arXiv subjects

Hao Xu

Publications and source records attributed to Hao Xu.

At least 109 records · Page 6Linked to original sources

arXiv2026

LaS-Comp: Zero-shot 3D Completion with Latent-Spatial Consistency

This paper introduces LaS-Comp, a zero-shot and category-agnostic approach that leverages the rich geometric priors of 3D foundation models to enable 3D shape completion across diverse types of partial observations. Our contributions are threefold: First, \ourname{} harnesses these powerful generative priors for completion through a complementary two-stage design: (i) an explicit replacement stage that preserves the partial observation geometry to ensure faithful completion; and (ii) an implicit refinement stage ensures seamless boundaries between the observed and synthesized regions. Second, our framework is training-free and compatible with different 3D foundation models. Third, we introduce Omni-Comp, a comprehensive benchmark combining real-world and synthetic data with diverse and challenging partial patterns, enabling a more thorough and realistic evaluation. Both quantitative and qualitative experiments demonstrate that our approach outperforms previous state-of-the-art approaches. Our code and data will be available at \href{https://github.com/DavidYan2001/LaS-Comp}{LaS-Comp}.

cs.CV↗

arXiv2026

VRM: Teaching Reward Models to Understand Authentic Human Preferences

Large Language Models (LLMs) have achieved remarkable success across diverse natural language tasks, yet the reward models employed for aligning LLMs often encounter challenges of reward hacking, where the approaches predominantly rely on directly mapping prompt-response pairs to scalar scores, which may inadvertently capture spurious correlations rather than authentic human preferences. In contrast, human evaluation employs a sophisticated process that initially weighs the relative importance of multiple high-dimensional objectives according to the prompt context, subsequently evaluating response quality through low-dimensional semantic features such as logical coherence and contextual appropriateness. Motivated by this consideration, we propose VRM, i.e., Variational Reward Modeling, a novel framework that explicitly models the evaluation process of human preference judgments by incorporating both high-dimensional objective weights and low-dimensional semantic features as latent variables, which are inferred through variational inference techniques. Additionally, we provide a theoretical analysis showing that VRM can achieve a tighter generalization error bound compared to the traditional reward model. Extensive experiments on benchmark datasets demonstrate that VRM significantly outperforms existing methods in capturing authentic human preferences.

cs.CL↗

arXiv2026

Magnetic Signature of Chiral Phonons Revealed by Neutron Spectroscopy in Ferrimagnetic Fe$_{1.75}$Zn$_{0.25}$Mo$_3$O$_8$

Lattice vibrations can carry angular momentum and magnetic moments under broken inversion or time-reversal symmetry, forming so-called chiral phonons. While such excitations have been explored in nonmagnetic systems via optical probes, their direct detection in magnetic materials and coupling to spin excitations remain largely unexplored. Here, using neutron spectroscopy, sensitive to both nuclear and magnetic scattering, we reveal the magnetic signature of chiral phonons in ferrimagnetic Fe$_{1.75}$Zn$_{0.25}$Mo$_3$O$_8$ with Curie temperature $T_{\rm C}\sim49$ K. Below $T_{\rm C}$, we observe enhanced magnetic scattering of phonons at small momenta, arising from strong magnon-phonon coupling. In addition, out-of-plane intensity modulation, phonon mode splitting, and field-induced Zeeman shifts are observed, all closely associated with the ferrimagnetic order. These features vanish above $T_{\rm C}$, where phonon spectra are dominated by nuclear scattering. These observations demonstrate the existence of chiral phonons carrying substantial magnetic moments that directly contribute to magnetic scattering, and establish neutron spectroscopy as a powerful, momentum-resolved probe of their magnetic character.

