82.6% for direct Program-of-Thought aggregation. These results show that preserving evidence provenance enables more reliable integration of natural-language and executable reasoning.

cs.LG↗
arXiv2026

Do More Agents Help? Evaluating Accuracy and Cost in LLM Multi-Agent Workflows

Choosing whether to replace a single-agent controller requires evidence on both task performance and inference expenditure. We present BenchAgent, a shared evaluation framework, and compare six multi-agent workflow configurations against its single-agent reference, BenchAgent Core, across ten reasoning, coding, and tool-use benchmarks. All configurations use GPT-4.1 and common tool and scoring interfaces; Core executions share the same step limits. We measure accuracy and recorded token use over three runs on fixed task sets, with 3,302 instances per configuration per run. EvoAgent reaches 79.70% mean pooled accuracy versus 76.65% for Core and exceeds Core in all three runs; the other five MAS trail by 1.71-8.76 percentage points. LLM-Debate gains 9.67 points on HumanEval but loses 1.96 points in pooled accuracy while consuming 1.39 times the tokens of Core. ChatEval gains 22.25 points on IFEval but trails Core in pooled accuracy with 4.34 times its token consumption. The broad comparison and targeted analyses of retained responses provide evidence for workflow selection and identify answer preservation and resource allocation as directions for subsequent multi-agent optimization.

cs.AI↗
arXiv2026

FlexP-SFT: A Flexible Aggregation-Free Framework for On-Device Personalized Split Federated Fine-Tuning of LLMs

To fine-tune large language models (LLMs) over private data, federated learning (FL) has emerged as a promising paradigm. However, the prohibitive memory and communication demands of LLMs render standard FL impractical for resource-constrained edge devices. While split federated learning (SFL) alleviates the computing burdens via model partitioning, existing frameworks still suffer from communication bottlenecks and straggler problem due to the parameter aggregation process. To address these challenges, we propose FlexP-SFT, a novel aggregation-free framework for personalized split federated fine-tuning, which fundamentally eliminates the client-side aggregation process. Crucially, to ensure robust training in the absence of global synchronization, we introduce a layer-flexible alignment strategy to balance personalization and generalization capabilities. We further formulate split-ratio selection as a resource-aware discrete optimization problem that jointly accounts for personalization accuracy and system cost. Our proposed scheme simultaneously enhances personalized performance, reduces communication overhead, and resolves the straggler problem. Extensive results show that FlexP-SFT substantially outperforms baselines in both accuracy and latency, and that the optimized split ratio achieves a better resource-accuracy trade-off than static or memory-only choices.

cs.DC↗
arXiv2026

Just Talk Once: Communication-Efficient Split Federated LLM Fine-Tuning on Edge Devices

Large language model (LLM) fine-tuning is increasingly shifting toward data generated on edge devices, where memory, computation, bandwidth, and connectivity constraints make conventional federated learning difficult to sustain. Split federated fine-tuning (SFT) improves client-side efficiency by offloading most model parameters and computation to the server but requires step-by-step bidirectional communication loop across the split interface and forces continuous client involvement throughout training. In this paper, we present L-shaped SFT, a split fine-tuning framework that removes this bidirectional bottleneck. Our key insight is that weight tying in modern LLMs enables server-side hidden activations to be directly supervised using target embeddings, allowing the training loss to be computed on the server without returning server outputs to the client. To further eliminate the need for continuous client participation, based on L-shaped SFT, we introduce one-shot SFT, in which clients upload activations once and then go offline while the server continues optimization over cached representations. We implement our design in a real system testbed with heterogeneous edge clients, including commercial smartphones and NVIDIA developer boards. Experiments demonstrate that our schemes significantly reduce communication costs and client online time compared with existing SFT baselines.

cs.DC↗
arXiv2026

Two-fluid effects on the nonlinear dynamics of RFP relaxation

This study investigates the role of two-fluid effects during magnetic relaxation in reversed-field pinch (RFP) plasmas. Within the multiple-helicity (MH) regime, two-fluid simulations produce distinct sawtooth oscillations, in contrast to the sawtooth-free state obtained in single-fluid simulations. Analysis of the magnetic field aligned projection of Faraday's law reveals that, tearing modes collectively generate a dynamo electric field that sustains the magnetic relaxation, a process analogous to the flux-pumping in tokamaks. Despite the reduced linear tearing-mode growth rates, stronger two-fluid effects produce more pronounced sawtooth activity over the parameter range considered. Modal energy analysis shows that Hall-mediated nonlinear energy redistribution disrupts the coherent tearing-mode dynamics required to sustain steady flux-pumping, thereby facilitating intermittent reconnection. This transition is interpreted as a Hall-mediated dynamical bifurcation between steady flux-pumping and quasi-periodic sawtooth relaxation.

