arXiv subjects

Zihao Wang

Publications and source records attributed to Zihao Wang.

At least 127 records · Page 7Linked to original sources

arXiv2026

Efficient Reinforcement Learning for Large Language Models with Intrinsic Exploration

Reinforcement learning with verifiable rewards (RLVR) has improved the reasoning ability of large language models, yet training remains costly because many rollouts contribute little to optimization, considering the amount of computation required. This study investigates how simply leveraging intrinsic data properties, almost free benefit during training, can improve data efficiency for RLVR. We propose PREPO with two complementary components. First, we adopt prompt perplexity as an indicator of model adaptability in learning, enabling the model to progress from well-understood contexts to more challenging ones. Second, we amplify the discrepancy among the rollouts by differentiating their relative entropy, and prioritize sequences that exhibit a higher degree of exploration. Together, these mechanisms reduce rollout demand while preserving competitive performance. On the Qwen and Llama models, PREPO achieves effective results on mathematical reasoning benchmarks with up to 3 times fewer rollouts than the baselines. Beyond empirical gains, we provide theoretical and in-depth analyses explaining the underlying rationale of our method to improve the data efficiency of RLVR.

cs.LG↗

arXiv2026

Interplay of Quantum Size Effect and Tensile Strain on Surface Morphology of Sn(100) Islands

The quantum size effect (QSE) and strain effect are two key factors influencing the surface morphology of thin films, which can increase film surface roughness through QSE-induced thickness oscillation and strain-induced island formation, respectively. Surface roughness usually manifests in the early stages of film growth and diminishes beyond a critical thickness. In this work, we employ molecular beam epitaxy (MBE) to grow Sn(100) islands with varying thickness N on bilayer graphene-terminated 6H-SiC(0001) substrates. Scanning tunneling microscopy and spectroscopy measurements reveal an inverse surface roughness effect that highlights the interplay of QSE and misfit strain in shaping the surface morphology of Sn(100) islands. For N =< 10, the islands exhibit flat surfaces, while for N >= 26, the island surfaces become corrugated and patterned. For the intermediate range, i.e., 12 =< N =<24, both flat and patterned surfaces coexist, with the percentage coverage of the patterned surface oscillating as a function of N. By performing density functional theory calculations, we demonstrate that the unusual surface pattern evolution in our MBE-grown Sn(100) islands is a result of the interplay between QSE-induced surface roughing and tensile strain-induced smoothening effect.

cond-mat.mtrl-sci↗

arXiv2026

Non-negative matrix factorization algorithms generally improve topic model fits

In an effort to develop topic modeling methods that can be quickly applied to large data sets, we revisit the problem of maximum-likelihood estimation in topic models. It is known, at least informally, that maximum-likelihood estimation in topic models is closely related to non-negative matrix factorization (NMF). Yet, to our knowledge, this relationship has not been exploited previously to fit topic models. We show that recent advances in NMF optimization methods can be leveraged to fit topic models very efficiently, often resulting in much better fits and in less time than existing algorithms for topic models. We also formally make the connection between the NMF optimization problem and maximum-likelihood estimation for the topic model, and using this result we show that the expectation maximization (EM) algorithm for the topic model is essentially the same as the classic multiplicative updates for NMF (the only difference being that the operations are performed in a different order). Our methods are implemented in the R package fastTopics.

stat.ML↗

arXiv2026

Overcoming sensitivity-bandwidth trade-off in mid-infrared spectroscopy by a microresonator-anchored swept laser

Optical frequency combs have revolutionized high-precision spectroscopy, yet an intrinsic trade-off between spectroscopic signal-to-noise ratio (sSNR) and measurement bandwidth ($B$) fundamentally constrains sensitive, broadband measurements. While broadband swept lasers offer a potential solution, generating broadband, ultrafast and linearly sweeping lasers with a narrow linewidth remains a significant challenge, particularly in the fingerprint mid-infrared (mid-IR) band. Here we overcome this limitation by using a microresonator-anchored ultrafast sweeping Fourier domain mode-locked (FDML) laser for mid-IR spectroscopy. We introduce a dual-microresonator-anchor approach: a microcomb provides frequency calibration and a high-Q microresonator resolves the instantaneous FDML lasing lineshape. The strategy enables accurate correction of the FDML laser's sweep nonlinearity and broad linewidth in the near-IR, allowing the FDML laser to function as a high-fidelity mid-IR light via difference frequency generation. The system achieves a record sSNR$\times$$B$ of 1.3$\times$10$^5$ THz$\cdot \sqrt{\rm Hz}$ and methane sensing precision of 9 ppb$\cdot$m$\cdot$$\sqrt{\rm s}$, while retaining GHz resolution to distinguish methane isotope. We further demonstrate broadband, coherent swept laser phase spectroscopy in the mid-IR, tolerating losses up to 78 dB. This work leverages advances in integrated photonics to overcome the fundamental limitations of precision spectroscopy, paving the way for next-generation, broadband, and ultra-sensitive mid-IR spectroscopic sensing systems.

