Searcharxiv⌕ Search

arXiv subjects

Yang Xu

Publications and source records attributed to Yang Xu.

At least 73 records · Page 4Linked to original sources
arXiv2026

ReLoRA: Knowledge-Reusing Adaptation for Fast Rollout of Evolving LLM Services

Large Language Models (LLMs) are increasingly deployed as continuously evolving services, where frequent base-model updates may invalidate previously deployed task-specific Low-Rank Adaptation (LoRA) adapters. For service providers managing numerous downstream model services, retraining each LoRA adapter from scratch for every updated base model is computationally prohibitive and delays service rollout. Meanwhile, the simpler alternative, i.e., naively applying the original LoRA adapter to the updated base model, often leads to degraded service quality due to adapter-backbone incompatibility. To address this problem, we propose ReLoRA, a knowledge-reusing re-adaptation framework that efficiently restores service-ready LoRA adapters for evolving LLM services while preserving or improving task performance. Specifically, ReLoRA comprises two key optimization steps: 1) Adaptive LoRA initialization leverages Bayesian optimization to construct a compatibility-aware starting point by fusing information from both the previously deployed task adapter and the base model's evolution; 2) Fine-tuning with scheduled regularization first rapidly steers the adapter to a high-quality region via strong regularization, followed by relaxed regularization for task-specific refinement. This design enables rapid service-quality recovery with reduced re-adaptation overhead. Extensive experiments demonstrate that ReLoRA reduces time-to-readiness by up to 8.9$\times$ and improves accuracy by up to 4.6\% compared to baselines.

cs.LG↗
arXiv2026

Pelican-Unify 1.0: A Unified Embodied Intelligence Model for Understanding, Reasoning, Imagination and Action

We present Pelican-Unify 1.0, the first embodied foundation model trained according to the principle of unification. Pelican-Unify 1.0 uses a single VLM as a unified understanding module, mapping scenes, instructions, visual contexts, and action histories into a shared semantic space. The same VLM also serves as a unified reasoning module, autoregressively producing task-, action-, and future-oriented chains of thought in a single forward pass and projecting the final hidden state into a dense latent variable. A Unified Future Generator (UFG) then conditions on this latent variable and jointly generates future videos and future actions through two modality-specific output heads within the same denoising process. The language, video, and action losses are all backpropagated into the shared representation, enabling the model to jointly optimize understanding, reasoning, imagination, and action during training, rather than training three isolated expert systems. Experiments demonstrate that unification does not imply compromise. With a single checkpoint, Pelican-Unify 1.0 achieves strong performance across all three capabilities: 64.7 on eight VLM benchmarks, the best among comparable-scale models; 66.03 on WorldArena, ranking first; and 93.5 on RoboTwin, the second-best average among compared action methods. These results show that the unified paradigm succeeds in preserving specialist strength while bringing understanding, reasoning, imagination, and action into one model.

cs.RO↗
arXiv2026

Destructive interference of second harmonic generation in AA stacked MoTe$_2$/WSe$_2$

The stacking configuration of two-dimensional materials critically governs their optical and electronic responses. Monolayer transition-metal dichalcogenides (TMDC) lack inversion symmetry and exhibit exciton-enhanced second-harmonic generation (SHG). In TMDC bilayers, 60° (0°) stacking is conventionally expected to suppress (enhance) SHG owing to destructive (constructive) interference of the layer-resolved nonlinear polarizations. Here, we report an unconventional destructive SHG interference in nearly 0°-stacked (AA-stacked) MoTe2/WSe2 heterobilayers using two independent probes: atomic-resolution imaging and stacking-sensitive exciton hybridization measurements. Supported by ab initio GW and Bethe-Salpeter equation calculations, we show that distinct two-photon resonances associated with the WSe2 C exciton and the MoTe2 D exciton generate a nearly $π$ phase difference ($Δϕ$) in their second-order nonlinear susceptibilities $χ^{(2)}$, leading to the anomalous destructive interference. We further demonstrate that in small-angle twisted MoTe2/WSe2, the SHG polarization state is governed by the interplay between twist angle $α$ and phase difference $Δϕ$, and can be mapped onto trajectories on the Poincaré sphere. At excitation energies satisfying $Δϕ$ + 3$α$ = 180°, the SHG output becomes nearly circularly polarized (ellipticity ~ 0.91) and undergoes an abrupt 90° azimuthal rotation, corresponding to a geometric polarization singularity in the parameter space. Our findings open new routes for exciton-resonance engineered nonlinear photonics and stacking-resolved optical functionality in moiré materials.

