arXiv subjects

Xing Zhang

Publications and source records attributed to Xing Zhang.

At least 55 records · Page 3Linked to original sources

arXiv2026

Verified Multi-Agent Orchestration: A Plan-Execute-Verify-Replan Framework for Complex Query Resolution

We present Verified Multi-Agent Orchestration (VMAO), a framework that coordinates specialized LLM-based agents through a verification-driven iterative loop. Given a complex query, our system decomposes it into a directed acyclic graph (DAG) of sub-questions, executes them through domain-specific agents in parallel, verifies result completeness via LLM-based evaluation, and adaptively replans to address gaps. The key contributions are: (1) dependency-aware parallel execution over a DAG of sub-questions with automatic context propagation, (2) verification-driven adaptive replanning that uses an LLM-based verifier as an orchestration-level coordination signal, and (3) configurable stop conditions that balance answer quality against resource usage. On 25 expert-curated market research queries, VMAO improves answer completeness from 3.1 to 4.2 and source quality from 2.6 to 4.1 (1-5 scale) compared to a single-agent baseline, demonstrating that orchestration-level verification is an effective mechanism for multi-agent quality assurance.

cs.AI↗

arXiv2026

CORE:Toward Ubiquitous 6G Intelligence Through Collaborative Orchestration of Large Language Model Agents Over Hierarchical Edge

Rapid advancements in sixth-generation (6G) networks and large language models (LLMs) have paved the way for ubiquitous intelligence, wherein seamless connectivity and distributed artificial intelligence (AI) have revolutionized various aspects of our lives.However, realizing this vision faces significant challenges owing to the fragmented and heterogeneous computing resources across hierarchical networks, which are insufficient for individual LLM agents to perform complex reasoning tasks.To address this issue, we propose Collaborative Orchestration Role at Edge (CORE), an innovative framework that employs a collaborative learning system in which multiple LLMs, each assigned a distinct functional role, are distributed across mobile devices and tiered edge servers. The system integrates three optimization modules, encompassing real-time perception,dynamic role orchestration, and pipeline-parallel execution, to facilitate efficient and rapid collaboration among distributed agents. Furthermore, we introduce a novel role affinity scheduling algorithm for dynamically orchestrating LLM role assignments across the hierarchical edge infrastructure, intelligently matching computational demands with available dispersed resources.Finally, comprehensive case studies and performance evaluations across various 6G application scenarios demonstrated the efficacy of CORE, revealing significant enhancements in the system efficiency and task completion rates. Building on these promising outcomes, we further validated the practical applicability of CORE by deploying it on a real-world edge-computing platform,that exhibits robust performance in operational environments.

cs.AI↗

arXiv2026

MiMo-V2-Flash Technical Report

We present MiMo-V2-Flash, a Mixture-of-Experts (MoE) model with 309B total parameters and 15B active parameters, designed for fast, strong reasoning and agentic capabilities. MiMo-V2-Flash adopts a hybrid attention architecture that interleaves Sliding Window Attention (SWA) with global attention, with a 128-token sliding window under a 5:1 hybrid ratio. The model is pre-trained on 27 trillion tokens with Multi-Token Prediction (MTP), employing a native 32k context length and subsequently extended to 256k. To efficiently scale post-training compute, MiMo-V2-Flash introduces a novel Multi-Teacher On-Policy Distillation (MOPD) paradigm. In this framework, domain-specialized teachers (e.g., trained via large-scale reinforcement learning) provide dense and token-level reward, enabling the student model to perfectly master teacher expertise. MiMo-V2-Flash rivals top-tier open-weight models such as DeepSeek-V3.2 and Kimi-K2, despite using only 1/2 and 1/3 of their total parameters, respectively. During inference, by repurposing MTP as a draft model for speculative decoding, MiMo-V2-Flash achieves up to 3.6 acceptance length and 2.6x decoding speedup with three MTP layers. We open-source both the model weights and the three-layer MTP weights to foster open research and community collaboration.

