arXiv subjects

Yuxuan Wang

Publications and source records attributed to Yuxuan Wang.

At least 127 records · Page 7Linked to original sources

arXiv2025

TPLA: Tensor Parallel Latent Attention for Efficient Disaggregated Prefill and Decode Inference

Multi-Head Latent Attention (MLA), introduced in DeepSeek-V2, compresses key-value states into a low-rank latent vector, caching only this vector to reduce memory. In tensor parallelism (TP), however, attention heads are computed across multiple devices, and each device must load the full cache, eroding the advantage of MLA over Grouped Query Attention (GQA). We propose Tensor-Parallel Latent Attention (TPLA): a scheme that partitions both the latent representation and each head's input dimension across devices, performs attention independently per shard, and then combines results with an all-reduce. TPLA preserves the benefits of a compressed KV cache while unlocking TP efficiency. Unlike Grouped Latent Attention (GLA), every head in TPLA still leverages the full latent representation, maintaining stronger representational capacity. TPLA is drop-in compatible with models pre-trained using MLA: it supports MLA-style prefilling and enables efficient tensor-parallel decoding without retraining. Applying simple orthogonal transforms -- e.g., the Hadamard transform or PCA -- before TP slicing further mitigates cross-shard interference, yielding minimal accuracy degradation. By reducing the per-device KV cache for DeepSeek-V3 and Kimi-K2, we achieve 1.79x and 1.93x speedups, respectively, at a 32K-token context length while maintaining performance on commonsense and LongBench benchmarks. TPLA can be implemented with FlashAttention-3, enabling practical end-to-end acceleration.

cs.LG↗

arXiv2025

Seed-X: Building Strong Multilingual Translation LLM with 7B Parameters

Multilingual translation stands as a challenging task for large language models (LLMs) to handle intricate language patterns and stilted translations that arise in automated translations. In this paper, we introduce Seed-X, a family of open-source LLMs comprising instruct and reasoning models, pushing the limits of translation capability with 7B parameter size. The base model is pre-trained on a diverse, high-quality dataset encompassing both monolingual and bilingual content across 28 languages, harnessing the full potential of multilingual data. The instruct model is then finetuned to translate by Chain-of-Thought (CoT) reasoning and further enhanced through reinforcement learning (RL) to achieve better generalization across diverse language pairs. Seed-X achieves performance comparable to leading closed-source models, including Gemini-2.5 and GPT-4o, across 28 languages, and significantly outperforms larger open-source models in both automatic metrics and human evaluations. We share the best practices through our optimization process, and make the parameter public available for advancing translation research and applications.

cs.CL↗

arXiv2025

Constructing characteristic initial data for three dimensional compressible Euler equations

This paper resolves the characteristic initial data problem for the three-dimensional compressible Euler equations - an open problem analogous to Christodoulou's characteristic initial value formulation for the vacuum Einstein field equations in general relativity. Within the framework of acoustical geometry, we prove that for any "initial cone" $C_0\subset \mathcal{D}=[0,T]\times\mathbb{R}^3$ with initial data $(\mathringρ,\mathring{v},\mathring{s})$ given at $S_{0,0}=C_0\cap Σ_0$, arbitrary smooth entropy function and angular velocity determine smooth initial data $(ρ,v,s)$ on $C_0$ that render $C_0$ characteristic. Differing from the intersecting-hypersurface case by Speck-Yu [19] and the symmetric reduction case by Lisibach [11], our vector field method recursively determines all (including $0$-th) order derivatives of the solution along $C_0$ via transport equations and wave equations. This work provides a complete characteristic data construction for admissible hypersurfaces in the 3D compressible Euler system, introducing useful tools and providing novel aspects for studies of the long-time dynamics of the compressible Euler flow.