cond-mat.mes-hall↗
arXiv2026

H-OmniStereo: Zero-Shot Omnidirectional Stereo Matching with Heading-Aligned Normal Priors

Stereo matching on top-bottom equirectangular images provides an effective framework for full-surround perception, as vertically aligned epipolar lines enable the use of advanced perspective stereo architectures that are largely driven by large-scale datasets and monocular priors. However, the performance of such adaptations is severely limited by the scarcity of omnidirectional stereo datasets and the degradation of perspective monocular priors under spherical distortions. To address these challenges, we propose H-OmniStereo, a zero-shot omnidirectional stereo matching framework. First, we construct high-quality synthetic dataset comprising over 2.8 million top-bottom equirectangular stereo pairs to scale up training. Second, we introduce an equirectangular monocular normal estimator, specifically operating in a heading-aligned coordinate system. Beyond providing distortion-robust and cross-view-consistent geometric priors for establishing reliable correspondences in stereo matching, this design boosts training efficiency and accommodates train-test FoV mismatches. Extensive experiments show that our approach achieves higher accuracy than existing methods on out-of-domain datasets and successfully generalizes to real-world consumer camera setups using a single model. The model and dataset will be released at https://github.com/JIANG-CX/H-OmniStereo.

cs.CV↗
arXiv2026

Escape from Callback Hell! A New Programming Paradigm for Network Simulation

Network simulation plays a crucial role in both networking research and industry. Existing commonly-used Discrete Event Simulations (DES) are based on callback mechanisms for discrete event (DE). However, due to the inability of callbacks to naturally simulate network events, programs in network simulation cannot be written in a sequential workflow. This leads to inherent complexity and poor maintainability, resulting in stack ripping and callback hell. These problems significantly increase simulation development workloads and introduce substantial cognitive loads associated with programming and debugging. To enable more efficient development of network simulation and facilitate the rapid evaluation and evolution of network functions, we propose a novel development paradigm for network simulation named ``CoDES" (\textbf{Co}routine-based \textbf{DES}). To the best of our knowledge, we are the first to focus on optimizing the network simulation development process rather than performance based on the coroutine mechanism. We implement a new network simulation framework based on CoDES that is capable of naturally simulating network events and effectively address key system challenges related to correctness, functionality, compatibility, and overhead. It enables developers to create sequential workflows for network programs and simplifies the code structure, thus reducing development workloads while enhancing code readability and maintainability. We apply this paradigm to a commonly used network simulator, NS-3 to implement Message Passing Interface (MPI), High Precision Congestion Control (HPCC), and Routing Information Protocol (RIP), achieving up to 62.3\% and 82.6\% reduction in code volume and structure complexity without sacrificing simulation accuracy, extending execution time or increasing runtime memory of simulation.