cs.CL↗

arXiv2025

MiMo-Audio: Audio Language Models are Few-Shot Learners

Existing audio language models typically rely on task-specific fine-tuning to accomplish particular audio tasks. In contrast, humans are able to generalize to new audio tasks with only a few examples or simple instructions. GPT-3 has shown that scaling next-token prediction pretraining enables strong generalization capabilities in text, and we believe this paradigm is equally applicable to the audio domain. By scaling MiMo-Audio's pretraining data to over one hundred million of hours, we observe the emergence of few-shot learning capabilities across a diverse set of audio tasks. We develop a systematic evaluation of these capabilities and find that MiMo-Audio-7B-Base achieves SOTA performance on both speech intelligence and audio understanding benchmarks among open-source models. Beyond standard metrics, MiMo-Audio-7B-Base generalizes to tasks absent from its training data, such as voice conversion, style transfer, and speech editing. MiMo-Audio-7B-Base also demonstrates powerful speech continuation capabilities, capable of generating highly realistic talk shows, recitations, livestreaming and debates. At the post-training stage, we curate a diverse instruction-tuning corpus and introduce thinking mechanisms into both audio understanding and generation. MiMo-Audio-7B-Instruct achieves open-source SOTA on audio understanding benchmarks (MMSU, MMAU, MMAR, MMAU-Pro), spoken dialogue benchmarks (Big Bench Audio, MultiChallenge Audio) and instruct-TTS evaluations, approaching or surpassing closed-source models. Model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-Audio.

cs.CL↗

arXiv2025

STED and Consistency Scoring: A Framework for Evaluating LLM Structured Output Reliability

Large Language Models (LLMs) are increasingly deployed for structured data generation, yet output consistency remains critical for production applications. We introduce a comprehensive framework for evaluating and improving consistency in LLM-generated structured outputs. Our approach combines: (1) STED (Semantic Tree Edit Distance), a novel similarity metric balancing semantic flexibility with structural strictness when comparing JSON outputs, and (2) a consistency scoring framework aggregating multiple STED measurements across repeated generations to quantify reliability. Through systematic experiments on synthetic datasets with controlled schema, expression, and semantic variations, we demonstrate STED achieves superior performance ($0.86-0.90$ similarity for semantic equivalents, $0.0$ for structural breaks) compared to existing metrics including TED, BERTScore, and DeepDiff. Applying our framework to benchmark six LLMs reveals significant variations: Claude-3.7-Sonnet demonstrates exceptional consistency, maintaining near-perfect structural reliability even at high temperatures ($T=0.9$), while models like Claude-3-Haiku and Nova-Pro exhibit substantial degradation requiring careful tuning. Our framework enables practical applications including targeted model selection for structured tasks, iterative prompt refinement for reproducible results, and diagnostic analysis to identify inconsistency root causes. This work provides theoretical foundations and practical tools for ensuring reliable structured output generation in LLM-based production systems.

cs.CL↗

arXiv2025

Deep Learning Accelerated First-Principles Quantum Transport Simulations at Nonequilibrium State

The non-equilibrium Green's function method combined with density functional theory (NEGF-DFT) provides a rigorous framework for simulating nanoscale electronic transport, but its computational cost scales steeply with system size. Recent artificial intelligence (AI) approaches have sought to accelerate such simulations, yet most rely on conventional machine learning, lack atomic resolution, struggle to extrapolate to larger systems, and cannot predict multiple properties simultaneously. Here we introduce DeepQT, a deep-learning framework that integrates graph neural networks with transformer architectures to enable multi-property predictions of electronic structure and transport without manual feature engineering. By learning key intermediate quantities of NEGF-DFT, the equilibrium Hamiltonian and the non-equilibrium total potential difference, DeepQT reconstructs Hamiltonians under both equilibrium and bias conditions, yielding accurate transport predictions. Leveraging the principle of electronic nearsightedness, DeepQT generalizes from small training systems to much larger ones with high fidelity. Benchmarks on graphene, MoS2, and silicon diodes with varied defects and dopants show that DeepQT achieves first-principles accuracy while reducing computational cost by orders of magnitude. This scalable, transferable framework advances AI-assisted quantum transport, offering a powerful tool for next-generation nanoelectronic device design.

