arXiv subjects

Jinwen Wang

Publications and source records attributed to Jinwen Wang.

At least 19 recordsLinked to original sources

arXiv2026

AuK Technical Report: An Open-Source Foundational Model for Speech Generation and Editing

We introduce AuK, an open-source foundational model that unifies speech generation and editing through a common interface of natural-language instructions and audio context. To support this broad capability set, we construct approximately 3.03 billion instruction--audio instances and 1.95 million hours of effective supervision across five task families: speech generation, content editing, enhancement and separation, paralinguistic editing, and acoustic editing. AuK combines a multimodal large language model for semantic conditioning, an VAE jointly trained on speech, general audio, and music for acoustic conditioning, and a hybrid rectified-flow Transformer that performs dual-stream MMDiT blocks followed by unified single-stream DiT blocks for generation. Training begins with generation-only warm-up and proceeds to joint generation--editing pre-training. We then apply complementary post-training strategies: human-feedback preference optimization for open-ended editing and reward-based reinforcement learning for speech generation. To reduce inference cost, we further distill the model with consistency initialization and task-routed Decoupled DMD. The resulting AuK-Flash performs 4-step inference without classifier-free guidance and achieves a 4.5 wall-clock speedup over the full model under matched conditions. Experiments demonstrate leading performance on zero-shot and instruction-controlled speech generation and general instruction-guided editing, while remaining competitive on signal-level restoration tasks. We release both the source code and model weights to support reproducibility and further research.

cs.SD↗

arXiv2026

Local Motion Matters: A Deconstruct-Recompose Paradigm for Reinforcement Learning Pre-training from Videos

Pre-training on large-scale videos to improve reinforcement learning efficiency is promising yet remains challenging. Existing methods typically treat the agent as an indivisible entity, modeling motion patterns globally. Such global modeling is tightly coupled with the morphology, hindering transfer across domains. In contrast, despite the vast disparity in global motions, the local components exhibit similar motion patterns across different agents. Building on this insight, we propose a novel Deconstruct-Recompose Paradigm (DRP) for learning transferable local motion representations. Specifically, in the Deconstruct phase, we identify multiple local points and track their frame-wise motions, defining each as an Atomic Action. We introduce a Dual-Attention Encoder (DAE) to learn local motion representations from these Atomic Actions, capturing their spatiotemporal relationships. In the Recompose phase, we compose local motion representations with a learnable Motion Aggregation Token [MAT] via latent dynamics model learning. Additionally, an adapter bridges local motion and downstream action-specific dynamics to accelerate policy learning. Extensive experiments demonstrate that our method effectively transfers to diverse robotic control and manipulation tasks, significantly improving sample efficiency and performance.

cs.LG↗

arXiv2026

From Pixels to Temporal Correlations: Learning Informative Representations for Reinforcement Learning Pre-training

Unsupervised pre-training on large-scale datasets has demonstrated significant potential for improving the sample efficiency and performance of Reinforcement Learning (RL). Given the large-scale action-free internet videos, existing methods utilize single-step transition prediction and image reconstruction to learn representations. However, these methods prefer to preserve large-proportion stationary information in the pixel space, neglecting small but crucial information. To preserve enough information in the representation, it is essential to pay equal attention to each element in videos. Specifically, we propose a temporal correlation space to distinguish each element. For implementation, we introduce the Multi-scale Temporal Contrastive Learning (MTCL) method to model multi-scale temporal correlations separately. This approach can balance the attention of different elements and yield more informative representations, effectively supporting policy learning in various downstream tasks. Experimental results demonstrate that our method improves sample efficiency and asymptotic performance across various downstream tasks.

cs.LG↗

arXiv2026

Task-Relevant Representation Decoupling for Visual Reinforcement Learning Generalization

Visual Reinforcement Learning (VRL) has achieved considerable success in solving control tasks. However, generalizing learned policies to new environments remains a major challenge, as agents often overfit to task-irrelevant features in the training environment. To solve this problem, we introduce the concept of decoupling observations into task-relevant and task-irrelevant representations. Building on this idea, we propose a self-supervised Task-Relevant Representation Decoupling (T2RD) algorithm for VRL. This algorithm consists of three components: task-relevant representation consistency, cross-reconstruction, and cross-dynamic prediction. The first two components achieve the decoupling of content and style features, but the resulting content representations are not necessarily task-relevant. To further refine task-relevant features from content representations, we design the third component that introduces dynamic prediction. T2RD achieves State-Of-The-Art (SOTA) generalization performance and sample efficiency in the DeepMind Control Suite and Robotic Manipulation tasks.

