arXiv subjects

Di Zhang

Publications and source records attributed to Di Zhang.

At least 199 records · Page 11Linked to original sources

arXiv2025

ExGes: Expressive Human Motion Retrieval and Modulation for Audio-Driven Gesture Synthesis

Audio-driven human gesture synthesis is a crucial task with broad applications in virtual avatars, human-computer interaction, and creative content generation. Despite notable progress, existing methods often produce gestures that are coarse, lack expressiveness, and fail to fully align with audio semantics. To address these challenges, we propose ExGes, a novel retrieval-enhanced diffusion framework with three key designs: (1) a Motion Base Construction, which builds a gesture library using training dataset; (2) a Motion Retrieval Module, employing constrative learning and momentum distillation for fine-grained reference poses retreiving; and (3) a Precision Control Module, integrating partial masking and stochastic masking to enable flexible and fine-grained control. Experimental evaluations on BEAT2 demonstrate that ExGes reduces Fréchet Gesture Distance by 6.2\% and improves motion diversity by 5.3\% over EMAGE, with user studies revealing a 71.3\% preference for its naturalness and semantic relevance. Code will be released upon acceptance.

cs.CV↗

arXiv2025

Cafe-Talk: Generating 3D Talking Face Animation with Multimodal Coarse- and Fine-grained Control

Speech-driven 3D talking face method should offer both accurate lip synchronization and controllable expressions. Previous methods solely adopt discrete emotion labels to globally control expressions throughout sequences while limiting flexible fine-grained facial control within the spatiotemporal domain. We propose a diffusion-transformer-based 3D talking face generation model, Cafe-Talk, which simultaneously incorporates coarse- and fine-grained multimodal control conditions. Nevertheless, the entanglement of multiple conditions challenges achieving satisfying performance. To disentangle speech audio and fine-grained conditions, we employ a two-stage training pipeline. Specifically, Cafe-Talk is initially trained using only speech audio and coarse-grained conditions. Then, a proposed fine-grained control adapter gradually adds fine-grained instructions represented by action units (AUs), preventing unfavorable speech-lip synchronization. To disentangle coarse- and fine-grained conditions, we design a swap-label training mechanism, which enables the dominance of the fine-grained conditions. We also devise a mask-based CFG technique to regulate the occurrence and intensity of fine-grained control. In addition, a text-based detector is introduced with text-AU alignment to enable natural language user input and further support multimodal control. Extensive experimental results prove that Cafe-Talk achieves state-of-the-art lip synchronization and expressiveness performance and receives wide acceptance in fine-grained control in user studies. Project page: https://harryxd2018.github.io/cafe-talk/

cs.CV↗

arXiv2025

An Adaptive Sampling-based Progressive Hedging Algorithm for Stochastic Programming

The progressive hedging algorithm (PHA) is a cornerstone among algorithms for large-scale stochastic programming problems. However, its traditional implementation is hindered by some limitations, including the requirement to solve all scenario subproblems in each iteration, reliance on an explicit probability distribution, and a convergence process that is highly sensitive to the choice of certain penalty parameters. This paper introduces a sampling-based PHA which aims to overcome these limitations. Our approach employs a dynamic selection process for the number of scenario subproblems solved per iteration. It incorporates adaptive sequential sampling for determining sample sizes, a stochastic conjugate subgradient method for direction finding, and a line-search technique to update the dual variables. Experimental results demonstrate that this novel algorithm not only addresses the bottlenecks of the conventional PHA but also potentially surpasses its scalability, representing a substantial improvement in the field of stochastic programming.

math.OC↗

arXiv2025

Anomalous Meets Topological Hall Effect in Cr2Ge2Te6 Heterostructures

Introducing topologically protected skyrmions in graphene holds significant importance for developing high-speed, low-energy spintronic devices. Here, we present a centrosymmetric ferromagnetic graphene/trilayer Cr2Ge2Te6/graphene heterostructure, demonstrating the anomalous and topological Hall effect due to the magnetic proximity effect. Through gate voltage control, we effectively tune the emergence and size of skyrmions. Micromagnetic simulations reveal the formation of skyrmions and antiskyrmions, which respond differently to external magnetic fields, leading to oscillations in the topological Hall signal. Our findings provide a novel pathway for the formation and manipulation of skyrmions in centrosymmetric two-dimensional magnetic systems, offering significant insights for developing topological spintronics.

cond-mat.mes-hall↗

arXiv2025

LivePortrait: Efficient Portrait Animation with Stitching and Retargeting Control