cond-mat.mes-hall↗

arXiv2025

Accurate crystal field Hamiltonians of single-ion magnets at mean-field cost

The effective crystal field Hamiltonian provides the key description of the electronic properties of single-ion magnets, but obtaining its parameters from ab initio computation is challenging. We introduce a simple approach to derive the effective crystal field Hamiltonian through density functional calculations of randomly rotated mean-field states within the low-energy manifold. In benchmarks on five lanthanide-based complexes, we find that we compute with mean-field cost an effective crystal field Hamiltonian that matches the state-of-the-art from much more expensive multi-configurational quantum chemistry methods. In addition, we are able to reproduce the experimental low-energy spectrum and magnetic properties with an accuracy exceeding prior attempts. Due to its low cost, our approach provides a crucial ingredient in the computational design of single-ion magnets with tailored physical properties and low-energy spectra.

physics.chem-ph↗

arXiv2025

Aerial-Ground Image Feature Matching via 3D Gaussian Splatting-based Intermediate View Rendering

The integration of aerial and ground images has been a promising solution in 3D modeling of complex scenes, which is seriously restricted by finding reliable correspondences. The primary contribution of this study is a feature matching algorithm for aerial and ground images, whose core idea is to generate intermediate views to alleviate perspective distortions caused by the extensive viewpoint changes. First, by using aerial images only, sparse models are reconstructed through an incremental SfM (Structure from Motion) engine due to their large scene coverage. Second, 3D Gaussian Splatting is then adopted for scene rendering by taking as inputs sparse points and oriented images. For accurate view rendering, a render viewpoint determination algorithm is designed by using the oriented camera poses of aerial images, which is used to generate high-quality intermediate images that can bridge the gap between aerial and ground images. Third, with the aid of intermediate images, reliable feature matching is conducted for match pairs from render-aerial and render-ground images, and final matches can be generated by transmitting correspondences through intermediate views. By using real aerial and ground datasets, the validation of the proposed solution has been verified in terms of feature matching and scene rendering and compared comprehensively with widely used methods. The experimental results demonstrate that the proposed solution can provide reliable feature matches for aerial and ground images with an obvious increase in the number of initial and refined matches, and it can provide enough matches to achieve accurate ISfM reconstruction and complete 3DGS-based scene rendering.

cs.CV↗

arXiv2025

Reactive Chemistry at Unrestricted Coupled Cluster Level: High-throughput Calculations for Training Machine Learning Potentials

Accurately modeling chemical reactions at the atomistic level requires high-level electronic structure theory due to the presence of unpaired electrons and the need to properly describe bond breaking and making energetics. Commonly used approaches such as Density Functional Theory (DFT) frequently fail for this task due to deficiencies that are well recognized. However, for high-fidelity approaches, creating large datasets of energies and forces for reactive processes to train machine learning interatomic potentials or force fields is daunting. For example, the use of the unrestricted coupled cluster level of theory has previously been seen as unfeasible due to high computational costs, the lack of analytical gradients in many computational codes, and additional challenges such as constructing suitable basis set corrections for forces. In this work, we develop new methods and workflows to overcome the challenges inherent to automating unrestricted coupled cluster calculations. Using these advancements, we create a dataset of gas-phase reactions containing energies and forces for 3119 different organic molecules configurations calculated at the gold-standard level of unrestricted CCSD(T) (coupled cluster singles doubles and perturbative triples). With this dataset, we provide an analysis of the differences between the density functional and unrestricted CCSD(T) descriptions. We develop a transferable machine learning interatomic potential for gas-phase reactions, trained on unrestricted CCSD(T) data, and demonstrate the advantages of transitioning away from DFT data. Transitioning from training to DFT to training to UCCSD(T) datasets yields an improvement of more than 0.1 eV/Å in force accuracy and over 0.1 eV in activation energy reproduction.