physics.optics↗

arXiv2026

2DESR: a two-dimensional Fourier-space gyrokinetic eigenvalue code for the ion-temperature-gradient modes in tokamaks

A two-dimensional (2D) gyrokinetic eigenvalue solver, 2DESR, has been developed to solve the 2D gyrokinetic eigenvalue problem in the poloidal Fourier space for the ion-temperature-gradient (ITG) modes in tokamaks. With full kinetic effects of ions retained, the 2D gyrokinetic eigenvalue equations in the poloidal Fourier space have been derived and numerically solved in the 2DESR code. In the linear ITG Cyclone test with adiabatic electrons, the 2DESR code benchmarks well against the gyrokinetic initial-value codes GENE and NLT. It is found that two branches of ITG modes coexist in the system.

physics.plasm-ph↗

arXiv2026

The THESAN-ZOOM project: Mystery N/O more -- uncovering the origin of peculiar chemical abundances and a not-so-fundamental metallicity relation at $3<z<12$

We present an analysis of metallicities and chemical abundances at $3 -0.6$) without the need for exotic yields in our chemical network. Instead, bursty star formation naturally generates order-of-magnitude excursions in N/O on $\lesssim$100 Myr timescales due to temporally differential galactic winds; after a starburst, stellar feedback expels gas, leaving a large population of asymptotic-giant-branch stars to dominate the enrichment of the relatively low-mass interstellar medium. NRGs lie below the main sequence and typically exhibit $\mathrm{EW}[H$β$]\lesssim40$ Å, in apparent tension with observed high-EW NRGs. This tension is reconciled if observed NRGs are in the initial stages of a subsequent starburst, illuminating previously enriched gas, which is supported by the finding of high SFR surface density nitrogen-rich giant molecular clouds.

astro-ph.GA↗

arXiv2026

MergePipe: A Budget-Aware Parameter Management System for Scalable LLM Merging

Large language model (LLM) merging has become a key technique in modern LLM development pipelines, enabling the integration of multiple task- or domain-specific expert models without retraining. However, as the number of experts grows, existing merging implementations treat model parameters as unstructured files and execute merges in a stateless, one-shot manner, leading to excessive disk I/O, redundant parameter scans, and poor scalability. In this paper, we present \textbf{MergePipe}, a parameter management system for scalable LLM merging. MergePipe is the first system that treats LLM merging as a data management and execution problem, and introduces a catalog-driven abstraction over model parameters, merge plans, and execution lineage. At its core, MergePipe employs a cost-aware planner that explicitly models expert parameter I/O and enforces user-specified I/O budgets, followed by a streaming execution engine that materializes merged models under transactional guarantees. Our key insight is that while base model reads and output writes are unavoidable, expert parameter reads dominate merge cost and constitute the primary optimization target. By making expert access budget-aware throughout planning and execution, MergePipe mitigates the $O(K)$ I/O growth of naive pipelines and achieves predictable scaling behavior. Experiments show that MergePipe reduces total I/O by up to an order of magnitude and delivers up to $11\times$ end-to-end speedups (up to 90\% wall-time reduction) over state-of-the-art LLM merging pipelines.

cs.DB↗

arXiv2026

Neural Networks Learn Generic Multi-Index Models Near Information-Theoretic Limit