cond-mat.str-el↗

arXiv2026

Huawei Cloud Model-as-a-Service on the CloudMatrix384 SuperPod

Scaled-out MoE LLMs and scaled-up SuperPods create new systems challenges for production Model-as-a-Service (MaaS), requiring disaggregation, low-latency communication, and decentralized serving. This report presents xDeepServe, the production serving system behind Huawei Cloud's MaaS offering on CloudMatrix384, a 48-server SuperPod with 384 Ascend 910C chips connected by a high-bandwidth UB fabric and global shared memory. It serves models including DeepSeek, Kimi, GLM, Qwen, and MiniMax, among others. xDeepServe is built around Transformerless, a disaggregated execution architecture that decomposes transformer inference into modular units -- attention, feedforward, and MoE -- and supports disaggregated Prefill-Decode and MoE-Attention deployments. To enable disaggregation, we develop XCCL, a memory-semantic communication layer providing microsecond-level point-to-point and scalable all-to-all primitives, and we extend FlowServe with decentralized DP groups and techniques to mitigate stragglers and synchronization variance. In a peak decoding configuration, xDeepServe reaches 2400 tokens/s per Ascend 910C chip at ~50ms time-per-output-token (TPOT).

cs.DC↗

arXiv2026

Learning Hierarchical Sparse Transform Coding for 3DGS Compression

Current 3DGS compression methods largely forego the neural analysis-synthesis transform, which is a crucial component in learned signal compression systems. As a result, redundancy removal is left solely to the entropy coder, overburdening the entropy coding module and reducing rate-distortion (R-D) performance. To fix this critical omission, we propose a training-time transform coding (TTC) method that adds the analysis-synthesis transform and optimizes it jointly with the 3DGS representation and entropy model. Concretely, we adopt a hierarchical design: a channel-wise KLT for decorrelation and energy compaction, followed by a sparsity-aware neural transform that reconstructs the KLT residuals with minimal parameter and computational overhead. Experiments show that our method delivers strong R-D performance with fast decoding, offering a favorable BD-rate-decoding-time trade-off over SOTA 3DGS compressors.

cs.CV↗

arXiv2026

UAV-Enabled Short-Packet Communication via Fluid Antenna Systems

This paper develops a framework for analyzing UAV-enabled short-packet communication, leveraging fluid antenna system (FAS)-assisted relaying networks. Operating in the short-packet regime and focusing on challenging urban environments, we derive novel, closed-form expressions for the block error rate (BLER). This is achieved by modeling the spatially correlated Nakagami-$m$ fading link via a tractable eigenvalue-based approach. A high-signal-to-noise ratio (SNR) asymptotic analysis is also presented, revealing the system's fundamental diversity order. Building on this analysis, we formulate a novel energy efficiency (EE) maximization problem that, unlike idealized models, uniquely incorporates the non-trivial time and energy overheads of FAS port selection. An efficient hierarchical algorithm is proposed to jointly optimize key system parameters. Numerical results validate our analysis, demonstrating that while FAS provides substantial power gains, the operational overhead creates a critical trade-off. This trade-off dictates an optimal number of FAS ports and a non-trivial optimal UAV deployment altitude, governed by the balance between blockage and path loss. This work provides key insights for FAS-aided UAV communications.

cs.IT↗

arXiv2026

Are you going to finish that? A Practical Study of the Partial Token Problem

Language models (LMs) are trained over sequences of tokens, whereas users interact with LMs via text. This mismatch gives rise to the partial token problem, which occurs when a user ends their prompt in the middle of the expected next-token, leading to distorted next-token predictions. Although this issue has been studied using arbitrary character prefixes, its prevalence and severity in realistic prompts respecting word boundaries remains underexplored. In this work, we identify three domains where token and "word" boundaries often do not line up: languages that do not use whitespace, highly compounding languages, and code. In Chinese, for example, up to 25% of word boundaries do not line up with token boundaries, making even natural, word-complete prompts susceptible to this problem. We systematically construct semantically natural prompts ending with a partial tokens; in experiments, we find that they comprise a serious failure mode: frontier LMs consistently place three orders of magnitude less probability on the correct continuation compared to when the prompt is "backed-off" to be token-aligned. This degradation does not diminish with scale and often worsens for larger models. Finally, we evaluate inference-time mitigations to the partial token problem and validate the effectiveness of recent exact solutions. Overall, we demonstrate the scale and severity of probability distortion caused by tokenization in realistic use cases, and provide practical recommentions for model inference providers.