cs.NI↗
arXiv2026

StyleVAR: Controllable Image Style Transfer via Visual Autoregressive Modeling

We build on the Visual Autoregressive Modeling (VAR) framework and formulate style transfer as conditional discrete sequence modeling in a learned latent space. Images are decomposed into multi-scale representations and tokenized into discrete codes by a VQ-VAE; a transformer then autoregressively models the distribution of target tokens conditioned on style and content tokens. To inject style and content information, we introduce a blended cross-attention mechanism in which the evolving target representation attends to its own history, while style and content features act as queries that decide which aspects of this history to emphasize. A scale-dependent blending coefficient controls the relative influence of style and content at each stage, encouraging the synthesized representation to align with both the content structure and the style texture without breaking the autoregressive continuity of VAR. We train StyleVAR in two stages from a pretrained VAR checkpoint: supervised fine-tuning on a large triplet dataset of content--style--target images, followed by reinforcement fine-tuning with Group Relative Policy Optimization (GRPO) against a DreamSim-based perceptual reward, with per-action normalization weighting to rebalance credit across VAR's multi-scale hierarchy. Across three benchmarks spanning in-, near-, and out-of-distribution regimes, StyleVAR consistently outperforms an AdaIN baseline on Style Loss, Content Loss, LPIPS, SSIM, DreamSim, and CLIP similarity, and the GRPO stage yields further gains over the SFT checkpoint, most notably on the reward-aligned perceptual metrics. Qualitatively, the method transfers texture while maintaining semantic structure, especially for landscapes and architectural scenes, while a generalization gap on internet images and difficulty with human faces highlight the need for better content diversity and stronger structural priors.

cs.CV↗
arXiv2026

Core-Halo Decomposition: Decentralizing Large-Scale Fixed-Point Problems

We study solving large-scale fixed-point equation \(x^\star=\bar F(x^\star)\) with decomposition. Standard strict decomposition assigns each agent a disjoint block and evaluates updates using only owned coordinates. For most operators, however, a block update may depend on variables outside the block. Truncating these dependencies by strict decomposition changes the mean operator and creates structural bias that cannot be removed by more samples, smaller stepsizes, or additional consensus. We therefore propose Core-Halo decomposition, which separates write ownership from read-only evaluation context: each agent updates its own core and reads from an overlapping halo. By aligning the Core-Halo decomposition with the block-dependence structure of $\bar F$, the original fixed-point problem can be implemented faithfully in a decentralized multi-agent system. We further characterize the fundamental obstruction faced by strict decomposition through a Bellman closure condition and a blockwise bias lower bound, showing that local-only updates can alter the original fixed-point operator. Finally, we conduct extensive experiments across a range of application settings, and demonstrate that Core-Halo achieves near-centralized performance while retaining the parallelism benefits of decentralization.

stat.ML↗
arXiv2026

LoHGNet: Infrared Small Target Detection through Lorentz Geometric Encoding with High-Order Relation Learning

Infrared small target detection (IRSTD) remains challenging due to the scarcity of useful target cues and the presence of severe background clutter. Most current methods rely on conventional feature learning and local interaction modeling, where features are represented in Euclidean space. However, such designs may still be limited in describing the subtle differences of weak targets and the contextual relations between targets and backgrounds. To address these limitations, we propose LoHGNet, an IRSTD network that integrates Lorentz geometric encoding with high-order relation learning. By introducing Lorentz manifold based feature learning, LoHGNet offers a different feature representation from conventional IRSTD methods and provides new discriminative cues for IRSTD. Specifically, a Lorentz encoding branch is constructed with the Geometric Attention Guided Lorentz Residual Convolution Module (GA-LRCM) to perform feature modeling under hyperbolic geometric constraints and enhance the hierarchical geometric representation capability of weak targets. Subsequently, the hyperbolic features are mapped into the Euclidean tangent space through logarithmic mapping, and a High-Order Relation Learning Module (HORL) is designed to model the high-order contextual dependencies between targets and backgrounds via hypergraph construction, thereby improving target discrimination in complex backgrounds. Experimental results on three datasets demonstrate that the proposed LoHGNet achieves competitive performance in both detection accuracy and adaptability to complex scenes. The code will be available at https://github.com/Kingwin97.

cs.CV↗
arXiv2026

Persistent-Transient Policy Evaluation for Markov Chains via Minimal Peripheral Quotients

