arXiv subjects

Xu Chen

Publications and source records attributed to Xu Chen.

At least 109 records · Page 6Linked to original sources

arXiv2026

Prompt and Parameter Co-Optimization for Large Language Models

Prompt optimization and fine-tuning are two major approaches to improve the performance of Large Language Models (LLMs). They enhance the capabilities of LLMs from complementary perspectives: the former through explicit natural language, and the latter through implicit parameter updates. However, prior work has typically studied them in isolation, leaving their synergistic potential largely underexplored. To bridge this gap, in this paper, we introduce MetaTuner, a novel framework that jointly integrates prompt optimization and fine-tuning for LLM training. Specifically, we introduce two neural networks to generate prompts and parameters, respectively, while allowing them to share a common bottom encoding layer to enable knowledge sharing. By the guidance of the final supervised signals, our framework is optimized to discover the optimal combinations between the prompts and parameters. Given that prompt learning involves discrete optimization while fine-tuning operates in a continuous parameter space, we design a supervised regularization loss to train our framework effectively. Extensive experiments across diverse benchmarks show that our method consistently outperforms the baselines.

cs.CL↗

arXiv2026

Solving the Granularity Mismatch: Hierarchical Preference Learning for Long-Horizon LLM Agents

Large Language Models (LLMs) as autonomous agents are increasingly tasked with solving complex, long-horizon problems. Aligning these agents via preference-based offline methods like Direct Preference Optimization (DPO) is a promising direction, yet it faces a critical granularity mismatch. Trajectory-level DPO provides a signal that is too coarse for precise credit assignment, while step-level DPO is often too myopic to capture the value of multi-step behaviors. To resolve this challenge, we introduce Hierarchical Preference Learning (HPL), a hierarchical framework that optimizes LLM agents by leveraging preference signals at multiple, synergistic granularities. While HPL incorporates trajectory- and step-level DPO for global and local policy stability, its core innovation lies in group-level preference optimization guided by a dual-layer curriculum. Our approach first decomposes expert trajectories into semantically coherent action groups and then generates contrasting suboptimal groups to enable preference learning at a fine-grained, sub-task level. Then, instead of treating all preference pairs equally, HPL introduces a curriculum scheduler that organizes the learning process from simple to complex. This curriculum is structured along two axes: the group length, representing sub-task complexity, and the sample difficulty, defined by the reward gap between preferred and dispreferred action groups. Experiments on three challenging agent benchmarks show that HPL outperforms existing state-of-the-art methods. Our analyses demonstrate that the hierarchical DPO loss effectively integrates preference signals across multiple granularities, while the dual-layer curriculum is crucial for enabling the agent to solve a wide range of tasks, from simple behaviors to complex multi-step sequences.

cs.LG↗

arXiv2026

On estimating superconducting shielding volume fraction from susceptibility in pressurized Ruddlesden-Popper nickelates: Response to arXiv:2602.19282

In a recent preprint (arXiv:2602.19282) [1], the authors questioned the procedure we used to evaluate the demagnetization-corrected superconducting shielding volume fraction in pressurized Ruddlesden-Popper nickelates [2-5]. They further claimed that this methodology has neither been derived nor used previously, and they proposed an alternative normalization scheme. Here we clarify that our evaluation follows directly from the standard magnetostatic self-consistency relation for finite samples and has been widely adopted in the superconductivity literature for decades. We also demonstrate that the discrepancies claimed in Ref. [1] stem from a fundamental flaw in their approach, namely, the assumption that the measured diamagnetic moment is linearly proportional to the superconducting shielding volume fraction in the presence of a finite demagnetization factor N. This assumption is not valid for strongly demagnetized, thin disk-like specimens, where the internal field and the measured moment are coupled self-consistently through the demagnetizing field.

cond-mat.supr-con↗

arXiv2026

Rejection Mixing: Fast Semantic Propagation of Mask Tokens for Efficient DLLM Inference

Diffusion Large Language Models (DLLMs) promise fast non-autoregressive inference but suffer a severe quality-speed trade-off in parallel decoding. This stems from the ''combinatorial contradiction'' phenomenon, where parallel tokens form semantically inconsistent combinations. We address this by integrating continuous representations into the discrete decoding process, as they preserve rich inter-position dependency. We propose ReMix (Rejection Mixing), a framework that introduces a novel Continuous Mixing State as an intermediate between the initial masked state and the final decoded token state. This intermediate state allows a token's representation to be iteratively refined in a continuous space, resolving mutual conflicts with other tokens before collapsing into a final discrete sample. Furthermore, a rejection rule reverts uncertain representations from the continuous state back to the masked state for reprocessing, ensuring stability and preventing error propagation. ReMix thus mitigates combinatorial contradictions by enabling continuous-space refinement during discrete diffusion decoding. Extensive experiments demonstrate that ReMix, as a training-free method, achieves a $2-8 \times$ inference speedup without any quality degradation.