In deep learning, a central issue is to understand how neural networks efficiently learn high-dimensional features. To this end, we explore the gradient descent learning of a general Gaussian Multi-index model $f(\boldsymbol{x})=g(\boldsymbol{U}\boldsymbol{x})$ with hidden subspace $\boldsymbol{U}\in \mathbb{R}^{r\times d}$, which is the canonical setup to study representation learning. We prove that under generic non-degenerate assumptions on the link function, a standard two-layer neural network trained via layer-wise gradient descent can agnostically learn the target with $o_d(1)$ test error using $\widetilde{\mathcal{O}}(d)$ samples and $\widetilde{\mathcal{O}}(d^2)$ time. The sample and time complexity both align with the information-theoretic limit up to leading order and are therefore optimal. During the first stage of gradient descent learning, the proof proceeds via showing that the inner weights can perform a power-iteration process. This process implicitly mimics a spectral start for the whole span of the hidden subspace and eventually eliminates finite-sample noise and recovers this span. It surprisingly indicates that optimal results can only be achieved if the first layer is trained for more than $\mathcal{O}(1)$ steps. This work demonstrates the ability of neural networks to effectively learn hierarchical functions with respect to both sample and time efficiency.

stat.ML↗

arXiv2026

RePPL: Recalibrating Perplexity by Uncertainty in Semantic Propagation and Language Generation for Explainable QA Hallucination Detection

Large Language Models (LLMs) have become powerful, but hallucinations remain a vital obstacle to their trustworthy use. Previous works improved the capability of hallucination detection by measuring uncertainty. But they can not explain the provenance behind why hallucinations occur, particularly in identifying which part of the inputs tends to trigger hallucinations. Recent works on the prompt attack indicate that uncertainty exists in semantic propagation, where attention mechanisms gradually fuse local token information into high-level semantics across layers. Meanwhile, uncertainty also emerges in language generation, due to its probability-based selection of high-level semantics for sampled generations. Based on that, we propose RePPL to recalibrate uncertainty measurement by these two aspects, which dispatches explainable uncertainty scores to each token and aggregates in Perplexity-style Log-Average form as a total score. Experiments show that it achieves the best comprehensive detection performance across various QA datasets on advanced models (average AUC of 0.833), and it is capable of producing token-level uncertainty scores as explanations of hallucination.

cs.CL↗

arXiv2026

Adaptive Domain Shift in Diffusion Models for Cross-Modality Image Translation

Cross-modal image translation remains brittle and inefficient. Standard diffusion approaches often rely on a single, global linear transfer between domains. We find that this shortcut forces the sampler to traverse off-manifold, high-cost regions, inflating the correction burden and inviting semantic drift. We refer to this shared failure mode as fixed-schedule domain transfer. In this paper, we embed domain-shift dynamics directly into the generative process. Our model predicts a spatially varying mixing field at every reverse step and injects an explicit, target-consistent restoration term into the drift. This in-step guidance keeps large updates on-manifold and shifts the model's role from global alignment to local residual correction. We provide a continuous-time formulation with an exact solution form and derive a practical first-order sampler that preserves marginal consistency. Empirically, across translation tasks in medical imaging, remote sensing, and electroluminescence semantic mapping, our framework improves structural fidelity and semantic consistency while converging in fewer denoising steps.

cs.CV↗

arXiv2026

Causal Bayesian Optimization via Exogenous Distribution Learning

Maximizing a target variable as an operational objective within a structural causal model is a fundamental problem. Causal Bayesian Optimization (CBO) approaches typically achieve this either by performing interventions that modify the causal structure to increase the reward or by introducing action nodes to endogenous variables, thereby adjusting the data-generating mechanisms to meet the objective. In this paper, we propose a novel method that learns the distribution of exogenous variables-an aspect often ignored or marginalized through expectation in existing CBO frameworks. By modeling the exogenous distribution, we enhance the approximation fidelity of the data-generating structural causal models (SCMs) used in surrogate models, which are commonly trained on limited observational data. Furthermore, the ability to recover exogenous variables enables the application of our approach to more general causal structures beyond the confines of Additive Noise Models (ANMs) and single-mode Gaussian, allowing the use of more expressive priors for context noise. We incorporate the learned exogenous distribution into a new CBO method, demonstrating its advantages across diverse datasets and application scenarios.

cs.LG↗

arXiv2026

Do Reasoning Models Enhance Embedding Models?