cs.CL↗

arXiv2026

EgoHandICL: Egocentric 3D Hand Reconstruction with In-Context Learning

Robust 3D hand reconstruction in egocentric vision is challenging due to depth ambiguity, self-occlusion, and complex hand-object interactions. Prior methods mitigate these issues by scaling training data or adding auxiliary cues, but they often struggle in unseen contexts. We present EgoHandICL, the first in-context learning (ICL) framework for 3D hand reconstruction that improves semantic alignment, visual consistency, and robustness under challenging egocentric conditions. EgoHandICL introduces complementary exemplar retrieval guided by vision-language models (VLMs), an ICL-tailored tokenizer for multimodal context, and a masked autoencoder (MAE)-based architecture trained with hand-guided geometric and perceptual objectives. Experiments on ARCTIC and EgoExo4D show consistent gains over state-of-the-art methods. We also demonstrate real-world generalization and improve EgoVLM hand-object interaction reasoning by using reconstructed hands as visual prompts. Code and data: https://github.com/Nicous20/EgoHandICL

cs.CV↗

arXiv2026

Simulating Dual-Pixel Images From Ray Tracing For Depth Estimation

Many studies utilize dual-pixel (DP) sensor phase characteristics for various applications, such as depth estimation and deblurring. However, since the DP image features are entirely determined by the camera hardware, DP-depth paired datasets are very scarce, especially when performing depth estimation on customized cameras. To overcome this, studies simulate DP images using ideal optical system models. However, these simulations often violate real optical propagation laws, leading to poor generalization to real DP data. To address this, we investigate the domain gap between simulated and real DP data, and propose solutions using the Simulating DP images from ray tracing (Sdirt) scheme. The Sdirt generates realistic DP images via ray tracing and integrates them into the depth estimation training pipeline. Experimental results show that models trained with Sdirt-simulated images generalize better to real DP data. The code and collected datasets will be available at github.com/LinYark/Sdirt

cs.CV↗

arXiv2026

PAS-Mamba: Phase-Amplitude-Spatial State Space Model for MRI Reconstruction

Joint feature modeling in both the spatial and frequency domains has become a mainstream approach in MRI reconstruction. However, existing methods generally treat the frequency domain as a whole, neglecting the differences in the information carried by its internal components. According to Fourier transform theory, phase and amplitude represent different types of information in the image. Our spectrum swapping experiments show that magnitude mainly reflects pixel-level intensity, while phase predominantly governs image structure. To prevent interference between phase and magnitude feature learning caused by unified frequency-domain modeling, we propose the Phase-Amplitude-Spatial State Space Model (PAS-Mamba) for MRI Reconstruction, a framework that decouples phase and magnitude modeling in the frequency domain and combines it with image-domain features for better reconstruction. In the image domain, LocalMamba preserves spatial locality to sharpen fine anatomical details. In frequency domain, we disentangle amplitude and phase into two specialized branches to avoid representational coupling. To respect the concentric geometry of frequency information, we propose Circular Frequency Domain Scanning (CFDS) to serialize features from low to high frequencies. Finally, a Dual-Domain Complementary Fusion Module (DDCFM) adaptively fuses amplitude phase representations and enables bidirectional exchange between frequency and image domains, delivering superior reconstruction. Extensive experiments on the IXI and fastMRI knee datasets show that PAS-Mamba consistently outperforms state of the art reconstruction methods.

cs.CV↗

arXiv2026

Learning to Factorize and Adapt: A Versatile Approach Toward Universal Spatio-Temporal Foundation Models

Spatio-Temporal (ST) Foundation Models (STFMs) promise cross-dataset generalization, yet joint ST pretraining is computationally expensive and grapples with the heterogeneity of domain-specific spatial patterns. Substantially extending our preliminary conference version, we present FactoST-v2, an enhanced factorized framework redesigned for full weight transfer and arbitrary-length generalization. FactoST-v2 decouples universal temporal learning from domain-specific spatial adaptation. The first stage pretrains a minimalist encoder-only backbone using randomized sequence masking to capture invariant temporal dynamics, enabling probabilistic quantile prediction across variable horizons. The second stage employs a streamlined adapter to rapidly inject spatial awareness via meta adaptive learning and prompting. Comprehensive evaluations across diverse domains demonstrate that FactoST-v2 achieves state-of-the-art accuracy with linear efficiency - significantly outperforming existing foundation models in zero-shot and few-shot scenarios while rivaling domain-specific expert baselines. This factorized paradigm offers a practical, scalable path toward truly universal STFMs. Code is available at https://github.com/CityMind-Lab/FactoST.

cs.LG↗

arXiv2026

How well can off-the-shelf LLMs elucidate molecular structures from mass spectra using chain-of-thought reasoning?

