arXiv subjects

Zeyu Liu

Publications and source records attributed to Zeyu Liu.

At least 73 records · Page 4Linked to original sources

arXiv2025

InfiMed-Foundation: Pioneering Advanced Multimodal Medical Models with Compute-Efficient Pre-Training and Multi-Stage Fine-Tuning

Multimodal large language models (MLLMs) have shown remarkable potential in various domains, yet their application in the medical field is hindered by several challenges. General-purpose MLLMs often lack the specialized knowledge required for medical tasks, leading to uncertain or hallucinatory responses. Knowledge distillation from advanced models struggles to capture domain-specific expertise in radiology and pharmacology. Additionally, the computational cost of continual pretraining with large-scale medical data poses significant efficiency challenges. To address these issues, we propose InfiMed-Foundation-1.7B and InfiMed-Foundation-4B, two medical-specific MLLMs designed to deliver state-of-the-art performance in medical applications. We combined high-quality general-purpose and medical multimodal data and proposed a novel five-dimensional quality assessment framework to curate high-quality multimodal medical datasets. We employ low-to-high image resolution and multimodal sequence packing to enhance training efficiency, enabling the integration of extensive medical data. Furthermore, a three-stage supervised fine-tuning process ensures effective knowledge extraction for complex medical tasks. Evaluated on the MedEvalKit framework, InfiMed-Foundation-1.7B outperforms Qwen2.5VL-3B, while InfiMed-Foundation-4B surpasses HuatuoGPT-V-7B and MedGemma-27B-IT, demonstrating superior performance in medical visual question answering and diagnostic tasks. By addressing key challenges in data quality, training efficiency, and domain-specific knowledge extraction, our work paves the way for more reliable and effective AI-driven solutions in healthcare. InfiMed-Foundation-4B model is available at \href{https://huggingface.co/InfiX-ai/InfiMed-Foundation-4B}{InfiMed-Foundation-4B}.

cs.AI↗

arXiv2025

STAR: A Fast and Robust Rigid Registration Framework for Serial Histopathological Images

Registration of serial whole-slide histopathological images (WSIs) is critical for enabling direct comparison across diverse stains and for preparing paired datasets in artificial intelligence (AI) workflows such as virtual staining and biomarker prediction. While existing methods often rely on complex deformable or deep learning approaches that are computationally intensive and difficult to reproduce, lightweight rigid frameworks-sufficient for many consecutive-section scenarios-remain underdeveloped. We introduce STAR (Serial Tissue Alignment for Rigid registration), a fast and robust open-source framework for multi-WSI alignment. STAR integrates stain-conditioned preprocessing with a hierarchical coarse-to-fine correlation strategy, adaptive kernel scaling, and built-in quality control, achieving reliable rigid registration across heterogeneous tissue types and staining protocols, including hematoxylin-eosin (H&E), special histochemical stains (e.g., PAS, PASM, Masson's), and immunohistochemical (IHC) markers (e.g., CD31, KI67). Evaluated on the ANHIR 2019 and ACROBAT 2022 datasets spanning multiple organs and scanning conditions, STAR consistently produced stable alignments within minutes per slide, demonstrating robustness to cross-stain variability and partial tissue overlap. Beyond benchmarks, we present case studies on H&E-IHC alignment, construction of multi-IHC panels, and typical failure modes, underscoring both utility and limitations. Released as an open and lightweight tool, STAR provides a reproducible baseline that lowers the barrier for clinical adoption and enables large-scale paired data preparation for next-generation computational pathology.

cs.CV↗

arXiv2025

DreamStory: Open-Domain Story Visualization by LLM-Guided Multi-Subject Consistent Diffusion

Story visualization aims to create visually compelling images or videos corresponding to textual narratives. Despite recent advances in diffusion models yielding promising results, existing methods still struggle to create a coherent sequence of subject-consistent frames based solely on a story. To this end, we propose DreamStory, an automatic open-domain story visualization framework by leveraging the LLMs and a novel multi-subject consistent diffusion model. DreamStory consists of (1) an LLM acting as a story director and (2) an innovative Multi-Subject consistent Diffusion model (MSD) for generating consistent multi-subject across the images. First, DreamStory employs the LLM to generate descriptive prompts for subjects and scenes aligned with the story, annotating each scene's subjects for subsequent subject-consistent generation. Second, DreamStory utilizes these detailed subject descriptions to create portraits of the subjects, with these portraits and their corresponding textual information serving as multimodal anchors (guidance). Finally, the MSD uses these multimodal anchors to generate story scenes with consistent multi-subject. Specifically, the MSD includes Masked Mutual Self-Attention (MMSA) and Masked Mutual Cross-Attention (MMCA) modules. MMSA and MMCA modules ensure appearance and semantic consistency with reference images and text, respectively. Both modules employ masking mechanisms to prevent subject blending. To validate our approach and promote progress in story visualization, we established a benchmark, DS-500, which can assess the overall performance of the story visualization framework, subject-identification accuracy, and the consistency of the generation model. Extensive experiments validate the effectiveness of DreamStory in both subjective and objective evaluations. Please visit our project homepage at https://dream-xyz.github.io/dreamstory.