Portrait Animation aims to synthesize a lifelike video from a single source image, using it as an appearance reference, with motion (i.e., facial expressions and head pose) derived from a driving video, audio, text, or generation. Instead of following mainstream diffusion-based methods, we explore and extend the potential of the implicit-keypoint-based framework, which effectively balances computational efficiency and controllability. Building upon this, we develop a video-driven portrait animation framework named LivePortrait with a focus on better generalization, controllability, and efficiency for practical usage. To enhance the generation quality and generalization ability, we scale up the training data to about 69 million high-quality frames, adopt a mixed image-video training strategy, upgrade the network architecture, and design better motion transformation and optimization objectives. Additionally, we discover that compact implicit keypoints can effectively represent a kind of blendshapes and meticulously propose a stitching and two retargeting modules, which utilize a small MLP with negligible computational overhead, to enhance the controllability. Experimental results demonstrate the efficacy of our framework even compared to diffusion-based methods. The generation speed remarkably reaches 12.8ms on an RTX 4090 GPU with PyTorch. The inference code and models are available at https://github.com/KwaiVGI/LivePortrait

cs.CV↗

arXiv2025

Be a Multitude to Itself: A Prompt Evolution Framework for Red Teaming

Large Language Models (LLMs) have gained increasing attention for their remarkable capacity, alongside concerns about safety arising from their potential to produce harmful content. Red teaming aims to find prompts that could elicit harmful responses from LLMs, and is essential to discover and mitigate safety risks before real-world deployment. However, manual red teaming is both time-consuming and expensive, rendering it unscalable. In this paper, we propose RTPE, a scalable evolution framework to evolve red teaming prompts across both breadth and depth dimensions, facilitating the automatic generation of numerous high-quality and diverse red teaming prompts. Specifically, in-breadth evolving employs a novel enhanced in-context learning method to create a multitude of quality prompts, whereas in-depth evolving applies customized transformation operations to enhance both content and form of prompts, thereby increasing diversity. Extensive experiments demonstrate that RTPE surpasses existing representative automatic red teaming methods on both attack success rate and diversity. In addition, based on 4,800 red teaming prompts created by RTPE, we further provide a systematic analysis of 8 representative LLMs across 8 sensitive topics.

cs.CL↗

arXiv2025

SPPD: Self-training with Process Preference Learning Using Dynamic Value Margin

Recently, enhancing the numerical and logical reasoning capability of Large Language Models (LLMs) has emerged as a research hotspot. Existing methods face several limitations: inference-phase techniques (e.g., Chain of Thoughts) rely on prompt selection and the pretrained knowledge; sentence-level Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO) struggle with step-wise mathematical correctness and depend on stronger models distillation or human annotations; while Reinforcement Learning (RL) approaches incur high GPU memory costs and unstable training. To address these, we propose \textbf{S}elf-training framework integrating \textbf{P}rocess \textbf{P}reference learning using \textbf{D}ynamic value margin (SPPD). SPPD leverages a process-based Markov Decision Process (MDP) and Bellman optimality equation to derive \textbf{dynamic value margin} on step-level preference optimization, which employs tree-based self-sampling on model responses \textbf{without any distillation} from other models. Furthermore, we theoretically prove that SPPD is \textbf{equivalent to on-policy policy gradient methods} under reward constraints. Experiments on 7B-scale models demonstrate superior performance across in-domain and out-domain mathematical benchmarks. We open-source our code at \href{https://anonymous.4open.science/r/SSDPO-D-DCDD}{https://anonymous.4open.science/r/SPPD-DCDD}.

cs.AI↗

arXiv2025

iMOVE: Instance-Motion-Aware Video Understanding

Enhancing the fine-grained instance spatiotemporal motion perception capabilities of Video Large Language Models is crucial for improving their temporal and general video understanding. However, current models struggle to perceive detailed and complex instance motions. To address these challenges, we have made improvements from both data and model perspectives. In terms of data, we have meticulously curated iMOVE-IT, the first large-scale instance-motion-aware video instruction-tuning dataset. This dataset is enriched with comprehensive instance motion annotations and spatiotemporal mutual-supervision tasks, providing extensive training for the model's instance-motion-awareness. Building on this foundation, we introduce iMOVE, an instance-motion-aware video foundation model that utilizes Event-aware Spatiotemporal Efficient Modeling to retain informative instance spatiotemporal motion details while maintaining computational efficiency. It also incorporates Relative Spatiotemporal Position Tokens to ensure awareness of instance spatiotemporal positions. Evaluations indicate that iMOVE excels not only in video temporal understanding and general video understanding but also demonstrates significant advantages in long-term video understanding.