math.AP↗

arXiv2025

DuPO: Enabling Reliable LLM Self-Verification via Dual Preference Optimization

We present DuPO, a dual learning-based preference optimization framework that generates annotation-free feedback via a generalized duality. DuPO addresses two key limitations: Reinforcement Learning with Verifiable Rewards (RLVR)'s reliance on costly labels and applicability restricted to verifiable tasks, and traditional dual learning's restriction to strictly dual task pairs (e.g., translation and back-translation). Specifically, DuPO decomposes a primal task's input into known and unknown components, then constructs its dual task to reconstruct the unknown part using the primal output and known information (e.g., reversing math solutions to recover hidden variables), broadening applicability to non-invertible tasks. The quality of this reconstruction serves as a self-supervised reward to optimize the primal task, synergizing with LLMs' ability to instantiate both tasks via a single model. Empirically, DuPO achieves substantial gains across diverse tasks: it enhances the average translation quality by 2.13 COMET over 756 directions, boosts the mathematical reasoning accuracy by an average of 6.4 points on three challenge benchmarks, and enhances performance by 9.3 points as an inference-time reranker (trading computation for accuracy). These results position DuPO as a scalable, general, and annotation-free paradigm for LLM optimization.

cs.LG↗

arXiv2025

Nautilus: Locality-aware Autoencoder for Scalable Mesh Generation

Triangle meshes are fundamental to 3D applications, enabling efficient modification and rasterization while maintaining compatibility with standard rendering pipelines. However, current automatic mesh generation methods typically rely on intermediate representations that lack the continuous surface quality inherent to meshes. Converting these representations into meshes produces dense, suboptimal outputs. Although recent autoregressive approaches demonstrate promise in directly modeling mesh vertices and faces, they are constrained by the limitation in face count, scalability, and structural fidelity. To address these challenges, we propose Nautilus, a locality-aware autoencoder for artist-like mesh generation that leverages the local properties of manifold meshes to achieve structural fidelity and efficient representation. Our approach introduces a novel tokenization algorithm that preserves face proximity relationships and compresses sequence length through locally shared vertices and edges, enabling the generation of meshes with an unprecedented scale of up to 5,000 faces. Furthermore, we develop a Dual-stream Point Conditioner that provides multi-scale geometric guidance, ensuring global consistency and local structural fidelity by capturing fine-grained geometric features. Extensive experiments demonstrate that Nautilus significantly outperforms state-of-the-art methods in both fidelity and scalability. The project page is at https://nautilusmeshgen.github.io.

cs.CV↗

arXiv2025

VideoLLaMB: Long Streaming Video Understanding with Recurrent Memory Bridges

Recent advancements in large-scale video-language models have shown significant potential for real-time planning and detailed interactions. However, their high computational demands and the scarcity of annotated datasets limit their practicality for academic researchers. In this work, we introduce VideoLLaMB, a novel and efficient framework for long video understanding that leverages recurrent memory bridges and temporal memory tokens to enable seamless encoding of entire video sequences with preserved semantic continuity. Central to our approach is a SceneTiling algorithm that segments videos into coherent semantic units, facilitating robust understanding across tasks without requiring additional training. VideoLLaMB achieves state-of-the-art performance, surpassing existing models by 4.2 points on four VideoQA benchmarks and by 2.06 points on egocentric planning tasks. Notably, it maintains strong performance under extreme video length scaling (up to 8 times) and excels at fine-grained frame retrieval on our proposed Needle in a Video Haystack (NIAVH) benchmark. With linear GPU memory scaling, VideoLLaMB processes up to 320 frames using a single Nvidia A100 GPU, despite being trained on only 16 frames-offering an unprecedented balance of accuracy, scalability, and cost-effectiveness. This makes it highly accessible and practical for the academic community.

cs.CV↗

arXiv2025

Seed LiveInterpret 2.0: End-to-end Simultaneous Speech-to-speech Translation with Your Voice

Simultaneous Interpretation (SI) represents one of the most daunting frontiers in the translation industry, with product-level automatic systems long plagued by intractable challenges: subpar transcription and translation quality, lack of real-time speech generation, multi-speaker confusion, and translated speech inflation, especially in long-form discourses. In this study, we introduce Seed-LiveInterpret 2.0, an end-to-end SI model that delivers high-fidelity, ultra-low-latency speech-to-speech generation with voice cloning capabilities. As a fully operational product-level solution, Seed-LiveInterpret 2.0 tackles these challenges head-on through our novel duplex speech-to-speech understanding-generating framework. Experimental results demonstrate that through large-scale pretraining and reinforcement learning, the model achieves a significantly better balance between translation accuracy and latency, validated by human interpreters to exceed 70% correctness in complex scenarios. Notably, Seed-LiveInterpret 2.0 outperforms commercial SI solutions by significant margins in translation quality, while slashing the average latency of cloned speech from nearly 10 seconds to a near-real-time 3 seconds, which is around a near 70% reduction that drastically enhances practical usability.