cs.CV↗

arXiv2025

Weil polynomials of abelian varieties over finite fields

In this paper, we investigate Weil polynomials and their relationship with isogeny classes of abelian varieties over finite fields. We give a necessary condition for a degree 12 polynomial with integer coefficients to be a Weil polynomial. Moreover, we provide explicit criteria that determine when a Weil polynomial of degree 14 occurs as the characteristic polynomial of a Frobenius endomorphism acting on an abelian variety.

math.NT↗

arXiv2025

Infi-MMR: Curriculum-based Unlocking Multimodal Reasoning via Phased Reinforcement Learning in Multimodal Small Language Models

Recent advancements in large language models (LLMs) have demonstrated substantial progress in reasoning capabilities, such as DeepSeek-R1, which leverages rule-based reinforcement learning to enhance logical reasoning significantly. However, extending these achievements to multimodal large language models (MLLMs) presents critical challenges, which are frequently more pronounced for Multimodal Small Language Models (MSLMs) given their typically weaker foundational reasoning abilities: (1) the scarcity of high-quality multimodal reasoning datasets, (2) the degradation of reasoning capabilities due to the integration of visual processing, and (3) the risk that direct application of reinforcement learning may produce complex yet incorrect reasoning processes. To address these challenges, we design a novel framework Infi-MMR to systematically unlock the reasoning potential of MSLMs through a curriculum of three carefully structured phases and propose our multimodal reasoning model Infi-MMR-3B. The first phase, Foundational Reasoning Activation, leverages high-quality textual reasoning datasets to activate and strengthen the model's logical reasoning capabilities. The second phase, Cross-Modal Reasoning Adaptation, utilizes caption-augmented multimodal data to facilitate the progressive transfer of reasoning skills to multimodal contexts. The third phase, Multimodal Reasoning Enhancement, employs curated, caption-free multimodal data to mitigate linguistic biases and promote robust cross-modal reasoning. Infi-MMR-3B achieves both state-of-the-art multimodal math reasoning ability (43.68% on MathVerse testmini, 27.04% on MathVision test, and 21.33% on OlympiadBench) and general reasoning ability (67.2% on MathVista testmini). Resources are available at https://huggingface.co/Reallm-Labs/Infi-MMR-3B.

cs.AI↗

arXiv2025

Accelerating the Evolution of Personalized Automated Lane Change through Lesson Learning

Personalization is crucial for the widespread adoption of advanced driver assistance system. To match up with each user's preference, the online evolution capability is a must. However, conventional evolution methods learn from naturalistic driving data, which requires a lot computing power and cannot be applied online. To address this challenge, this paper proposes a lesson learning approach: learning from driver's takeover interventions. By leveraging online takeover data, the driving zone is generated to ensure perceived safety using Gaussian discriminant analysis. Real-time corrections to trajectory planning rewards are enacted through apprenticeship learning. Guided by the objective of optimizing rewards within the constraints of the driving zone, this approach employs model predictive control for trajectory planning. This lesson learning framework is highlighted for its faster evolution capability, adeptness at experience accumulating, assurance of perceived safety, and computational efficiency. Simulation results demonstrate that the proposed system consistently achieves a successful customization without further takeover interventions. Accumulated experience yields a 24% enhancement in evolution efficiency. The average number of learning iterations is only 13.8. The average computation time is 0.08 seconds.

cs.LG↗

arXiv2025

Semiregular tessellation of electronic lattices in untwisted bilayer graphene under anisotropic strain gradients