cs.CV↗

arXiv2025

Finedeep: Mitigating Sparse Activation in Dense LLMs via Multi-Layer Fine-Grained Experts

Large language models have demonstrated exceptional performance across a wide range of tasks. However, dense models usually suffer from sparse activation, where many activation values tend towards zero (i.e., being inactivated). We argue that this could restrict the efficient exploration of model representation space. To mitigate this issue, we propose Finedeep, a deep-layered fine-grained expert architecture for dense models. Our framework partitions the feed-forward neural network layers of traditional dense models into small experts, arranges them across multiple sub-layers. A novel routing mechanism is proposed to determine each expert's contribution. We conduct extensive experiments across various model sizes, demonstrating that our approach significantly outperforms traditional dense architectures in terms of perplexity and benchmark performance while maintaining a comparable number of parameters and floating-point operations. Moreover, we find that Finedeep achieves optimal results when balancing depth and width, specifically by adjusting the number of expert sub-layers and the number of experts per sub-layer. Empirical results confirm that Finedeep effectively alleviates sparse activation and efficiently utilizes representation capacity in dense models.

cs.CL↗

arXiv2025

TaskGalaxy: Scaling Multi-modal Instruction Fine-tuning with Tens of Thousands Vision Task Types

Multimodal visual language models are gaining prominence in open-world applications, driven by advancements in model architectures, training techniques, and high-quality data. However, their performance is often limited by insufficient task-specific data, leading to poor generalization and biased outputs. Existing efforts to increase task diversity in fine-tuning datasets are hindered by the labor-intensive process of manual task labeling, which typically produces only a few hundred task types. To address this, we propose TaskGalaxy, a large-scale multimodal instruction fine-tuning dataset comprising 19,227 hierarchical task types and 413,648 samples. TaskGalaxy utilizes GPT-4o to enrich task diversity by expanding from a small set of manually defined tasks, with CLIP and GPT-4o filtering those that best match open-source images, and generating relevant question-answer pairs. Multiple models are employed to ensure sample quality. This automated process enhances both task diversity and data quality, reducing manual intervention. Incorporating TaskGalaxy into LLaVA-v1.5 and InternVL-Chat-v1.0 models shows substantial performance improvements across 16 benchmarks, demonstrating the critical importance of task diversity. TaskGalaxy is publicly released at https://github.com/Kwai-YuanQi/TaskGalaxy.

cs.CV↗

arXiv2025

MM-RLHF: The Next Step Forward in Multimodal LLM Alignment

Despite notable advancements in Multimodal Large Language Models (MLLMs), most state-of-the-art models have not undergone thorough alignment with human preferences. This gap exists because current alignment research has primarily achieved progress in specific areas (e.g., hallucination reduction), while the broader question of whether aligning models with human preferences can systematically enhance MLLM capability remains largely unexplored. To this end, we introduce MM-RLHF, a dataset containing $\mathbf{120k}$ fine-grained, human-annotated preference comparison pairs. This dataset represents a substantial advancement over existing resources, offering superior size, diversity, annotation granularity, and quality. Leveraging this dataset, we propose several key innovations to improve both the quality of reward models and the efficiency of alignment algorithms. Notably, we introduce a Critique-Based Reward Model, which generates critiques of model outputs before assigning scores, offering enhanced interpretability and more informative feedback compared to traditional scalar reward mechanisms. Additionally, we propose Dynamic Reward Scaling, a method that adjusts the loss weight of each sample according to the reward signal, thereby optimizing the use of high-quality comparison pairs. Our approach is rigorously evaluated across $\mathbf{10}$ distinct dimensions and $\mathbf{27}$ benchmarks, with results demonstrating significant and consistent improvements in model performance. Specifically, fine-tuning LLaVA-ov-7B with MM-RLHF and our alignment algorithm leads to a $\mathbf{19.5}$% increase in conversational abilities and a $\mathbf{60}$% improvement in safety. We have open-sourced the preference dataset, reward model, training and evaluation code, as well as reward modeling and safety benchmarks. For more details, please visit our project page: https://mm-rlhf.github.io.

cs.CL↗

arXiv2025

CineMaster: A 3D-Aware and Controllable Framework for Cinematic Text-to-Video Generation