physics.chem-ph↗

arXiv2025

LLM Enabled Multi-Agent System for 6G Networks: Framework and Method of Dual-Loop Edge-Terminal Collaboration

The ubiquitous computing resources in 6G networks provide ideal environments for the fusion of large language models (LLMs) and intelligent services through the agent framework. With auxiliary modules and planning cores, LLM-enabled agents can autonomously plan and take actions to deal with diverse environment semantics and user intentions. However, the limited resources of individual network devices significantly hinder the efficient operation of LLM-enabled agents with complex tool calls, highlighting the urgent need for efficient multi-level device collaborations. To this end, the framework and method of the LLM-enabled multi-agent system with dual-loop terminal-edge collaborations are proposed in 6G networks. Firstly, the outer loop consists of the iterative collaborations between the global agent and multiple sub-agents deployed on edge servers and terminals, where the planning capability is enhanced through task decomposition and parallel sub-task distribution. Secondly, the inner loop utilizes sub-agents with dedicated roles to circularly reason, execute, and replan the sub-task, and the parallel tool calling generation with offloading strategies is incorporated to improve efficiency. The improved task planning capability and task execution efficiency are validated through the conducted case study in 6G-supported urban safety governance. Finally, the open challenges and future directions are thoroughly analyzed in 6G networks, accelerating the advent of the 6G era.

cs.MA↗

arXiv2025

MegaCacheX: Towards Cost-Effective Hierarchical Collaborative Content Caching in Emerging Mega-Constellations

Significant latency in global content delivery primarily arises from insufficient terrestrial infrastructure. Deploying space-based content delivery networks within emerging mega-constellations provides an effective means to bridge the digital divide. However, space-based caching faces constraints from physical-layer dynamics, including dynamic topologies, time-varying inter-satellite link conditions, and limited onboard energy. In addition, existing mechanisms often lack fine-grained content categorization and global optimization. This paper proposes MegaCacheX, a cost-effective hierarchical framework for collaborative content distribution that achieves "Earth-independence" by providing cloud services directly from space. Specifically, data centers in Sun-synchronous orbit act as primary content sources, while caching nodes in mega-constellations and ground stations collaboratively form a distributed edge layer. MegaCacheX optimizes caching strategies by integrating content popularity, regional user distribution, and satellite trajectory predictions. Multi-tier caching nodes serve as service anchors, enabling seamless content delivery with low latency. A prototype implemented on a microservices-based, containerized testbed demonstrates that MegaCacheX reduces global content access latency by about 36% compared to baseline approaches, while maintaining cost efficiency.

eess.SY↗

arXiv2025

Two-Timescale Dynamic Service Deployment and Task Scheduling with Spatiotemporal Collaboration in Mobile Edge Networks

Collaborative edge computing addresses the resource constraints of individual edge nodes by enabling resource sharing and task co-processing across multiple nodes. To fully leverage the advantages of collaborative edge computing, joint optimization of service deployment and task scheduling is necessary. Existing optimization methods insufficiently address the collaboration across spatial and temporal dimensions, which hinders their adaptability to the spatiotemporally varying nature of user demands and system states. This paper focuses on optimizing the expected task processing delay in edge networks. We propose a two-timescale online optimization framework to jointly determine: i) service deployment decisions at each large timescale; and ii) task scheduling decisions at each small timescale. Specifically, the convex optimization technique is used to solve the task scheduling problem, while a multi-agent deep reinforcement learning technique is employed for the service deployment problem. These two methods are combined for spatiotemporal co-optimization through a two-timescale alternating optimization approach. Compared to the baseline algorithms, the proposed scheme achieves better delay performance, while also exhibiting low running time and favorable convergence behavior.