We study fixed-policy evaluation for finite Markov chains that may be reducible and periodic. Classical evaluation methods with gain and bias decomposition are not always diagnostic: the gain records only invariant Cesàro averages, while persistent phase-dependent behavior is absorbed into the bias together with genuinely transient effects. We identify the real peripheral invariant subspace $\mathcal{K}(P)$ of the transition matrix $P$ as the source of this ambiguity. Quotienting by $\mathcal{K}(P)$ is the minimal exact quotient that removes all non-decaying modes and makes the remaining dynamics strictly stable. After choosing a gauge projection $Π$ with kernel $\mathcal{K}(P)$, the reward admits a unique decomposition $r = g_Π^\star + (I-P)v_Π^\star$, where $g_Π^\star$ is a persistent regime profile and $v_Π^\star$ is a gauge-fixed transient component. An exact comparison with classical normalized gain and bias shows that the new pair reallocates the same information so that all persistent modes are represented in $g_Π^\star$ and $v_Π^\star$ is transient. This decomposition reconstructs finite-horizon returns, recovers statewise average reward, admits a transient-cost interpretation, and yields a stable estimator under a generative model.

stat.ML↗
arXiv2026

When Is the Same Model Not the Same Service? A Measurement Study of Hosted Open-Weight LLM APIs

Open-weight large language models (LLMs) are usually named as model artifacts, but production users often consume them as hosted API services. This paper argues that the operational unit is a service object: a provider-specific, time-varying endpoint defined by model variant, protocol behavior, context capacity, listed price, latency and throughput distribution, reliability, and task feasibility. Using sampled request logs, provider metadata, compatibility probes, pricing snapshots, and continuous latency measurements collected by AI Ping during Q4 2025, we study how this service layer changes the meaning of "the same model." Three empirical patterns emerge. First, observed demand is concentrated but persistent across versions: in the displayed family aggregate, the largest family carries 32.0% of relative demand and the top five carry 87.4%, with a Gini coefficient of 0.693, while older variants remain active after newer releases. Second, supply and use separate: provider listing breadth does not imply realized adoption, and listed prices are more anchored than latency, throughput, context length, protocol support, and error semantics. Third, task mix matters: applications induce different token-length regimes, so provider choice is a constrained decision over provider-model-task-time tuples rather than a lookup by model name. In two representative counterfactuals under observed feasibility constraints, routing lowers Qwen3-32B cost by 37.8% and raises DeepSeek-V3.2 average throughput by about 90% relative to direct official access. The results support a measurement view of hosted open-weight LLMs as heterogeneous services, not static catalog entries. We open-source the measurement methodology and reproduction artifacts at https://github.com/haoruilee/llm_api_measurement_study to support result reproduction.

cs.PF↗
arXiv2026

AGPO: Asymmetric Group Policy Optimization for Verifiable Reasoning and Search Ads Relevance at JD

Reinforcement Learning with Verifiable Rewards (RLVR) has demonstrated notable success in enhancing the reasoning performance of large language models (LLMs). However, recent studies reveal that while current RLVR methods improve sampling efficiency towards correct paths, they do not elicit fundamentally new reasoning patterns. Instead, the reasoning capability boundary of trained models often narrows compared to their base models, with base models achieving higher coverage at large sample sizes. In this work, we propose Asymmetric Group Policy Optimization (AGPO) to counteract this boundary shrinkage. AGPO adopts a negative-dominant reinforcement strategy to suppress incorrect reasoning paths, maintaining the base model's exploration capacity. For positive reinforcement, AGPO adopts a group advantage mechanism, which scales positive updates based on intra-group variance, allowing the model to focus on rare correct paths while suppressing updates from trivial paths. Our experiments on five mathematical benchmarks demonstrate that AGPO achieves state-of-the-art accuracy while consistently improving pass@$k$ performance at scale. In a large-scale industrial application for search ads relevance optimization, AGPO effectively enhances the quality of the data annotation, leading to substantial performance gains in downstream student models.

cs.AI↗
arXiv2026

Towards Reliable LLM Evaluation: Correcting the Winner's Curse in Adaptive Benchmarking