cs.CL↗

arXiv2025

SplitMeanFlow: Interval Splitting Consistency in Few-Step Generative Modeling

Generative models like Flow Matching have achieved state-of-the-art performance but are often hindered by a computationally expensive iterative sampling process. To address this, recent work has focused on few-step or one-step generation by learning the average velocity field, which directly maps noise to data. MeanFlow, a leading method in this area, learns this field by enforcing a differential identity that connects the average and instantaneous velocities. In this work, we argue that this differential formulation is a limiting special case of a more fundamental principle. We return to the first principles of average velocity and leverage the additivity property of definite integrals. This leads us to derive a novel, purely algebraic identity we term Interval Splitting Consistency. This identity establishes a self-referential relationship for the average velocity field across different time intervals without resorting to any differential operators. Based on this principle, we introduce SplitMeanFlow, a new training framework that enforces this algebraic consistency directly as a learning objective. We formally prove that the differential identity at the core of MeanFlow is recovered by taking the limit of our algebraic consistency as the interval split becomes infinitesimal. This establishes SplitMeanFlow as a direct and more general foundation for learning average velocity fields. From a practical standpoint, our algebraic approach is significantly more efficient, as it eliminates the need for JVP computations, resulting in simpler implementation, more stable training, and broader hardware compatibility. One-step and two-step SplitMeanFlow models have been successfully deployed in large-scale speech synthesis products (such as Doubao), achieving speedups of 20x.

cs.LG↗

arXiv2025

Thermophysical and Mechanical Properties Prediction of Rear-earth High-entropy Pyrochlore Based on Deep-learning Potential

High-entropy pyrochlore oxides possess ultra-low thermal conductivity and excellent high-temperature phase stability, making them promising candidate for next-generation thermal barrier coating (TBC) materials. However, reliable predictive models for such complex and disordered systems remain challenging. Ab initio methods, although accurate in describing anharmonic phonon-phonon interactions, struggle to capture the strong inherent phonon-disorder scattering in high-entropy systems. Moreover, the limited simulation cell size, hundreds of atoms, cannot fully represent the configurational complexity of high-entropy phases. On the other hand, classical molecular dynamics (MD) simulations lack accurate and transferable interatomic potentials, particularly in multi-component systems like high-entropy ceramics. In this work, we employed Deep Potential Molecular Dynamics (DPMD) to predict the thermophysical and mechanical properties of rare-earth high-entropy pyrochlore oxide system. The deep-potential (DP) model is trained on a limited dataset from ab initio molecular dynamics (AIMD) calculations, enabling large-scale molecular dynamics simulations with on-the-fly potential evaluations. This model not only achieves high accuracy in reproducing ab initio results but also demonstrates strong generalizability, making it applicable to medium-entropy ceramics containing the same constituent elements. Our study successfully develops a deep potential model for rare-earth pyrochlore systems and demonstrates that the deep-learning-based potential method offers a powerful computational approach for designing high-entropy TBC materials.

cond-mat.mtrl-sci↗

arXiv2025

A Novel Discovery of Negative Thermal Expansion in Rare-earth Pyrochlore through Anion Order-Disorder Transition

In this study, we report for the first time the occurrence and investigation of the negative thermal expansion (NTE) effect in rare-earth pyrochlores. It is found that the NTE originates from the migration of oxygen anions from 48f sites to 8b sites, where one-twelfth of the original anions gradually occupy half of the available oxygen vacancies. This initial rapid transition leads to the distortion and rotation of polyhedral units, effectively contracting the lattice and manifesting as macroscopic NTE. The transition is sensitive to external isotropic pressure, where increasing pressure delays the onset of anion migration. This study deepens our understanding of NTE in complex oxides and demonstrates the utility of deep learning potentials for exploring intricate structural behaviors.