State-of-the-art embedding models are increasingly derived from decoder-only Large Language Model (LLM) backbones adapted via contrastive learning. Given the emergence of reasoning models trained via Reinforcement Learning with Verifiable Rewards (RLVR), a natural question arises: do enhanced reasoning translate to superior semantic representations when these models serve as embedding initializations? Contrary to expectation, our evaluation on MTEB and BRIGHT reveals a **null effect**: embedding models initialized from RLVR-tuned backbones yield no consistent performance advantage over their base counterparts when subjected to identical training recipes. To unpack this paradox, we introduce **H**ierarchical **R**epresentation **S**imilarity **A**nalysis (HRSA), a framework that decomposes similarity across representation, geometry, and function levels. HRSA reveals that while RLVR induces irreversible latent manifold's local geometry reorganization and reversible coordinate basis drift, it preserves the global manifold geometry and linear readout. Consequently, subsequent contrastive learning drives strong alignment between base- and reasoning-initialized models, a phenomenon we term **Manifold Realignment**. Empirically, our findings suggest that unlike Supervised Fine-Tuning (SFT), RLVR optimizes trajectories within an existing semantic landscape rather than fundamentally restructuring the landscape itself.

cs.AI↗

arXiv2026

Structure-constrained Language-informed Diffusion Model for Unpaired Low-dose Computed Tomography Angiography Reconstruction

The application of iodinated contrast media (ICM) improves the sensitivity and specificity of computed tomography (CT) for a wide range of clinical indications. However, overdose of ICM can cause problems such as kidney damage and life-threatening allergic reactions. Deep learning methods can generate CT images of normal-dose ICM from low-dose ICM, reducing the required dose while maintaining diagnostic power. However, existing methods are difficult to realize accurate enhancement with incompletely paired images, mainly because of the limited ability of the model to recognize specific structures. To overcome this limitation, we propose a Structure-constrained Language-informed Diffusion Model (SLDM), a unified medical generation model that integrates structural synergy and spatial intelligence. First, the structural prior information of the image is effectively extracted to constrain the model inference process, thus ensuring structural consistency in the enhancement process. Subsequently, semantic supervision strategy with spatial intelligence is introduced, which integrates the functions of visual perception and spatial reasoning, thus prompting the model to achieve accurate enhancement. Finally, the subtraction angiography enhancement module is applied, which serves to improve the contrast of the ICM agent region to suitable interval for observation. Qualitative analysis of visual comparison and quantitative results of several metrics demonstrate the effectiveness of our method in angiographic reconstruction for low-dose contrast medium CT angiography.

cs.CV↗

arXiv2026

Unlocking Scaling Law in Industrial Recommendation Systems with a Three-step Paradigm based Large User Model

Recent advancements in autoregressive Large Language Models (LLMs) have achieved significant milestones, largely attributed to their scalability, often referred to as the "scaling law". Inspired by these achievements, there has been a growing interest in adapting LLMs for Recommendation Systems (RecSys) by reformulating RecSys tasks into generative problems. However, these End-to-End Generative Recommendation (E2E-GR) methods tend to prioritize idealized goals, often at the expense of the practical advantages offered by traditional Deep Learning based Recommendation Models (DLRMs) in terms of in features, architecture, and practices. This disparity between idealized goals and practical needs introduces several challenges and limitations, locking the scaling law in industrial RecSys. In this paper, we introduce a large user model (LUM) that addresses these limitations through a three-step paradigm, designed to meet the stringent requirements of industrial settings while unlocking the potential for scalable recommendations. Our extensive experimental evaluations demonstrate that LUM outperforms both state-of-the-art DLRMs and E2E-GR approaches. Notably, LUM exhibits excellent scalability, with performance improvements observed as the model scales up to 7 billion parameters. Additionally, we have successfully deployed LUM in an industrial application, where it achieved significant gains in an A/B test, further validating its effectiveness and practicality.

cs.IR↗

arXiv2026

StoxLSTM: A Stochastic Extended Long Short-Term Memory Network for Time Series Forecasting

The Extended Long Short-Term Memory (xLSTM) network has demonstrated strong capability in modeling complex long-term dependencies in time series data. Despite its success, the deterministic architecture of xLSTM limits its representational capacity and forecasting performance, especially on challenging real-world time series datasets characterized by inherent uncertainty, stochasticity, and complex hierarchical latent dynamics. In this work, we propose StoxLSTM, a stochastic xLSTM within a designed state space modeling framework, which integrates latent stochastic variables directly into the recurrent units to effectively model deep latent temporal dynamics and uncertainty. The designed state space model follows an efficient non-autoregressive generative approach, achieving strong predictive performance without complex modifications to the original xLSTM architecture. Extensive experiments on publicly available benchmark datasets demonstrate that StoxLSTM consistently outperforms state-of-the-art baselines, achieving superior performance and generalization.