Two-dimensional (2D) moiré superlattices have emerged as a versatile platform for uncovering exotic quantum phases, many of which arise in bilayer systems exhibiting Archimedean tessellation patterns such as triangular, hexagonal, and kagome lattices. Here, we propose a strategy to engineer semiregular tessellation patterns in untwisted bilayer graphene by applying anisotropic epitaxial tensile strain (AETS) along crystallographic directions. Through force-field and first-principles calculations, we demonstrate that AETS can induce a rich variety of semiregular tessellation geometries, including truncated hextille, prismatic pentagon, and brick-phase arrangements. The characteristic electronic bands (Dirac and flat bands) of the lattice models associated with these semiregular tessellations are observed near the Fermi level, arising from interlayer interactions generated by the redistribution of specific stacking registries (AB, BA, and SP). Furthermore, the electronic kagome, distorted Lieb, brick-like, and one-dimensional stripe lattices captured in real-space confirm the tunable nature of the semiregular tessellation lattices enabled by AETS. Our study identifies AETS as a promising new degree of freedom in moiré engineering, offering a reproducible and scalable platform for exploring exotic electronic lattices in moiré systems.

cond-mat.mes-hall↗

arXiv2025

Rabi Oscillations of Strongly Driven Bose Polarons

Understanding the dynamical behavior of quasiparticles is essential for uncovering novel quantum many-body phenomena. Among these phenomena, the polaron in ultracold atomic gases has attracted considerable interest due to its precise controllability. By engineering the underlying Hamiltonian, polarons serve as a versatile platform for studying both equilibrium properties and non-equilibrium dynamics. In this work, we investigate the quantum dynamics of strongly driven Bose polarons, where minority atoms are described as mobile impurities with spin-1/2, and majority atoms are bosons. The spin-$\uparrow$ impurity interacts with majority atoms through a tunable scattering length $a$, while the spin-$\downarrow$ impurity remains non-interacting. After turning on the Rabi coupling, we calculate the evolution of the total magnetization using a trial wavefunction. We identify the exhibition of anomalous Rabi oscillation and steady-state magnetization for $a>0$, due to the interplay between attractive and repulsive polarons. Our results provide a concrete example that illustrates how Rabi oscillations are dressed by system-environment coupling.

cond-mat.quant-gas↗

arXiv2025

A stacky approach to prismatic crystals via $q$-prism charts

Let $Y$ be a locally complete intersection over $\mathcal{O}_K$ containing a $p$-power root of unity $ζ_p$. We classify the derived category of prismatic crystals on the absolute prismatic site of $Y$ by studying quasi-coherent complexes on the prismatization of $Y$ via $q$-prism charts. We also develop a Galois descent mechanism to remove the assumption on $\mathcal{O}_K$. As an application, we classify quasi-coherent complexes on the Cartier-Witt stack and give a purely algebraic calculation of the cohomology of the structure sheaf on the absolute prismatic site of $\mathbb{Z}_p$. Along the way, for $Y$ a locally complete intersection over $\overline{A}$ with $A$ lying over a $q$-prism, we classify quasi-coherent complexes on the relative prismatization of $Y$.

math.AG↗

arXiv2025

UnPuzzle: A Unified Framework for Pathology Image Analysis

Pathology image analysis plays a pivotal role in medical diagnosis, with deep learning techniques significantly advancing diagnostic accuracy and research. While numerous studies have been conducted to address specific pathological tasks, the lack of standardization in pre-processing methods and model/database architectures complicates fair comparisons across different approaches. This highlights the need for a unified pipeline and comprehensive benchmarks to enable consistent evaluation and accelerate research progress. In this paper, we present UnPuzzle, a novel and unified framework for pathological AI research that covers a broad range of pathology tasks with benchmark results. From high-level to low-level, upstream to downstream tasks, UnPuzzle offers a modular pipeline that encompasses data pre-processing, model composition,taskconfiguration,andexperimentconduction.Specifically, it facilitates efficient benchmarking for both Whole Slide Images (WSIs) and Region of Interest (ROI) tasks. Moreover, the framework supports variouslearningparadigms,includingself-supervisedlearning,multi-task learning,andmulti-modallearning,enablingcomprehensivedevelopment of pathology AI models. Through extensive benchmarking across multiple datasets, we demonstrate the effectiveness of UnPuzzle in streamlining pathology AI research and promoting reproducibility. We envision UnPuzzle as a cornerstone for future advancements in pathology AI, providing a more accessible, transparent, and standardized approach to model evaluation. The UnPuzzle repository is publicly available at https://github.com/Puzzle-AI/UnPuzzle.

eess.IV↗

arXiv2025

InfiR : Crafting Effective Small Language Models and Multimodal Small Language Models in Reasoning

Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) have made significant advancements in reasoning capabilities. However, they still face challenges such as high computational demands and privacy concerns. This paper focuses on developing efficient Small Language Models (SLMs) and Multimodal Small Language Models (MSLMs) that retain competitive reasoning abilities. We introduce a novel training pipeline that enhances reasoning capabilities and facilitates deployment on edge devices, achieving state-of-the-art performance while minimizing development costs. \InfR~ aims to advance AI systems by improving reasoning, reducing adoption barriers, and addressing privacy concerns through smaller model sizes. Resources are available at https://github. com/Reallm-Labs/InfiR.

cs.CL↗

arXiv2025

The Case of FBA as a DEX Processing Model

We investigate the welfare loss of continuous and discrete order matching models in blockchain-based decentralized exchanges (DEX) that utilize order books to record outstanding orders. Continuous processing matches each incoming transaction against the current order book. The discrete processing model, i.e., frequent batch auction (FBA), executes transactions discretely in batches with a uniform price double auction: Orders are first matched according to price, then the exact transaction order if competing orders specify the same price. We find that FBA imposes less welfare loss and provides better liquidity than continuous processing in typical scenarios, e.g., when few parties are privately informed about asset valuations. Even otherwise, it achieves better social welfare and liquidity provision in the following settings: when price takers and public information reflecting asset value changes arrive sufficiently frequently compared to private information, when the priority fees (for faster transaction inclusion into blockchains) are small, or when the market is more balanced on both buy and sell sides. Our empirical analysis on the BTC-USD and ETH-USD transactions on a DEX named dYdX indicates that FBA can reduce transaction costs by $21\%-37\%$.

cs.CR↗

arXiv2025

Kagome electronic states in gradient-strained untwisted graphene bilayers

Moiré superlattices in twisted homo-bilayers have revealed exotic electronic states, including unconventional superconductivity and correlated insulating phases. However, their fabrication process often introduces moiré disorders, hindering reproducibility and experimental control. Here, we propose an alternative approach using gradient strain to construct moiré superlattices in untwisted bilayer graphene (gs-BLG). Through force-field and first-principles calculations, we show that gs-BLG exhibits kagome-like interlayer-spacing distributions and strain-tunable kagome electronic bands. The competition between interlayer coupling and in-plane strain relaxation leads to distinct structural deformations, giving rise to three forms of diatomic kagome lattices: subtle, pronounced, and distorted. Kagome electronic bands are identified near the Fermi level in their band structures. Modulating strain gradients enables tailoring bandwidths and signs of hopping parameters of these kagome bands, providing a versatile platform for studying exotic electronic phases. Our findings establish gradient strain as an alternative to twist engineering, opening an avenue for exploring emergent electronic phases in graphene-based systems.

cond-mat.mes-hall↗

arXiv2025

Scrambling Enabled Entropy Accumulation in Open Quantum Systems

In closed quantum many-body systems, initially localized information spreads throughout the system and becomes highly complex. This phenomenon, known as information scrambling, is closely related to entropy growth and quantum thermalization. Recent studies have shown that dissipation in open systems can hinder information scrambling, driving the system into a dissipative phase when the system-bath coupling is strong. However, the signature of this scrambling transition in entropy dynamics remains unexplored. In this work, we unveil a novel phenomenon in open quantum systems, termed entropy accumulation, which occurs exclusively within the scrambling phase. We consider a setup in which a probe is weakly coupled to a system that is already interacting with a bath. We calculate the increase in the second Rényi entropy induced by an external impulse on the system, after tracing out the probe. Despite the system-probe coupling being weak, the entropy continues to increase and eventually saturates at a finite value due to operator growth. In contrast, the entropy increase is limited by the coupling strength in the dissipative phase. The theoretical prediction is derived from both general arguments and an explicit example using generalized Boltzmann equations. Our results offer new insights into the intriguing relationship between entropy dynamics and information scrambling in open quantum systems.