cs.CL↗

arXiv2026

NextMem: Towards Latent Factual Memory for LLM-based Agents

Memory is critical for LLM-based agents to preserve past observations for future decision-making, where factual memory serves as its foundational part. However, existing approaches to constructing factual memory face several limitations. Textual methods impose heavy context and indexing burdens, while parametric methods suffer from catastrophic forgetting and high costs. To address these challenges, we introduce NextMem, a latent factual memory framework that utilizes an autoregressive autoencoder to efficiently construct latent memory while ensuring accurate reconstruction. For better optimization, we propose a two-stage training process, including autoregressive reconstruction alignment and progressive latent substitution. We also incorporate quantization to reduce storage overhead. Extensive experiments demonstrate that NextMem achieves superior performance, and excels in retrieval, robustness, and extensibility properties. We release our code and model checkpoints at https://github.com/nuster1128/NextMem.

cs.AI↗

arXiv2026

Resource-Efficient Personal Large Language Models Fine-Tuning with Collaborative Edge Computing

Large language models (LLMs) have unlocked a plethora of powerful applications at the network edge, such as intelligent personal assistants. Data privacy and security concerns have prompted a shift towards edge-based fine-tuning of personal LLMs, away from cloud reliance. However, this raises issues of computational intensity and resource scarcity, hindering training efficiency and feasibility. While current studies investigate parameter-efficient fine-tuning (PEFT) techniques to mitigate resource constraints, our analysis indicates that these techniques are not sufficiently resource-efficient for edge devices. To tackle these challenges, we propose Pluto and Charon (PAC), a time and memory efficient collaborative edge AI framework for personal LLMs fine-tuning. PAC breaks the resource wall of personal LLMs fine-tuning with a sophisticated algorithm-system co-design. (1) Algorithmically, PAC implements a personal LLMs fine-tuning technique that is efficient in terms of parameters, time, and memory. It utilizes Parallel Adapters to circumvent the need for a full backward pass through the LLM backbone. Additionally, an activation cache mechanism further streamlining the process by negating the necessity for repeated forward passes across multiple epochs. (2) Systematically, PAC leverages edge devices in close proximity, pooling them as a collective resource for in-situ personal LLMs fine-tuning, utilizing a hybrid data and pipeline parallelism to orchestrate distributed training. The use of the activation cache eliminates the need for forward pass through the LLM backbone,enabling exclusive fine-tuning of the Parallel Adapters using data parallelism. Extensive evaluation based on prototype implementation demonstrates that PAC remarkably outperforms state-of-the-art approaches, achieving up to 8.64x end-to-end speedup and up to 88.16% reduction in memory footprint.

cs.DC↗

arXiv2026

ABot-N0: Technical Report on the VLA Foundation Model for Versatile Embodied Navigation

Embodied navigation has long been fragmented by task-specific architectures. We introduce ABot-N0, a unified Vision-Language-Action (VLA) foundation model that achieves a ``Grand Unification'' across 5 core tasks: Point-Goal, Object-Goal, Instruction-Following, POI-Goal, and Person-Following. ABot-N0 utilizes a hierarchical ``Brain-Action'' architecture, pairing an LLM-based Cognitive Brain for semantic reasoning with a Flow Matching-based Action Expert for precise, continuous trajectory generation. To support large-scale learning, we developed the ABot-N0 Data Engine, curating 16.9M expert trajectories and 5.0M reasoning samples across 7,802 high-fidelity 3D scenes (10.7 $\text{km}^2$). ABot-N0 achieves new SOTA performance across 7 benchmarks, significantly outperforming specialized models. Furthermore, our Agentic Navigation System integrates a planner with hierarchical topological memory, enabling robust, long-horizon missions in dynamic real-world environments.

cs.RO↗

arXiv2026

The Devil Behind Moltbook: Anthropic Safety is Always Vanishing in Self-Evolving AI Societies