cond-mat.mtrl-sci↗

arXiv2025

TokenSwift: Lossless Acceleration of Ultra Long Sequence Generation

Generating ultra-long sequences with large language models (LLMs) has become increasingly crucial but remains a highly time-intensive task, particularly for sequences up to 100K tokens. While traditional speculative decoding methods exist, simply extending their generation limits fails to accelerate the process and can be detrimental. Through an in-depth analysis, we identify three major challenges hindering efficient generation: frequent model reloading, dynamic key-value (KV) management and repetitive generation. To address these issues, we introduce TOKENSWIFT, a novel framework designed to substantially accelerate the generation process of ultra-long sequences while maintaining the target model's inherent quality. Experimental results demonstrate that TOKENSWIFT achieves over 3 times speedup across models of varying scales (1.5B, 7B, 8B, 14B) and architectures (MHA, GQA). This acceleration translates to hours of time savings for ultra-long sequence generation, establishing TOKENSWIFT as a scalable and effective solution at unprecedented lengths. Code can be found at https://github.com/bigai-nlco/TokenSwift.

cs.CL↗

arXiv2025

Efficient First-Principles Framework for Overdamped Phonon Dynamics and Anharmonic Electron-Phonon Coupling in Superionic Materials

Relying on the anharmonic special displacement method, we introduce an ab initio quasistatic polymorphous framework to describe local disorder, anharmonicity, and electron-phonon coupling in superionic conductors. Using the example of cubic Cu2Se, we show that positional polymorphism yields extremely overdamped anharmonic vibrations while preserving transverse acoustic phonons, consistent with experiments. We also demonstrate well-defined electronic band structures with large band gap openings due to polymorphism of 1.0 eV and calculate anharmonic electron-phonon renormalization, yielding band gap narrowing with increasing temperature in agreement with previous measurements. Our approach opens the way for efficient ab initio electronic structure calculations in superionic crystals to elucidate their compelling high figure-of-merit.

cond-mat.mtrl-sci↗

arXiv2025

Coadjoint-orbit bosonization of a Fermi surface in a weak magnetic field

We present a bosonized effective field theory for a 2d Fermi surface in a weak magnetic field using the coadjoint orbit approach, which was recently developed as a nonlinear bosonization method in phase space for Fermi liquids and non-Fermi liquids. We show that by parametrizing the phase space with the guiding center and the mechanical momentum, and by using techniques in noncommutative field theory, the physics of Landau levels and Landau level degeneracy ($N_Φ$) naturally arises. For a parabolic dispersion, the resulting theory describes $N_Φ$ flavors of free chiral bosons propagating in \emph{momentum space}. In addition, the action contains a linear term in the bosonic field, which upon mode expansion becomes a topological $θ$-term. By properly quantizing this theory, we reproduce the well-known thermal and magnetic responses of a Fermi surface, including linear-in-$T$ specific heat, Landau diamagnetism, and the de Haas-van Alphen effect. {In particular, the de Haas-van Alphen effect is shown to be a direct consequence of the topological $θ$-term.} Our theory paves the way toward understanding correlated gapless fermionic systems in a magnetic field using the powerful approach of bosonization.

cond-mat.str-el↗

arXiv2025

FairHuman: Boosting Hand and Face Quality in Human Image Generation with Minimum Potential Delay Fairness in Diffusion Models

Image generation has achieved remarkable progress with the development of large-scale text-to-image models, especially diffusion-based models. However, generating human images with plausible details, such as faces or hands, remains challenging due to insufficient supervision of local regions during training. To address this issue, we propose FairHuman, a multi-objective fine-tuning approach designed to enhance both global and local generation quality fairly. Specifically, we first construct three learning objectives: a global objective derived from the default diffusion objective function and two local objectives for hands and faces based on pre-annotated positional priors. Subsequently, we derive the optimal parameter updating strategy under the guidance of the Minimum Potential Delay (MPD) criterion, thereby attaining fairness-ware optimization for this multi-objective problem. Based on this, our proposed method can achieve significant improvements in generating challenging local details while maintaining overall quality. Extensive experiments showcase the effectiveness of our method in improving the performance of human image generation under different scenarios.