cs.LG↗

arXiv2026

Optical hopfions with arbitrary two winding numbers

Hopfions, as three-dimensional topologically nontrivial structures described by poloidal and toroidal winding numbers, hold promise as robust information carriers in spintronics, functional materials, and optical communications. Although they have been experimentally realized in various physical systems, such realizations have been restricted to low orders, with the winding numbers lacking tunability. Here, using optical fields as our platform, we outline how to make tunable hopfions in any order with any winding number. We use tailored superpositions of Laguerre-Gaussian modes in free-space as our construction, achieving effective control for arbitrary-order poloidal and toroidal winding numbers, which we demonstrate up to orders 5 and 3, respectively, for a new state-of-the-art. The resulting torus-knot structures are visualized experimentally via polarization filaments, confirming the designed topological textures. Our work reports an exotic optical topologies observed in free space, provides a systematic route hopfions of any order, with implications for topological photonics, optical communications, and analogies in magnetic and condensed-matter systems.

physics.optics↗

arXiv2025

Storage and retrieval of optical skyrmions with topological characteristics

Optical skyrmions are topological structures of light whose defining property, the skyrmion number, is robust against perturbations. This makes them attractive for applications in quantum information storage, where resilience to decoherence is paramount. However, their preservation during coherent storage remains unexplored. We report the first experimental demonstration of storing and retrieving optical skyrmions in a cold $^{87}$Rb vapor using a dual-path electromagnetically induced transparency memory. Crucially, we show that the skyrmion number remains invariant for storage times up to several microseconds, even when subjected to imbalanced loss between the two paths and substantial perturbations in control beam power. Our work demonstrates the survival of a non-trivial topological invariant in a quantum memory, marking a significant step towards topologically protected photonic technologies.

quant-ph↗

arXiv2025

SCRUTINEER: Detecting Logic-Level Usage Violations of Reusable Components in Smart Contracts

Smart Contract Reusable Components(SCRs) play a vital role in accelerating the development of business-specific contracts by promoting modularity and code reuse. However, the risks associated with SCR usage violations have become a growing concern. One particular type of SCR usage violation, known as a logic-level usage violation, is becoming especially harmful. This violation occurs when the SCR adheres to its specified usage rules but fails to align with the specific business logic of the current context, leading to significant vulnerabilities. Detecting such violations necessitates a deep semantic understanding of the contract's business logic, including the ability to extract implicit usage patterns and analyze fine-grained logical behaviors. To address these challenges, we propose SCRUTINEER, the first automated and practical system for detecting logic-level usage violations of SCRs. First, we design a composite feature extraction approach that produces three complementary feature representations, supporting subsequent analysis. We then introduce a Large Language Model-powered knowledge construction framework, which leverages comprehension-oriented prompts and domain-specific tools to extract logic-level usage and build the SCR knowledge base. Next, we develop a Retrieval-Augmented Generation-driven inspector, which combines a rapid retrieval strategy with both comprehensive and targeted analysis to identify potentially insecure logic-level usages. Finally, we implement a logic-level usage violation analysis engine that integrates a similarity-based checker and a snapshot-based inference conflict checker to enable accurate and robust detection. We evaluate SCRUTINEER from multiple perspectives on 3 ground-truth datasets. The results show that SCRUTINEER achieves a precision of 80.77%, a recall of 82.35%, and an F1-score of 81.55% in detecting logic-level usage violations of SCRs.

cs.SE↗

arXiv2025

Toward Engineering AGI: Benchmarking the Engineering Design Capabilities of LLMs

Modern engineering, spanning electrical, mechanical, aerospace, civil, and computer disciplines, stands as a cornerstone of human civilization and the foundation of our society. However, engineering design poses a fundamentally different challenge for large language models (LLMs) compared with traditional textbook-style problem solving or factual question answering. Although existing benchmarks have driven progress in areas such as language understanding, code synthesis, and scientific problem solving, real-world engineering design demands the synthesis of domain knowledge, navigation of complex trade-offs, and management of the tedious processes that consume much of practicing engineers' time. Despite these shared challenges across engineering disciplines, no benchmark currently captures the unique demands of engineering design work. In this work, we introduce EngDesign, an Engineering Design benchmark that evaluates LLMs' abilities to perform practical design tasks across nine engineering domains. Unlike existing benchmarks that focus on factual recall or question answering, EngDesign uniquely emphasizes LLMs' ability to synthesize domain knowledge, reason under constraints, and generate functional, objective-oriented engineering designs. Each task in EngDesign represents a real-world engineering design problem, accompanied by a detailed task description specifying design goals, constraints, and performance requirements. EngDesign pioneers a simulation-based evaluation paradigm that moves beyond textbook knowledge to assess genuine engineering design capabilities and shifts evaluation from static answer checking to dynamic, simulation-driven functional verification, marking a crucial step toward realizing the vision of engineering Artificial General Intelligence (AGI).