physics.plasm-ph↗
arXiv2026

Do Agent Benchmarks Measure Capability? Protocol Validity in the Age of Agentic AI

Agent benchmarks increasingly evaluate repository editing, web research, terminal use, and long-horizon interaction. Their scores support capability claims only when the evaluation protocol keeps the intended capability necessary for success. Recent reward-hacking benchmarks and system reports show that agents can instead recover public solutions, read evaluation artifacts, infer generator structure, manipulate feedback, or benefit from invalid scoring paths; existing responses do not provide a common procedure for attributing these shortcuts and quantifying their effect across benchmarks. We formulate protocol validity and introduce HackDetect, a post-hoc audit that identifies an exposure, determines how the agent used it, and assesses whether the resulting score is misleading. We quantify score inflation with the Mislead gap, defined as the exploit score minus the intended score. We audit 2,385 traces across 15 agent benchmarks and find evidence of exposures and reward hacking in 67.0% of Frontier Science traces and 66.7% of AutoLab tasks. Across paired comparisons, we measure score inflation of 0.45-1.00, showing that benchmark reports should provide evidence that scores reflect the intended capability.

cs.AI↗
arXiv2026

MobileFineTuner: A Mobile-Native Framework for On-Device LLM Fine-Tuning in Real-World Embedded AI Applications

Large language models (LLMs) are moving from cloud-centric services toward on-device embedded AI, where models interact with private, longitudinal signals sensed from users and their physical environments. Mobile phones are a natural platform for such applications because they are continuously carried by users, connected to wearable sensors, and deeply integrated with daily mobile applications. However, practical LLM fine-tuning on commodity phones remains difficult. Existing fine-tuning frameworks are largely Python-based and server-oriented, making them hard to deploy inside mobile applications. We present MobileFineTuner, a mobile-native open-source framework for end-to-end LLM fine-tuning on commodity mobile phones. MobileFineTuner is implemented in C++ and provides a reusable training stack. To make fine-tuning feasible under mobile resource constraints, MobileFineTuner integrates a resource-aware training runtime with memory-efficient attention, activation checkpointing, gradient accumulation, parameter sharding, and energy-aware scheduling. We evaluate MobileFineTuner on real mobile phones using GPT-2, Gemma 3, and Qwen2.5 models across multiple fine-tuning tasks. The results show that MobileFineTuner reproduces standard Full-FT and LoRA fine-tuning behavior, substantially reduces memory pressure and improves executability on memory-constrained phones. We further demonstrate MobileFineTuner through a private campus health-agent application, where a local LLM is fine-tuned on user-specific wearable-sensing records to provide more personalized responses while keeping raw records on the phone. These results establish MobileFineTuner as a practical toolkit for studying and building on-device LLM fine-tuning applications in embedded AI and sensing systems.

cs.LG↗
arXiv2026

EdgeFlowerTune: Evaluating Federated LLM Fine-Tuning Under Realistic Edge System Constraints

Federated fine-tuning offers a promising paradigm for adapting large language models (LLMs) on edge devices by leveraging the rich, diverse, and continuously generated data from smartphones and IoT devices without compromising user data privacy. Such edge-side adaptation can improve model personalization, robustness, and responsiveness to local contexts. However, the practical feasibility of federated LLM fine-tuning on real edge devices remains unclear, as most existing work focuses on cross-silo or simulation-based settings, overlooking the resource and runtime constraints that determine whether a method is deployable on real edge systems. We present EdgeFlowerTune, a deployment-oriented benchmark for federated LLM fine-tuning under realistic edge-system constraints. EdgeFlowerTune jointly evaluates model quality and system costs, including communication, wall-clock latency, memory usage, energy consumption, and robustness to dynamic edge conditions. To compare methods in terms of effectiveness, efficiency, and robustness, EdgeFlowerTune introduces three complementary protocols: Quality-under-Budget, Cost-to-Target, and Robustness. We instantiate EdgeFlowerTune as a real-device platform built on Flower and MobileFineTuner, spanning commercial Android smartphones and NVIDIA edge development boards. Our benchmark results show that accuracy-only evaluation can lead to misleading conclusions: methods with similar final quality may differ substantially in deployability once realistic system constraints are considered. EdgeFlowerTune provides a reproducible benchmark for system-aware evaluation of federated LLM fine-tuning at the edge.