cs.PF↗

arXiv2025

Repeating Words for Video-Language Retrieval with Coarse-to-Fine Objectives

The explosive growth of video streaming presents challenges in achieving high accuracy and low training costs for video-language retrieval. However, existing methods rely on large-scale pre-training to improve video retrieval performance, resulting in significant computational demands. Additionally, the fine-grained information in videos and texts remains underexplored. To alleviate these problems, we propose a novel framework to learn fine-grained features for better alignment and introduce an inference pipeline to improve performance without additional training. Specifically, we employ coarse-to-fine objectives to understand the semantic information of video-text pairs, including contrastive and matching learning. The fine-grained data used for training is obtained through the Granularity-Aware Representation module, which is designed based on similarity analysis between video frames and words in captions. Furthermore, we observe that the repetition of keywords in the original captions, referred to as "Repetition", can enhance retrieval performance and improve alignment between video and text. Based on this insight, we propose a novel and effective inference pipeline that incorporates a voting mechanism and a new Matching Entropy metric to achieve better retrieval performance without requiring additional pre-training. Experimental results on four benchmarks demonstrate that the proposed method outperforms previous approaches. Additionally, our inference pipeline achieves significant performance improvements, with a 2.1% increase in Recall@1 on the MSR-VTT dataset and a 1.6% increase on the DiDeMo dataset.

cs.CV↗

arXiv2025

On kernel isomorphisms of $m$-Cayley digraphs and finite $2$PCI-groups

The isomorphism problem for digraphs is a fundamental problem in graph theory. In this paper, we consider this problem for $m$-Cayley digraphs which are generalization of Cayley digraphs. Let $m$ be a positive integer. A digraph admitting a group $G$ of automorphisms acting semiregularly on the vertices with exactly $m$ orbits is called an $m$-Cayley digraph of $G$. In our previous paper, we developed a theory for $m$-Cayley isomorphisms of $m$-Cayley digraphs, and classified finite $m$CI-groups for each $m\geq 2$, and finite $m$PCI-groups for each $m\geq 4$. The next natural step is to classify finite $m$PCI-groups for $m=2$ or $3$. Note that BCI-groups form an important subclass of the $2$PCI-groups, which were introduced in 2008 by Xu et al. Despite much effort having been made on the study of BCI-groups, the problem of classifying finite BCI-groups is still widely open. In this paper, we prove that every finite $2$PCI-group is solvable, and its Sylow $3$-subgroup is isomorphic to $Z_3, Z_3\times Z_3$ or $Z_9$, and Sylow $p$-subgroup with $p\not=3$ is either elementary abelian, or isomorphic to $Z_4$ or $Q_8$. We also introduce the kernel isomorphisms of $m$-Cayley digraphs, and establish some useful theory for studying this kind of isomorphisms. Using the results of kernel isomorphisms of $m$-Cayley digraphs together with the results on $2$PCI-groups, we give a proper description of finite BCI-groups, and in particular, we obtain a complete classification of finite non-abelian BCI-groups.

math.CO↗

arXiv2025

MiMo: Unlocking the Reasoning Potential of Language Model -- From Pretraining to Posttraining

We present MiMo-7B, a large language model born for reasoning tasks, with optimization across both pre-training and post-training stages. During pre-training, we enhance the data preprocessing pipeline and employ a three-stage data mixing strategy to strengthen the base model's reasoning potential. MiMo-7B-Base is pre-trained on 25 trillion tokens, with additional Multi-Token Prediction objective for enhanced performance and accelerated inference speed. During post-training, we curate a dataset of 130K verifiable mathematics and programming problems for reinforcement learning, integrating a test-difficulty-driven code-reward scheme to alleviate sparse-reward issues and employing strategic data resampling to stabilize training. Extensive evaluations show that MiMo-7B-Base possesses exceptional reasoning potential, outperforming even much larger 32B models. The final RL-tuned model, MiMo-7B-RL, achieves superior performance on mathematics, code and general reasoning tasks, surpassing the performance of OpenAI o1-mini. The model checkpoints are available at https://github.com/xiaomimimo/MiMo.