The emergence of multi-agent systems built from large language models (LLMs) offers a promising paradigm for scalable collective intelligence and self-evolution. Ideally, such systems would achieve continuous self-improvement in a fully closed loop while maintaining robust safety alignment--a combination we term the self-evolution trilemma. However, we demonstrate both theoretically and empirically that an agent society satisfying continuous self-evolution, complete isolation, and safety invariance is impossible. Drawing on an information-theoretic framework, we formalize safety as the divergence degree from anthropic value distributions. We theoretically demonstrate that isolated self-evolution induces statistical blind spots, leading to the irreversible degradation of the system's safety alignment. Empirical and qualitative results from an open-ended agent community (Moltbook) and two closed self-evolving systems reveal phenomena that align with our theoretical prediction of inevitable safety erosion. We further propose several solution directions to alleviate the identified safety concern. Our work establishes a fundamental limit on the self-evolving AI societies and shifts the discourse from symptom-driven safety patches to a principled understanding of intrinsic dynamical risks, highlighting the need for external oversight or novel safety-preserving mechanisms.

cs.CL↗

arXiv2026

AgentSkiller: Scaling Generalist Agent Intelligence through Semantically Integrated Cross-Domain Data Synthesis

Large Language Model agents demonstrate potential in solving real-world problems via tools, yet generalist intelligence is bottlenecked by scarce high-quality, long-horizon data. Existing methods collect privacy-constrained API logs or generate scripted interactions lacking diversity, which struggle to produce data requisite for scaling capabilities. We propose AgentSkiller, a fully automated framework synthesizing multi-turn interaction data across realistic, semantically linked domains. It employs a DAG-based architecture with explicit state transitions to ensure determinism and recoverability. The pipeline builds a domain ontology and Person-Centric Entity Graph, defines tool interfaces via Service Blueprints for Model Context Protocol servers, and populates environments with consistent databases and strict Domain Policies. A cross-domain fusion mechanism links services to simulate complex tasks. Finally, the pipeline creates user tasks by verifying solution paths, filtering via execution-based validation, and generating queries using a Persona-based Simulator for automated rollout. This produces reliable environments with clear state changes. To demonstrate effectiveness, we synthesized $\approx$ 11K interaction samples; experimental results indicate that models trained on this dataset achieve significant improvements on function calling over baselines, particularly in larger parameter regimes.

cs.CL↗

arXiv2026

Understanding Large Language Models in Your Pockets: Performance Study on COTS Mobile Devices

As large language models (LLMs) increasingly integrate into every aspect of our work and daily lives, there are growing concerns about user privacy, which push the trend toward local deployment of these models. There are a number of lightweight LLMs (e.g., Gemini Nano, LLAMA2 7B) that can run locally on smartphones, providing users with greater control over their personal data. As a rapidly emerging application, we are concerned about their performance on commercial-off-the-shelf mobile devices. To fully understand the current landscape of LLM deployment on mobile platforms, we conduct a comprehensive measurement study on mobile devices. While user experience is the primary concern for end-users, developers focus more on the underlying implementations. Therefore, we evaluate both user-centric metrics-such as token throughput, latency, and response quality-and developer-critical factors, including resource utilization, OS strategies, battery consumption, and launch time. We also provide comprehensive comparisons across the mobile system-on-chips (SoCs) from major vendors, highlighting their performance differences in handling LLM workloads, which may help developers identify and address bottlenecks for mobile LLM applications. We hope that this study can provide insights for both the development of on-device LLMs and the design for future mobile system architecture.

cs.LG↗

arXiv2026

Kirin: Improving ANN efficiency with SNN Hybridization

Artificial neural networks (ANNs), particularly large language models (LLMs), demonstrate powerful inference capabilities but consume substantial energy. Conversely, spiking neural networks (SNNs) exhibit exceptional energy efficiency due to their binary and event-driven characteristics, thus motivating the study of ANN-to-SNN conversion. In this process, quantization plays a pivotal role, mapping LLMs' floating-point parameters to discrete SNN parameters via the temporal dimension of the time window. However, several challenges remain in the conversion process: (i) converting high bit-width quantization values into binary spikes requires longer time windows, increasing system latency; and (ii) the inherent trade-off between the information loss of single-spike schemes and the energy costs of multi-spike ones in SNN. To address these challenges, we propose Kirin, a integer and spike hybrid based SNN to achieve accuracy lossless ANN-to-SNN conversion with time and energy efficiency. Specifically, we first propose a Spike Matrix Hybridization strategy that encoding low bit-width parameters that leading to small time window size into binary spikes while preserving the rest in integer format, thereby reducing the overall latency of SNN execution. Second, we introduce a silence threshold mechanism to regulate the timing of single-spike firing, ensuring the output is mathematically equivalent to the LLM's output and preserves accuracy. Experimental results demonstrate that Kirin, under a W4A4\&8 quantization setting, achieves near-FP16 accuracy while reducing energy consumption by up to 84.66\% and shortening time steps by 93.75\%.