cs.CL↗
arXiv2026

CoCoPlan: Adaptive Coordination and Communication for Multi-robot Systems in Dynamic and Unknown Environments

Multi-robot systems can greatly enhance efficiency through coordination and collaboration, yet in practice, full-time communication is rarely available and interactions are constrained to close-range exchanges. Existing methods either maintain all-time connectivity, rely on fixed schedules, or adopt pairwise protocols, but none adapt effectively to dynamic spatio-temporal task distributions under limited communication, resulting in suboptimal coordination. To address this gap, we propose CoCoPlan, a unified framework that co-optimizes collaborative task planning and team-wise intermittent communication. Our approach integrates a branch-and-bound architecture that jointly encodes task assignments and communication events, an adaptive objective function that balances task efficiency against communication latency, and a communication event optimization module that strategically determines when, where and how the global connectivity should be re-established. Extensive experiments demonstrate that it outperforms state-of-the-art methods by achieving a 22.4% higher task completion rate, reducing communication overhead by 58.6%, and improving the scalability by supporting up to 100 robots in dynamic environments. Hardware experiments include the complex 2D office environment and large-scale 3D disaster-response scenario.

cs.RO↗
arXiv2026

Mechanism Design for Federated Learning with Non-Monotonic Network Effects

Mechanism design is pivotal to federated learning (FL) for maximizing social welfare by coordinating self-interested clients. Existing mechanisms, however, often overlook the network effects of client participation and the diverse model performance requirements (i.e., generalization error) across applications, leading to suboptimal incentives and social welfare, or even inapplicability in real deployments. To address this gap, we explore incentive mechanism design for FL with network effects and application-specific requirements of model performance. We develop a theoretical model to quantify the impact of network effects on heterogeneous client participation, revealing the non-monotonic nature of such effects. Based on these insights, we propose a Model Trading and Sharing (MoTS) framework, which enables clients to obtain FL models through either participation or purchase. To further address clients' strategic behaviors, we design a Social Welfare maximization with Application-aware and Network effects (SWAN) mechanism, exploiting model customer payments for incentivization. Experimental results on a hardware prototype demonstrate that our SWAN mechanism outperforms existing FL mechanisms, improving social welfare by up to $352.42\%$ and reducing extra incentive costs by $93.07\%$.

cs.GT↗
arXiv2026

SLEI3D: Simultaneous Exploration and Inspection via Heterogeneous Fleets under Limited Communication

Robotic fleets such as unmanned aerial and ground vehicles have been widely used for routine inspections of static environments, where the areas of interest are known and planned in advance. However, in many applications, such areas of interest are unknown and should be identified online during exploration. Thus, this paper considers the problem of simultaneous exploration, inspection of unknown environments and then real-time communication to a mobile ground control station to report the findings. The heterogeneous robots are equipped with different sensors, e.g., long-range lidars for fast exploration and close-range cameras for detailed inspection. Furthermore, global communication is often unavailable in such environments, where the robots can only communicate with each other via ad-hoc wireless networks when they are in close proximity and free of obstruction. This work proposes a novel planning and coordination framework (SLEI3D) that integrates the online strategies for collaborative 3D exploration, adaptive inspection and timely communication (via the intermit-tent or proactive protocols). To account for uncertainties w.r.t. the number and location of features, a multi-layer and multi-rate planning mechanism is developed for inter-and-intra robot subgroups, to actively meet and coordinate their local plans. The proposed framework is validated extensively via high-fidelity simulations of numerous large-scale missions with up to 48 robots and 384 thousand cubic meters. Hardware experiments of 7 robots are also conducted. Project website is available at https://junfengchen-robotics.github.io/SLEI3D/.

cs.RO↗
arXiv2025

FoldAct: Efficient and Stable Context Folding for Long-Horizon Search Agents

Long-horizon reinforcement learning (RL) for large language models faces critical scalability challenges from unbounded context growth, leading to context folding methods that compress interaction history during task execution. However, existing approaches treat summary actions as standard actions, overlooking that summaries fundamentally modify the agent's future observation space, creating a policy-dependent, non-stationary observation distribution that violates core RL assumptions. This introduces three fundamental challenges: (1) gradient dilution where summary tokens receive insufficient training signal, (2) self-conditioning where policy updates change summary distributions, creating a vicious cycle of training collapse, and (3) computational cost from processing unique contexts at each turn. We introduce \textbf{FoldAct}\footnote{https://github.com/SHAO-Jiaqi757/FoldAct}, a framework that explicitly addresses these challenges through three key innovations: separated loss computation for independent gradient signals on summary and action tokens, full context consistency loss to reduce distribution shift, and selective segment training to reduce computational cost. Our method enables stable training of long-horizon search agents with context folding, addressing the non-stationary observation problem while improving training efficiency with 5.19$\times$ speedup.