Mass spectrometry (MS) is a powerful analytical technique for identifying small molecules, yet determining complete molecular structures directly from tandem mass spectra (MS/MS) remains a long-standing challenge due to complex fragmentation patterns and the vast diversity of chemical space. Recent progress in large language models (LLMs) has shown promise for reasoning-intensive scientific tasks, but their capability for chemical interpretation is still unclear. In this work, we introduce a Chain-of-Thought (CoT) prompting framework and benchmark that evaluate how LLMs reason about mass spectral data to predict molecular structures. We formalize expert chemists' reasoning steps-such as double bond equivalent (DBE) analysis, neutral loss identification, and fragment assembly-into structured prompts and assess multiple state-of-the-art LLMs (Claude-3.5-Sonnet, GPT-4o-mini, and Llama-3 series) in a zero-shot setting using the MassSpecGym dataset. Our evaluation across metrics of SMILES validity, formula consistency, and structural similarity reveals that while LLMs can produce syntactically valid and partially plausible structures, they fail to achieve chemical accuracy or link reasoning to correct molecular predictions. These findings highlight both the interpretive potential and the current limitations of LLM-based reasoning for molecular elucidation, providing a foundation for future work that combines domain knowledge and reinforcement learning to achieve chemically grounded AI reasoning.

cs.CL↗

arXiv2026

Infini-gram mini: Exact n-gram Search at the Internet Scale with FM-Index

Language models are trained mainly on massive text data from the Internet, and it becomes increasingly important to understand this data source. Exact-match search engines enable searching in large text corpora - counting string appearances and retrieving the enclosing documents - yet the high storage overhead hinders their application on Internet-scale data. We present infini-gram mini, an efficient and scalable system that can make petabyte-level text corpora searchable. Based on the FM-index data structure (Ferragina and Manzini, 2000), which simultaneously indexes and compresses text, our system creates indexes with size only 44% of the corpus. Infini-gram mini greatly improves upon the best existing implementation of FM-index in terms of indexing speed (18$\times$) and memory use during both indexing (3.2$\times$ reduction) and querying (down to a negligible amount). We index 83TB of Internet text in 99 days with a single CPU node with 128 vCPUs (or 19 hours if using 137 such nodes). We show one important use case of infini-gram mini in a large-scale analysis of benchmark contamination. We find several core LM evaluation benchmarks to be heavily contaminated in Internet crawls (up to 74.2% in GSM8K), which could lead to overestimating the capabilities of language models if trained on such data. We host a benchmark contamination bulletin to share the contamination rate of many core and community-contributed benchmarks. We also release a web interface and an API endpoint to serve general search queries on infini-gram mini indexes.

cs.CL↗

arXiv2026

Effective dark matter component presents a robust signature of negative pressure by the DESI observations

Comprehensive cosmological analysis of an effective non-standard dark matter(NSDM) model, characterized by an equation of state $w_{\mathrm{dm}} = w_2 a^2$, which allows for mild deviations from the previously assumed pressureless cold dark matter, is elaborated in the present work. This effective description framework is the scenarios that matter contents coupled to three distinct single-parameter dynamical dark energy models: i.e, the thawing scalar field, the Modified Emergent Dark Energy(MEDE) scenario, and the constant-$w$ model. We constrain these frameworks by using the latest cosmological probes, including the Planck 2018 Cosmic Microwave Background(CMB) distance priors, the Baryon Acoustic Oscillation(BAO) measurements from the Data Release 2 of the Dark Energy Spectroscopic Instrument(DESI), and three compilations of Type Ia Supernovae(SN Ia) namely the Dark Energy Survey Year 5 (DESY5) compilation, the Union3 compilation, and the PantheonPlus (PP) sample. Across all three dark energy scenarios and all dataset combinations, we find a consistent preference for negative values of the parameter $w_2$. Furthermore, this result is robust against the choice of dark energy parametrization, suggesting a model-independent deviation from "standard" cold dark matter. This result indicates that the dark matter fluid possesses a small but non-vanishing negative pressure, meaning a non-cold nature. While the inferred Hubble constant $H_0$ remains consistent with the Planck $Λ$CDM value and does not fully alleviate the $H_0$ tension with local measurements, the persistent detection of $w_2 < 0$ across a wide range of independent cosmological probes provides compelling evidence for new physics in the dark matter sector -- suggesting that dark matter may be better described as an effective fluid endowed with a mild negative pressure, rather than as a perfectly cold, pressureless substance.