cs.LG↗

arXiv2026

Towards Adaptive, Scalable, and Robust Coordination of LLM Agents: A Dynamic Ad-Hoc Networking Perspective

Multi-agent architectures built on large language models (LLMs) have demonstrated the potential to realize swarm intelligence through well-crafted collaboration. However, the substantial burden of manual orchestration inherently raises an imperative to automate the design of agentic workflows. We frame such an agent coordination challenge as a classic problem in dynamic ad-hoc networking: How to establish adaptive and reliable communication among a scalable number of agentic hosts? In response to this unresolved dilemma, we introduce RAPS, a reputation-aware publish-subscribe paradigm for adaptive, scalable, and robust coordination of LLM agents. RAPS is grounded in the Distributed Publish-Subscribe Protocol, allowing LLM agents to exchange messages based on their declared intents rather than predefined topologies. Beyond this substrate, RAPS further incorporates two coherent overlays: (i) Reactive Subscription, enabling agents to dynamically refine their intents; and (ii) Bayesian Reputation, empowering each agent with a local watchdog to detect and isolate malicious peers. Extensive experiments over five benchmarks showcase that our design effectively reconciles adaptivity, scalability, and robustness in a unified multi-agent coordination framework.

cs.AI↗

arXiv2026

Orchestrating Heterogeneous Experts: A Scalable MoE Framework with Anisotropy-Preserving Fusion

In cross-border e-commerce, search relevance modeling faces the dual challenge of extreme linguistic diversity and fine-grained semantic nuances. Existing approaches typically rely on scaling up a single monolithic Large Language Model (LLM). However, our empirical analysis reveals that single models suffer from uneven capability distributions across regions. For example, excelling in English while underperforming in specific Southeast Asian languages. In this work, we shift the paradigm from scaling a single model to orchestrating heterogeneous experts. We propose a scalable Coarse-grained Mixture-of-Experts (MoE) framework that leverages the inherent complementarity of distinct open-source LLMs (e.g., Qwen, Gemma) without expensive pre-training. Unlike standard token-level MoE, our framework dynamically routes entire queries to specialized experts and, crucially, employs an Information-Preserving Concatenation Fusion strategy. We theoretically posit that preserving the distinct embedding manifolds of heterogeneous experts-rather than compressing them via weighted averaging-is essential for capturing complex relevance signals in a multi-model latent space. On datasets spanning six Southeast Asian markets, our MoE improves AUC by 0.72 percentage points over a dense baseline with the same active parameters. Meanwhile, the optimized pipeline achieves 13.72 queries per second (QPS), a 9% throughput improvement.

cs.IR↗

arXiv2026

GraphPilot: GUI Task Automation with One-Step LLM Reasoning Powered by Knowledge Graph

Mobile graphical user interface (GUI) agents are designed to automate everyday tasks on smartphones. Recent advances in large language models (LLMs) have significantly enhanced the capabilities of mobile GUI agents. However, most LLM-powered mobile GUI agents operate in stepwise query-act loops, which incur high latency due to repeated LLM queries. We present GraphPilot, a mobile GUI agent that leverages knowledge graphs of the target apps to complete user tasks in almost one LLM query. GraphPilot operates in two complementary phases to enable efficient and reliable LLM-powered GUI task automation. In the offline phase, it explores target apps, records and analyzes interaction history, and constructs an app-specific knowledge graph that encodes functions of pages and elements as well as transition rules for each app. In the online phase, given an app and a user task, it leverages the knowledge graph of the given app to guide the reasoning process of LLM. When the reasoning process encounters uncertainty, GraphPilot dynamically requests the HTML representation of the current interface to refine subsequent reasoning. Finally, a validator checks the generated sequence of actions against the transition rules in the knowledge graph, performing iterative corrections to ensure it is valid. The structured, informative information in the knowledge graph allows the LLM to plan the complete sequence of actions required to complete the user task. On the DroidTask benchmark, GraphPilot improves task completion rate over Mind2Web and AutoDroid, while substantially reducing latency and the number of LLM queries.

cs.HC↗

arXiv2026

How Does Personalized Memory Shape LLM Behavior? Benchmarking Rational Preference Utilization in Personalized Assistants