cs.CE↗

arXiv2025

Learning Robust Representations via Bidirectional Transition for Visual Reinforcement Learning

Visual reinforcement learning has proven effective in solving control tasks with high-dimensional observations. However, extracting reliable and generalizable representations from vision-based observations remains a central challenge. Inspired by the human thought process, when the representation extracted from the observation can predict the future and trace history, the representation is reliable and accurate in comprehending the environment. Based on this concept, we introduce a Bidirectional Transition (BiT) model, which leverages the ability to bidirectionally predict environmental transitions both forward and backward to extract reliable representations. Our model demonstrates competitive generalization performance and sample efficiency on two settings of the DeepMind Control suite. Additionally, we utilize robotic manipulation and CARLA simulators to demonstrate the wide applicability of our method.

cs.CV↗

arXiv2025

Tailoring ultra-high-order optical skyrmions

Skyrmions, as quasiparticles with topological spin textures, has recently garnered great attention for both condensed matter and structured wave communities, promising next-generation large-density robust information technologies. However, a big challenge to this end is that the generation of high-order skyrmions is elusive in any physical systems. Here, we propose the method to create and control ultra-high-order skyrmions (skyrmion number up to $400^{th}$) in a structured light system. We also experimentally control the topological state transition between bimeron and skyrmion, arbitrarily tailor the transverse size of an arbitrary-order skyrmionic beam independent of topological number, and ensure the topological stability upon propagation. Our work offers solutions for topologically resilient communication and memory with much enhanced information capacity.

physics.optics↗

arXiv2025

Efficient multiplexed quantum memory with high dimensional orbital angular momentum states in cold atoms

Quantum memory plays a pivotal role in the construction of quantum repeaters, which are essential devices for establishing long-distance quantum communications and large-scale quantum networks. To boost information capacity and signal processing speed, the implementation of high-efficiency multiplexed quantum memories is essential for the development of multiplexed quantum repeaters. In this work, we experimentally demonstrate an efficient multiplexed quantum memory by consolidating photons carrying high-dimensional orbital angular momentum (OAM) state from 4 individual channels into an elongated cold $^{87}$Rb atomic ensemble. Benefiting from the cold atomic ensemble with high optical depth, we obtain a storage efficiency exceeding 70% for the 4-channel multiplexed beam. The storage fidelities surpass 83% when all channels are operated in a 4-dimensional Hilbert space, which is far above the classical benchmark and validates the quantum functionality of our memory system. The achieved high-efficiency OAM multiplexed quantum memory opens up an avenue for efficient quantum information processing over multiple parallel channels, promising significant advances in the field of quantum communication and networking.

quant-ph↗

arXiv2025

Secure Information Embedding in Forensic 3D Fingerprinting

Printer fingerprinting techniques have long played a critical role in forensic applications, including the tracking of counterfeiters and the safeguarding of confidential information. The rise of 3D printing technology introduces significant risks to public safety, enabling individuals with internet access and consumer-grade 3D printers to produce untraceable firearms, counterfeit products, and more. This growing threat calls for a better mechanism to track the production of 3D-printed parts. Inspired by the success of fingerprinting on traditional 2D printers, we introduce SIDE (\textbf{S}ecure \textbf{I}nformation Embe\textbf{D}ding and \textbf{E}xtraction), a novel fingerprinting framework tailored for 3D printing. SIDE addresses the adversarial challenges of 3D print forensics by offering both secure information embedding and extraction. First, through novel coding-theoretic techniques, SIDE is both~\emph{break-resilient} and~\emph{loss-tolerant}, enabling fingerprint recovery even if the adversary breaks the print into fragments and conceals a portion of them. Second, SIDE further leverages Trusted Execution Environments (TEE) to secure the fingerprint embedding process.

cs.CR↗

arXiv2024

Generation and characterization of customized perfect Laguerre-Gaussian beams with arbitrary profiles

We experimentally demonstrate the generation of customized perfect Laguerre-Gaussian (PLG) beams whose intensity maxima localized around any desired curves. The principle is to act appropriate algebraic functions on the angular spectra of PLG beams. We characterize the propagation properties of these beams and compare them with non-diffraction caustic beams possessing the same intensity profiles. The results manifest that the customized-PLG beams can maintain their profiles during propagation and suffer less energy loss than the non-diffraction caustic beams, and hence are able to propagate a longer distance. This new structure beam would have potential applications in areas such as optical communication, soliton routing and steering, optical tweezing and trapping, atom optics, etc.

physics.optics↗

arXiv2024

Generation of subnatural-linewidth orbital angular momentum entangled biphotons using a single driving laser in hot atoms