Adaptive prompt and program search makes LLM evaluation selection-sensitive. Once benchmark items are reused inside tuning, the observed winner's score need not estimate the fresh-data performance of the full tune-then-deploy procedure. We study inference for this procedure-level target under explicit tuning budgets. We propose SIREN, a selection-aware repeated-split reporting protocol that freezes the post-search shortlist, separates splitwise selection from held-out evaluation, and uses an item-level Gaussian multiplier bootstrap for uncertainty quantification. In a fixed-shortlist regime with smooth stabilized selection, the estimator admits a first-order item-level representation, and the bootstrap yields valid simultaneous inference on a finite budget grid. This supports confidence intervals for procedure-performance curves and pre-specified equal-budget and cross-budget comparisons. Controlled simulations and MMLU-Pro tuning experiments show that winner-based reporting can be optimistic and can change deployment conclusions, while SIREN remains close to the finite-sample reporting target.

stat.ML↗
arXiv2026

Online Bayesian Calibration under Gradual and Abrupt System Changes

Bayesian model calibration is central to digital twins and computer experiments, as it aligns model outputs with field observations by estimating calibration parameters and correcting systematic model bias. Classical Bayesian calibration introduces latent parameters and a discrepancy function to model bias, but suffers from parameter--discrepancy confounding and is typically formulated as an offline procedure under a stationary data-generating assumption. These limitations are restrictive in modern digital twin applications, where systems evolve over time and may exhibit gradual drift and abrupt regime shifts. While data assimilation methods enable sequential updates, they generally do not explicitly model systematic bias and are less effective under abrupt changes. We propose Bayesian Recursive Projected Calibration (BRPC), an online Bayesian calibration framework for streaming data under simulator mismatch and nonstationarity. BRPC extends projected calibration to the online setting by separating a discrepancy-free particle update for calibration parameters from a conditional Gaussian process update for discrepancy, preserving identifiability while enabling bias-aware adaptation under gradual system evolution. To handle abrupt changes, BRPC is integrated with restart mechanisms that detect regime shifts and reset the calibration process. We establish theoretical guarantees for both components, including tracking performance under gradual evolution and false-alarm and detection behavior for restart mechanisms. Empirical studies on synthetic and plant-simulation benchmarks show that BRPC improves calibration accuracy under gradual changes, while restart-augmented BRPC further improves robustness and predictive performance under abrupt regime shifts compared to sliding-window Bayesian calibration and data assimilation baselines.

cs.LG↗
arXiv2026

Motion-Adapter: A Diffusion Model Adapter for Text-to-Motion Generation of Compound Actions

Recent advances in generative motion synthesis have enabled the production of realistic human motions from diverse input modalities. However, synthesizing compound actions from texts, which integrate multiple concurrent actions into coherent full-body sequences, remains a major challenge. We identify two key limitations in current text-to-motion diffusion models: (i) catastrophic neglect, where earlier actions are overwritten by later ones due to improper handling of temporal information, and (ii) attention collapse, which arises from excessive feature fusion in cross-attention mechanisms. As a result, existing approaches often depend on overly detailed textual descriptions (e.g., raising right hand), explicit body-part specifications (e.g., editing the upper body), or the use of large language models (LLMs) for body-part interpretation. These strategies lead to deficient semantic representations of physical structures and kinematic mechanisms, limiting the ability to incorporate natural behaviors such as greeting while walking. To address these issues, we propose the Motion-Adapter, a plug-and-play module that guides text-to-motion diffusion models in generating compound actions by computing decoupled cross-attention maps, which serve as structural masks during the denoising process. Extensive experiments demonstrate that our method consistently produces more faithful and coherent compound motions across diverse textual prompts, surpassing state-of-the-art approaches.