astro-ph.CO↗

arXiv2026

Towards Open Diversity-Aware Social Interactions

Social Media and the Internet have catalyzed an unprecedented potential for exposure to human diversity in terms of demographics, talents, opinions, knowledge, and the like. However, this potential has not come with new, much-needed, instruments and skills to harness it. This paper presents our work on promoting richer and deeper social relations through the design and development of the "Internet of Us", an online platform that uses diversity-aware Artificial Intelligence to mediate and empower human social interactions. We discuss the multiple facets of diversity in social settings, the multidisciplinary work that is required to reap the benefits of diversity, and the vision for a diversity-aware hybrid human-AI society.

cs.HC↗

arXiv2025

AncientBench: Towards Comprehensive Evaluation on Excavated and Transmitted Chinese Corpora

Comprehension of ancient texts plays an important role in archaeology and understanding of Chinese history and civilization. The rapid development of large language models needs benchmarks that can evaluate their comprehension of ancient characters. Existing Chinese benchmarks are mostly targeted at modern Chinese and transmitted documents in ancient Chinese, but the part of excavated documents in ancient Chinese is not covered. To meet this need, we propose the AncientBench, which aims to evaluate the comprehension of ancient characters, especially in the scenario of excavated documents. The AncientBench is divided into four dimensions, which correspond to the four competencies of ancient character comprehension: glyph comprehension, pronunciation comprehension, meaning comprehension, and contextual comprehension. The benchmark also contains ten tasks, including radical, phonetic radical, homophone, cloze, translation, and more, providing a comprehensive framework for evaluation. We convened archaeological researchers to conduct experimental evaluations, proposed an ancient model as baseline, and conducted extensive experiments on the currently best-performing large language models. The experimental results reveal the great potential of large language models in ancient textual scenarios as well as the gap with humans. Our research aims to promote the development and application of large language models in the field of archaeology and ancient Chinese language.

cs.CL↗

arXiv2025

Hessian spectrum at the global minimum and topology trivialization of locally isotropic Gaussian random fields

We study the energy landscape near the ground state of a model of a single particle in a random potential with trivial topology. More precisely, we find the large dimensional limit of the Hessian spectrum at the global minimum of the Hamiltonian $X_N(x) +\frac\mu2 \|x\|^2, x\in\mathbb{R}^N,$ when $μ$ is above the phase transition threshold so that the system has only one critical point. Here $X_N$ is a locally isotropic Gaussian random field. The same idea is also applied to study the more general model of elastic manifold. In the replica symmetric regime, our results confirm the predictions of Fyodorov and Le Doussal made in 2018 and 2020 using the replica method.

math.PR↗

arXiv2025

Measurement-Induced Perturbations of Hausdorff Dimension in Quantum Paths

In a seminal paper, Abbott et al. analyzed the relationship between a particle's trajectory and the resolution of position measurements performed by an observer at fixed time intervals. They predicted that quantum paths exhibit a universal Hausdorff dimension that transitions from $d=2$ to $d=1$ as the momentum of the particle increases. However, although measurements were assumed to occur at intervals of time, the calculations only involved evaluating the expectation value of operators for the free evolution of wave function within a single interval, with no actual physical measurements performed. In this work we investigate how quantum measurements alter the fractal geometry of quantum particle paths. By modelling sequential measurements using Gaussian wave packets for both the particle and the apparatus, we reveal that the dynamics of the measurement change the roughness of the path and shift the emergent Hausdorff dimension towards a lower value in nonselective evolution. For selective evolution, feedback control forces must be introduced to counteract stochastic wave function collapse, stabilising trajectories and enabling dimensionality to be tuned. When the contribution of the measurement approaches zero, our result reduces to that of Abbott et al. Our work can thus be regarded as a more realistic formulation of their approach, and it connects theoretical quantum fractality with measurement physics, quantifying how detectors reshape spacetime statistics at quantum scales.

quant-ph↗