quant-ph↗

arXiv2025

Mapping Transient Structures of Cyclo[18]Carbon by Computational X-Ray Spectra

The structure of cyclo[18]carbon (C$_{18}$), whether in its polyynic form with bond length alternation (BLA) or its cumulenic form without BLA, has long fascinated researchers, even prior to its successful synthesis. Recent studies suggest a polyynic ground state and a cumulenic transient state; however, the dynamics remain unclear and lack experimental validation. This study presents a first-principles theoretical investigation of the bond lengths ($R_1$ and $R_2$) dependent two-dimensional potential energy surfaces (PESs) of C$_{18}$, concentrating on the ground state and carbon 1s ionized and excited states. We examine the potential of X-ray spectra for determining bond lengths and monitoring transient structures, finding that both X-ray photoelectron (XPS) and absorption (XAS) spectra are sensitive to these variations. Utilizing a library of ground-state minimum structures optimized with 14 different functionals, we observe that core binding energies predicted with the $ω$B97XD functional can vary by 0.9 eV (290.3--291.2 eV). Unlike the ground state PES, which predicts minima at alternating bond lengths, the C1s ionized state PES predicts minima with equivalent bond lengths. In the XAS spectra, peaks 1$π^*$ and 2$π^*$ show a redshift with increasing bond lengths along the line where $R_1 = R_2$. Additionally, increasing $R_2$ (with $R_1$ fixed) results in an initial redshift followed by a blueshift, minimizing at $R_1 = R_2$. Major peaks indicate that both 1$π^*$ and 2$π^*$ arise from two channels: C1s$\rightarrowπ^*_{z}$ (out-of-plane) and C1s$\rightarrowπ^*_{xy}$ (in-plane) transitions at coinciding energies.

physics.chem-ph↗

arXiv2024

Fleximo: Towards Flexible Text-to-Human Motion Video Generation

Current methods for generating human motion videos rely on extracting pose sequences from reference videos, which restricts flexibility and control. Additionally, due to the limitations of pose detection techniques, the extracted pose sequences can sometimes be inaccurate, leading to low-quality video outputs. We introduce a novel task aimed at generating human motion videos solely from reference images and natural language. This approach offers greater flexibility and ease of use, as text is more accessible than the desired guidance videos. However, training an end-to-end model for this task requires millions of high-quality text and human motion video pairs, which are challenging to obtain. To address this, we propose a new framework called Fleximo, which leverages large-scale pre-trained text-to-3D motion models. This approach is not straightforward, as the text-generated skeletons may not consistently match the scale of the reference image and may lack detailed information. To overcome these challenges, we introduce an anchor point based rescale method and design a skeleton adapter to fill in missing details and bridge the gap between text-to-motion and motion-to-video generation. We also propose a video refinement process to further enhance video quality. A large language model (LLM) is employed to decompose natural language into discrete motion sequences, enabling the generation of motion videos of any desired length. To assess the performance of Fleximo, we introduce a new benchmark called MotionBench, which includes 400 videos across 20 identities and 20 motions. We also propose a new metric, MotionScore, to evaluate the accuracy of motion following. Both qualitative and quantitative results demonstrate that our method outperforms existing text-conditioned image-to-video generation methods. All code and model weights will be made publicly available.

cs.CV↗

arXiv2024

Diagnosing Strong-to-Weak Symmetry Breaking via Wightman Correlators

Symmetry plays a fundamental role in quantum many-body physics, and a central concept is spontaneous symmetry breaking, which imposes crucial constraints on the possible quantum phases and their transitions. Recent developments have extended the discussion of symmetry and its breaking to mixed states, enhancing our understanding of novel quantum phases that have no counterpart in pure states. Specific attention has been paid to scenarios where a strongly symmetric density matrix exhibits spontaneous symmetry breaking to weak symmetry, characterized by the fidelity correlator. In this work, we propose the Wightman correlator as an alternative diagnostic tool. This construction relies on the introduction of the thermofield double state for a generic density matrix, which maps the strong symmetry of the density matrix to the doubled symmetry of the pure state, allowing the Wightman correlator to emerge naturally as a standard probe of symmetry breaking. We prove the equivalence between the Wightman function and the fidelity correlator in defining strong-to-weak symmetry breaking, and examine explicit examples involving spin glasses, thermal density matrices, and the decohered Ising model. Additionally, we discuss a susceptibility interpretation of the Wightman correlator.

quant-ph↗

arXiv2024

Experimental observation of ballistic to diffusive transition in AlN thin films

Bulk AlN possesses high thermal conductivity due to long phonon mean-free-paths, high group velocity, and long lifetimes. However, the thermal transport scenario becomes very different in a thin AlN film due to phonon-defect and phonon-boundary scattering. Herein, we report experimental observation of ballistic to diffusive transition in a series of AlN thin films (1.6 - 2440 nm) grown on sapphire substrates. The ballistic transport is characterized by constant thermal resistance as a function of film thickness due to phonon scattering by defects and boundaries. In this transport regime, phonons possess very small group velocities and lifetimes. The lifetime of the optical phonons increases by more than an order of magnitude in the diffusive regime, however, remains nearly constant afterwards. Our study is important for understanding the details of nano and microscale thermal transport in a highly conductive material.

physics.app-ph↗