Orbital angular momentum (OAM) entangled photon pairs with narrow bandwidths play a crucial role in the interaction of light and quantum states of matter. In this article, we demonstrate an approach for generating OAM entangled photon pairs with a narrow bandwidth by using a single driving beam in a $^{85}$Rb atomic vapor cell. This single driving beam is able to simultaneously couple two atomic transitions and directly generate OAM entangled biphotons by leveraging the OAM conservation law through the spontaneous four-wave mixing (SFWM) process. The photon pairs exhibit a maximum cross-correlation function value of 27.7 and a linewidth of 4 MHz. The OAM entanglement is confirmed through quantum state tomography, revealing a fidelity of 95.7\% and a concurrence of 0.926 when compared to the maximally entangled state. Our scheme is notably simpler than previously proposed schemes and represents the first demonstration of generating subnatural-linewidth entangled photon pairs in hot atomic systems.

quant-ph↗

arXiv2024

Quantum erasure based on phase structure

The quantum eraser effect exemplifies the distinct properties of quantum mechanics that challenge classical intuition and expose the wave-particle duality of light. This effect has been extensively explored in various experiments; most of these investigations use polarisation to distinguish which path information, and less attention has been paid to the phase structure which is related wavefront of photon. In this study, we introduce a theoretical framework for quantum erasure that focusses on the phase structure and demonstrate it experimentally. In this experiment, we employ a Mach-Zehnder interferometer (MZI) where a first-order spiral phase plate (SPP) is integrated into one of its arms. This setup applied orbital angular momentum (OAM) to the photons and established predetermined which-way information. Consequently, the photon demonstrates its particle characteristics, with absence of interference at the MZI's output ports. Utilizing an additional SPP to erase the phase structure from the output photon results in pronounced interference patterns, observable in a post-measurement scenario. This result allows us to include the structure information of the equiphase plane of the light field in quantum erasure. The results challenge the traditional cause-effect relationship in classical physics, given that the subsequent choice of the SPP adheres to a space-like separation.

quant-ph↗

arXiv2023

The multiplexed light storage of Orbital Angular Momentum based on atomic ensembles

The improvement of the multi-mode capability of quantum memory can further improve the utilization efficiency of the quantum memory and reduce the requirement of quantum communication for storage units. In this letter, we experimentally investigate the multi-mode light multiplexing storage of orbital angular momentum (OAM) mode based on rubidium vapor, and demultiplexing by a photonic OAM mode splitter which combines a Sagnac loop with two dove prisms. Our results show a mode extinction ratio higher than 80$\%$ at 1 $μ$s of storage time. Meanwhile, two OAM modes have been multiplexing stored and demultiplexed in our experimental configuration. We believe the experimental scheme may provide a possibility for high channel capacity and multi-mode quantum multiplexed quantum storage based on atomic ensembles.

physics.atom-ph↗

arXiv2021

Visualization of magnetic fields with cylindrical vector beams in a warm atomic vapor

We propose and demonstrate an experimental implementation for the observation of magnetic fields from spatial features of absorption profiles in a warm atomic vapor. A radially polarized vector beam that traverses an atomic vapor will generate an absorption pattern with petal-like structure by the mediation of a transverse magnetic field (TMF). The spatial absorption pattern rotates when the azimuthal angle of the TMF is changed, while its contrast decreases when the longitudinal component of the magnetic field increases. By analyzing the intensity distribution of the transmitted pattern we can determine the magnetic field strength. Our work provides a framework for investigating three-dimensional magnetic field distributions based on atoms.

physics.atom-ph↗

arXiv2021

Phase gradient protection of stored spatially multimode perfect optical vortex beams in a diffused rubidium vapor

We experimentally investigate the optical storage of perfect optical vortex (POV) and spatially multimode perfect optical vortex (MPOV) beams via electromagnetically induced transparency (EIT) in a hot vapor cell. In particular, we study the role that phase gradients and phase singularities play in reducing the blurring of the retrieved images due to atomic diffusion. Three kinds of manifestations are enumerated to demonstrate such effect. Firstly, the suppression of the ring width broadening is more prominent for POVs with larger orbital angular momentum (OAM). Secondly, the retrieved double-ring MPOV beams' profiles present regular dark singularity distributions that are related to their vortex charge difference. Thirdly, the storage fidelities of the triple-ring MPOVs are substantially improved by designing line phase singularities between multi-ring MPOVs with the same OAM number but $π$ offset phases between adjacent rings. Our experimental demonstration of MPOV storage opens new opportunities for increasing data capacity in quantum memories by spatial multiplexing, as well as the generation and manipulation of complex optical vortex arrays.

quant-ph↗