cs.LG↗

arXiv2026

Quantum Hall Effect at 0.002T

Graphene enables precise carrier-density control via gating, making it an ideal platform for studying electronic interactions. However, sample inhomogeneities often limit access to the low-density regimes where these interactions dominate. Enhancing carrier mobility is therefore crucial for exploring fundamental properties and developing device applications. Here, we demonstrate a significant reduction in external inhomogeneity using a double-layer graphene architecture separated by an ultra-thin hexagonal boron nitride layer. Mutual screening between the layers reduces scattering from random Coulomb potentials, resulting in a quantum mobility exceeding. Shubnikov de-Haas oscillations emerge at magnetic fields below 1 mT, while integer quantum Hall features are observed at 0.002T. Furthermore, we identify a fractional quantum Hall plateau at a filling factor of at 2T. These results demonstrate the platform's suitability for investigating strongly correlated electronic phases in graphene-based heterostructures.

cond-mat.mes-hall↗

arXiv2026

CREATE: Cross-Layer Resilience Characterization and Optimization for Efficient yet Reliable Embodied AI Systems

Embodied Artificial Intelligence (AI) has recently attracted significant attention as it bridges AI with the physical world. Modern embodied AI systems often combine a Large Language Model (LLM)-based planner for high-level task planning and a reinforcement learning (RL)-based controller for low-level action generation, enabling embodied agents to tackle complex tasks in real-world environments. However, deploying embodied agents remains challenging due to their high computation requirements, especially for battery-powered local devices. Although techniques like lowering operating voltage can improve energy efficiency, they can introduce bit errors and result in task failures. In this work, we propose CREATE, a general design principle that leverages heterogeneous resilience at different layers for synergistic energy-reliability co-optimization. For the first time, we conduct a comprehensive error injection study on modern embodied AI systems and observe an inherent but heterogeneous fault tolerance. Building upon these insights, we develop an anomaly detection and clearance mechanism at the circuit level to eliminate outlier errors. At the model level, we propose a weight-rotation-enhanced planning algorithm to improve the fault tolerance of the LLM-based planner. Furthermore, we introduce an application-level technique, autonomy-adaptive voltage scaling, to dynamically adjust the operating voltage of the controllers. The voltage scaling circuit is co-designed to enable online voltage adjustment. Extensive experiments demonstrate that without compromising task quality, CREATE achieves 40.6% computational energy savings on average over nominal-voltage baselines and 35.0% over prior-art techniques. This further leads to 29.5% to 37.3% chip-level energy savings and approximately a 15% to 30% improvement in battery life.

cs.AR↗

arXiv2026

Dynamic Large Concept Models: Latent Reasoning in an Adaptive Semantic Space

Large Language Models (LLMs) apply uniform computation to all tokens, despite language exhibiting highly non-uniform information density. This token-uniform regime wastes capacity on locally predictable spans while under-allocating computation to semantically critical transitions. We propose $\textbf{Dynamic Large Concept Models (DLCM)}$, a hierarchical language modeling framework that learns semantic boundaries from latent representations and shifts computation from tokens to a compressed concept space where reasoning is more efficient. DLCM discovers variable-length concepts end-to-end without relying on predefined linguistic units. Hierarchical compression fundamentally changes scaling behavior. We introduce the first $\textbf{compression-aware scaling law}$, which disentangles token-level capacity, concept-level reasoning capacity, and compression ratio, enabling principled compute allocation under fixed FLOPs. To stably train this heterogeneous architecture, we further develop a $\textbf{decoupled $μ$P parametrization}$ that supports zero-shot hyperparameter transfer across widths and compression regimes. At a practical setting ($R=4$, corresponding to an average of four tokens per concept), DLCM reallocates roughly one-third of inference compute into a higher-capacity reasoning backbone, achieving a $\textbf{+2.69$\%$ average improvement}$ across 12 zero-shot benchmarks under matched inference FLOPs.

cs.LG↗