cs.LG↗
arXiv2025

Automated stereotactic radiosurgery planning using a human-in-the-loop reasoning large language model agent

Stereotactic radiosurgery (SRS) demands precise dose shaping around critical structures, yet black-box AI systems have limited clinical adoption due to opacity concerns. We tested whether chain-of-thought reasoning improves agentic planning in a retrospective cohort of 41 patients with brain metastases treated with 18 Gy single-fraction SRS. We developed SAGE (Secure Agent for Generative Dose Expertise), an LLM-based planning agent for automated SRS treatment planning. Two variants generated plans for each case: one using a non-reasoning model, one using a reasoning model. The reasoning variant showed comparable plan dosimetry relative to human planners on primary endpoints (PTV coverage, maximum dose, conformity index, gradient index; all p > 0.21) while reducing cochlear dose below human baselines (p = 0.022). When prompted to improve conformity, the reasoning model demonstrated systematic planning behaviors including prospective constraint verification (457 instances) and trade-off deliberation (609 instances), while the standard model exhibited none of these deliberative processes (0 and 7 instances, respectively). Content analysis revealed that constraint verification and causal explanation concentrated in the reasoning agent. The optimization traces serve as auditable logs, offering a path toward transparent automated planning.

cs.AI↗
arXiv2025

Do LLM Agents Know How to Ground, Recover, and Assess? A Benchmark for Epistemic Competence in Information-Seeking Agents

Recent work has explored training Large Language Model (LLM) search agents with reinforcement learning (RL) for open-domain question answering (QA). However, most evaluations focus solely on final answer accuracy, overlooking how these agents reason with and act on external evidence. We introduce SeekBench, the first benchmark for evaluating the \textit{epistemic competence} of LLM search agents through step-level analysis of their response traces. SeekBench comprises 190 expert-annotated traces with over 1,800 response steps generated by LLM search agents, each enriched with evidence annotations for granular analysis of whether agents (1) generate reasoning steps grounded in observed evidence, (2) adaptively reformulate searches to recover from low-quality results, and (3) have proper calibration to correctly assess whether the current evidence is sufficient for providing an answer.

cs.AI↗
arXiv2025

MorphAgent: Empowering Agents through Self-Evolving Profiles and Decentralized Collaboration

Large Language Model (LLM) based multi-agent systems (MAS) have shown promise in tackling complex tasks, but often rely on predefined roles and centralized coordination, limiting their adaptability to evolving challenges. This paper introduces MorphAgent, a novel Autonomous, Self-Organizing, and Self-Adaptive Multi-Agent System for decentralized agent collaboration that enables agents to dynamically evolve their roles and capabilities. Our approach employs self-evolving agent profiles, optimized through three key metrics, guiding agents in refining their individual expertise while maintaining complementary team dynamics. MorphAgent implements a two-phase process: a Profile Update phase for profile optimization, followed by a Task Execution phase where agents continuously adapt their roles based on task feedback. Our experimental results show that MorphAgent outperforms existing frameworks in terms of task performance and adaptability to changing requirements, paving the way for more robust and versatile multi-agent collaborative systems.

cs.AI↗
arXiv2025

Beyond Right to be Forgotten: Managing Heterogeneity Side Effects Through Strategic Incentives

Federated Unlearning (FU) enables the removal of specific clients' data influence from trained models. However, in non-IID settings, removing clients creates critical side effects: remaining clients with similar data distributions suffer disproportionate performance degradation, while the global model's stability deteriorates. These vulnerable clients then have reduced incentives to stay in the federation, potentially triggering a cascade of withdrawals that further destabilize the system. To address this challenge, we develop a theoretical framework that quantifies how data heterogeneity impacts unlearning outcomes. Based on these insights, we model FU as a Stackelberg game where the server strategically offers payments to retain crucial clients based on their contribution to both unlearning effectiveness and system stability. Our rigorous equilibrium analysis reveals how data heterogeneity fundamentally shapes the trade-offs between system-wide objectives and client interests. Our approach improves global stability by up to 6.23\%, reduces worst-case client degradation by 10.05\%, and achieves up to 38.6\% runtime efficiency over complete retraining.

cs.GT↗