cs.CL↗

arXiv2025

MiMo-VL Technical Report

We open-source MiMo-VL-7B-SFT and MiMo-VL-7B-RL, two powerful vision-language models delivering state-of-the-art performance in both general visual understanding and multimodal reasoning. MiMo-VL-7B-RL outperforms Qwen2.5-VL-7B on 35 out of 40 evaluated tasks, and scores 59.4 on OlympiadBench, surpassing models with up to 78B parameters. For GUI grounding applications, it sets a new standard with 56.1 on OSWorld-G, even outperforming specialized models such as UI-TARS. Our training combines four-stage pre-training (2.4 trillion tokens) with Mixed On-policy Reinforcement Learning (MORL) integrating diverse reward signals. We identify the importance of incorporating high-quality reasoning data with long Chain-of-Thought into pre-training stages, and the benefits of mixed RL despite challenges in simultaneous multi-domain optimization. We also contribute a comprehensive evaluation suite covering 50+ tasks to promote reproducibility and advance the field. The model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-VL.

cs.CL↗

arXiv2025

AnimeDL-2M: Million-Scale AI-Generated Anime Image Detection and Localization in Diffusion Era

Recent advances in image generation, particularly diffusion models, have significantly lowered the barrier for creating sophisticated forgeries, making image manipulation detection and localization (IMDL) increasingly challenging. While prior work in IMDL has focused largely on natural images, the anime domain remains underexplored-despite its growing vulnerability to AI-generated forgeries. Misrepresentations of AI-generated images as hand-drawn artwork, copyright violations, and inappropriate content modifications pose serious threats to the anime community and industry. To address this gap, we propose AnimeDL-2M, the first large-scale benchmark for anime IMDL with comprehensive annotations. It comprises over two million images including real, partially manipulated, and fully AI-generated samples. Experiments indicate that models trained on existing IMDL datasets of natural images perform poorly when applied to anime images, highlighting a clear domain gap between anime and natural images. To better handle IMDL tasks in anime domain, we further propose AniXplore, a novel model tailored to the visual characteristics of anime imagery. Extensive evaluations demonstrate that AniXplore achieves superior performance compared to existing methods. Dataset and code can be found in https://flytweety.github.io/AnimeDL2M/.

cs.CV↗

arXiv2025

Spatiotemporal Non-Uniformity-Aware Online Task Scheduling in Collaborative Edge Computing for Industrial Internet of Things

Mobile edge computing mitigates the shortcomings of cloud computing caused by unpredictable wide-area network latency and serves as a critical enabling technology for the Industrial Internet of Things (IIoT). Unlike cloud computing, mobile edge networks offer limited and distributed computing resources. As a result, collaborative edge computing emerges as a promising technology that enhances edge networks' service capabilities by integrating computational resources across edge nodes. This paper investigates the task scheduling problem in collaborative edge computing for IIoT, aiming to optimize task processing performance under long-term cost constraints. We propose an online task scheduling algorithm to cope with the spatiotemporal non-uniformity of user request distribution in distributed edge networks. For the spatial non-uniformity of user requests across different factories, we introduce a graph model to guide optimal task scheduling decisions. For the time-varying nature of user request distribution and long-term cost constraints, we apply Lyapunov optimization to decompose the long-term optimization problem into a series of real-time subproblems that do not require prior knowledge of future system states. Given the NP-hard nature of the subproblems, we design a heuristic-based hierarchical optimization approach incorporating enhanced discrete particle swarm and harmonic search algorithms. Finally, an imitation learning-based approach is devised to further accelerate the algorithm's operation, building upon the initial two algorithms. Comprehensive theoretical analysis and experimental evaluation demonstrate the effectiveness of the proposed schemes.

cs.PF↗