In this work, we present CineMaster, a novel framework for 3D-aware and controllable text-to-video generation. Our goal is to empower users with comparable controllability as professional film directors: precise placement of objects within the scene, flexible manipulation of both objects and camera in 3D space, and intuitive layout control over the rendered frames. To achieve this, CineMaster operates in two stages. In the first stage, we design an interactive workflow that allows users to intuitively construct 3D-aware conditional signals by positioning object bounding boxes and defining camera movements within the 3D space. In the second stage, these control signals--comprising rendered depth maps, camera trajectories and object class labels--serve as the guidance for a text-to-video diffusion model, ensuring to generate the user-intended video content. Furthermore, to overcome the scarcity of in-the-wild datasets with 3D object motion and camera pose annotations, we carefully establish an automated data annotation pipeline that extracts 3D bounding boxes and camera trajectories from large-scale video data. Extensive qualitative and quantitative experiments demonstrate that CineMaster significantly outperforms existing methods and implements prominent 3D-aware text-to-video generation. Project page: https://cinemaster-dev.github.io/.

cs.CV↗

arXiv2025

Why Do Weak-Binding M-N-C Single-Atom Catalysts Possess Anomalously High Oxygen Reduction Activity?

Single-atom catalysts (SACs) with metal-nitrogen-carbon (M-N-C) structures are widely recognized as promising candidates in oxygen reduction reactions (ORR). According to the classical Sabatier principle, optimal 3d metal catalysts, such as Fe/Co-N-C, achieve superior catalytic performance due to the moderate binding strength. However, the substantial ORR activity demonstrated by weakly binding M-N-C catalysts such as NiCu-N-C challenges current understandings, emphasizing the need to explore new underlying mechanisms. In this work, we integrated a pH-field coupled microkinetic model with detailed experimental electron state analyses to verify a novel key step in the ORR reaction pathway of weak-binding SACs-the oxygen adsorption at the metal-N bridge site. This step significantly altered the adsorption scaling relations, electric field responses, and solvation effects, further impacting the key kinetic reaction barrier from HOO* to O* and pH-dependent performance. Synchrotron spectra analysis further provides evidence for the new weak-binding M-N-C model, showing an increase in electron density on the anti-bonding pi orbitals of N atoms in weak-binding M-N-C catalysts and confirming the presence of N-O bonds. These findings redefine the understanding of weak-binding M-N-C catalyst behavior, opening up new perspectives for their application in clean energy.

physics.chem-ph↗

arXiv2025

Universal Catalyst Design Framework for Electrochemical Hydrogen Peroxide Synthesis Facilitated by Local Atomic Environment Descriptors

Developing a universal and precise design framework is crucial to search high-performance catalysts, but it remains a giant challenge due to the diverse structures and sites across various types of catalysts. To address this challenge, herein, we developed a novel framework by the refined local atomic environment descriptors (i.e., weighted Atomic Center Symmetry Function, wACSF) combined with machine learning (ML), microkinetic modeling, and computational high-throughput screening. This framework is successfully integrated into the Digital Catalysis Database (DigCat), enabling efficient screening for 2e- water oxidation reaction (2e- WOR) catalysts across four material categories (i.e., metal alloys, metal oxides and perovskites, and single-atom catalysts) within a ML model. The proposed wACSF descriptors integrating both geometric and chemical features are proven effective in predicting the adsorption free energies with ML. Excitingly, based on the wACSF descriptors, the ML models accurately predict the adsorption free energies of hydroxyl (ΔGOH*) and oxygen (ΔGO*) for such a wide range of catalysts, achieving R2 values of 0.84 and 0.91, respectively. Through density functional theory calculations and microkinetic modeling, a universal 2e- WOR microkinetic volcano model was derived with excellent agreement with experimental observations reported to date, which was further used to rapidly screen high-performance catalysts with the input of ML-predicted ΔGOH*. Most importantly, this universal framework can significantly improve the efficiency of catalyst design by considering multiple types of materials at the same time, which can dramatically accelerate the screening of high-performance catalysts.

cond-mat.mtrl-sci↗

arXiv2025

Cyclic Delay-Doppler Shift: A Simple Transmit Diversity Technique for Delay-Doppler Waveforms in Doubly Selective Channels