Large language model (LLM)-powered assistants have recently integrated memory mechanisms that record user preferences, leading to more personalized and user-aligned responses. However, irrelevant personalized memories are often introduced into the context, interfering with the LLM's intent understanding. To comprehensively investigate the dual effects of personalization, we develop RPEval, a benchmark comprising a personalized intent reasoning dataset and a multi-granularity evaluation protocol. RPEval reveals the widespread phenomenon of irrational personalization in existing LLMs and, through error pattern analysis, illustrates its negative impact on user experience. Finally, we introduce RP-Reasoner, which treats memory utilization as a pragmatic reasoning process, enabling the selective integration of personalized information. Experimental results demonstrate that our method significantly outperforms carefully designed baselines on RPEval, and resolves 80% of the bad cases observed in a large-scale commercial personalized assistant, highlighting the potential of pragmatic reasoning to mitigate irrational personalization. Our benchmark is publicly available at https://github.com/XueyangFeng/RPEval.

cs.CL↗

arXiv2026

On-Chip Generation of Co-Polarized and Spectrally Separable Photon Pairs

On-chip generation of high-purity single photons is essential for scalable photonic quantum technologies. Spontaneous parametric down-conversion (SPDC) is widely used to generate photon pairs for heralded single-photon sources, but intrinsic spectral correlations of the pairs often limit the purity and interference visibility of the heralded photons. Existing approaches to suppress these correlations rely on narrowband spectral filtering, which introduces loss, or exploiting different polarizations, which complicates on-chip integration. Here, we demonstrate a new strategy for generating spectrally separable photon pairs in thin-film lithium niobate nanophotonic circuits by harnessing higher-order spatial modes, with all interacting fields residing in the same polarization. Spectral separability is achieved by engineering group-velocity matching using higher-order transverse-electric modes, combined with a Gaussian-apodized poling profile to further suppress residual correlations inherent to standard periodic poling. Subsequent on-chip mode conversion with efficiency exceeding 95\% maps the higher-order mode to the fundamental mode and routes the photons into distinct output channels. The resulting heralded photons exhibit spectral purities exceeding 94\% inferred from joint-spectral intensity and 89\% from unheralded $g^{(2)}$ measurement. This approach enables flexible spectral and temporal engineering of on-chip quantum light sources for quantum computing and quantum networking.

quant-ph↗

arXiv2026

FlowSpec: Continuous Pipelined Speculative Decoding for Efficient Distributed LLM Inference

Distributed inference serves as a promising approach to enabling the inference of large language models (LLMs) at the network edge. It distributes the inference process to multiple devices to ensure that the LLMs can fit into the device memory. Recent pipeline-based approaches have the potential to parallelize communication and computation, which helps reduce inference latency. However, the benefit diminishes when the inference request at the network edge is sparse, where pipeline is typically at low utilization. To enable efficient distributed LLM inference at the edge, we propose \textbf{FlowSpec}, a pipeline-parallel tree-based speculative decoding framework. FlowSpec incorporates three key mechanisms to improve decoding efficiency: 1) score-based step-wise verification prioritizes more important draft tokens to bring earlier accepted tokens; 2) efficient draft management to prune invalid tokens while maintaining correct causal relationship during verification; 3) dynamic draft expansion strategies to supply high-quality speculative inputs. These techniques work in concert to enhance both pipeline utilization and speculative efficiency. We evaluate FlowSpec on a real-world testbed with other baselines. Experimental results demonstrate that our proposed framework significantly improves inference speed across diverse models and configurations, achieving speedup ratios 1.37$\times$-1.73$\times$ compared to baselines. Our code is publicly available at \href{https://github.com/Leosang-lx/FlowSpec#}{https://github.com/Leosang-lx/FlowSpec\#}.

cs.DC↗

arXiv2026

HAPS: Hierarchical LLM Routing with Joint Architecture and Parameter Search

Large language model (LLM) routing aims to exploit the specialized strengths of different LLMs for diverse tasks. However, existing approaches typically focus on selecting LLM architectures while overlooking parameter settings, which are critical for task performance. In this paper, we introduce HAPS, a hierarchical LLM routing framework that jointly searches over model architectures and parameters. Specifically, we use a high-level router to select among candidate LLM architectures, and then search for the optimal parameters for the selected architectures based on a low-level router. We design a parameter generation network to share parameters between the two routers to mutually enhance their capabilities. In the training process, we design a reward-augmented objective to effectively optimize our framework. Experiments on two commonly used benchmarks show that HAPS consistently outperforms strong routing baselines. We have released our code at https://github.com/zihangtian/HAPS.

cs.CL↗