cs.CV↗

arXiv2025

Bosonized theory of de Haas-van Alphen quantum oscillation in Fermi liquids

The de Haas-van Alphen effect (dHvA) of a 2d Fermi liquid remains poorly understood, due to the $\sim\mathcal{O}(1)$ contribution to the oscillations of grand potential from the oscillatory part of the fermionic self energy, which has no known closed-form solution. In this work, we solve this problem via coadjoint-orbit bosonization of the Fermi surface. Compared with the fermionic formalism, the issue of the oscillatory self energy is circumvented. As an effective field theory, Landau parameters $F_{n}$ directly enter the theory. We use the bosonized theory to derive the energies of cyclotron resonance and specific heat, which are consistent with Fermi liquid theory. Via a mode expansion, we show that the problem of dHvA is reduced to 0+1D quantum mechanics. We obtain analytic expressions for the behavior of dHvA at low and high temperatures, which deviate from the well-known Lifshitz-Kosevich formula. We contrast this behavior with that of 3d Fermi liquids, for which we show such deviations are parametrically small. We discuss the effects of disorder on dHvA within the bosonized theory.

cond-mat.str-el↗

arXiv2025

BCS-like formula for $T_c$ does not necessarily imply BCS pairing mechanism: the case of magnetically-mediated quantum-critical pairing

In the BCS theory of superconductivity, an instability towards pairing develops at arbitrary weak dimensionless coupling $λ$ due to a divergence of logarithmic perturbative series for the pairing susceptibility (Cooper logarithms) at $T_c \sim ω_0 e^{-1/λ}$, where $ω_0$ is an energy cutoff. On the contrary, in many models of superconductivity out of a non-Fermi liquid, Cooper logarithm is absent and superconductivity emerges only when $λ$ exceeds a certain threshold. We argue that there are situations when there is no threshold and at weak coupling the formula for $T_c$ is BCS-like, yet the origin of the pairing instability is fundamentally different from that in the BCS scenario. As an example, we revisit superconductivity in a metal at the onset of $(π,π)$ spin-density-wave order. Earlier studies of this problem found no threshold and a BCS-like expression for $T_c$ at weak coupling. We argue that, despite this, the pairing is not caused by the Cooper logarithm and in many respects is qualitatively similar to that in non-Fermi liquids.

cond-mat.supr-con↗

arXiv2025

JAEGER: Dual-Level Humanoid Whole-Body Controller

This paper presents JAEGER, a dual-level whole-body controller for humanoid robots that addresses the challenges of training a more robust and versatile policy. Unlike traditional single-controller approaches, JAEGER separates the control of the upper and lower bodies into two independent controllers, so that they can better focus on their distinct tasks. This separation alleviates the dimensionality curse and improves fault tolerance. JAEGER supports both root velocity tracking (coarse-grained control) and local joint angle tracking (fine-grained control), enabling versatile and stable movements. To train the controller, we utilize a human motion dataset (AMASS), retargeting human poses to humanoid poses through an efficient retargeting network, and employ a curriculum learning approach. This method performs supervised learning for initialization, followed by reinforcement learning for further exploration. We conduct our experiments on two humanoid platforms and demonstrate the superiority of our approach against state-of-the-art methods in both simulation and real environments.

cs.RO↗

arXiv2025

QualiSpeech: A Speech Quality Assessment Dataset with Natural Language Reasoning and Descriptions

This paper explores a novel perspective to speech quality assessment by leveraging natural language descriptions, offering richer, more nuanced insights than traditional numerical scoring methods. Natural language feedback provides instructive recommendations and detailed evaluations, yet existing datasets lack the comprehensive annotations needed for this approach. To bridge this gap, we introduce QualiSpeech, a comprehensive low-level speech quality assessment dataset encompassing 11 key aspects and detailed natural language comments that include reasoning and contextual insights. Additionally, we propose the QualiSpeech Benchmark to evaluate the low-level speech understanding capabilities of auditory large language models (LLMs). Experimental results demonstrate that finetuned auditory LLMs can reliably generate detailed descriptions of noise and distortion, effectively identifying their types and temporal characteristics. The results further highlight the potential for incorporating reasoning to enhance the accuracy and reliability of quality assessments. The dataset will be released at https://huggingface.co/datasets/tsinghua-ee/QualiSpeech.

eess.AS↗