Delay-Doppler waveform design has been considered as a promising solution to achieve reliable communication under high-mobility channels for the space-air-ground-integrated networks (SAGIN). In this paper, we introduce the cyclic delay-Doppler shift (CDDS) technique for delay-Doppler waveforms to extract transmit diversity in doubly selective channels. Two simple CDDS schemes, named time-domain CDDS (TD-CDDS) and modulation-domain CDDS (MD-CDDS), are proposed in the setting of multiple-input multiple-output (MIMO). We demonstrate the applications of CDDS on two representative delay-Doppler waveforms, namely orthogonal time frequency space (OTFS) and affine frequency division multiplexing (AFDM), by deriving their corresponding CDDS matrices. Furthermore, we prove theoretically and experimentally that CDDS can provide OTFS and AFDM with full transmit diversity gain on most occasions.

eess.SP↗

arXiv2024

Towards Precise Scaling Laws for Video Diffusion Transformers

Achieving optimal performance of video diffusion transformers within given data and compute budget is crucial due to their high training costs. This necessitates precisely determining the optimal model size and training hyperparameters before large-scale training. While scaling laws are employed in language models to predict performance, their existence and accurate derivation in visual generation models remain underexplored. In this paper, we systematically analyze scaling laws for video diffusion transformers and confirm their presence. Moreover, we discover that, unlike language models, video diffusion models are more sensitive to learning rate and batch size, two hyperparameters often not precisely modeled. To address this, we propose a new scaling law that predicts optimal hyperparameters for any model size and compute budget. Under these optimal settings, we achieve comparable performance and reduce inference costs by 40.1% compared to conventional scaling methods, within a compute budget of 1e10 TFlops. Furthermore, we establish a more generalized and precise relationship among validation loss, any model size, and compute budget. This enables performance prediction for non-optimal model sizes, which may also be appealed under practical inference cost constraints, achieving a better trade-off.

cs.CV↗

arXiv2024

The Key Steps and Distinct Performance Trends of Pyrrolic vs. Pyridinic M-N-C Catalysts in Electrocatalytic Nitrate Reduction

Electrochemical nitrate reduction reaction(NO3RR)offers a sustainable route for ambient ammonia synthesis. While metal-nitrogen-carbon (M-N-C) single-atom catalysts have emerged as promising candidates for NO3RR, the structure-activity relations underlying their catalytic behavior remain to be elucidated. Through systematic analysis of reported experimental data and pH-field coupled microkinetic modelling on a reversible hydrogen electrode (RHE) scale, we reveal that the coordination-dependent activity originates from distinct scaling relations governed by metal-intermediate interactions. M-N-Pyrrolic catalysts demonstrate higher turnover frequencies for ammonia production, whereas M-N-Pyridinic catalysts exhibit broader activity ranges across the activity volcano plot. Meanwhile, the adsorption and protonation of nitrate, which is a step often dismissed and/or assumed to be simultaneous in many previous reports, is identified to be the rate-determining step (RDS) in NO3RR. Remarkably, our subsequent experimental validation confirms the theoretical predictions under both neutral and alkaline conditions. This study offers a comprehensive mechanistic framework for interpreting the electrocatalytic activity of M-N-C catalysts in NO3RR, showing that a classical thermodynamic limiting-potential model is not sufficiently accurate to capture the RDS and the catalytic performance trends of different materials (even on M-N-Pyrrolic and M-N-Pyridinic catalysts). These findings provide brand new insights into the reaction mechanism of NO3RR and establish fundamental design principles for electrocatalytic ammonia synthesis.

physics.chem-ph↗

arXiv2024

Owl-1: Omni World Model for Consistent Long Video Generation

Video generation models (VGMs) have received extensive attention recently and serve as promising candidates for general-purpose large vision models. While they can only generate short videos each time, existing methods achieve long video generation by iteratively calling the VGMs, using the last-frame output as the condition for the next-round generation. However, the last frame only contains short-term fine-grained information about the scene, resulting in inconsistency in the long horizon. To address this, we propose an Omni World modeL (Owl-1) to produce long-term coherent and comprehensive conditions for consistent long video generation. As videos are observations of the underlying evolving world, we propose to model the long-term developments in a latent space and use VGMs to film them into videos. Specifically, we represent the world with a latent state variable which can be decoded into explicit video observations. These observations serve as a basis for anticipating temporal dynamics which in turn update the state variable. The interaction between evolving dynamics and persistent state enhances the diversity and consistency of the long videos. Extensive experiments show that Owl-1 achieves comparable performance with SOTA methods on VBench-I2V and VBench-Long, validating its ability to generate high-quality video observations. Code: https://github.com/huang-yh/Owl.

cs.CV↗