cs.CV↗
arXiv2026

SVOM Science User Support Services at Chinese Science Center

The Chinese-French SVOM (Space-based Multi-band Astronomical Variable Objects Monitor) mission is dedicated to the study of gamma-ray bursts (GRBs) from the distant universe. A key component of the SVOM Chinese Ground Segment, the Science User Support Services (SUSS) provides comprehensive support for the mission's scientific operations. SUSS consists of two integral pillars: a suite of specialized software tools that automate key workflows, and a dedicated User Support Team that delivers expert-led, human services. These human-delivered services include operational coordination across telescope networks, direct technical assistance to astronomers, user training, and proactive problem-solving throughout the observation lifecycle. This paper focuses on the organization of SVOM scientific operations and the role of SUSS in facilitating these tasks. We provide a detailed description of the SUSS software architecture and its functionalities, encompassing the General Platform, the Burst Advocate (BA) support tools for GRB counterpart identification, the Target of Opportunity (ToO) support tools, and the General Program (GP) support tools. The structure and services provided by the user support team at the Chinese Science Center (CSC) are also elaborated. Furthermore, we evaluate the performance of SUSS during its first operational year, assessing its effectiveness in fulfilling user requirements. The evaluation offers valuable insights to guide future user support strategies and software enhancements, ultimately enabling better service for the SVOM scientific community.

astro-ph.IM↗
arXiv2026

Overview of Ground-based Wide-Angle Cameras array

As one of the key ground-based facilities of the Chinese-French SVOM mission, the main scientific objectives of the Ground-based Wide Angle Camera array (GWAC) are to detect prompt optical emission of gamma-ray bursts or other short duration astronomical transients on a second-scale temporal resolution. GWAC is located at Xinglong observatory, China, and consists of 10 mounts and 40 cameras, providing a joint field of view of about 3600 square degrees.The detection ability is 16 magnitude in 10 seconds of exposure time in the visual band under the condition of the new moon phase. Here, we give an overview of GWAC and introduce the science motivation of the project, as well as the performance of the hardware and the software. The observation strategies and the data processing are briefly presented. The early sciences in the last 5 years since the first light are summarized.

astro-ph.IM↗
arXiv2026

Alert Chain and Observation Planning for Ground Wide Angle Camera Network

The Ground Wide Angle Camera Network (GWAC-N) is a robotic telescope network. It consists of ten wide-field core telescopes (GWAC-A) and two 60cm narrow-field rapid follow-up telescopes (GWAC-F60A/B). The primary scientific goal of GWAC-N is to detect optical counterparts of gamma-ray bursts (GRBs) discovered by the SVOM satellite. This is achieved through synchronized monitoring with the GWAC-A array. Upon receiving a GRB trigger alert, the network conducts rapid, multi-band follow-up observations using the GWAC-F60A/B telescopes. The two-stage observation process involves many telescopes, making manual control impractical. Automated operations are therefore essential. They are realized through an integrated alert processing chain and an automated observation scheduling and dispatching mechanism. To enable this, we employ the SVOM Follow-up Observation Coordinating Service (FOCS) and GWAC-N's Automatic Observation Management (AOM) system. This paper presents the integrated alert processing workflow. It also describes the formulation of observation strategies, and the scheduling and execution of observations enabled by FOCS and AOM.

astro-ph.IM↗
arXiv2026

SVOM/VT: Overview of data processing and GRB identifications with X-band data

VT (the Visible Telescope) is an optical telescope onboard the SVOM (Space-based Multi-band Astronomical Variable Objects Monitor) mission, specifically designed to detect optical counterparts of gamma-ray bursts (GRBs), study their afterglows, and select high-redshift candidates. It performs rapid follow-up observations simultaneously in two channels either via autonomous platform slewing or Target of Opportunity (ToO) observations. The science images acquired by VT and transmitted via the X-band downlink system are designated as VT X-band data. This paper provides an overview of GRB optical afterglow identifications with VT and describes the ground-based processing pipeline for VT X-band data, including preprocessing, astrometric calibration, and photometry. Up to 2025 December 3, VT has followed up 111 GRBs triggered by SVOM or external missions. The overall detection rate of optical counterparts is approximately 75%. Specifically, for bursts detected by SVOM/ECLAIRs, the detection rate is 77% when observed by VT within 30 minutes after the burst. A slightly higher detection rate of 81% is achieved for GRBs triggered by external missions through rapid ToO observations with a mid-time of less than 3 hours.

astro-ph.HE↗